wip

Prefix public symbols with "p_".
User lowercase for all type aliases.
Clean up some comments.

.gitignore

@@ -2,7 +2,8 @@
 /.yardoc
 /_yardoc/
 /coverage/
-/doc/
+/dist/
+/gen/
 /pkg/
 /spec/reports/
 /tmp/

CHANGELOG.md

@@ -0,0 +1,3 @@
+## v1.0.0
+
+- Initial release

Gemfile

@@ -2,3 +2,6 @@ source "https://rubygems.org"
 
 gem "rake"
 gem "rspec"
+gem "rdoc"
+gem "redcarpet"
+gem "syntax"

Gemfile.lock

@@ -2,27 +2,37 @@ GEM
   remote: https://rubygems.org/
   specs:
     diff-lcs (1.5.0)
+    psych (5.1.0)
+      stringio
     rake (13.0.6)
-    rspec (3.11.0)
+    rdoc (6.5.0)
-      rspec-core (~> 3.11.0)
+      psych (>= 4.0.0)
-      rspec-expectations (~> 3.11.0)
+    redcarpet (3.6.0)
-      rspec-mocks (~> 3.11.0)
+    rspec (3.12.0)
-    rspec-core (3.11.0)
+      rspec-core (~> 3.12.0)
-      rspec-support (~> 3.11.0)
+      rspec-expectations (~> 3.12.0)
-    rspec-expectations (3.11.0)
+      rspec-mocks (~> 3.12.0)
+    rspec-core (3.12.2)
+      rspec-support (~> 3.12.0)
+    rspec-expectations (3.12.3)
       diff-lcs (>= 1.2.0, < 2.0)
-      rspec-support (~> 3.11.0)
+      rspec-support (~> 3.12.0)
-    rspec-mocks (3.11.1)
+    rspec-mocks (3.12.6)
       diff-lcs (>= 1.2.0, < 2.0)
-      rspec-support (~> 3.11.0)
+      rspec-support (~> 3.12.0)
-    rspec-support (3.11.0)
+    rspec-support (3.12.1)
+    stringio (3.0.7)
+    syntax (1.2.2)
 
 PLATFORMS
   ruby
 
 DEPENDENCIES
   rake
+  rdoc
+  redcarpet
   rspec
+  syntax
 
 BUNDLED WITH
-   2.4.0.dev
+   2.3.7

LICENSE.txt

@@ -1,6 +1,6 @@
 The MIT License (MIT)
 
-Copyright (c) 2010-2022 Josh Holtrop
+Copyright (c) 2010-2023 Josh Holtrop
 
 Permission is hereby granted, free of charge, to any person obtaining a copy
 of this software and associated documentation files (the "Software"), to deal

Rakefile

@@ -1,5 +1,9 @@
 require "rspec/core/rake_task"
 
+task :build_dist do
+  sh "ruby rb/build_dist.rb"
+end
+
 RSpec::Core::RakeTask.new(:spec, :example_pattern) do |task, args|
   if args.example_pattern
     task.rspec_opts = %W[-e "#{args.example_pattern}" -f documentation]
@@ -7,3 +11,8 @@ RSpec::Core::RakeTask.new(:spec, :example_pattern) do |task, args|
 end
 
 task :default => :spec
+
+desc "Build user guide"
+task :user_guide do
+  system("ruby", "-Ilib", "rb/gen_user_guide.rb")
+end

assets/parser.c.erb

@@ -0,0 +1,938 @@
+#include "<%= File.basename(output_file).sub(%r{\.[a-z]+$}, "") %>.h"
+#include <stdbool.h>
+#include <stdlib.h>
+#include <string.h>
+
+/**************************************************************************
+ * User code blocks
+ *************************************************************************/
+
+<% @grammar.code_blocks.each do |code| %>
+<%= code %>
+<% end %>
+
+/**************************************************************************
+ * Private types
+ *************************************************************************/
+
+<% if @grammar.prefix.upcase != "P_" %>
+/* Result codes. */
+#define P_SUCCESS 0u
+#define P_DECODE_ERROR 1u
+#define P_UNEXPECTED_INPUT 2u
+#define P_UNEXPECTED_TOKEN 3u
+#define P_DROP 4u
+#define P_EOF 5u
+<% end %>
+
+/* An invalid ID value. */
+#define INVALID_ID ((size_t)-1)
+
+/**************************************************************************
+ * State initialization
+ *************************************************************************/
+
+/**
+ * Initialize lexer/parser context structure.
+ *
+ * @param[out] context
+ *   Lexer/parser context structure.
+ * @param input
+ *   Text input.
+ * @param input_length
+ *   Text input length.
+ */
+void <%= @grammar.prefix %>context_init(<%= @grammar.prefix %>context_t * context, uint8_t const * input, size_t input_length)
+{
+    /* New default-initialized context structure. */
+    <%= @grammar.prefix %>context_t newcontext = {0};
+
+    /* Lexer initialization. */
+    newcontext.input = input;
+    newcontext.input_length = input_length;
+    newcontext.mode = <%= @lexer.mode_id("default") %>;
+
+    /* Copy to the user's context structure. */
+    *context = newcontext;
+}
+
+/**************************************************************************
+ * Decoder
+ *************************************************************************/
+
+/**
+ * Decode a UTF-8 code point.
+ *
+ * @param input
+ *   Text input to decode.
+ * @param input_length
+ *   Input text length.
+ * @param[out] out_code_point
+ *   The decoded code point is stored here if the return value is P_SUCCESS.
+ * @param[out] out_code_point_length
+ *   The number of bytes the code point used is stored here if the return value
+ *   is P_SUCCESS.
+ *
+ * @retval P_SUCCESS on a successful code point decode
+ * @retval P_DECODE_ERROR when an encoding error is observed
+ * @retval P_EOF when the end of the text input is reached
+ */
+size_t <%= @grammar.prefix %>decode_code_point(uint8_t const * input, size_t input_length,
+    <%= @grammar.prefix %>code_point_t * out_code_point, uint8_t * out_code_point_length)
+{
+    if (input_length == 0u)
+    {
+        return P_EOF;
+    }
+    char c = input[0];
+    <%= @grammar.prefix %>code_point_t code_point;
+    uint8_t code_point_length;
+    if ((c & 0x80u) == 0u)
+    {
+        code_point = c;
+        code_point_length = 1u;
+    }
+    else
+    {
+        uint8_t following_bytes;
+        if ((c & 0xE0u) == 0xC0u)
+        {
+            code_point = c & 0x1Fu;
+            following_bytes = 1u;
+        }
+        else if ((c & 0xF0u) == 0xE0u)
+        {
+            code_point = c & 0x0Fu;
+            following_bytes = 2u;
+        }
+        else if ((c & 0xF8u) == 0xF0u)
+        {
+            code_point = c & 0x07u;
+            following_bytes = 3u;
+        }
+        else if ((c & 0xFCu) == 0xF8u)
+        {
+            code_point = c & 0x03u;
+            following_bytes = 4u;
+        }
+        else if ((c & 0xFEu) == 0xFCu)
+        {
+            code_point = c & 0x01u;
+            following_bytes = 5u;
+        }
+        else
+        {
+            return P_DECODE_ERROR;
+        }
+        if (input_length <= following_bytes)
+        {
+            return P_DECODE_ERROR;
+        }
+        code_point_length = (uint8_t)(following_bytes + 1u);
+        for (size_t i = 0u; i < following_bytes; i++)
+        {
+            char b = input[i + 1u];
+            if ((b & 0xC0u) != 0x80u)
+            {
+                return P_DECODE_ERROR;
+            }
+            code_point = (code_point << 6u) | (b & 0x3Fu);
+        }
+    }
+    *out_code_point = code_point;
+    *out_code_point_length = code_point_length;
+    return P_SUCCESS;
+}
+
+/**************************************************************************
+ * Lexer
+ *************************************************************************/
+
+/** Lexer state ID type. */
+typedef <%= get_type_for(@lexer.state_table.size) %> lexer_state_id_t;
+
+/** Invalid lexer state ID. */
+#define INVALID_LEXER_STATE_ID <%= @lexer.state_table.size %>u
+
+/** Lexer user code ID type. */
+<% user_code_id_count = (@grammar.patterns.map(&:code_id).compact.max || 0) + 1 %>
+typedef <%= get_type_for(user_code_id_count) %> lexer_user_code_id_t;
+
+/** Invalid lexer user code ID. */
+#define INVALID_USER_CODE_ID <%= user_code_id_count %>u
+
+/**
+ * Lexer transition table entry.
+ *
+ * An incoming code point matching the range for a transition entry will cause
+ * the lexer to progress to the destination state.
+ */
+typedef struct
+{
+    /** First code point in the range for this transition. */
+    <%= @grammar.prefix %>code_point_t first;
+
+    /** Last code point in the range for this transition. */
+    <%= @grammar.prefix %>code_point_t last;
+
+    /** Destination lexer state ID for this transition. */
+    lexer_state_id_t destination_state;
+} lexer_transition_t;
+
+/** Lexer state table entry. */
+typedef struct
+{
+    /** Index to the transition table for this state. */
+    <%= get_type_for(@lexer.transition_table.size - 1) %> transition_table_index;
+
+    /** Number of transition table entries for this state. */
+    <%= get_type_for(@lexer.state_table.map {|ste| ste[:n_transitions]}.max) %> n_transitions;
+
+    /** Lexer token formed at this state. */
+    <%= @grammar.prefix %>token_t token;
+
+    /** Lexer user code ID to execute at this state. */
+    lexer_user_code_id_t code_id;
+
+    /** Whether this state matches a lexer pattern. */
+    bool accepts;
+} lexer_state_t;
+
+/** Lexer mode table entry. */
+typedef struct
+{
+    /** Offset in the state table to be used for this mode. */
+    uint32_t state_table_offset;
+} lexer_mode_t;
+
+/**
+ * Lexer match info structure.
+ *
+ * This structure holds output values from the lexer upon a successful pattern
+ * match.
+ */
+typedef struct
+{
+    /** Number of bytes of input text used to match. */
+    size_t length;
+
+    /** Input text position delta. */
+    <%= @grammar.prefix %>position_t delta_position;
+
+    /** Accepting lexer state from the match. */
+    lexer_state_t const * accepting_state;
+} lexer_match_info_t;
+
+/** Lexer transition table. */
+static lexer_transition_t lexer_transition_table[] = {
+<% @lexer.transition_table.each do |transition_table_entry| %>
+    {<%= transition_table_entry[:first] %>u, <%= transition_table_entry[:last] %>u, <%= transition_table_entry[:destination] %>u},
+<% end %>
+};
+
+/** Lexer state table. */
+static lexer_state_t lexer_state_table[] = {
+<% @lexer.state_table.each do |state_table_entry| %>
+    {<%= state_table_entry[:transition_table_index] %>u, <%= state_table_entry[:n_transitions] %>u, <%= state_table_entry[:token] || "INVALID_TOKEN_ID" %>, <%= state_table_entry[:code_id] || "INVALID_USER_CODE_ID" %>, <%= state_table_entry[:accepts] %>},
+<% end %>
+};
+
+/** Lexer mode table. */
+static lexer_mode_t lexer_mode_table[] = {
+<% @lexer.mode_table.each do |mode_table_entry| %>
+    {<%= mode_table_entry[:state_table_offset] %>},
+<% end %>
+};
+
+/**
+ * Execute user code associated with a lexer pattern.
+ *
+ * @param context
+ *   Lexer/parser context structure.
+ * @param code_id
+ *   The ID of the user code block to execute.
+ * @param match
+ *   Matched text for this pattern.
+ * @param match_length
+ *   Matched text length.
+ * @param out_token_info
+ *   Lexer token info in progress.
+ *
+ * @return Token to accept, or invalid token if the user code does
+ *   not explicitly return a token.
+ */
+static <%= @grammar.prefix %>token_t lexer_user_code(<%= @grammar.prefix %>context_t * context,
+    lexer_user_code_id_t code_id, uint8_t const * match,
+    size_t match_length, <%= @grammar.prefix %>token_info_t * out_token_info)
+{
+    switch (code_id)
+    {
+<% @grammar.patterns.each do |pattern| %>
+<%   if pattern.code_id %>
+    case <%= pattern.code_id %>u: {
+<%= expand_code(pattern.code, false, nil, pattern) %>
+    } break;
+<%   end %>
+<% end %>
+    default: break;
+    }
+
+    return INVALID_TOKEN_ID;
+}
+
+/**
+ * Check if there is a transition from the current lexer state to another
+ * based on the given input code point.
+ *
+ * @param current_state
+ *   Current lexer state.
+ * @param code_point
+ *   Input code point.
+ *
+ * @return Lexer state to transition to, or INVALID_LEXER_STATE_ID if none.
+ */
+static lexer_state_id_t check_lexer_transition(uint32_t current_state, uint32_t code_point)
+{
+    uint32_t transition_table_index = lexer_state_table[current_state].transition_table_index;
+    for (uint32_t i = 0u; i < lexer_state_table[current_state].n_transitions; i++)
+    {
+        if ((lexer_transition_table[transition_table_index + i].first <= code_point) &&
+            (code_point <= lexer_transition_table[transition_table_index + i].last))
+        {
+            return lexer_transition_table[transition_table_index + i].destination_state;
+        }
+    }
+    return INVALID_LEXER_STATE_ID;
+}
+
+/**
+ * Find the longest lexer pattern match at the current position.
+ *
+ * @param context
+ *   Lexer/parser context structure.
+ * @param[out] out_token_info
+ *   The lexed token information is stored here if the return value is
+ *   P_SUCCESS.
+ *
+ * @reval P_SUCCESS
+ *   A token was successfully lexed.
+ * @reval P_DECODE_ERROR
+ *   The decoder encountered invalid text encoding.
+ * @reval P_UNEXPECTED_INPUT
+ *   Input text does not match any lexer pattern.
+ * @retval P_EOF
+ *   The end of the text input was reached.
+ */
+static size_t find_longest_match(<%= @grammar.prefix %>context_t * context,
+    lexer_match_info_t * out_match_info, size_t * out_unexpected_input_length)
+{
+    lexer_match_info_t longest_match = {0};
+    lexer_match_info_t attempt_match = {0};
+    *out_match_info = longest_match;
+    uint32_t current_state = lexer_mode_table[context->mode].state_table_offset;
+    for (;;)
+    {
+        size_t const input_index = context->input_index + attempt_match.length;
+        uint8_t const * input = &context->input[input_index];
+        size_t input_length = context->input_length - input_index;
+        <%= @grammar.prefix %>code_point_t code_point;
+        uint8_t code_point_length;
+        size_t result = <%= @grammar.prefix %>decode_code_point(input, input_length, &code_point, &code_point_length);
+        switch (result)
+        {
+        case P_SUCCESS:
+            lexer_state_id_t transition_state = check_lexer_transition(current_state, code_point);
+            if (transition_state != INVALID_LEXER_STATE_ID)
+            {
+                attempt_match.length += code_point_length;
+                if (code_point == '\n')
+                {
+                    attempt_match.delta_position.row++;
+                    attempt_match.delta_position.col = 0u;
+                }
+                else
+                {
+                    attempt_match.delta_position.col++;
+                }
+                current_state = transition_state;
+                if (lexer_state_table[current_state].accepts)
+                {
+                    attempt_match.accepting_state = &lexer_state_table[current_state];
+                    longest_match = attempt_match;
+                }
+            }
+            else if (longest_match.length > 0)
+            {
+                *out_match_info = longest_match;
+                return P_SUCCESS;
+            }
+            else
+            {
+                *out_unexpected_input_length = attempt_match.length + code_point_length;
+                return P_UNEXPECTED_INPUT;
+            }
+            break;
+
+        case P_EOF:
+            /* We hit EOF. */
+            if (longest_match.length > 0)
+            {
+                /* We have a match, so use it. */
+                *out_match_info = longest_match;
+                return P_SUCCESS;
+            }
+            else if (attempt_match.length != 0)
+            {
+                /* There is a partial match - error! */
+                *out_unexpected_input_length = attempt_match.length;
+                return P_UNEXPECTED_INPUT;
+            }
+            else
+            {
+                /* Valid EOF return. */
+                return P_EOF;
+            }
+            break;
+
+        case P_DECODE_ERROR:
+            /* If we see a decode error, we may be partially in the middle of
+             * matching a pattern, so return the attempted match info so that
+             * the input text position can be updated. */
+            *out_match_info = attempt_match;
+            return result;
+
+        default:
+            return result;
+        }
+    }
+}
+
+/**
+ * Attempt to lex the next token in the input stream.
+ *
+ * @param context
+ *   Lexer/parser context structure.
+ * @param[out] out_token_info
+ *   The lexed token information is stored here if the return value is
+ *   P_SUCCESS.
+ *
+ * @reval P_SUCCESS
+ *   A token was successfully lexed.
+ * @reval P_DECODE_ERROR
+ *   The decoder encountered invalid text encoding.
+ * @reval P_UNEXPECTED_INPUT
+ *   Input text does not match any lexer pattern.
+ * @retval P_DROP
+ *   A drop pattern was matched so the lexer should continue.
+ */
+static size_t attempt_lex_token(<%= @grammar.prefix %>context_t * context, <%= @grammar.prefix %>token_info_t * out_token_info)
+{
+    <%= @grammar.prefix %>token_info_t token_info = {0};
+    token_info.position = context->text_position;
+    token_info.token = INVALID_TOKEN_ID;
+    *out_token_info = token_info; // TODO: remove
+    lexer_match_info_t match_info;
+    size_t unexpected_input_length;
+    size_t result = find_longest_match(context, &match_info, &unexpected_input_length);
+    switch (result)
+    {
+    case P_SUCCESS:
+        <%= @grammar.prefix %>token_t token_to_accept = match_info.accepting_state->token;
+        if (match_info.accepting_state->code_id != INVALID_USER_CODE_ID)
+        {
+            uint8_t const * match = &context->input[context->input_index];
+            <%= @grammar.prefix %>token_t user_code_token = lexer_user_code(context,
+                match_info.accepting_state->code_id, match, match_info.length, &token_info);
+            /* An invalid token returned from lexer_user_code() means that the
+             * user code did not explicitly return a token. So only override
+             * the token to return if the user code does explicitly return a
+             * token. */
+            if (user_code_token != INVALID_TOKEN_ID)
+            {
+                token_to_accept = user_code_token;
+            }
+        }
+
+        /* Update the input position tracking. */
+        context->input_index += match_info.length;
+        context->text_position.row += match_info.delta_position.row;
+        if (match_info.delta_position.row != 0u)
+        {
+            context->text_position.col = match_info.delta_position.col;
+        }
+        else
+        {
+            context->text_position.col += match_info.delta_position.col;
+        }
+
+        if (token_to_accept == INVALID_TOKEN_ID)
+        {
+            return P_DROP;
+        }
+        token_info.token = token_to_accept;
+        token_info.length = match_info.length;
+        *out_token_info = token_info;
+        return P_SUCCESS;
+
+    case P_EOF:
+        token_info.token = TOKEN___EOF;
+        *out_token_info = token_info;
+        return P_SUCCESS;
+
+    case P_DECODE_ERROR:
+        /* Update the input position tracking. */
+        context->input_index += match_info.length;
+        context->text_position.row += match_info.delta_position.row;
+        if (match_info.delta_position.row != 0u)
+        {
+            context->text_position.col = match_info.delta_position.col;
+        }
+        else
+        {
+            context->text_position.col += match_info.delta_position.col;
+        }
+        return result;
+
+    default:
+        return result;
+    }
+}
+
+/**
+ * Lex the next token in the input stream.
+ *
+ * @param context
+ *   Lexer/parser context structure.
+ * @param[out] out_token_info
+ *   The lexed token information is stored here if the return value is
+ *   P_SUCCESS.
+ *
+ * @reval P_SUCCESS
+ *   A token was successfully lexed.
+ * @reval P_DECODE_ERROR
+ *   The decoder encountered invalid text encoding.
+ * @reval P_UNEXPECTED_INPUT
+ *   Input text does not match any lexer pattern.
+ */
+size_t <%= @grammar.prefix %>lex(<%= @grammar.prefix %>context_t * context, <%= @grammar.prefix %>token_info_t * out_token_info)
+{
+    for (;;)
+    {
+        size_t result = attempt_lex_token(context, out_token_info);
+        if (result != P_DROP)
+        {
+            return result;
+        }
+    }
+}
+
+/**************************************************************************
+ * Parser
+ *************************************************************************/
+
+/** Reduce ID type. */
+typedef <%= get_type_for(@parser.reduce_table.size) %> reduce_id_t;
+
+/**
+ * A symbol ID can hold either a token ID or a rule set ID.
+ *
+ * Token IDs and rule set IDs share the same namespace, with rule set IDs
+ * beginning after token IDs end.
+ */
+typedef <%= get_type_for(@parser.rule_sets.map(&:last).map(&:id).max) %> symbol_id_t;
+
+/** Parser state ID type. */
+typedef <%= get_type_for(@parser.state_table.size) %> parser_state_id_t;
+
+/** Parser rule ID type. */
+typedef <%= get_type_for(@grammar.rules.size) %> rule_id_t;
+
+/** Parser shift ID type. */
+typedef <%= get_type_for(@parser.shift_table.size) %> shift_id_t;
+
+/** Shift table entry. */
+typedef struct
+{
+    /** Token or rule set ID. */
+    symbol_id_t symbol_id;
+
+    /** Parser state to shift to. */
+    parser_state_id_t state_id;
+} shift_t;
+
+/** Reduce table entry. */
+typedef struct
+{
+    /** Lookahead token. */
+    <%= @grammar.prefix %>token_t token;
+
+    /**
+     * Rule ID.
+     *
+     * This is used to execute the parser user code block associated with a
+     * grammar rule.
+     */
+    rule_id_t rule;
+
+    /**
+     * Rule set ID.
+     *
+     * This is used as the new top symbol ID of the parse stack after this
+     * reduce action.
+     */
+    symbol_id_t rule_set;
+
+    /**
+     * Number of states leading to this reduce action.
+     *
+     * This is the number of entries popped from the parse stack after this
+     * reduce action.
+     */
+    parser_state_id_t n_states;
+} reduce_t;
+
+/** Parser state entry. */
+typedef struct
+{
+    /** First shift table entry for this parser state. */
+    shift_id_t shift_table_index;
+
+    /** Number of shift table entries for this parser state. */
+    shift_id_t n_shift_entries;
+
+    /** First reduce table entry for this parser state. */
+    reduce_id_t reduce_table_index;
+
+    /** Number of reduce table entries for this parser state. */
+    reduce_id_t n_reduce_entries;
+} parser_state_t;
+
+/**
+ * Structure to hold a state ID and value pair.
+ *
+ * A stack of these structures makes up the parse stack.
+ */
+typedef struct
+{
+    /** Parser state ID. */
+    size_t state_id;
+
+    /** Parser value from this state. */
+    <%= @grammar.prefix %>value_t pvalue;
+} state_value_t;
+
+/** Parser shift table. */
+static const shift_t parser_shift_table[] = {
+<%   @parser.shift_table.each do |shift| %>
+    {<%= shift[:symbol_id] %>u, <%= shift[:state_id] %>u},
+<%   end %>
+};
+
+/** Parser reduce table. */
+static const reduce_t parser_reduce_table[] = {
+<%   @parser.reduce_table.each do |reduce| %>
+    {<%= reduce[:token_id] %>u, <%= reduce[:rule_id] %>u, <%= reduce[:rule_set_id] %>u, <%= reduce[:n_states] %>u},
+<%   end %>
+};
+
+/** Parser state table. */
+static const parser_state_t parser_state_table[] = {
+<%   @parser.state_table.each do |state| %>
+    {<%= state[:shift_index] %>u, <%= state[:n_shifts] %>u, <%= state[:reduce_index] %>u, <%= state[:n_reduces] %>u},
+<%   end %>
+};
+
+/* state_values stack functionality */
+
+/** state_values stack type. */
+typedef struct
+{
+    size_t length;
+    size_t capacity;
+    state_value_t * entries;
+} state_values_stack_t;
+
+/**
+ * Initialize state_values stack structure.
+ *
+ * @param stack
+ *   state_values stack structure.
+ */
+static void state_values_stack_init(state_values_stack_t * stack)
+{
+    const size_t initial_capacity = 10u;
+    stack->length = 0u;
+    stack->capacity = initial_capacity;
+    stack->entries = (state_value_t *)malloc(initial_capacity * sizeof(state_value_t));
+}
+
+/**
+ * Index a state_values stack.
+ *
+ * @param stack
+ *   state_values stack structure.
+ * @param index
+ *   Index to the stack.
+ *
+ * @return Pointer to the state value structure at the given index.
+ */
+static state_value_t * state_values_stack_index(state_values_stack_t * stack, int index)
+{
+    if (index >= 0)
+    {
+        return &stack->entries[index];
+    }
+    else
+    {
+        return &stack->entries[stack->length - (size_t)(unsigned int)(-index)];
+    }
+}
+
+/**
+ * Push a new state_value to the state_values stack.
+ *
+ * @param stack
+ *   state_values stack structure.
+ */
+static void state_values_stack_push(state_values_stack_t * stack)
+{
+    size_t const current_capacity = stack->capacity;
+    size_t const current_length = stack->length;
+    if (current_length >= current_capacity)
+    {
+        size_t const new_capacity = current_capacity * 2u;
+        state_value_t * new_entries = malloc(new_capacity * sizeof(state_value_t));
+        memcpy(new_entries, stack->entries, current_length * sizeof(state_value_t));
+        free(stack->entries);
+        stack->capacity = new_capacity;
+        stack->entries = new_entries;
+    }
+    memset(&stack->entries[current_length], 0, sizeof(state_value_t));
+    stack->length = current_length + 1u;
+}
+
+/**
+ * Pop entries from a state_values stack.
+ *
+ * @param stack
+ *   state_values stack structure.
+ * @param n
+ *   Number of states to pop.
+ */
+static void state_values_stack_pop(state_values_stack_t * stack, size_t n)
+{
+    stack->length -= n;
+}
+
+/**
+ * Free memory for a state_values stack structure.
+ *
+ * @param stack
+ *   state_values stack structure.
+ */
+static void state_values_stack_free(state_values_stack_t * stack)
+{
+     free(stack->entries);
+}
+
+/**
+ * Execute user code associated with a parser rule.
+ *
+ * @param rule The ID of the rule.
+ *
+ * @return Parse value.
+ */
+static <%= @grammar.prefix %>value_t parser_user_code(uint32_t rule, state_values_stack_t * statevalues, uint32_t n_states)
+{
+    <%= @grammar.prefix %>value_t _pvalue = {0};
+
+    switch (rule)
+    {
+<% @grammar.rules.each do |rule| %>
+<%   if rule.code %>
+    case <%= rule.id %>u: {
+<%= expand_code(rule.code, true, rule, nil) %>
+    } break;
+<%   end %>
+<% end %>
+    default: break;
+    }
+
+    return _pvalue;
+}
+
+/**
+ * Check if the parser should shift to a new state.
+ *
+ * @param state_id
+ *   Parser state ID.
+ * @param symbol_id
+ *   Incoming token/rule set ID.
+ *
+ * @return State to shift to, or INVALID_ID if none.
+ */
+static size_t check_shift(size_t state_id, size_t symbol_id)
+{
+    uint32_t start = parser_state_table[state_id].shift_table_index;
+    uint32_t end = start + parser_state_table[state_id].n_shift_entries;
+    for (uint32_t i = start; i < end; i++)
+    {
+        if (parser_shift_table[i].symbol_id == symbol_id)
+        {
+            return parser_shift_table[i].state_id;
+        }
+    }
+    return INVALID_ID;
+}
+
+/**
+ * Check if the parser should reduce to a new state.
+ *
+ * @param state_id
+ *   Parser state ID.
+ * @param token
+ *   Incoming token.
+ *
+ * @return State to reduce to, or INVALID_ID if none.
+ */
+static size_t check_reduce(size_t state_id, <%= @grammar.prefix %>token_t token)
+{
+    size_t start = parser_state_table[state_id].reduce_table_index;
+    size_t end = start + parser_state_table[state_id].n_reduce_entries;
+    for (size_t i = start; i < end; i++)
+    {
+        if ((parser_reduce_table[i].token == token) ||
+            (parser_reduce_table[i].token == INVALID_TOKEN_ID))
+        {
+            return i;
+        }
+    }
+    return INVALID_ID;
+}
+
+/**
+ * Run the parser.
+ *
+ * @param context
+ *   Lexer/parser context structure.
+ *
+ * @retval P_SUCCESS
+ *   The parser successfully matched the input text. The parse result value
+ *   can be accessed with <%= @grammar.prefix %>result().
+ * @retval P_UNEXPECTED_TOKEN
+ *   An unexpected token was encountered that does not match any grammar rule.
+ *   The value context->token holds the unexpected token.
+ * @reval P_DECODE_ERROR
+ *   The decoder encountered invalid text encoding.
+ * @reval P_UNEXPECTED_INPUT
+ *   Input text does not match any lexer pattern.
+ */
+size_t <%= @grammar.prefix %>parse(<%= @grammar.prefix %>context_t * context)
+{
+    <%= @grammar.prefix %>token_info_t token_info;
+    <%= @grammar.prefix %>token_t token = INVALID_TOKEN_ID;
+    state_values_stack_t statevalues;
+    size_t reduced_rule_set = INVALID_ID;
+    <%= @grammar.prefix %>value_t reduced_parser_value;
+    state_values_stack_init(&statevalues);
+    state_values_stack_push(&statevalues);
+    size_t result;
+    for (;;)
+    {
+        if (token == INVALID_TOKEN_ID)
+        {
+            size_t lexer_result = <%= @grammar.prefix %>lex(context, &token_info);
+            if (lexer_result != P_SUCCESS)
+            {
+                result = lexer_result;
+                break;
+            }
+            token = token_info.token;
+        }
+        size_t shift_state = INVALID_ID;
+        if (reduced_rule_set != INVALID_ID)
+        {
+            shift_state = check_shift(state_values_stack_index(&statevalues, -1)->state_id, reduced_rule_set);
+        }
+        if (shift_state == INVALID_ID)
+        {
+            shift_state = check_shift(state_values_stack_index(&statevalues, -1)->state_id, token);
+            if ((shift_state != INVALID_ID) && (token == TOKEN___EOF))
+            {
+                /* Successful parse. */
+                context->parse_result = state_values_stack_index(&statevalues, -1)->pvalue;
+                result = P_SUCCESS;
+                break;
+            }
+        }
+        if (shift_state != INVALID_ID)
+        {
+            /* We have something to shift. */
+            state_values_stack_push(&statevalues);
+            state_values_stack_index(&statevalues, -1)->state_id = shift_state;
+            if (reduced_rule_set == INVALID_ID)
+            {
+                /* We shifted a token, mark it consumed. */
+                token = INVALID_TOKEN_ID;
+                state_values_stack_index(&statevalues, -1)->pvalue = token_info.pvalue;
+            }
+            else
+            {
+                /* We shifted a RuleSet. */
+                state_values_stack_index(&statevalues, -1)->pvalue = reduced_parser_value;
+                <%= @grammar.prefix %>value_t new_parse_result = {0};
+                reduced_parser_value = new_parse_result;
+                reduced_rule_set = INVALID_ID;
+            }
+            continue;
+        }
+
+        size_t reduce_index = check_reduce(state_values_stack_index(&statevalues, -1)->state_id, token);
+        if (reduce_index != INVALID_ID)
+        {
+            /* We have something to reduce. */
+            reduced_parser_value = parser_user_code(parser_reduce_table[reduce_index].rule, &statevalues, parser_reduce_table[reduce_index].n_states);
+            reduced_rule_set = parser_reduce_table[reduce_index].rule_set;
+            state_values_stack_pop(&statevalues, parser_reduce_table[reduce_index].n_states);
+            continue;
+        }
+
+        /* A token was successfully lexed, so the input text position was
+         * advanced. However, this is an unexpected token, so we want to reset
+         * the context text position to point to the token rather than the text
+         * after it, so that if the caller wants to report the error position,
+         * it will point to the correct position of the unexpected token. */
+        context->text_position = token_info.position;
+        context->token = token;
+        result = P_UNEXPECTED_TOKEN;
+        break;
+    }
+    state_values_stack_free(&statevalues);
+    return result;
+}
+
+/**
+ * Get the parse result value.
+ *
+ * @param context
+ *   Lexer/parser context structure.
+ *
+ * @return Parse result value.
+ */
+<%= start_rule_type[1] %> <%= @grammar.prefix %>result(<%= @grammar.prefix %>context_t * context)
+{
+    return context->parse_result.v_<%= start_rule_type[0] %>;
+}
+
+/**
+ * Get the current text input position.
+ *
+ * @param context
+ *   Lexer/parser context structure.
+ *
+ * @return Current text position.
+ */
+<%= @grammar.prefix %>position_t <%= @grammar.prefix %>position(<%= @grammar.prefix %>context_t * context)
+{
+    return context->text_position;
+}

assets/parser.h.erb

@@ -0,0 +1,122 @@
+/**
+ * @file
+ *
+ * This file is generated by Propane.
+ */
+
+#pragma once
+
+#include <stdint.h>
+#include <stddef.h>
+
+/**************************************************************************
+ * Public types
+ *************************************************************************/
+
+/* Result codes. */
+#define <%= @grammar.prefix.upcase %>SUCCESS 0u
+#define <%= @grammar.prefix.upcase %>DECODE_ERROR 1u
+#define <%= @grammar.prefix.upcase %>UNEXPECTED_INPUT 2u
+#define <%= @grammar.prefix.upcase %>UNEXPECTED_TOKEN 3u
+#define <%= @grammar.prefix.upcase %>DROP 4u
+#define <%= @grammar.prefix.upcase %>EOF 5u
+
+/** Token type. */
+typedef <%= get_type_for(@grammar.invalid_token_id) %> <%= @grammar.prefix %>token_t;
+
+/** Token IDs. */
+<% @grammar.tokens.each_with_index do |token, index| %>
+#define TOKEN_<%= token.code_name %> <%= index %>u
+<%   unless token.id == index %>
+<%     raise "Token ID (#{token.id}) does not match index (#{index}) for token #{token.name}!" %>
+<%   end %>
+<% end %>
+#define INVALID_TOKEN_ID <%= @grammar.invalid_token_id %>u
+
+/** Code point type. */
+typedef uint32_t <%= @grammar.prefix %>code_point_t;
+
+/** Parser values type(s). */
+typedef union
+{
+<% @grammar.ptypes.each do |name, typestring| %>
+    <%= typestring %> v_<%= name %>;
+<% end %>
+} <%= @grammar.prefix %>value_t;
+
+/**
+ * A structure to keep track of parser position.
+ *
+ * This is useful for reporting errors, etc...
+ */
+typedef struct
+{
+    /** Input text row (0-based). */
+    uint32_t row;
+
+    /** Input text column (0-based). */
+    uint32_t col;
+} <%= @grammar.prefix %>position_t;
+
+/** Lexed token information. */
+typedef struct
+{
+    /** Text position where the token was found. */
+    <%= @grammar.prefix %>position_t position;
+
+    /** Number of input bytes used by the token. */
+    size_t length;
+
+    /** Token that was lexed. */
+    <%= @grammar.prefix %>token_t token;
+
+    /** Parser value associated with the token. */
+    <%= @grammar.prefix %>value_t pvalue;
+} <%= @grammar.prefix %>token_info_t;
+
+/**
+ * Lexer and parser context.
+ *
+ * The user must allocate an instance of this structure and pass it to any
+ * public API function.
+ */
+typedef struct
+{
+    /* Lexer context data. */
+
+    /** Input text. */
+    uint8_t const * input;
+
+    /** Input text length. */
+    size_t input_length;
+
+    /** Input text index (byte offset). */
+    size_t input_index;
+
+    /** Input text position (row/column). */
+    <%= @grammar.prefix %>position_t text_position;
+
+    /** Current lexer mode. */
+    size_t mode;
+
+    /* Parser context data. */
+
+    /** Parse result value. */
+    <%= @grammar.prefix %>value_t parse_result;
+
+    /** Unexpected token received. */
+    <%= @grammar.prefix %>token_t token;
+} <%= @grammar.prefix %>context_t;
+
+void <%= @grammar.prefix %>context_init(<%= @grammar.prefix %>context_t * context, uint8_t const * input, size_t input_length);
+
+size_t <%= @grammar.prefix %>decode_code_point(uint8_t const * input, size_t input_length,
+        <%= @grammar.prefix %>code_point_t * out_code_point, uint8_t * out_code_point_length);
+
+size_t <%= @grammar.prefix %>lex(<%= @grammar.prefix %>context_t * context, <%= @grammar.prefix %>token_info_t * out_token_info);
+
+size_t <%= @grammar.prefix %>parse(<%= @grammar.prefix %>context_t * context);
+
+<%= start_rule_type[1] %> <%= @grammar.prefix %>result(<%= @grammar.prefix %>context_t * context);
+
+<%= @grammar.prefix %>position_t <%= @grammar.prefix %>position(<%= @grammar.prefix %>context_t * context);

doc/user_guide.md

@@ -0,0 +1,130 @@
+${remove}
+WARNING: This user guide is meant to be preprocessed and rendered by a custom
+script.
+The markdown source file is not intended to be viewed directly and will not
+include all intended content.
+${/remove}
+
+#> Overview
+
+Propane is an LR Parser Generator (LPG) which:
+
+  * accepts LR(0), SLR, and LALR grammars
+  * generates a built-in lexer to tokenize input
+  * supports UTF-8 lexer inputs
+  * generates a table-driven parser to parse input in linear time
+  * is MIT-licensed
+  * is distributable as a standalone Ruby script
+  * supports D language
+
+#> Installation
+
+Propane is designed to be distributed as a stand-alone single file script that
+can be copied into and versioned in a project's source tree.
+The only requirement to run Propane is that the system has a Ruby interpreter
+installed.
+The latest release can be downloaded from [https://github.com/holtrop/propane/releases](https://github.com/holtrop/propane/releases).
+Simply copy the `propane` executable script into the desired location within
+the project to be built (typically the root of the repository) and mark it
+executable.
+
+#> Command Line Usage
+
+Propane is typically invoked from the command-line as `./propane`.
+
+    Usage: ./propane [options] <input-file> <output-file>
+    Options:
+      --log LOG   Write log file
+      --version   Show program version and exit
+      -h, --help  Show this usage and exit
+
+The user must specify the path to a Propane input grammar file and a path to an
+output file.
+The generated source code will be written to the output file.
+If a log file path is specified, Propane will write a log file containing
+detailed information about the parser states and transitions.
+
+#> Propane Grammar File
+
+A Propane grammar file provides Propane with the patterns, tokens, grammar
+rules, and user code blocks from which to build the generated lexer and parser.
+
+Example grammar file:
+
+```
+<<
+import std.math;
+>>
+
+ptype ulong;
+
+token plus /\\+/;
+token times /\\*/;
+token power /\\*\\*/;
+token integer /\\d+/ <<
+  ulong v;
+  foreach (c; match)
+  {
+    v *= 10;
+    v += (c - '0');
+  }
+  $$ = v;
+>>
+token lparen /\\(/;
+token rparen /\\)/;
+drop /\\s+/;
+
+Start -> E1 <<
+  $$ = $1;
+>>
+E1 -> E2 <<
+  $$ = $1;
+>>
+E1 -> E1 plus E2 <<
+  $$ = $1 + $3;
+>>
+E2 -> E3 <<
+  $$ = $1;
+>>
+E2 -> E2 times E3 <<
+  $$ = $1 * $3;
+>>
+E3 -> E4 <<
+  $$ = $1;
+>>
+E3 -> E3 power E4 <<
+  $$ = pow($1, $3);
+>>
+E4 -> integer <<
+  $$ = $1;
+>>
+E4 -> lparen E1 rparen <<
+  $$ = $2;
+>>
+```
+
+#> License
+
+Propane is licensed under the terms of the MIT License:
+
+```
+${include LICENSE.txt}
+```
+
+#> Contributing
+
+Propane is developed on [github](https://github.com/holtrop/propane).
+
+Issues may be submitted to [https://github.com/holtrop/propane/issues](https://github.com/holtrop/propane/issues).
+
+Pull requests may be submitted as well:
+
+  1. Fork it
+  2. Create your feature branch (`git checkout -b my-new-feature`)
+  3. Commit your changes (`git commit -am 'Add some feature'`)
+  4. Push to the branch (`git push origin my-new-feature`)
+  5. Create new Pull Request
+
+#> Change Log
+
+${changelog}

lib/propane.rb

@@ -1,5 +1,6 @@
 require "erb"
 require "set"
+require "stringio"
 require_relative "propane/cli"
 require_relative "propane/code_point_range"
 require_relative "propane/fa"
@@ -12,28 +13,18 @@ require_relative "propane/lexer/dfa"
 require_relative "propane/parser"
 require_relative "propane/parser/item"
 require_relative "propane/parser/item_set"
+require_relative "propane/pattern"
 require_relative "propane/regex"
 require_relative "propane/regex/nfa"
 require_relative "propane/regex/unit"
 require_relative "propane/rule_set"
 require_relative "propane/rule"
 require_relative "propane/token"
+require_relative "propane/util"
 require_relative "propane/version"
 
 class Propane
 
-  # EOF.
-  TOKEN_EOF = 0xFFFFFFFC
-
-  # Decoding error.
-  TOKEN_DECODE_ERROR = 0xFFFFFFFD
-
-  # Token ID for a "dropped" token.
-  TOKEN_DROP = 0xFFFFFFFE
-
-  # Invalid token ID.
-  TOKEN_NONE = 0xFFFFFFFF
-
   class Error < RuntimeError
   end
 

lib/propane/generator.rb

@@ -5,40 +5,108 @@ class Propane
     def initialize(grammar, output_file, log_file)
       @grammar = grammar
       @output_file = output_file
-      @log_file = log_file
+      if log_file
+        @log = File.open(log_file, "wb")
+      else
+        @log = StringIO.new
+      end
       @classname = @grammar.classname || File.basename(output_file).sub(%r{[^a-zA-Z0-9].*}, "").capitalize
+      @language =
+        if output_file =~ /\.([a-z]+)$/
+          $1
+        else
+          "d"
+        end
       process_grammar!
     end
 
     def generate
-      erb = ERB.new(File.read(File.join(File.dirname(File.expand_path(__FILE__)), "../../assets/parser.d.erb")), trim_mode: "<>")
+      extensions = [@language]
+      if @language == "c"
+        extensions += %w[h]
+      end
+      extensions.each do |extension|
+        erb = ERB.new(File.read(File.join(File.dirname(File.expand_path(__FILE__)), "../../assets/parser.#{extension}.erb")), trim_mode: "<>")
+        output_file = @output_file.sub(%r{\.[a-z]+$}, ".#{extension}")
         result = erb.result(binding.clone)
-      File.open(@output_file, "wb") do |fh|
+        File.open(output_file, "wb") do |fh|
           fh.write(result)
         end
       end
+      @log.close
+    end
 
     private
 
     def process_grammar!
+      # Assign default pattern mode to patterns without a mode assigned.
+      found_default = false
+      @grammar.patterns.each do |pattern|
+        if pattern.mode.nil?
+          pattern.mode = "default"
+          found_default = true
+        end
+        pattern.ptypename ||= "default"
+      end
+      unless found_default
+        raise Error.new("No patterns found for default mode")
+      end
+      # Add EOF token.
+      @grammar.tokens << Token.new("$EOF", nil, nil)
       tokens_by_name = {}
-      @grammar.tokens.each do |token|
+      @grammar.tokens.each_with_index do |token, token_id|
+        # Assign token ID.
+        token.id = token_id
+        # Set default ptypename if none given.
+        token.ptypename ||= "default"
+        # Check for token name conflicts.
         if tokens_by_name.include?(token.name)
           raise Error.new("Duplicate token name #{token.name.inspect}")
         end
         tokens_by_name[token.name] = token
       end
+      # Check for user start rule.
+      unless @grammar.rules.find {|rule| rule.name == "Start"}
+        raise Error.new("Start rule not found")
+      end
+      # Add "real" start rule.
+      @grammar.rules.unshift(Rule.new("$Start", ["Start", "$EOF"], nil, nil, nil))
       rule_sets = {}
-      @grammar.rules.each do |rule|
+      rule_set_id = @grammar.tokens.size
+      @grammar.rules.each_with_index do |rule, rule_id|
+        # Assign rule ID.
+        rule.id = rule_id
+        # Check for token/rule name conflict.
         if tokens_by_name.include?(rule.name)
           raise Error.new("Rule name collides with token name #{rule.name.inspect}")
         end
-        rule_sets[rule.name] ||= RuleSet.new(rule.name)
+        # Build rule sets of all rules with the same name.
-        rule_sets[rule.name] << rule
+        unless rule_sets[rule.name]
+          rule_sets[rule.name] = RuleSet.new(rule.name, rule_set_id)
+          rule_set_id += 1
         end
-      unless rule_sets["Start"]
+        rule_set = rule_sets[rule.name]
-        raise Error.new("Start rule not found")
+        if rule_set.ptypename && rule.ptypename && rule_set.ptypename != rule.ptypename
+          raise Error.new("Conflicting ptypes for rule #{rule.name}")
         end
+        rule_set.ptypename ||= rule.ptypename
+        rule.rule_set = rule_set
+        rule_set << rule
+      end
+      rule_sets.each do |name, rule_set|
+        rule_set.ptypename ||= "default"
+        # Assign rule set ptypenames back to rules.
+        rule_set.rules.each do |rule|
+          rule.ptypename = rule_set.ptypename
+        end
+      end
+      # Generate lexer user code IDs for lexer patterns with user code blocks.
+      @grammar.patterns.select do |pattern|
+        pattern.code
+      end.each_with_index do |pattern, code_id|
+        pattern.code_id = code_id
+      end
+      # Map rule components from names to Token/RuleSet objects.
       @grammar.rules.each do |rule|
         rule.components.map! do |component|
           if tokens_by_name[component]
@@ -50,8 +118,166 @@ class Propane
           end
         end
       end
-      @lexer = Lexer.new(@grammar.tokens, @grammar.drop_tokens)
+      determine_possibly_empty_rulesets!(rule_sets)
-      @parser = Parser.new(rule_sets["Start"])
+      # Generate the lexer.
+      @lexer = Lexer.new(@grammar)
+      # Generate the parser.
+      @parser = Parser.new(@grammar, rule_sets, @log)
+    end
+
+    # Determine which grammar rules could expand to empty sequences.
+    #
+    # @param rule_sets [Hash]
+    #   RuleSets.
+    #
+    # @return [void]
+    def determine_possibly_empty_rulesets!(rule_sets)
+      begin
+        newly_discovered_empty_rulesets = false
+        rule_sets.each do |name, rule_set|
+          unless rule_set.could_be_empty?
+            if could_rule_set_be_empty?(rule_set)
+              newly_discovered_empty_rulesets = true
+              rule_set.could_be_empty = true
+            end
+          end
+        end
+      end while newly_discovered_empty_rulesets
+    end
+
+    # Determine whether a RuleSet could be empty.
+    #
+    # @param rule_set [RuleSet]
+    #   RuleSet to test.
+    #
+    # @return [Boolean]
+    #   Whether the RuleSet could be empty.
+    def could_rule_set_be_empty?(rule_set)
+      rule_set.rules.any? do |rule|
+        could_rule_be_empty?(rule)
+      end
+    end
+
+    # Determine whether a Rule could be empty.
+    #
+    # @param rule [Rule]
+    #   Rule to test.
+    #
+    # @return [Boolean]
+    #   Whether the Rule could be empty.
+    def could_rule_be_empty?(rule)
+      i = 0
+      loop do
+        if i == rule.components.size
+          return true
+        end
+        if rule.components[i].is_a?(Token)
+          return false
+        end
+        if !rule.components[i].could_be_empty?
+          return false
+        end
+        i += 1
+      end
+    end
+
+    # Expand expansions in user code block.
+    #
+    # @param code [String]
+    #   User code block.
+    # @param parser [Boolean]
+    #   Whether the user code is for the parser or lexer.
+    # @param rule [Rule, nil]
+    #   The Rule associated with the user code if user code is for the parser.
+    # @param pattern [Pattern, nil]
+    #   The Pattern associated with the user code if user code is for the lexer.
+    #
+    # @return [String]
+    #   Expanded user code block.
+    def expand_code(code, parser, rule, pattern)
+      code = code.gsub(/\$token\(([$\w]+)\)/) do |match|
+        "TOKEN_#{Token.code_name($1)}"
+      end
+      if parser
+        code = code.gsub(/\$\$/) do |match|
+          "_pvalue.v_#{rule.ptypename}"
+        end
+        code = code.gsub(/\$(\d+)/) do |match|
+          index = $1.to_i
+          case @language
+          when "c"
+            "state_values_stack_index(statevalues, -1 - (int)n_states + #{index})->pvalue.v_#{rule.components[index - 1].ptypename}"
+          when "d"
+            "statevalues[$-1-n_states+#{index}].pvalue.v_#{rule.components[index - 1].ptypename}"
+          end
+        end
+      else
+        code = code.gsub(/\$\$/) do |match|
+          case @language
+          when "c"
+            "out_token_info->pvalue.v_#{pattern.ptypename}"
+          when "d"
+            "out_token_info.pvalue.v_#{pattern.ptypename}"
+          end
+        end
+        code = code.gsub(/\$mode\(([a-zA-Z_][a-zA-Z_0-9]*)\)/) do |match|
+          mode_name = $1
+          mode_id = @lexer.mode_id(mode_name)
+          unless mode_id
+            raise Error.new("Lexer mode '#{mode_name}' not found")
+          end
+          case @language
+          when "c"
+            "context->mode = #{mode_id}u"
+          when "d"
+            "context.mode = #{mode_id}u"
+          end
+        end
+      end
+      code
+    end
+
+    # Get the parser value type for the start rule.
+    #
+    # @return [Array<String>]
+    #   Start rule parser value type name and type string.
+    def start_rule_type
+      start_rule = @grammar.rules.find do |rule|
+        rule.name == "Start"
+      end
+      [start_rule.ptypename, @grammar.ptypes[start_rule.ptypename]]
+    end
+
+    # Get an unsigned integer type that can hold the given maximum value.
+    #
+    # @param max [Integer]
+    #   Maximum value to store.
+    #
+    # @return [String]
+    #   Type.
+    def get_type_for(max)
+      if max <= 0xFF
+        case @language
+        when "c"
+          "uint8_t"
+        when "d"
+          "ubyte"
+        end
+      elsif max <= 0xFFFF
+        case @language
+        when "c"
+          "uint16_t"
+        when "d"
+          "ushort"
+        end
+      else
+        case @language
+        when "c"
+          "uint32_t"
+        else
+          "uint"
+        end
+      end
     end
 
   end

lib/propane/grammar.rb

@@ -2,60 +2,259 @@ class Propane
 
   class Grammar
 
+    # Reserve identifiers beginning with a double-underscore for internal use.
+    IDENTIFIER_REGEX = /(?:[a-zA-Z]|_[a-zA-Z0-9])[a-zA-Z_0-9]*/
+
     attr_reader :classname
-    attr_reader :drop_tokens
     attr_reader :modulename
+    attr_reader :patterns
     attr_reader :rules
     attr_reader :tokens
+    attr_reader :code_blocks
+    attr_reader :ptypes
+    attr_reader :prefix
 
     def initialize(input)
+      @patterns = []
       @tokens = []
-      @drop_tokens = []
       @rules = []
-      input = input.gsub("\r\n", "\n")
+      @code_blocks = []
-      parse_grammar(input)
+      @line_number = 1
+      @next_line_number = @line_number
+      @mode = nil
+      @input = input.gsub("\r\n", "\n")
+      @ptypes = {"default" => "void *"}
+      @prefix = "p_"
+      parse_grammar!
+    end
+
+    def ptype
+      @ptypes["default"]
+    end
+
+    def invalid_token_id
+      @tokens.size
     end
 
     private
 
-    def parse_grammar(input)
+    def parse_grammar!
-      line_number = 1
+      while @input.size > 0
-      while !input.empty?
+        parse_statement!
-        if sliced = input.slice!(/\A\s+/)
-          # Skip white space.
-        elsif sliced = input.slice!(/\A#.*\n/)
-          # Skip comment lines.
-        elsif sliced = input.slice!(/\Amodule\s+(\S+)\n/)
-          @modulename = $1
-        elsif sliced = input.slice!(/\Aclass\s+(\S+)\n/)
-          @classname = $1
-        elsif sliced = input.slice!(/\Atoken\s+(\S+)(?:\s+(\S+))?\n/)
-          name, pattern = $1, $2
-          if pattern.nil?
-            pattern = name
       end
-          unless name =~ /^[a-zA-Z_][a-zA-Z_0-9]*$/
-            raise Error.new("Invalid token name #{name.inspect}")
     end
-          @tokens << Token.new(name: name, pattern: pattern, id: @tokens.size, line_number: line_number)
+
-        elsif sliced = input.slice!(/\Adrop\s+(\S+)\n/)
+    def parse_statement!
-          pattern = $1
+      if parse_white_space!
-          @drop_tokens << Token.new(pattern: pattern, line_number: line_number)
+      elsif parse_comment_line!
-        elsif sliced = input.slice!(/\A(\S+)\s*->\s*(.*?)(?:;|<<\n(.*?)^>>\n)/m)
+      elsif @mode.nil? && parse_mode_label!
-          rule_name, components, code = $1, $2, $3
+      elsif parse_module_statement!
-          unless rule_name =~ /^[a-zA-Z_][a-zA-Z_0-9]*$/
+      elsif parse_class_statement!
-            raise Error.new("Invalid rule name #{name.inspect}")
+      elsif parse_ptype_statement!
-          end
+      elsif parse_pattern_statement!
-          components = components.strip.split(/\s+/)
+      elsif parse_token_statement!
-          # Reserve rule ID 0 for the "real" start rule.
+      elsif parse_tokenid_statement!
-          @rules << Rule.new(rule_name, components, code, line_number, @rules.size + 1)
+      elsif parse_drop_statement!
+      elsif parse_rule_statement!
+      elsif parse_code_block_statement!
+      elsif parse_prefix_statement!
       else
-          if input.size > 25
+        if @input.size > 25
-            input = input.slice(0..20) + "..."
+          @input = @input.slice(0..20) + "..."
         end
-          raise Error.new("Unexpected grammar input at line #{line_number}: #{input.chomp}")
+        raise Error.new("Unexpected grammar input at line #{@line_number}: #{@input.chomp}")
       end
-        line_number += sliced.count("\n")
+    end
+
+    def parse_mode_label!
+      if md = consume!(/(#{IDENTIFIER_REGEX})\s*:/)
+        @mode = md[1]
+      end
+    end
+
+    def parse_white_space!
+      consume!(/\s+/)
+    end
+
+    def parse_comment_line!
+      consume!(/#.*\n/)
+    end
+
+    def parse_module_statement!
+      if consume!(/module\s+/)
+        md = consume!(/([\w.]+)\s*/, "expected module name")
+        @modulename = md[1]
+        consume!(/;/, "expected `;'")
+        @mode = nil
+        true
+      end
+    end
+
+    def parse_class_statement!
+      if consume!(/class\s+/)
+        md = consume!(/([\w.]+)\s*/, "expected class name")
+        @classname = md[1]
+        consume!(/;/, "expected `;'")
+        @mode = nil
+        true
+      end
+    end
+
+    def parse_ptype_statement!
+      if consume!(/ptype\s+/)
+        name = "default"
+        if md = consume!(/(#{IDENTIFIER_REGEX})\s*=\s*/)
+          name = md[1]
+        end
+        md = consume!(/([^;]+);/, "expected parser result type expression")
+        @ptypes[name] = md[1].strip
+      end
+    end
+
+    def parse_token_statement!
+      if consume!(/token\s+/)
+        md = consume!(/(#{IDENTIFIER_REGEX})\s*/, "expected token name")
+        name = md[1]
+        if md = consume!(/\((#{IDENTIFIER_REGEX})\)\s*/)
+          ptypename = md[1]
+        end
+        pattern = parse_pattern! || name
+        consume!(/\s+/)
+        unless code = parse_code_block!
+          consume!(/;/, "expected pattern or `;' or code block")
+        end
+        token = Token.new(name, ptypename, @line_number)
+        @tokens << token
+        pattern = Pattern.new(pattern: pattern, token: token, line_number: @line_number, code: code, mode: @mode, ptypename: ptypename)
+        @patterns << pattern
+        @mode = nil
+        true
+      end
+    end
+
+    def parse_tokenid_statement!
+      if md = consume!(/tokenid\s+/)
+        md = consume!(/(#{IDENTIFIER_REGEX})\s*/, "expected token name")
+        name = md[1]
+        if md = consume!(/\((#{IDENTIFIER_REGEX})\)\s*/)
+          ptypename = md[1]
+        end
+        consume!(/;/, "expected `;'");
+        token = Token.new(name, ptypename, @line_number)
+        @tokens << token
+        @mode = nil
+        true
+      end
+    end
+
+    def parse_drop_statement!
+      if md = consume!(/drop\s+/)
+        pattern = parse_pattern!
+        unless pattern
+          raise Error.new("Line #{@line_number}: expected pattern to follow `drop'")
+        end
+        consume!(/\s+/)
+        consume!(/;/, "expected `;'")
+        @patterns << Pattern.new(pattern: pattern, line_number: @line_number, mode: @mode)
+        @mode = nil
+        true
+      end
+    end
+
+    def parse_rule_statement!
+      if md = consume!(/(#{IDENTIFIER_REGEX})\s*(?:\((#{IDENTIFIER_REGEX})\))?\s*->\s*/)
+        rule_name, ptypename = *md[1, 2]
+        md = consume!(/((?:#{IDENTIFIER_REGEX}\s*)*)\s*/, "expected rule component list")
+        components = md[1].strip.split(/\s+/)
+        unless code = parse_code_block!
+          consume!(/;/, "expected pattern or `;' or code block")
+        end
+        @rules << Rule.new(rule_name, components, code, ptypename, @line_number)
+        @mode = nil
+        true
+      end
+    end
+
+    def parse_pattern_statement!
+      if pattern = parse_pattern!
+        consume!(/\s+/)
+        if md = consume!(/\((#{IDENTIFIER_REGEX})\)\s*/)
+          ptypename = md[1]
+        end
+        unless code = parse_code_block!
+          raise Error.new("Line #{@line_number}: expected code block to follow pattern")
+        end
+        @patterns << Pattern.new(pattern: pattern, line_number: @line_number, code: code, mode: @mode, ptypename: ptypename)
+        @mode = nil
+        true
+      end
+    end
+
+    def parse_code_block_statement!
+      if code = parse_code_block!
+        @code_blocks << code
+        @mode = nil
+        true
+      end
+    end
+
+    def parse_prefix_statement!
+      if md = consume!(/prefix\s+(#{IDENTIFIER_REGEX})\s*;/)
+        @prefix = md[1]
+        true
+      end
+    end
+
+    def parse_pattern!
+      if md = consume!(%r{/})
+        pattern = ""
+        while !consume!(%r{/})
+          if consume!(%r{\\})
+            pattern += "\\"
+            if md = consume!(%r{(.)})
+              pattern += md[1]
+            else
+              raise Error.new("Line #{@line_number}: unterminated escape sequence")
+            end
+          elsif md = consume!(%r{(.)})
+            pattern += md[1]
+          end
+        end
+        pattern
+      end
+    end
+
+    def parse_code_block!
+      if md = consume!(/<<\n(.*?)^>>\n/m)
+        md[1]
+      end
+    end
+
+    # Check if the input string matches the given regex.
+    #
+    # If so, remove the match from the input string, and update the line
+    # number. If the regex is not matched and an error message is provided,
+    # the error is raised.
+    #
+    # @param regex [Regexp]
+    #   Regex to attempt to match.
+    # @param error_message [String, nil]
+    #   Error message to display if the regex is not matched. If nil and the
+    #   regex is not matched, an error is not raised.
+    #
+    # @return [MatchData, nil]
+    #   MatchData for the given regex if it was matched and removed from the
+    #   input.
+    def consume!(regex, error_message = nil)
+      @line_number = @next_line_number
+      if md = @input.match(/\A#{regex}/)
+        @input.slice!(0, md[0].size)
+        @next_line_number += md[0].count("\n")
+        md
+      elsif error_message
+        raise Error.new("Line #{@line_number}: Error: #{error_message}")
+      else
+        false
       end
     end
 

lib/propane/lexer.rb

@@ -1,41 +1,67 @@
 class Propane
   class Lexer
 
-    # @return [DFA]
+    attr_reader :state_table
-    #   Lexer DFA.
+    attr_reader :transition_table
-    attr_accessor :dfa
+    attr_reader :mode_table
 
-    def initialize(tokens, drop_tokens)
+    def initialize(grammar)
-      @dfa = DFA.new(tokens, drop_tokens)
+      @grammar = grammar
+      build_tables!
     end
 
-    def build_tables
+    # Get ID for a mode.
-      transition_table = []
+    #
-      state_table = []
+    # @param mode_name [String]
-      states = @dfa.enumerate
+    #   Mode name.
+    #
+    # @return [Integer, nil]
+    #   Mode ID.
+    def mode_id(mode_name)
+      if mode_info = @modes[mode_name]
+        mode_info[:id]
+      end
+    end
+
+    private
+
+    def build_tables!
+      @modes = @grammar.patterns.group_by do |pattern|
+        pattern.mode
+      end.transform_values do |patterns|
+        {dfa: DFA.new(patterns)}
+      end
+      @modes.each_with_index do |(mode_name, mode_info), index|
+        mode_info[:id] = index
+      end
+      @state_table = []
+      @transition_table = []
+      @mode_table = []
+      @modes.each do |mode_name, mode_info|
+        state_table_offset = @state_table.size
+        @mode_table << {
+          state_table_offset: state_table_offset,
+        }
+        states = mode_info[:dfa].enumerate
         states.each do |state, id|
-        accepts =
+          token = state.accepts && state.accepts.token && state.accepts.token.id
-          if state.accepts.nil?
+          code_id = state.accepts && state.accepts.code_id && state.accepts.code_id
-            TOKEN_NONE
+          @state_table << {
-          elsif state.accepts.drop?
+            transition_table_index: @transition_table.size,
-            TOKEN_DROP
-          else
-            state.accepts.id
-          end
-        state_table << {
-          transition_table_index: transition_table.size,
             n_transitions: state.transitions.size,
-          accepts: accepts,
+            accepts: !!state.accepts,
+            token: token,
+            code_id: code_id,
           }
           state.transitions.each do |transition|
-          transition_table << {
+            @transition_table << {
               first: transition.code_point_range.first,
               last: transition.code_point_range.last,
-            destination: states[transition.destination],
+              destination: states[transition.destination] + state_table_offset,
             }
           end
         end
-      [transition_table, state_table]
+      end
     end
 
   end

lib/propane/lexer/dfa.rb

@@ -3,11 +3,11 @@ class Propane
 
     class DFA < FA
 
-      def initialize(tokens, drop_tokens)
+      def initialize(patterns)
         super()
         start_nfa = Regex::NFA.new
-        (tokens + drop_tokens).each do |token|
+        patterns.each do |pattern|
-          start_nfa.start_state.add_transition(nil, token.nfa.start_state)
+          start_nfa.start_state.add_transition(nil, pattern.nfa.start_state)
         end
         @nfa_state_sets = {}
         @states = []
@@ -40,7 +40,7 @@ class Propane
           nfa_state_set.each do |nfa_state|
             if nfa_state.accepts
               if state.accepts
-                if nfa_state.accepts.id < state.accepts.id
+                if nfa_state.accepts.line_number < state.accepts.line_number
                   state.accepts = nfa_state.accepts
                 end
               else

lib/propane/parser.rb

@@ -2,74 +2,91 @@ class Propane
 
   class Parser
 
-    def initialize(start_rule_set)
+    attr_reader :state_table
-      @eof_token = Token.new(name: "$", id: TOKEN_EOF)
+    attr_reader :shift_table
-      start_rule = Rule.new("$$", [start_rule_set, @eof_token], nil, nil, 0)
+    attr_reader :reduce_table
+    attr_reader :rule_sets
+
+    def initialize(grammar, rule_sets, log)
+      @grammar = grammar
+      @rule_sets = rule_sets
+      @log = log
       @item_sets = []
       @item_sets_set = {}
-      start_item = Item.new(start_rule, 0)
+      start_item = Item.new(grammar.rules.first, 0)
-      eval_item_sets = Set.new
+      eval_item_sets = Set[ItemSet.new([start_item])]
-      eval_item_sets << ItemSet.new([start_item])
 
       while eval_item_sets.size > 0
-        this_eval_item_sets = eval_item_sets
+        item_set = eval_item_sets.first
-        eval_item_sets = Set.new
+        eval_item_sets.delete(item_set)
-        this_eval_item_sets.each do |item_set|
         unless @item_sets_set.include?(item_set)
           item_set.id = @item_sets.size
           @item_sets << item_set
           @item_sets_set[item_set] = item_set
           item_set.following_symbols.each do |following_symbol|
-              unless following_symbol == @eof_token
+            unless following_symbol.name == "$EOF"
               following_set = item_set.build_following_item_set(following_symbol)
               eval_item_sets << following_set
             end
           end
         end
       end
-      end
 
       @item_sets.each do |item_set|
         process_item_set(item_set)
-        puts "Item set #{item_set.id}:"
-        ids = item_set.in_sets.map(&:id)
-        if ids.size > 0
-          puts "    (in from #{ids.join(", ")})"
-        end
-        puts item_set
-        item_set.following_item_set.each do |following_symbol, following_item_set|
-          puts " #{following_symbol.name} => #{following_item_set.id}"
-        end
-        puts
-      end
       end
 
-    def build_tables
+      build_reduce_actions!
-      shift_table = []
+      write_log!
-      state_table = []
+      build_tables!
-      @item_sets.each do |item_set|
-        shift_entries = item_set.following_symbols.select do |following_symbol|
-          following_symbol.is_a?(Token)
-        end.map do |following_symbol|
-          {
-            token_id: following_symbol.id,
-            state_id: item_set.following_item_set[following_symbol].id,
-          }
-        end
-        state_table << {
-          shift_index: shift_table.size,
-          n_shifts: shift_entries.size,
-        }
-        shift_table += shift_entries
-      end
-      [state_table, shift_table]
     end
 
     private
 
+    def build_tables!
+      @state_table = []
+      @shift_table = []
+      @reduce_table = []
+      @item_sets.each do |item_set|
+        shift_entries = item_set.following_symbols.map do |following_symbol|
+          state_id =
+            if following_symbol.name == "$EOF"
+              0
+            else
+              item_set.following_item_set[following_symbol].id
+            end
+          {
+            symbol_id: following_symbol.id,
+            state_id: state_id,
+          }
+        end
+        reduce_entries =
+          case ra = item_set.reduce_actions
+          when Rule
+            [{token_id: @grammar.invalid_token_id, rule_id: ra.id,
+              rule_set_id: ra.rule_set.id, n_states: ra.components.size}]
+          when Hash
+            ra.map do |token, rule|
+              {token_id: token.id, rule_id: rule.id,
+               rule_set_id: rule.rule_set.id, n_states: rule.components.size}
+            end
+          else
+            []
+          end
+        @state_table << {
+          shift_index: @shift_table.size,
+          n_shifts: shift_entries.size,
+          reduce_index: @reduce_table.size,
+          n_reduces: reduce_entries.size,
+        }
+        @shift_table += shift_entries
+        @reduce_table += reduce_entries
+      end
+    end
+
     def process_item_set(item_set)
       item_set.following_symbols.each do |following_symbol|
-        unless following_symbol == @eof_token
+        unless following_symbol.name == "$EOF"
           following_set = @item_sets_set[item_set.build_following_item_set(following_symbol)]
           item_set.following_item_set[following_symbol] = following_set
           following_set.in_sets << item_set
@@ -77,6 +94,168 @@ class Propane
       end
     end
 
+    # Build the reduce actions for each ItemSet.
+    #
+    # @return [void]
+    def build_reduce_actions!
+      @item_sets.each do |item_set|
+        item_set.reduce_actions = build_reduce_actions_for_item_set(item_set)
+      end
+    end
+
+    # Build the reduce actions for a single item set (parser state).
+    #
+    # @param item_set [ItemSet]
+    #   ItemSet (parser state)
+    #
+    # @return [nil, Rule, Hash]
+    #   If no reduce actions are possible for the given item set, nil.
+    #   If only one reduce action is possible for the given item set, the Rule
+    #   to reduce.
+    #   Otherwise, a mapping of lookahead Tokens to the Rules to reduce.
+    def build_reduce_actions_for_item_set(item_set)
+      # To build the reduce actions, we start by looking at any
+      # "complete" items, i.e., items where the parse position is at the
+      # end of a rule. These are the only rules that are candidates for
+      # reduction in the current ItemSet.
+      reduce_rules = Set.new(item_set.items.select(&:complete?).map(&:rule))
+
+      # If there are no rules to reduce for this ItemSet, we're done here.
+      return nil if reduce_rules.size == 0
+
+      # If there is exactly one rule to reduce for this ItemSet, then do not
+      # figure out the lookaheads; just reduce it.
+      return reduce_rules.first if reduce_rules.size == 1
+
+      # Otherwise, we have more than one possible rule to reduce.
+
+      # We will be looking for all possible tokens that can follow instances of
+      # these rules. Rather than looking through the entire grammar for the
+      # possible following tokens, we will only look in the item sets leading
+      # up to this one. This restriction gives us a more precise lookahead set,
+      # and allows us to parse LALR grammars.
+      item_sets = item_set.leading_item_sets
+      reduce_rules.reduce({}) do |reduce_actions, reduce_rule|
+        lookahead_tokens_for_rule = build_lookahead_tokens_to_reduce(reduce_rule, item_sets)
+        lookahead_tokens_for_rule.each do |lookahead_token|
+          if existing_reduce_rule = reduce_actions[lookahead_token]
+            raise Error.new("Error: reduce/reduce conflict between rule #{existing_reduce_rule.id} (#{existing_reduce_rule.name}) and rule #{reduce_rule.id} (#{reduce_rule.name})")
+          end
+          reduce_actions[lookahead_token] = reduce_rule
+        end
+        reduce_actions
+      end
+    end
+
+    # Build the set of lookahead Tokens that should cause the given Rule to be
+    # reduced in the given context of ItemSets.
+    #
+    # @param rule [Rule]
+    #   Rule to reduce.
+    # @param item_sets [Set<ItemSet>]
+    #   ItemSets to consider for the context in which to reduce this Rule.
+    #
+    # @return [Set<Token>]
+    #   Possible lookahead Tokens for the given Rule within the context of the
+    #   given ItemSets.
+    def build_lookahead_tokens_to_reduce(rule, item_sets)
+      # We need to look for possible following tokens for this reduce rule. We
+      # do this by looking for tokens that follow the reduce rule, or the
+      # start token set for any other rule that follows the reduce rule.
+      # While doing this, the following situations could arise:
+      # 1. We may come across a following rule that could be empty. In this
+      #    case, in addition to the start token set for that rule, we must also
+      #    continue to the next following symbol after the potentially empty
+      #    rule and continue the search for potential following tokens.
+      # 2. We may reach the end of a rule that was not one of the original
+      #    reduce rules. In this case, we must also search for all potential
+      #    following tokens for this rule as well.
+      lookahead_tokens = Set.new
+      rule_sets_to_check_after = [rule.rule_set]
+      checked_rule_sets = Set.new
+      while !rule_sets_to_check_after.empty?
+        rule_set = rule_sets_to_check_after.slice!(0)
+        checked_rule_sets << rule_set
+        # For each RuleSet we're checking, we're going to look through all
+        # items in the item sets of interest and gather all possible following
+        # tokens to form the lookahead token set.
+        item_sets.each do |item_set|
+          item_set.items.each do |item|
+            if item.following_symbol == rule_set
+              (1..).each do |offset|
+                case symbol = item.following_symbol(offset)
+                when nil
+                  rule_set = item.rule.rule_set
+                  unless checked_rule_sets.include?(rule_set)
+                    rule_sets_to_check_after << rule_set
+                  end
+                  break
+                when Token
+                  lookahead_tokens << symbol
+                  break
+                when RuleSet
+                  lookahead_tokens += symbol.start_token_set
+                  unless symbol.could_be_empty?
+                    break
+                  end
+                end
+              end
+            end
+          end
+        end
+      end
+      lookahead_tokens
+    end
+
+    def write_log!
+      @log.puts Util.banner("Parser Rules")
+      @grammar.rules.each do |rule|
+        @log.puts
+        @log.puts "Rule #{rule.id}:"
+        @log.puts "  #{rule}"
+      end
+
+      @log.puts
+      @log.puts Util.banner("Parser Tokens")
+      @log.puts
+      @grammar.tokens.each do |token|
+        @log.puts "Token #{token.id}: #{token.name}"
+      end
+
+      @log.puts
+      @log.puts Util.banner("Parser Rule Sets")
+      @rule_sets.each do |rule_set_name, rule_set|
+        @log.puts
+        @log.puts "Rule Set #{rule_set.id}: #{rule_set_name}"
+        @log.puts "  Start token set: #{rule_set.start_token_set.map(&:name).join(", ")}"
+      end
+
+      @log.puts
+      @log.puts Util.banner("Parser States")
+      @item_sets.each do |item_set|
+        @log.puts
+        @log.puts "State #{item_set.id}:"
+        @log.puts item_set.to_s.gsub(/^/, "  ")
+        incoming_ids = item_set.in_sets.map(&:id)
+        @log.puts
+        @log.puts "  Incoming states: #{incoming_ids.join(", ")}"
+        @log.puts "  Outgoing states:"
+        item_set.following_item_set.each do |following_symbol, following_item_set|
+          @log.puts "    #{following_symbol.name} => #{following_item_set.id}"
+        end
+        @log.puts
+        @log.puts "  Reduce actions:"
+        case item_set.reduce_actions
+        when Rule
+          @log.puts "    * => rule #{item_set.reduce_actions.id}, rule set #{@rule_sets[item_set.reduce_actions.name].id} (#{item_set.reduce_actions.name})"
+        when Hash
+          item_set.reduce_actions.each do |token, rule|
+            @log.puts "    lookahead #{token.name} => #{rule.name} (#{rule.id}), rule set ##{rule.rule_set.id}"
+          end
+        end
+      end
+    end
+
   end
 
 end

lib/propane/parser/item.rb

@@ -72,15 +72,40 @@ class Propane
         end
       end
 
+      # Return whether the item is "complete", meaning that the parse position
+      # marker is at the end of the rule.
+      #
+      # @return [Boolean]
+      #   Whether the item is "complete".
+      def complete?
+        @position == @rule.components.size
+      end
+
       # Get the following symbol for the Item.
       #
       # That is, the symbol which follows the parse position marker in the
       # current Item.
       #
+      # @param offset [Integer]
+      #   Offset from current parse position to examine.
+      #
       # @return [Token, RuleSet, nil]
       #   Following symbol for the Item.
-      def following_symbol
+      def following_symbol(offset = 0)
-        @rule.components[@position]
+        @rule.components[@position + offset]
+      end
+
+      # Get the previous symbol for the Item.
+      #
+      # That is, the symbol which precedes the parse position marker in the
+      # current Item.
+      #
+      # @return [Token, RuleSet, nil]
+      #   Previous symbol for this Item.
+      def previous_symbol
+        if @position > 0
+          @rule.components[@position - 1]
+        end
       end
 
       # Get whether this Item is followed by the provided symbol.

lib/propane/parser/item_set.rb

@@ -17,10 +17,14 @@ class Propane
       #   Maps a following symbol to its ItemSet.
       attr_reader :following_item_set
 
-      # @return [Set]
+      # @return [Set<ItemSet>]
-      #   Item sets leading to this item set.
+      #   ItemSets leading to this item set.
       attr_reader :in_sets
 
+      # @return [nil, Rule, Hash]
+      #   Reduce actions, mapping lookahead tokens to rules.
+      attr_accessor :reduce_actions
+
       # Build an ItemSet.
       #
       # @param items [Array<Item>]
@@ -81,6 +85,18 @@ class Propane
         self == other
       end
 
+      # Set of ItemSets that lead to this ItemSet.
+      #
+      # This set includes this ItemSet.
+      #
+      # @return [Set<ItemSet>]
+      #   Set of all ItemSets that lead up to this ItemSet.
+      def leading_item_sets
+        @in_sets.reduce(Set[self]) do |result, item_set|
+          result + item_set.leading_item_sets
+        end
+      end
+
       # Represent the ItemSet as a String.
       #
       # @return [String]
@@ -98,17 +114,15 @@ class Propane
       #
       # @return [void]
       def close!
-        eval_items = @items
+        eval_items = @items.dup
         while eval_items.size > 0
-          this_eval_items = eval_items
+          item = eval_items.first
-          eval_items = Set.new
+          eval_items.delete(item)
-          this_eval_items.each do |item|
           item.closed_items.each do |new_item|
             unless @items.include?(new_item)
               eval_items << new_item
             end
           end
-          end
           @items += eval_items
         end
       end

lib/propane/pattern.rb

@@ -0,0 +1,65 @@
+class Propane
+
+  class Pattern
+
+    # @return [String, nil]
+    #   Code block to execute when the pattern is matched.
+    attr_reader :code
+
+    # @option options [Integer, nil] :code_id
+    #   Code block ID.
+    attr_accessor :code_id
+
+    # @return [String, nil]
+    #   Pattern.
+    attr_reader :pattern
+
+    # @return [Token, nil]
+    #   Token to be returned by this pattern.
+    attr_reader :token
+
+    # @return [Integer, nil]
+    #   Line number where the pattern was defined in the input grammar.
+    attr_reader :line_number
+
+    # @return [Regex::NFA, nil]
+    #   Regex NFA for matching the pattern.
+    attr_reader :nfa
+
+    # @return [String, nil]
+    #   Lexer mode for this pattern.
+    attr_accessor :mode
+
+    # @return [String, nil]
+    #   Parser value type name.
+    attr_accessor :ptypename
+
+    # Construct a Pattern.
+    #
+    # @param options [Hash]
+    #   Optional parameters.
+    # @option options [String, nil] :code
+    #   Code block to execute when the pattern is matched.
+    # @option options [String, nil] :pattern
+    #   Pattern.
+    # @option options [Token, nil] :token
+    #   Token to be returned by this pattern.
+    # @option options [Integer, nil] :line_number
+    #   Line number where the token was defined in the input grammar.
+    # @option options [String, nil] :mode
+    #   Lexer mode for this pattern.
+    def initialize(options)
+      @code = options[:code]
+      @pattern = options[:pattern]
+      @token = options[:token]
+      @line_number = options[:line_number]
+      @mode = options[:mode]
+      @ptypename = options[:ptypename]
+      regex = Regex.new(@pattern)
+      regex.nfa.end_state.accepts = self
+      @nfa = regex.nfa
+    end
+
+  end
+
+end

lib/propane/rule.rb

@@ -12,7 +12,11 @@ class Propane
 
     # @return [Integer]
     #   Rule ID.
-    attr_reader :id
+    attr_accessor :id
+
+    # @return [String, nil]
+    #   Parser type name.
+    attr_accessor :ptypename
 
     # @return [Integer]
     #   Line number where the rule was defined in the input grammar.
@@ -22,6 +26,10 @@ class Propane
     #   Rule name.
     attr_reader :name
 
+    # @return [RuleSet]
+    #   The RuleSet that this Rule is a part of.
+    attr_accessor :rule_set
+
     # Construct a Rule.
     #
     # @param name [String]
@@ -30,18 +38,36 @@ class Propane
     #   Rule components.
     # @param code [String]
     #   User code associated with the rule.
+    # @param ptypename [String, nil]
+    #   Parser type name.
     # @param line_number [Integer]
     #   Line number where the rule was defined in the input grammar.
-    # @param id [Integer]
+    def initialize(name, components, code, ptypename, line_number)
-    #   Rule ID.
-    def initialize(name, components, code, line_number, id)
       @name = name
       @components = components
       @code = code
-      @id = id
+      @ptypename = ptypename
       @line_number = line_number
     end
 
+    # Return whether the Rule is empty.
+    #
+    # A Rule is empty if it has no components.
+    #
+    # @return [Boolean]
+    #   Whether the Rule is empty.
+    def empty?
+      @components.empty?
+    end
+
+    # Represent the Rule as a String.
+    #
+    # @return [String]
+    #   Rule represented as a String.
+    def to_s
+      "#{@name} -> #{@components.map(&:name).join(" ")}"
+    end
+
   end
 
 end

lib/propane/rule_set.rb

@@ -2,19 +2,79 @@ class Propane
 
   class RuleSet
 
+    # @return [Integer]
+    #   ID of the RuleSet.
+    attr_reader :id
+
+    # @return [String]
+    #   Name of the RuleSet.
     attr_reader :name
 
+    # @return [String, nil]
+    #   Parser type name.
+    attr_accessor :ptypename
+
+    # @return [Array<Rule>]
+    #   Rules in the RuleSet.
     attr_reader :rules
 
-    def initialize(name)
+    # @return [Boolean]
+    #   Whether the RuleSet could expand to an empty sequence.
+    attr_writer :could_be_empty
+
+    # Construct a RuleSet.
+    #
+    # @param name [String]
+    #   Name of the RuleSet.
+    # @param id [Integer]
+    #   ID of the RuleSet.
+    def initialize(name, id)
+      @id = id
       @name = name
       @rules = []
+      @could_be_empty = false
     end
 
+    # Add a Rule to the RuleSet.
+    #
+    # @param rule [Rule]
+    #   Rule to add.
     def <<(rule)
       @rules << rule
     end
 
+    # Return whether any Rule in the RuleSet is empty.
+    #
+    # @return [Boolean]
+    #   Whether any rule in the RuleSet is empty.
+    def could_be_empty?
+      @could_be_empty
+    end
+
+    # Build the start token set for the RuleSet.
+    #
+    # @return [Set<Token>]
+    #   Start token set for the RuleSet.
+    def start_token_set
+      if @_start_token_set.nil?
+        @_start_token_set = Set.new
+        @rules.each do |rule|
+          rule.components.each do |component|
+            if component.is_a?(Token)
+              @_start_token_set << component
+              break
+            else
+              @_start_token_set += component.start_token_set
+              unless component.could_be_empty?
+                break
+              end
+            end
+          end
+        end
+      end
+      @_start_token_set
+    end
+
   end
 
 end

lib/propane/token.rb

@@ -2,60 +2,54 @@ class Propane
 
   class Token
 
+    class << self
+
+      # Name of the token to use in code (special characters replaced).
+      #
+      # @return [String]
+      #   Name of the token to use in code (special characters replaced).
+      def code_name(name)
+        name.sub(/^\$/, "__")
+      end
+
+    end
+
     # @return [String, nil]
     #   Token name.
     attr_reader :name
 
     # @return [String, nil]
-    #   Token pattern.
+    #   Parser value type name.
-    attr_reader :pattern
+    attr_accessor :ptypename
 
     # @return [Integer, nil]
     #   Token ID.
-    attr_reader :id
+    attr_accessor :id
 
     # @return [Integer, nil]
     #   Line number where the token was defined in the input grammar.
     attr_reader :line_number
 
-    # @return [Regex::NFA, nil]
-    #   Regex NFA for matching the token.
-    attr_reader :nfa
-
     # Construct a Token.
     #
-    # @param options [Hash]
+    # @param name [String, nil]
-    #   Optional parameters.
-    # @option options [String, nil] :name
     #   Token name.
-    # @option options [String, nil] :pattern
+    # @param ptypename [String, nil]
-    #   Token pattern.
+    #   Parser value type for this token.
-    # @option options [Integer, nil] :id
+    # @param line_number [Integer, nil]
-    #   Token ID.
-    # @option options [Integer, nil] :line_number
     #   Line number where the token was defined in the input grammar.
-    def initialize(options)
+    def initialize(name, ptypename, line_number)
-      @name = options[:name]
+      @name = name
-      @pattern = options[:pattern]
+      @ptypename = ptypename
-      @id = options[:id]
+      @line_number = line_number
-      @line_number = options[:line_number]
-      unless @pattern.nil?
-        regex = Regex.new(@pattern)
-        regex.nfa.end_state.accepts = self
-        @nfa = regex.nfa
-      end
     end
 
-    def c_name
+    # Name of the token to use in code (special characters replaced).
-      @name.upcase
-    end
-
-    # Whether the token is a drop token.
     #
-    # @return [Boolean]
+    # @return [String]
-    #   Whether the token is a drop token.
+    #   Name of the token to use in code (special characters replaced).
-    def drop?
+    def code_name
-      @name.nil?
+      self.class.code_name(@name)
     end
 
     def to_s

lib/propane/util.rb

@@ -0,0 +1,17 @@
+class Propane
+
+  # Utility methods.
+  module Util
+
+    class << self
+
+      def banner(message)
+        s = "*" * (message.size + 4)
+        "#{s}\n* #{message} *\n#{s}\n"
+      end
+
+    end
+
+  end
+
+end

rb/assets/user_guide.html.erb

@@ -0,0 +1,68 @@
+<html>
+  <head>
+    <title>Propane User Guide<%= subpage_title %> - Version <%= Propane::VERSION %></title>
+    <style>
+      body {
+        background-color: #CCC;
+        margin: 0px;
+        padding: 0px;
+      }
+      #body-content {
+        margin-left: auto;
+        margin-right: auto;
+        margin-top: 1em;
+        margin-bottom: 1em;
+        border: 1px solid black;
+        background-color: #FFF;
+        width: 120ex;
+        padding: 2ex;
+      }
+      a {
+        text-decoration: none;
+      }
+      .separator {
+        height: 2em;
+      }
+      .img_block_center {
+        display: block;
+        margin-left: auto;
+        margin-right: auto;
+      }
+      .page_nav {
+        width: 100%;
+        table-layout: fixed;
+      }
+      .page_nav_toc {
+        text-align: center;
+      }
+      .page_nav_next {
+        text-align: right;
+      }
+      .code {
+        padding-left: 2ex;
+        width: 116ex;
+        overflow-x: auto;
+      }
+      .ruby_code .normal {}
+      .ruby_code .comment { color: #005; font-style: italic; }
+      .ruby_code .keyword { color: #A00; font-weight: bold; }
+      .ruby_code .method { color: #077; }
+      .ruby_code .class { color: #074; }
+      .ruby_code .module { color: #050; }
+      .ruby_code .punct { color: #447; font-weight: bold; }
+      .ruby_code .symbol { color: #099; }
+      .ruby_code .string { color: #090; }
+      .ruby_code .char { color: #F07; }
+      .ruby_code .ident { color: #004; }
+      .ruby_code .constant { color: #07F; }
+      .ruby_code .regex { color: #B66; }
+      .ruby_code .number { color: #D55; }
+      .ruby_code .attribute { color: #377; }
+      .ruby_code .global { color: #3B7; }
+      .ruby_code .expr { color: #227; }
+    </style>
+  </head>
+  <body>
+    <div id="body-content"><%= content %></div>
+  </body>
+</html>

rb/build_dist.rb

@@ -0,0 +1,106 @@
+#!/usr/bin/env ruby
+
+require "fileutils"
+require "digest/md5"
+
+if File.read("lib/propane/version.rb") =~ /VERSION = "(.+)"/
+  VERSION = $1
+else
+  raise "Could not determine version."
+end
+PROG_NAME = "propane"
+START_FILE = "bin/#{PROG_NAME}"
+LIB_DIR = "lib"
+DIST = "dist"
+
+files_processed = {}
+combined_file = []
+
+combine_files = lambda do |file|
+  File.read(file, mode: "rb").each_line do |line|
+    if line =~ /^\s*require(?:_relative)?\s*"(.*)"$/
+      require_name = $1
+      if require_name =~ %r{^#{PROG_NAME}(?:/.*)?$}
+        path = "#{LIB_DIR}/#{require_name}.rb"
+        if File.exist?(path)
+          unless files_processed[path]
+            files_processed[path] = true
+            combine_files[path]
+          end
+        else
+          raise "require path #{path.inspect} not found"
+        end
+      else
+        combined_file << line
+      end
+    else
+      combined_file << line
+    end
+  end
+end
+
+combine_files[START_FILE]
+
+# Strip Ruby comment lines and empty lines to save some space, but do not
+# remove lines that are in heredoc sections. This isn't terribly robust to be
+# used in the wild, but works for the heredoc instances for this project.
+stripped = []
+heredoc_end = nil
+combined_file.each do |line|
+  if line =~ /<<-?([A-Z]+)/
+    heredoc_end = $1
+  end
+  if heredoc_end and line =~ /^\s*#{heredoc_end}/
+    heredoc_end = nil
+  end
+  if line !~ /#\sspecs/
+    if heredoc_end or line !~ /^\s*(#[^!].*)?$/
+      stripped << line
+    end
+  end
+end
+
+license = File.read("LICENSE.txt").gsub(/^(.*?)$/) do |line|
+  if line.size > 0
+    "# #{line}"
+  else
+    "#"
+  end
+end
+
+require "zlib"
+require "base64"
+compressed_script = Zlib::Deflate.deflate(stripped.join)
+encoded_compressed_script = Base64.encode64(compressed_script).gsub("\n", "")
+hash = Digest::MD5.hexdigest(encoded_compressed_script)
+
+FileUtils.rm_rf(DIST)
+FileUtils.mkdir_p(DIST)
+File.open("#{DIST}/#{PROG_NAME}", "wb", 0755) do |fh|
+  fh.write(<<EOF)
+#!/usr/bin/env ruby
+
+#{license}
+
+script = File.join(File.dirname(__FILE__), ".propane-#{VERSION}-#{hash}.rb")
+unless File.exist?(script)
+  if File.read(__FILE__, mode: "rb") =~ /^#==>(.*)/
+    require "zlib"
+    require "base64"
+    encoded_compressed = $1
+    unescaped_compressed = Base64.decode64(encoded_compressed)
+    inflated = Zlib::Inflate.inflate(unescaped_compressed)
+    File.open(script, "wb") do |fh|
+      fh.write(inflated)
+    end
+  else
+    raise "Could not decompress."
+  end
+end
+load script
+if __FILE__ == $0
+  Propane::CLI.new.run(ARGV)
+end
+#==>#{encoded_compressed_script}
+EOF
+end

rb/gen_user_guide.rb

@@ -0,0 +1,227 @@
+#!/usr/bin/env ruby
+
+require "erb"
+require "fileutils"
+require "redcarpet"
+require "syntax"
+require "syntax/convertors/html"
+require "propane/version"
+
+def load_file(file_name)
+  contents = File.read(file_name)
+  contents.gsub(/\$\{include (.*?)\}/) do |match|
+    include_file_name = $1
+    File.read(include_file_name)
+  end
+end
+
+class Generator
+  class Section
+    attr_reader :number
+    attr_reader :title
+    attr_reader :page
+    attr_reader :contents
+    attr_reader :anchor
+    def initialize(number, title, page, anchor)
+      @number = number
+      @title = title.gsub(/[`]/, "")
+      @page = page
+      @anchor = anchor
+      @contents = ""
+    end
+    def append(contents)
+      @contents += contents
+    end
+  end
+
+  class Page
+    attr_reader :name
+    attr_reader :title
+    attr_accessor :contents
+    def initialize(name, title, contents = "")
+      @name = name
+      @title = title
+      @contents = contents
+    end
+  end
+
+  def initialize(input, output_file, multi_page)
+    current_page =
+      if multi_page
+        nil
+      else
+        File.basename(output_file).sub(/\.html$/, "")
+      end
+    @sections = []
+    @current_section_number = [0]
+    @lines = input.lines
+    while @lines.size > 0
+      line = @lines.slice!(0)
+      if line =~ /^```(.*)$/
+        @sections.last.append(gather_code_section($1))
+      elsif line.chomp == "${remove}"
+        remove_section
+      elsif line =~ /^(#+)(>)?\s*(.*)$/
+        level_text, new_page_text, title_text = $1, $2, $3
+        level = $1.size
+        new_page = !new_page_text.nil?
+        section_number = get_next_section_number(level)
+        anchor = make_anchor(section_number, title_text)
+        if multi_page and (new_page or current_page.nil?)
+          current_page = anchor
+        end
+        @sections << Section.new(section_number, title_text, current_page, anchor)
+        @sections.last.append("#{level_text} #{section_number} #{title_text}")
+      elsif @sections.size > 0
+        @sections.last.append(line)
+      end
+    end
+
+    renderer = Redcarpet::Render::HTML.new
+    @markdown_renderer = Redcarpet::Markdown.new(renderer)
+    changelog = @markdown_renderer.render(File.read("CHANGELOG.md"))
+
+    @pages = [Page.new("toc", "Table of Contents", render_toc)]
+    @sections.each do |section|
+      unless @pages.last.name == section.page
+        @pages << Page.new(section.page, "#{section.number} #{section.title}")
+      end
+      @pages.last.contents += render_section(section)
+    end
+
+    @pages.each do |page|
+      page.contents.gsub!("${changelog}", changelog)
+      page.contents.gsub!(%r{\$\{#(.+?)\}}) do |match|
+        section_name = $1
+        href = get_link_to_section(section_name)
+        %[<a href="#{href}">#{section_name}</a>]
+      end
+    end
+
+    template = File.read("rb/assets/user_guide.html.erb")
+    erb = ERB.new(template, trim_mode: "<>")
+
+    if multi_page
+      @pages.each_with_index do |page, page_index|
+        subpage_title = " - #{page.title}"
+        page_nav_bar = render_page_nav_bar(page_index)
+        content = page_nav_bar + separator + page.contents + separator + page_nav_bar
+        html_result = erb.result(binding.clone)
+        File.open(File.join(output_file, "#{page.name}.html"), "w") do |fh|
+          fh.write(html_result)
+        end
+      end
+    else
+      subpage_title = ""
+      content = @pages.reduce("") do |result, page|
+        result + page.contents
+      end
+      html_result = erb.result(binding.clone)
+      File.open(output_file, "w") do |fh|
+        fh.write(html_result)
+      end
+    end
+  end
+
+  def separator
+    %[<div class="separator"></div>]
+  end
+
+  def render_page_nav_bar(page_index)
+    page_nav_prev =
+      if page_index > 1
+        %[<a href="#{@pages[page_index - 1].name}.html">&laquo; Prev<br/>#{@pages[page_index - 1].title}</a>]
+      else
+        ""
+      end
+    page_nav_toc =
+      if page_index > 0
+        %[<a href="toc.html">Table of Contents</a>]
+      else
+        ""
+      end
+    page_nav_next =
+      if page_index < @pages.size - 1
+        %[<a href="#{@pages[page_index + 1].name}.html">Next &raquo<br/>#{@pages[page_index + 1].title}</a>]
+      else
+        ""
+      end
+    %[<table class="page_nav"><tr>] + \
+    %[<td class="page_nav_prev">#{page_nav_prev}</td>] + \
+    %[<td class="page_nav_toc">#{page_nav_toc}</td>] + \
+    %[<td class="page_nav_next">#{page_nav_next}</td>] + \
+    %[</tr></table>]
+  end
+
+  def render_toc
+    toc_content = %[<h1>Table of Contents</h1>\n]
+    @sections.each do |section|
+      indent = section.number.split(".").size - 1
+      toc_content += %[<span style="padding-left: #{4 * indent}ex;">]
+      toc_content += %[<a href="#{section.page}.html##{section.anchor}">#{section.number} #{section.title}</a><br/>\n]
+      toc_content += %[</span>]
+    end
+    toc_content
+  end
+
+  def render_section(section)
+    %[<a name="#{section.anchor}" />] + \
+      @markdown_renderer.render(section.contents)
+  end
+
+  def make_anchor(section_number, section_title)
+    "s" + ("#{section_number} #{section_title}").gsub(/[^a-zA-Z0-9]/, "_")
+  end
+
+  def get_link_to_section(section_name)
+    section = @sections.find do |section|
+      section.title == section_name
+    end
+    raise "Could not find section #{section_name}" unless section
+    "#{section.page}.html##{section.anchor}"
+  end
+
+  def gather_code_section(syntax)
+    code = ""
+    loop do
+      line = @lines.slice!(0)
+      if line =~ /^```/
+        break
+      end
+      code += line
+    end
+    if syntax != ""
+      convertor = Syntax::Convertors::HTML.for_syntax(syntax)
+      %[<div class="code #{syntax}_code">\n#{convertor.convert(code)}\n</div>\n]
+    else
+      %[<div class="code">\n<pre>#{code}</pre>\n</div>\n]
+    end
+  end
+
+  def remove_section
+    loop do
+      line = @lines.slice!(0)
+      if line.chomp == "${/remove}"
+        break
+      end
+    end
+  end
+
+  def get_next_section_number(level)
+    if @current_section_number.size == level - 1
+      @current_section_number << 1
+    elsif @current_section_number.size >= level
+      @current_section_number[level - 1] += 1
+      @current_section_number.slice!(level, @current_section_number.size)
+    else
+      raise "Section level change from #{@current_section_number.size} to #{level}"
+    end
+    @current_section_number.join(".")
+  end
+end
+
+input = load_file("doc/user_guide.md")
+FileUtils.rm_rf("gen/user_guide")
+FileUtils.mkdir_p("gen/user_guide")
+Generator.new(input, "gen/user_guide/user_guide.html", false)
+Generator.new(input, "gen/user_guide", true)

spec/json_parser.propane

@@ -0,0 +1,179 @@
+<<
+import std.math;
+import json_types;
+private string string_value;
+>>
+
+ptype JSONValue;
+ptype array = JSONValue[];
+ptype dict = JSONValue[string];
+ptype string = string;
+
+drop /\s+/;
+token lbrace /\{/;
+token rbrace /\}/;
+token lbracket /\[/;
+token rbracket /\]/;
+token comma /,/;
+token colon /:/;
+token number /-?(0|[1-9][0-9]*)(\.[0-9]+)?([eE][-+]?[0-9]+)?/ <<
+    double n;
+    bool negative;
+    size_t i = 0u;
+    if (match[i] == '-')
+    {
+        negative = true;
+        i++;
+    }
+    while ('0' <= match[i] && match[i] <= '9')
+    {
+        n *= 10.0;
+        n += (match[i] - '0');
+        i++;
+    }
+    if (match[i] == '.')
+    {
+        i++;
+        double mult = 0.1;
+        while ('0' <= match[i] && match[i] <= '9')
+        {
+            n += mult * (match[i] - '0');
+            mult /= 10.0;
+            i++;
+        }
+    }
+    if (match[i] == 'e' || match[i] == 'E')
+    {
+        bool exp_negative;
+        i++;
+        if (match[i] == '-')
+        {
+            exp_negative = true;
+            i++;
+        }
+        else if (match[i] == '+')
+        {
+            i++;
+        }
+        long exp;
+        while ('0' <= match[i] && match[i] <= '9')
+        {
+            exp *= 10;
+            exp += (match[i] - '0');
+            i++;
+        }
+        if (exp_negative)
+        {
+            exp = -exp;
+        }
+        n = pow(n, exp);
+    }
+    if (negative)
+    {
+        n = -n;
+    }
+    $$ = new JSONNumber(n);
+>>
+token true <<
+  $$ = new JSONTrue();
+>>
+token false <<
+  $$ = new JSONFalse();
+>>
+token null <<
+  $$ = new JSONNull();
+>>
+/"/ <<
+  $mode(string);
+  string_value = "";
+>>
+string: token string (string) /"/ <<
+  $$ = string_value;
+  $mode(default);
+>>
+string: /\\"/ <<
+  string_value ~= "\"";
+>>
+string: /\\\\/ <<
+  string_value ~= "\\";
+>>
+string: /\\\// <<
+  string_value ~= "/";
+>>
+string: /\\b/ <<
+  string_value ~= "\b";
+>>
+string: /\\f/ <<
+  string_value ~= "\f";
+>>
+string: /\\n/ <<
+  string_value ~= "\n";
+>>
+string: /\\r/ <<
+  string_value ~= "\r";
+>>
+string: /\\t/ <<
+  string_value ~= "\t";
+>>
+string: /\\u[0-9a-fA-F]{4}/ <<
+  /* Not actually going to encode the code point for this example... */
+  string_value ~= "{" ~ match[2..6] ~ "}";
+>>
+string: /[^\\]/ <<
+  string_value ~= match;
+>>
+Start -> Value <<
+  $$ = $1;
+>>
+Value -> string <<
+  $$ = new JSONString($1);
+>>
+Value -> number <<
+  $$ = $1;
+>>
+Value -> Object <<
+  $$ = $1;
+>>
+Value -> Array <<
+  $$ = $1;
+>>
+Value -> true <<
+  $$ = $1;
+>>
+Value -> false <<
+  $$ = $1;
+>>
+Value -> null <<
+  $$ = $1;
+>>
+Object -> lbrace rbrace <<
+  $$ = new JSONObject();
+>>
+Object -> lbrace KeyValues rbrace <<
+  $$ = new JSONObject($2);
+>>
+KeyValues (dict) -> KeyValue <<
+  $$ = $1;
+>>
+KeyValues -> KeyValues comma KeyValue <<
+  foreach (key, value; $3)
+  {
+    $1[key] = value;
+  }
+  $$ = $1;
+>>
+KeyValue (dict) -> string colon Value <<
+  $$ = [$1: $3];
+>>
+Array -> lbracket rbracket <<
+  $$ = new JSONArray();
+>>
+Array -> lbracket Values rbracket <<
+  $$ = new JSONArray($2);
+>>
+Values (array) -> Value <<
+  $$ = [$1];
+>>
+Values -> Values comma Value <<
+  $$ = $1 ~ [$3];
+>>

spec/json_types.d

@@ -0,0 +1,63 @@
+class JSONValue
+{
+}
+
+class JSONObject : JSONValue
+{
+    JSONValue[string] value;
+
+    this()
+    {
+    }
+
+    this(JSONValue[string] value)
+    {
+        this.value = value;
+    }
+}
+
+class JSONArray : JSONValue
+{
+    JSONValue[] value;
+
+    this()
+    {
+    }
+
+    this(JSONValue[] value)
+    {
+        this.value = value;
+    }
+}
+
+class JSONNumber : JSONValue
+{
+    double value;
+
+    this(double value)
+    {
+        this.value = value;
+    }
+}
+
+class JSONString : JSONValue
+{
+    string value;
+
+    this(string value)
+    {
+        this.value = value;
+    }
+}
+
+class JSONTrue : JSONValue
+{
+}
+
+class JSONFalse : JSONValue
+{
+}
+
+class JSONNull : JSONValue
+{
+}

spec/propane/grammar_spec.rb

@@ -0,0 +1,233 @@
+class Propane
+  describe Grammar do
+    it "parses a user grammar" do
+      input = <<EOF
+# Comment line
+
+module a.b;
+class Foobar;
+ptype   XYZ  *  ;
+
+token while;
+
+token id
+  /[a-zA-Z_][a-zA-Z_0-9]*/;
+
+token token_with_code <<
+Code for the token
+>>
+
+tokenid token_with_no_pattern;
+
+drop /\\s+/;
+
+A -> B <<
+  a = 42;
+>>
+B -> C while id;
+B -> <<
+  b = 0;
+>>
+EOF
+      grammar = Grammar.new(input)
+      expect(grammar.classname).to eq "Foobar"
+      expect(grammar.modulename).to eq "a.b"
+      expect(grammar.ptype).to eq "XYZ  *"
+      expect(grammar.ptypes).to eq("default" => "XYZ  *")
+      expect(grammar.prefix).to eq "p_"
+
+      o = grammar.tokens.find {|token| token.name == "while"}
+      expect(o).to_not be_nil
+      expect(o.line_number).to eq 7
+
+      o = grammar.patterns.find {|pattern| pattern.token == o}
+      expect(o).to_not be_nil
+      expect(o.pattern).to eq "while"
+      expect(o.line_number).to eq 7
+      expect(o.code).to be_nil
+
+      o = grammar.tokens.find {|token| token.name == "id"}
+      expect(o).to_not be_nil
+      expect(o.line_number).to eq 10
+
+      o = grammar.patterns.find {|pattern| pattern.token == o}
+      expect(o).to_not be_nil
+      expect(o.pattern).to eq "[a-zA-Z_][a-zA-Z_0-9]*"
+      expect(o.line_number).to eq 10
+      expect(o.code).to be_nil
+
+      o = grammar.tokens.find {|token| token.name == "token_with_code"}
+      expect(o).to_not be_nil
+      expect(o.line_number).to eq 12
+
+      o = grammar.patterns.find {|pattern| pattern.token == o}
+      expect(o).to_not be_nil
+      expect(o.pattern).to eq "token_with_code"
+      expect(o.line_number).to eq 12
+      expect(o.code).to eq "Code for the token\n"
+
+      o = grammar.tokens.find {|token| token.name == "token_with_no_pattern"}
+      expect(o).to_not be_nil
+      expect(o.line_number).to eq 16
+
+      o = grammar.patterns.find {|pattern| pattern.token == o}
+      expect(o).to be_nil
+
+      o = grammar.patterns.find {|pattern| pattern.pattern == "\\s+"}
+      expect(o).to_not be_nil
+      expect(o.line_number).to eq 18
+      expect(o.token).to be_nil
+      expect(o.code).to be_nil
+
+      expect(grammar.rules.size).to eq 3
+
+      o = grammar.rules[0]
+      expect(o.name).to eq "A"
+      expect(o.components).to eq %w[B]
+      expect(o.line_number).to eq 20
+      expect(o.code).to eq "  a = 42;\n"
+
+      o = grammar.rules[1]
+      expect(o.name).to eq "B"
+      expect(o.components).to eq %w[C while id]
+      expect(o.line_number).to eq 23
+      expect(o.code).to be_nil
+
+      o = grammar.rules[2]
+      expect(o.name).to eq "B"
+      expect(o.components).to eq []
+      expect(o.line_number).to eq 24
+      expect(o.code).to eq "  b = 0;\n"
+    end
+
+    it "parses code segments with semicolons" do
+      input = <<EOF
+token code1 <<
+  a = b;
+  return c;
+>>
+
+token code2 <<
+  writeln("Hello there");
+>>
+
+tokenid token_with_no_pattern;
+
+prefix myparser_;
+EOF
+      grammar = Grammar.new(input)
+      expect(grammar.prefix).to eq "myparser_"
+
+      o = grammar.tokens.find {|token| token.name == "code1"}
+      expect(o).to_not be_nil
+      expect(o.line_number).to eq 1
+
+      o = grammar.patterns.find {|pattern| pattern.token == o}
+      expect(o).to_not be_nil
+      expect(o.code).to eq "  a = b;\n  return c;\n"
+
+      o = grammar.tokens.find {|token| token.name == "code2"}
+      expect(o).to_not be_nil
+      expect(o.line_number).to eq 6
+
+      o = grammar.patterns.find {|pattern| pattern.token == o}
+      expect(o).to_not be_nil
+      expect(o.code).to eq %[  writeln("Hello there");\n]
+    end
+
+    it "supports mode labels" do
+      input = <<EOF
+token a;
+m1: token b;
+/foo/ <<
+>>
+m2: /bar/ <<
+>>
+drop /q/;
+m3: drop /r/;
+EOF
+      grammar = Grammar.new(input)
+
+      o = grammar.tokens.find {|token| token.name == "a"}
+      expect(o).to_not be_nil
+
+      o = grammar.patterns.find {|pattern| pattern.token == o}
+      expect(o).to_not be_nil
+      expect(o.mode).to be_nil
+
+      o = grammar.tokens.find {|token| token.name == "b"}
+      expect(o).to_not be_nil
+
+      o = grammar.patterns.find {|pattern| pattern.token == o}
+      expect(o).to_not be_nil
+      expect(o.mode).to eq "m1"
+
+      o = grammar.patterns.find {|pattern| pattern.pattern == "foo"}
+      expect(o).to_not be_nil
+      expect(o.mode).to be_nil
+
+      o = grammar.patterns.find {|pattern| pattern.pattern == "bar"}
+      expect(o).to_not be_nil
+      expect(o.mode).to eq "m2"
+
+      o = grammar.patterns.find {|pattern| pattern.pattern == "q"}
+      expect(o).to_not be_nil
+      expect(o.mode).to be_nil
+
+      o = grammar.patterns.find {|pattern| pattern.pattern == "r"}
+      expect(o).to_not be_nil
+      expect(o.mode).to eq "m3"
+    end
+
+    it "allows assigning ptypes to tokens and rules" do
+      input = <<EOF
+ptype Subnode *;
+ptype string = char *;
+ptype integer = int;
+ptype node = Node *;
+
+token abc(string);
+token bar;
+tokenid int(integer);
+
+/xyz/ (string) <<
+>>
+
+/z28/ <<
+>>
+
+Start (node) -> R;
+R -> abc int;
+EOF
+      grammar = Grammar.new(input)
+
+      o = grammar.tokens.find {|token| token.name == "abc"}
+      expect(o).to_not be_nil
+      expect(o.ptypename).to eq "string"
+
+      o = grammar.tokens.find {|token| token.name == "bar"}
+      expect(o).to_not be_nil
+      expect(o.ptypename).to be_nil
+
+      o = grammar.tokens.find {|token| token.name == "int"}
+      expect(o).to_not be_nil
+      expect(o.ptypename).to eq "integer"
+
+      o = grammar.rules.find {|rule| rule.name == "Start"}
+      expect(o).to_not be_nil
+      expect(o.ptypename).to eq "node"
+
+      o = grammar.rules.find {|rule| rule.name == "R"}
+      expect(o).to_not be_nil
+      expect(o.ptypename).to be_nil
+
+      o = grammar.patterns.find {|pattern| pattern.pattern == "xyz"}
+      expect(o).to_not be_nil
+      expect(o.ptypename).to eq "string"
+
+      o = grammar.patterns.find {|pattern| pattern.pattern == "z28"}
+      expect(o).to_not be_nil
+      expect(o.ptypename).to be_nil
+    end
+  end
+end

spec/propane/lexer/dfa_spec.rb

@@ -35,7 +35,8 @@ class TestLexer
       end
     end
     if last_accepts
-      [last_accepts.name, last_s]
+      name = last_accepts.token ? last_accepts.token.name : nil
+      [name, last_s]
     end
   end
 
@@ -51,7 +52,7 @@ end
 
 def run(grammar, input)
   grammar = Propane::Grammar.new(grammar)
-  token_dfa = Propane::Lexer::DFA.new(grammar.tokens, grammar.drop_tokens)
+  token_dfa = Propane::Lexer::DFA.new(grammar.patterns)
   test_lexer = TestLexer.new(token_dfa)
   test_lexer.lex(input)
 end
@@ -59,7 +60,7 @@ end
 describe Propane::Lexer::DFA do
   it "lexes a simple token" do
     expect(run(<<EOF, "foo")).to eq [["foo", "foo"]]
-token foo
+token foo;
 EOF
   end
 
@@ -69,8 +70,8 @@ EOF
       ["bar", "bar"],
     ]
     expect(run(<<EOF, "foobar")).to eq expected
-token foo
+token foo;
-token bar
+token bar;
 EOF
   end
 
@@ -79,17 +80,17 @@ EOF
       ["identifier", "foobar"],
     ]
     expect(run(<<EOF, "foobar")).to eq expected
-token foo
+token foo;
-token bar
+token bar;
-token identifier [a-z]+
+token identifier /[a-z]+/;
 EOF
     expected = [
       ["plusplus", "++"],
       ["plus", "+"],
     ]
     expect(run(<<EOF, "+++")).to eq expected
-token plus \\+
+token plus /\\+/;
-token plusplus \\+\\+
+token plusplus /\\+\\+/;
 EOF
   end
 
@@ -100,9 +101,9 @@ EOF
       ["bar", "bar"],
     ]
     expect(run(<<EOF, "foo \tbar")).to eq expected
-token foo
+token foo;
-token bar
+token bar;
-token WS \\s+
+token WS /\\s+/;
 EOF
   end
 
@@ -113,9 +114,18 @@ EOF
       ["bar", "bar"],
     ]
     expect(run(<<EOF, "foo \tbar")).to eq expected
-token foo
+token foo;
-token bar
+token bar;
-drop \\s+
+drop /\\s+/;
+EOF
+  end
+
+  it "matches a semicolon" do
+    expected = [
+      ["semicolon", ";"],
+    ]
+    expect(run(<<EOF, ";")).to eq expected
+token semicolon /;/;
 EOF
   end
 end

spec/propane_spec.rb

@@ -1,23 +1,72 @@
 require "fileutils"
+require "open3"
+
+Results = Struct.new(:stdout, :stderr, :status)
 
 describe Propane do
-  def write_grammar(grammar)
+  def write_grammar(grammar, options = {})
-    File.write("spec/run/testparser.propane", grammar)
+    options[:name] ||= ""
+    File.write("spec/run/testparser#{options[:name]}.propane", grammar)
   end
 
-  def build_parser
+  def build_parser(options = {})
-    result = system(*%w[./propane.sh spec/run/testparser.propane spec/run/testparser.d])
+    options[:name] ||= ""
+    command = %W[./propane.sh spec/run/testparser#{options[:name]}.propane spec/run/testparser#{options[:name]}.#{options[:language]} --log spec/run/testparser#{options[:name]}.log]
+    if (options[:capture])
+      stdout, stderr, status = Open3.capture3(*command)
+      Results.new(stdout, stderr, status)
+    else
+      result = system(*command)
       expect(result).to be_truthy
     end
+  end
 
-  def compile(test_file)
+  def compile(test_files, options = {})
-    result = system(*%w[gdc -funittest -o spec/run/testparser spec/run/testparser.d], test_file)
+    test_files = Array(test_files)
+    options[:parsers] ||= [""]
+    parsers = options[:parsers].map do |name|
+      "spec/run/testparser#{name}.#{options[:language]}"
+    end
+    case options[:language]
+    when "c"
+      result = system(*%w[gcc -Wall -o spec/run/testparser -Ispec -Ispec/run], *parsers, *test_files, "spec/testutils.c", "-lm")
+    when "d"
+      result = system(*%w[ldc2 --unittest -of spec/run/testparser -Ispec], *parsers, *test_files, "spec/testutils.d")
+    end
     expect(result).to be_truthy
   end
 
   def run
-    result = system("spec/run/testparser")
+    stdout, stderr, status = Open3.capture3("spec/run/testparser")
-    expect(result).to be_truthy
+    File.binwrite("spec/run/.stderr", stderr)
+    File.binwrite("spec/run/.stdout", stdout)
+    stderr.sub!(/^.*modules passed unittests\n/, "")
+    Results.new(stdout, stderr, status)
+  end
+
+  def lines(str)
+    str.lines.map(&:chomp)
+  end
+
+  def verify_lines(lines, patterns)
+    if lines.is_a?(String)
+      lines = lines.lines.map(&:chomp)
+    end
+    patterns.each_with_index do |pattern, i|
+      found_index =
+        if pattern.is_a?(Regexp)
+          lines.find_index {|line| line =~ pattern}
+        else
+          lines.find_index do |line|
+            line.chomp == pattern.chomp
+          end
+        end
+      unless found_index
+        $stderr.puts "Lines:"
+        $stderr.puts lines
+        raise "A line matching #{pattern.inspect} (index #{i}) was not found."
+      end
+    end
   end
 
   before(:each) do
@@ -25,29 +74,77 @@ describe Propane do
     FileUtils.mkdir_p("spec/run")
   end
 
-  it "generates a D lexer" do
+  %w[d c].each do |language|
+
+    context "#{language.upcase} language" do
+
+      it "generates a lexer" do
         write_grammar <<EOF
-token int \\d+
+token int /\\d+/;
-token plus \\+
+token plus /\\+/;
-token times \\*
+token times /\\*/;
-drop \\s+
+drop /\\s+/;
 Start -> Foo;
 Foo -> int <<
 >>
 Foo -> plus <<
 >>
 EOF
-    build_parser
+        build_parser(language: language)
-    compile("spec/test_d_lexer.d")
+        compile("spec/test_lexer.#{language}", language: language)
-    run
+        results = run
+        expect(results.stderr).to eq ""
+        expect(results.status).to eq 0
+      end
+
+      it "detects a lexer error when an unknown character is seen" do
+        case language
+        when "c"
+          write_grammar <<EOF
+ptype int;
+token int /\\d+/ <<
+  int v = 0;
+  for (size_t i = 0u; i < match_length; i++)
+  {
+    v *= 10;
+    v += (match[i] - '0');
+  }
+  $$ = v;
+>>
+Start -> int <<
+  $$ = $1;
+>>
+EOF
+        when "d"
+          write_grammar <<EOF
+ptype int;
+token int /\\d+/ <<
+  int v;
+  foreach (c; match)
+  {
+    v *= 10;
+    v += (c - '0');
+  }
+  $$ = v;
+>>
+Start -> int <<
+  $$ = $1;
+>>
+EOF
+        end
+        build_parser(language: language)
+        compile("spec/test_lexer_unknown_character.#{language}", language: language)
+        results = run
+        expect(results.stderr).to eq ""
+        expect(results.status).to eq 0
       end
 
       it "generates a parser" do
         write_grammar <<EOF
-token plus \\+
+token plus /\\+/;
-token times \\*
+token times /\\*/;
-token zero 0
+token zero /0/;
-token one 1
+token one /1/;
 Start -> E;
 E -> E times B;
 E -> E plus B;
@@ -55,29 +152,545 @@ E -> B;
 B -> zero;
 B -> one;
 EOF
-    build_parser
+        build_parser(language: language)
+      end
+
+      it "generates a parser that does basic math - user guide example" do
+        case language
+        when "c"
+          write_grammar <<EOF
+<<
+#include <math.h>
+>>
+
+ptype size_t;
+
+token plus /\\+/;
+token times /\\*/;
+token power /\\*\\*/;
+token integer /\\d+/ <<
+  size_t v = 0u;
+  for (size_t i = 0u; i < match_length; i++)
+  {
+    v *= 10;
+    v += (match[i] - '0');
+  }
+  $$ = v;
+>>
+token lparen /\\(/;
+token rparen /\\)/;
+drop /\\s+/;
+
+Start -> E1 <<
+  $$ = $1;
+>>
+E1 -> E2 <<
+  $$ = $1;
+>>
+E1 -> E1 plus E2 <<
+  $$ = $1 + $3;
+>>
+E2 -> E3 <<
+  $$ = $1;
+>>
+E2 -> E2 times E3 <<
+  $$ = $1 * $3;
+>>
+E3 -> E4 <<
+  $$ = $1;
+>>
+E3 -> E3 power E4 <<
+  $$ = (size_t)pow($1, $3);
+>>
+E4 -> integer <<
+  $$ = $1;
+>>
+E4 -> lparen E1 rparen <<
+  $$ = $2;
+>>
+EOF
+        when "d"
+          write_grammar <<EOF
+<<
+import std.math;
+>>
+
+ptype ulong;
+
+token plus /\\+/;
+token times /\\*/;
+token power /\\*\\*/;
+token integer /\\d+/ <<
+  ulong v;
+  foreach (c; match)
+  {
+    v *= 10;
+    v += (c - '0');
+  }
+  $$ = v;
+>>
+token lparen /\\(/;
+token rparen /\\)/;
+drop /\\s+/;
+
+Start -> E1 <<
+  $$ = $1;
+>>
+E1 -> E2 <<
+  $$ = $1;
+>>
+E1 -> E1 plus E2 <<
+  $$ = $1 + $3;
+>>
+E2 -> E3 <<
+  $$ = $1;
+>>
+E2 -> E2 times E3 <<
+  $$ = $1 * $3;
+>>
+E3 -> E4 <<
+  $$ = $1;
+>>
+E3 -> E3 power E4 <<
+  $$ = pow($1, $3);
+>>
+E4 -> integer <<
+  $$ = $1;
+>>
+E4 -> lparen E1 rparen <<
+  $$ = $2;
+>>
+EOF
+        end
+        build_parser(language: language)
+        compile("spec/test_basic_math_grammar.#{language}", language: language)
+        results = run
+        expect(results.stderr).to eq ""
+        expect(results.status).to eq 0
+      end
+
+      it "generates an SLR parser" do
+        write_grammar <<EOF
+token one /1/;
+Start -> E;
+E -> one E;
+E -> one;
+EOF
+        build_parser(language: language)
       end
 
       it "distinguishes between multiple identical rules with lookahead symbol" do
         write_grammar <<EOF
-token a
+token a;
-token b
+token b;
 Start -> R1 a;
 Start -> R2 b;
 R1 -> a b;
 R2 -> a b;
 EOF
-    build_parser
+        build_parser(language: language)
+        compile("spec/test_parser_identical_rules_lookahead.#{language}", language: language)
+        results = run
+        expect(results.status).to eq 0
       end
 
       it "handles reducing a rule that could be arrived at from multiple states" do
         write_grammar <<EOF
-token a
+token a;
-token b
+token b;
+drop /\\s+/;
 Start -> a R1;
 Start -> b R1;
 R1 -> b;
 EOF
-    build_parser
+        build_parser(language: language)
+        compile("spec/test_parser_rule_from_multiple_states.#{language}", language: language)
+        results = run
+        expect(results.status).to eq 0
+      end
+
+      it "executes user code when matching lexer token" do
+        case language
+        when "c"
+          write_grammar <<EOF
+<<
+#include <stdio.h>
+>>
+token abc <<
+  printf("abc!\\n");
+>>
+token def;
+Start -> Abcs def;
+Abcs -> ;
+Abcs -> abc Abcs;
+EOF
+        when "d"
+          write_grammar <<EOF
+<<
+import std.stdio;
+>>
+token abc <<
+  writeln("abc!");
+>>
+token def;
+Start -> Abcs def;
+Abcs -> ;
+Abcs -> abc Abcs;
+EOF
+        end
+        build_parser(language: language)
+        compile("spec/test_user_code.#{language}", language: language)
+        results = run
+        expect(results.status).to eq 0
+        verify_lines(results.stdout, [
+          "abc!",
+          "pass1",
+          "abc!",
+          "abc!",
+          "pass2",
+        ])
+      end
+
+      it "supports a pattern statement" do
+        case language
+        when "c"
+          write_grammar <<EOF
+<<
+#include <stdio.h>
+>>
+token abc;
+/def/ <<
+  printf("def!\\n");
+>>
+Start -> abc;
+EOF
+        when "d"
+          write_grammar <<EOF
+<<
+import std.stdio;
+>>
+token abc;
+/def/ <<
+  writeln("def!");
+>>
+Start -> abc;
+EOF
+        end
+        build_parser(language: language)
+        compile("spec/test_pattern.#{language}", language: language)
+        results = run
+        expect(results.status).to eq 0
+        verify_lines(results.stdout, [
+          "def!",
+          "pass1",
+          "def!",
+          "def!",
+          "pass2",
+        ])
+      end
+
+      it "supports returning tokens from pattern code blocks" do
+        case language
+        when "c"
+          write_grammar <<EOF
+<<
+#include <stdio.h>
+>>
+token abc;
+/def/ <<
+  printf("def!\\n");
+>>
+/ghi/ <<
+  printf("ghi!\\n");
+  return $token(abc);
+>>
+Start -> abc;
+EOF
+        when "d"
+          write_grammar <<EOF
+<<
+import std.stdio;
+>>
+token abc;
+/def/ <<
+  writeln("def!");
+>>
+/ghi/ <<
+  writeln("ghi!");
+  return $token(abc);
+>>
+Start -> abc;
+EOF
+        end
+        build_parser(language: language)
+        compile("spec/test_return_token_from_pattern.#{language}", language: language)
+        results = run
+        expect(results.status).to eq 0
+        verify_lines(results.stdout, [
+          "def!",
+          "ghi!",
+          "def!",
+        ])
+      end
+
+      it "supports lexer modes" do
+        case language
+        when "c"
+          write_grammar <<EOF
+<<
+#include <stdio.h>
+>>
+token abc;
+token def;
+tokenid string;
+drop /\\s+/;
+/"/ <<
+  printf("begin string mode\\n");
+  $mode(string);
+>>
+string: /[^"]+/ <<
+  printf("captured string\\n");
+>>
+string: /"/ <<
+  $mode(default);
+  return $token(string);
+>>
+Start -> abc string def;
+EOF
+        when "d"
+          write_grammar <<EOF
+<<
+import std.stdio;
+>>
+token abc;
+token def;
+tokenid string;
+drop /\\s+/;
+/"/ <<
+  writeln("begin string mode");
+  $mode(string);
+>>
+string: /[^"]+/ <<
+  writeln("captured string");
+>>
+string: /"/ <<
+  $mode(default);
+  return $token(string);
+>>
+Start -> abc string def;
+EOF
+        end
+        build_parser(language: language)
+        compile("spec/test_lexer_modes.#{language}", language: language)
+        results = run
+        expect(results.status).to eq 0
+        verify_lines(results.stdout, [
+          "begin string mode",
+          "captured string",
+          "pass1",
+          "begin string mode",
+          "captured string",
+          "pass2",
+        ])
+      end
+
+      it "executes user code associated with a parser rule" do
+        case language
+        when "c"
+          write_grammar <<EOF
+<<
+#include <stdio.h>
+>>
+token a;
+token b;
+Start -> A B <<
+  printf("Start!\\n");
+>>
+A -> a <<
+  printf("A!\\n");
+>>
+B -> b <<
+  printf("B!\\n");
+>>
+EOF
+        when "d"
+          write_grammar <<EOF
+<<
+import std.stdio;
+>>
+token a;
+token b;
+Start -> A B <<
+  writeln("Start!");
+>>
+A -> a <<
+  writeln("A!");
+>>
+B -> b <<
+  writeln("B!");
+>>
+EOF
+        end
+        build_parser(language: language)
+        compile("spec/test_parser_rule_user_code.#{language}", language: language)
+        results = run
+        expect(results.status).to eq 0
+        verify_lines(results.stdout, [
+          "A!",
+          "B!",
+          "Start!",
+        ])
+      end
+
+      it "parses lists" do
+        write_grammar <<EOF
+ptype #{language == "c" ? "uint32_t" : "uint"};
+token a;
+Start -> As <<
+  $$ = $1;
+>>
+As -> <<
+  $$ = 0u;
+>>
+As -> As a <<
+  $$ = $1 + 1u;
+>>
+EOF
+        build_parser(language: language)
+        compile("spec/test_parsing_lists.#{language}", language: language)
+        results = run
+        expect(results.status).to eq 0
+        expect(results.stderr).to eq ""
+      end
+
+      it "fails to generate a parser for a LR(1) grammar that is not LALR" do
+        write_grammar <<EOF
+token a;
+token b;
+token c;
+token d;
+token e;
+Start -> a E c;
+Start -> a F d;
+Start -> b F c;
+Start -> b E d;
+E -> e;
+F -> e;
+EOF
+        results = build_parser(capture: true, language: language)
+        expect(results.status).to_not eq 0
+        expect(results.stderr).to match %r{reduce/reduce conflict.*\(E\).*\(F\)}
+      end
+
+      it "provides matched text to user code blocks" do
+        case language
+        when "c"
+          write_grammar <<EOF
+<<
+#include <stdio.h>
+#include <stdlib.h>
+>>
+token id /[a-zA-Z_][a-zA-Z0-9_]*/ <<
+  char * t = malloc(match_length + 1);
+  strncpy(t, (char *)match, match_length);
+  printf("Matched token is %s\\n", t);
+  free(t);
+>>
+Start -> id;
+EOF
+        when "d"
+          write_grammar <<EOF
+<<
+import std.stdio;
+>>
+token id /[a-zA-Z_][a-zA-Z0-9_]*/ <<
+  writeln("Matched token is ", match);
+>>
+Start -> id;
+EOF
+        end
+        build_parser(language: language)
+        compile("spec/test_lexer_match_text.#{language}", language: language)
+        results = run
+        expect(results.status).to eq 0
+        verify_lines(results.stdout, [
+          "Matched token is identifier_123",
+          "pass1",
+        ])
+      end
+
+      it "allows storing a result value for the lexer" do
+        case language
+        when "c"
+          write_grammar <<EOF
+ptype uint64_t;
+token word /[a-z]+/ <<
+  $$ = match_length;
+>>
+Start -> word <<
+  $$ = $1;
+>>
+EOF
+        when "d"
+          write_grammar <<EOF
+ptype ulong;
+token word /[a-z]+/ <<
+  $$ = match.length;
+>>
+Start -> word <<
+  $$ = $1;
+>>
+EOF
+        end
+        build_parser(language: language)
+        compile("spec/test_lexer_result_value.#{language}", language: language)
+        results = run
+        expect(results.stderr).to eq ""
+        expect(results.status).to eq 0
+      end
+
+      it "tracks position of parser errors" do
+        write_grammar <<EOF
+token a;
+token num /\\d+/;
+drop /\\s+/;
+Start -> a num Start;
+Start -> a num;
+EOF
+        build_parser(language: language)
+        compile("spec/test_error_positions.#{language}", language: language)
+        results = run
+        expect(results.stderr).to eq ""
+        expect(results.status).to eq 0
+      end
+
+      it "allows creating a JSON parser" do
+        write_grammar(File.read("spec/json_parser.propane"))
+        build_parser(language: language)
+        compile(["spec/test_parsing_json.#{language}", "spec/json_types.#{language}"], language: language)
+      end
+
+      it "allows generating multiple parsers in the same program" do
+        write_grammar(<<EOF, name: "myp1")
+prefix myp1_;
+token a;
+token num /\\d+/;
+drop /\\s+/;
+Start -> a num;
+EOF
+        build_parser(name: "myp1", language: language)
+        write_grammar(<<EOF, name: "myp2")
+prefix myp2_;
+token b;
+token c;
+Start -> b c b;
+EOF
+        build_parser(name: "myp2", language: language)
+        compile("spec/test_multiple_parsers.#{language}", parsers: %w[myp1 myp2], language: language)
+        results = run
+        expect(results.stderr).to eq ""
+        expect(results.status).to eq 0
+      end
+    end
   end
 end

spec/test_basic_math_grammar.c

@@ -0,0 +1,29 @@
+#include "testparser.h"
+#include "testutils.h"
+#include <string.h>
+
+int main()
+{
+    char const * input = "1 + 2 * 3 + 4";
+    p_context_t context;
+    p_context_init(&context, (uint8_t const *)input, strlen(input));
+    assert_eq(P_SUCCESS, p_parse(&context));
+    assert_eq(11, p_result(&context));
+
+    input = "1 * 2 ** 4 * 3";
+    p_context_init(&context, (uint8_t const *)input, strlen(input));
+    assert_eq(P_SUCCESS, p_parse(&context));
+    assert_eq(48, p_result(&context));
+
+    input = "(1 + 2) * 3 + 4";
+    p_context_init(&context, (uint8_t const *)input, strlen(input));
+    assert_eq(P_SUCCESS, p_parse(&context));
+    assert_eq(13, p_result(&context));
+
+    input = "(2 * 2) ** 3 + 4 + 5";
+    p_context_init(&context, (uint8_t const *)input, strlen(input));
+    assert_eq(P_SUCCESS, p_parse(&context));
+    assert_eq(73, p_result(&context));
+
+    return 0;
+}

spec/test_basic_math_grammar.d

@@ -0,0 +1,32 @@
+import testparser;
+import std.stdio;
+import testutils;
+
+int main()
+{
+    return 0;
+}
+
+unittest
+{
+    string input = "1 + 2 * 3 + 4";
+    p_context_t context;
+    p_context_init(&context, input);
+    assert_eq(P_SUCCESS, p_parse(&context));
+    assert_eq(11, p_result(&context));
+
+    input = "1 * 2 ** 4 * 3";
+    p_context_init(&context, input);
+    assert_eq(P_SUCCESS, p_parse(&context));
+    assert_eq(48, p_result(&context));
+
+    input = "(1 + 2) * 3 + 4";
+    p_context_init(&context, input);
+    assert_eq(P_SUCCESS, p_parse(&context));
+    assert_eq(13, p_result(&context));
+
+    input = "(2 * 2) ** 3 + 4 + 5";
+    p_context_init(&context, input);
+    assert_eq(P_SUCCESS, p_parse(&context));
+    assert_eq(73, p_result(&context));
+}

spec/test_d_lexer.d

@@ -1,84 +0,0 @@
-import testparser;
-import std.stdio;
-
-int main()
-{
-    return 0;
-}
-
-unittest
-{
-    alias DCP = Testparser.Decoder.DecodedCodePoint;
-    string inputstring = "5+\n 66";
-    const(ubyte) * input = cast(const(ubyte) *)inputstring.ptr;
-    size_t input_length = inputstring.length;
-    DCP dcp;
-    dcp = Testparser.Decoder.decode_code_point(input, input_length);
-    assert(dcp == DCP('5', 1u));
-    input += dcp.code_point_length;
-    input_length -= dcp.code_point_length;
-    dcp = Testparser.Decoder.decode_code_point(input, input_length);
-    assert(dcp == DCP('+', 1u));
-    input += dcp.code_point_length;
-    input_length -= dcp.code_point_length;
-    dcp = Testparser.Decoder.decode_code_point(input, input_length);
-    assert(dcp == DCP('\n', 1u));
-    input += dcp.code_point_length;
-    input_length -= dcp.code_point_length;
-    dcp = Testparser.Decoder.decode_code_point(input, input_length);
-    assert(dcp == DCP(' ', 1u));
-    input += dcp.code_point_length;
-    input_length -= dcp.code_point_length;
-    dcp = Testparser.Decoder.decode_code_point(input, input_length);
-    assert(dcp == DCP('6', 1u));
-    input += dcp.code_point_length;
-    input_length -= dcp.code_point_length;
-    dcp = Testparser.Decoder.decode_code_point(input, input_length);
-    assert(dcp == DCP('6', 1u));
-    input += dcp.code_point_length;
-    input_length -= dcp.code_point_length;
-    dcp = Testparser.Decoder.decode_code_point(input, input_length);
-    assert(dcp == DCP(Testparser.Decoder.CODE_POINT_EOF, 0u));
-
-    inputstring = "\xf0\x1f\x27\x21";
-    input = cast(const(ubyte) *)inputstring.ptr;
-    input_length = inputstring.length;
-    dcp = Testparser.Decoder.decode_code_point(input, input_length);
-    assert(dcp == DCP(0x1F9E1, 4u));
-
-    inputstring = "\xf0\x1f\x27";
-    input = cast(const(ubyte) *)inputstring.ptr;
-    input_length = inputstring.length;
-    dcp = Testparser.Decoder.decode_code_point(input, input_length);
-    assert(dcp == DCP(Testparser.Decoder.CODE_POINT_INVALID, 0u));
-
-    inputstring = "\xf0\x1f\x27\xFF";
-    input = cast(const(ubyte) *)inputstring.ptr;
-    input_length = inputstring.length;
-    dcp = Testparser.Decoder.decode_code_point(input, input_length);
-    assert(dcp == DCP(Testparser.Decoder.CODE_POINT_INVALID, 0u));
-
-    inputstring = "\xfe";
-    input = cast(const(ubyte) *)inputstring.ptr;
-    input_length = inputstring.length;
-    dcp = Testparser.Decoder.decode_code_point(input, input_length);
-    assert(dcp == DCP(Testparser.Decoder.CODE_POINT_INVALID, 0u));
-}
-
-unittest
-{
-    alias LT = Testparser.Lexer.LexedToken;
-    string input = "5 + 4 * \n677 + 567";
-    Testparser.Lexer lexer = new Testparser.Lexer(cast(const(ubyte) *)input.ptr, input.length);
-    assert(lexer.lex_token() == LT(0, 0, 1, Testparser.TOKEN_INT));
-    assert(lexer.lex_token() == LT(0, 2, 1, Testparser.TOKEN_PLUS));
-    assert(lexer.lex_token() == LT(0, 4, 1, Testparser.TOKEN_INT));
-    assert(lexer.lex_token() == LT(0, 6, 1, Testparser.TOKEN_TIMES));
-    assert(lexer.lex_token() == LT(1, 0, 3, Testparser.TOKEN_INT));
-    assert(lexer.lex_token() == LT(1, 4, 1, Testparser.TOKEN_PLUS));
-    assert(lexer.lex_token() == LT(1, 6, 3, Testparser.TOKEN_INT));
-    assert(lexer.lex_token() == LT(1, 9, 0, Testparser.TOKEN_EOF));
-
-    lexer = new Testparser.Lexer(null, 0u);
-    assert(lexer.lex_token() == LT(0, 0, 0, Testparser.TOKEN_EOF));
-}

spec/test_error_positions.c

@@ -0,0 +1,39 @@
+#include "testparser.h"
+#include <assert.h>
+#include <string.h>
+
+int main()
+{
+    char const * input = "a 42";
+    p_context_t context;
+    p_context_init(&context, (uint8_t const *)input, strlen(input));
+    assert(p_parse(&context) == P_SUCCESS);
+
+    input = "a\n123\na  a";
+    p_context_init(&context, (uint8_t const *)input, strlen(input));
+    assert(p_parse(&context) == P_UNEXPECTED_TOKEN);
+    assert(p_position(&context).row == 2);
+    assert(p_position(&context).col == 3);
+    assert(context.token == TOKEN_a);
+
+    input = "12";
+    p_context_init(&context, (uint8_t const *)input, strlen(input));
+    assert(p_parse(&context) == P_UNEXPECTED_TOKEN);
+    assert(p_position(&context).row == 0);
+    assert(p_position(&context).col == 0);
+    assert(context.token == TOKEN_num);
+
+    input = "a 12\n\nab";
+    p_context_init(&context, (uint8_t const *)input, strlen(input));
+    assert(p_parse(&context) == P_UNEXPECTED_INPUT);
+    assert(p_position(&context).row == 2);
+    assert(p_position(&context).col == 1);
+
+    input = "a 12\n\na\n\n77\na   \xAA";
+    p_context_init(&context, (uint8_t const *)input, strlen(input));
+    assert(p_parse(&context) == P_DECODE_ERROR);
+    assert(p_position(&context).row == 5);
+    assert(p_position(&context).col == 4);
+
+    return 0;
+}

spec/test_error_positions.d

@@ -0,0 +1,37 @@
+import testparser;
+import std.stdio;
+
+int main()
+{
+    return 0;
+}
+
+unittest
+{
+    string input = "a 42";
+    p_context_t context;
+    p_context_init(&context, input);
+    assert(p_parse(&context) == P_SUCCESS);
+
+    input = "a\n123\na  a";
+    p_context_init(&context, input);
+    assert(p_parse(&context) == P_UNEXPECTED_TOKEN);
+    assert(p_position(&context) == p_position_t(2, 3));
+    assert(context.token == TOKEN_a);
+
+    input = "12";
+    p_context_init(&context, input);
+    assert(p_parse(&context) == P_UNEXPECTED_TOKEN);
+    assert(p_position(&context) == p_position_t(0, 0));
+    assert(context.token == TOKEN_num);
+
+    input = "a 12\n\nab";
+    p_context_init(&context, input);
+    assert(p_parse(&context) == P_UNEXPECTED_INPUT);
+    assert(p_position(&context) == p_position_t(2, 1));
+
+    input = "a 12\n\na\n\n77\na   \xAA";
+    p_context_init(&context, input);
+    assert(p_parse(&context) == P_DECODE_ERROR);
+    assert(p_position(&context) == p_position_t(5, 4));
+}

spec/test_lexer.c

@@ -0,0 +1,92 @@
+#include "testparser.h"
+#include <assert.h>
+#include <string.h>
+
+int main()
+{
+    size_t result;
+    p_code_point_t code_point;
+    uint8_t code_point_length;
+
+    result = p_decode_code_point((uint8_t const *)"5", 1u, &code_point, &code_point_length);
+    assert(result == P_SUCCESS);
+    assert(code_point == '5');
+    assert(code_point_length == 1u);
+
+    result = p_decode_code_point((uint8_t const *)"", 0u, &code_point, &code_point_length);
+    assert(result == P_EOF);
+
+    result = p_decode_code_point((uint8_t const *)"\xC2\xA9", 2u, &code_point, &code_point_length);
+    assert(result == P_SUCCESS);
+    assert(code_point == 0xA9u);
+    assert(code_point_length == 2u);
+
+    result = p_decode_code_point((uint8_t const *)"\xf0\x9f\xa7\xa1", 4u, &code_point, &code_point_length);
+    assert(result == P_SUCCESS);
+    assert(code_point == 0x1F9E1u);
+    assert(code_point_length == 4u);
+
+    result = p_decode_code_point((uint8_t const *)"\xf0\x9f\x27", 3u, &code_point, &code_point_length);
+    assert(result == P_DECODE_ERROR);
+
+    result = p_decode_code_point((uint8_t const *)"\xf0\x9f\xa7\xFF", 4u, &code_point, &code_point_length);
+    assert(result == P_DECODE_ERROR);
+
+    result = p_decode_code_point((uint8_t const *)"\xfe", 1u, &code_point, &code_point_length);
+    assert(result == P_DECODE_ERROR);
+
+
+    p_token_info_t token_info;
+    char const * input = "5 + 4 * \n677 + 567";
+    p_context_t context;
+    p_context_init(&context, (uint8_t const *)input, strlen(input));
+    assert(p_lex(&context, &token_info) == P_SUCCESS);
+    assert(token_info.position.row == 0u);
+    assert(token_info.position.col == 0u);
+    assert(token_info.length == 1u);
+    assert(token_info.token == TOKEN_int);
+    assert(p_lex(&context, &token_info) == P_SUCCESS);
+    assert(token_info.position.row == 0u);
+    assert(token_info.position.col == 2u);
+    assert(token_info.length == 1u);
+    assert(token_info.token == TOKEN_plus);
+    assert(p_lex(&context, &token_info) == P_SUCCESS);
+    assert(token_info.position.row == 0u);
+    assert(token_info.position.col == 4u);
+    assert(token_info.length == 1u);
+    assert(token_info.token == TOKEN_int);
+    assert(p_lex(&context, &token_info) == P_SUCCESS);
+    assert(token_info.position.row == 0u);
+    assert(token_info.position.col == 6u);
+    assert(token_info.length == 1u);
+    assert(token_info.token == TOKEN_times);
+    assert(p_lex(&context, &token_info) == P_SUCCESS);
+    assert(token_info.position.row == 1u);
+    assert(token_info.position.col == 0u);
+    assert(token_info.length == 3u);
+    assert(token_info.token == TOKEN_int);
+    assert(p_lex(&context, &token_info) == P_SUCCESS);
+    assert(token_info.position.row == 1u);
+    assert(token_info.position.col == 4u);
+    assert(token_info.length == 1u);
+    assert(token_info.token == TOKEN_plus);
+    assert(p_lex(&context, &token_info) == P_SUCCESS);
+    assert(token_info.position.row == 1u);
+    assert(token_info.position.col == 6u);
+    assert(token_info.length == 3u);
+    assert(token_info.token == TOKEN_int);
+    assert(p_lex(&context, &token_info) == P_SUCCESS);
+    assert(token_info.position.row == 1u);
+    assert(token_info.position.col == 9u);
+    assert(token_info.length == 0u);
+    assert(token_info.token == TOKEN___EOF);
+
+    p_context_init(&context, (uint8_t const *)"", 0u);
+    assert(p_lex(&context, &token_info) == P_SUCCESS);
+    assert(token_info.position.row == 0u);
+    assert(token_info.position.col == 0u);
+    assert(token_info.length == 0u);
+    assert(token_info.token == TOKEN___EOF);
+
+    return 0;
+}

spec/test_lexer.d

@@ -0,0 +1,69 @@
+import testparser;
+import std.stdio;
+
+int main()
+{
+    return 0;
+}
+
+unittest
+{
+    size_t result;
+    p_code_point_t code_point;
+    ubyte code_point_length;
+
+    result = p_decode_code_point("5", &code_point, &code_point_length);
+    assert(result == P_SUCCESS);
+    assert(code_point == '5');
+    assert(code_point_length == 1u);
+
+    result = p_decode_code_point("", &code_point, &code_point_length);
+    assert(result == P_EOF);
+
+    result = p_decode_code_point("\xC2\xA9", &code_point, &code_point_length);
+    assert(result == P_SUCCESS);
+    assert(code_point == 0xA9u);
+    assert(code_point_length == 2u);
+
+    result = p_decode_code_point("\xf0\x9f\xa7\xa1", &code_point, &code_point_length);
+    assert(result == P_SUCCESS);
+    assert(code_point == 0x1F9E1u);
+    assert(code_point_length == 4u);
+
+    result = p_decode_code_point("\xf0\x9f\x27", &code_point, &code_point_length);
+    assert(result == P_DECODE_ERROR);
+
+    result = p_decode_code_point("\xf0\x9f\xa7\xFF", &code_point, &code_point_length);
+    assert(result == P_DECODE_ERROR);
+
+    result = p_decode_code_point("\xfe", &code_point, &code_point_length);
+    assert(result == P_DECODE_ERROR);
+}
+
+unittest
+{
+    p_token_info_t token_info;
+    string input = "5 + 4 * \n677 + 567";
+    p_context_t context;
+    p_context_init(&context, input);
+    assert(p_lex(&context, &token_info) == P_SUCCESS);
+    assert(token_info == p_token_info_t(p_position_t(0, 0), 1, TOKEN_int));
+    assert(p_lex(&context, &token_info) == P_SUCCESS);
+    assert(token_info == p_token_info_t(p_position_t(0, 2), 1, TOKEN_plus));
+    assert(p_lex(&context, &token_info) == P_SUCCESS);
+    assert(token_info == p_token_info_t(p_position_t(0, 4), 1, TOKEN_int));
+    assert(p_lex(&context, &token_info) == P_SUCCESS);
+    assert(token_info == p_token_info_t(p_position_t(0, 6), 1, TOKEN_times));
+    assert(p_lex(&context, &token_info) == P_SUCCESS);
+    assert(token_info == p_token_info_t(p_position_t(1, 0), 3, TOKEN_int));
+    assert(p_lex(&context, &token_info) == P_SUCCESS);
+    assert(token_info == p_token_info_t(p_position_t(1, 4), 1, TOKEN_plus));
+    assert(p_lex(&context, &token_info) == P_SUCCESS);
+    assert(token_info == p_token_info_t(p_position_t(1, 6), 3, TOKEN_int));
+    assert(p_lex(&context, &token_info) == P_SUCCESS);
+    assert(token_info == p_token_info_t(p_position_t(1, 9), 0, TOKEN___EOF));
+
+    p_context_init(&context, "");
+    assert(p_lex(&context, &token_info) == P_SUCCESS);
+    assert(token_info == p_token_info_t(p_position_t(0, 0), 0, TOKEN___EOF));
+}

spec/test_lexer_match_text.c

@@ -0,0 +1,15 @@
+#include "testparser.h"
+#include <assert.h>
+#include <string.h>
+#include <stdio.h>
+
+int main()
+{
+    char const * input = "identifier_123";
+    p_context_t context;
+    p_context_init(&context, (uint8_t const *)input, strlen(input));
+    assert(p_parse(&context) == P_SUCCESS);
+    printf("pass1\n");
+
+    return 0;
+}

spec/test_lexer_match_text.d

@@ -0,0 +1,16 @@
+import testparser;
+import std.stdio;
+
+int main()
+{
+    return 0;
+}
+
+unittest
+{
+    string input = `identifier_123`;
+    p_context_t context;
+    p_context_init(&context, input);
+    assert(p_parse(&context) == P_SUCCESS);
+    writeln("pass1");
+}

spec/test_lexer_modes.c

@@ -0,0 +1,20 @@
+#include "testparser.h"
+#include <assert.h>
+#include <string.h>
+#include <stdio.h>
+
+int main()
+{
+    char const * input = "abc \"a string\" def";
+    p_context_t context;
+    p_context_init(&context, (uint8_t const *)input, strlen(input));
+    assert(p_parse(&context) == P_SUCCESS);
+    printf("pass1\n");
+
+    input = "abc \"abc def\" def";
+    p_context_init(&context, (uint8_t const *)input, strlen(input));
+    assert(p_parse(&context) == P_SUCCESS);
+    printf("pass2\n");
+
+    return 0;
+}

spec/test_lexer_modes.d

@@ -0,0 +1,21 @@
+import testparser;
+import std.stdio;
+
+int main()
+{
+    return 0;
+}
+
+unittest
+{
+    string input = `abc "a string" def`;
+    p_context_t context;
+    p_context_init(&context, input);
+    assert(p_parse(&context) == P_SUCCESS);
+    writeln("pass1");
+
+    input = `abc "abc def" def`;
+    p_context_init(&context, input);
+    assert(p_parse(&context) == P_SUCCESS);
+    writeln("pass2");
+}

spec/test_lexer_result_value.c

@@ -0,0 +1,19 @@
+#include "testparser.h"
+#include <assert.h>
+#include <string.h>
+
+int main()
+{
+    char const * input = "x";
+    p_context_t context;
+    p_context_init(&context, (uint8_t const *)input, strlen(input));
+    assert(p_parse(&context) == P_SUCCESS);
+    assert(p_result(&context) == 1u);
+
+    input = "fabulous";
+    p_context_init(&context, (uint8_t const *)input, strlen(input));
+    assert(p_parse(&context) == P_SUCCESS);
+    assert(p_result(&context) == 8u);
+
+    return 0;
+}

spec/test_lexer_result_value.d

@@ -0,0 +1,21 @@
+import testparser;
+import std.stdio;
+
+int main()
+{
+    return 0;
+}
+
+unittest
+{
+    string input = `x`;
+    p_context_t context;
+    p_context_init(&context, input);
+    assert(p_parse(&context) == P_SUCCESS);
+    assert(p_result(&context) == 1u);
+
+    input = `fabulous`;
+    p_context_init(&context, input);
+    assert(p_parse(&context) == P_SUCCESS);
+    assert(p_result(&context) == 8u);
+}

spec/test_lexer_unknown_character.c

@@ -0,0 +1,18 @@
+#include "testparser.h"
+#include <assert.h>
+#include <string.h>
+
+int main()
+{
+    char const * input = "x";
+    p_context_t context;
+    p_context_init(&context, (uint8_t const *)input, strlen(input));
+    assert(p_parse(&context) == P_UNEXPECTED_INPUT);
+
+    input = "123";
+    p_context_init(&context, (uint8_t const *)input, strlen(input));
+    assert(p_parse(&context) == P_SUCCESS);
+    assert(p_result(&context) == 123u);
+
+    return 0;
+}

spec/test_lexer_unknown_character.d

@@ -0,0 +1,20 @@
+import testparser;
+import std.stdio;
+
+int main()
+{
+    return 0;
+}
+
+unittest
+{
+    string input = `x`;
+    p_context_t context;
+    p_context_init(&context, input);
+    assert(p_parse(&context) == P_UNEXPECTED_INPUT);
+
+    input = `123`;
+    p_context_init(&context, input);
+    assert(p_parse(&context) == P_SUCCESS);
+    assert(p_result(&context) == 123u);
+}

spec/test_multiple_parsers.c

@@ -0,0 +1,19 @@
+#include "testparsermyp1.h"
+#include "testparsermyp2.h"
+#include <assert.h>
+#include <string.h>
+
+int main()
+{
+    char const * input1 = "a\n1";
+    myp1_context_t context1;
+    myp1_context_init(&context1, (uint8_t const *)input1, strlen(input1));
+    assert(myp1_parse(&context1) == MYP1_SUCCESS);
+
+    char const * input2 = "bcb";
+    myp2_context_t context2;
+    myp2_context_init(&context2, (uint8_t const *)input2, strlen(input2));
+    assert(myp2_parse(&context2) == MYP2_SUCCESS);
+
+    return 0;
+}

spec/test_multiple_parsers.d

@@ -0,0 +1,21 @@
+import testparsermyp1;
+import testparsermyp2;
+import std.stdio;
+
+int main()
+{
+    return 0;
+}
+
+unittest
+{
+    string input1 = "a\n1";
+    myp1_context_t context1;
+    myp1_context_init(&context1, input1);
+    assert(myp1_parse(&context1) == MYP1_SUCCESS);
+
+    string input2 = "bcb";
+    myp2_context_t context2;
+    myp2_context_init(&context2, input2);
+    assert(myp2_parse(&context2) == MYP2_SUCCESS);
+}

spec/test_parser_identical_rules_lookahead.c

@@ -0,0 +1,17 @@
+#include "testparser.h"
+#include <string.h>
+#include <assert.h>
+
+int main()
+{
+    char const * input = "aba";
+    p_context_t context;
+    p_context_init(&context, (uint8_t const *)input, strlen(input));
+    assert(p_parse(&context) == P_SUCCESS);
+
+    input = "abb";
+    p_context_init(&context, (uint8_t const *)input, strlen(input));
+    assert(p_parse(&context) == P_SUCCESS);
+
+    return 0;
+}

spec/test_parser_identical_rules_lookahead.d

@@ -0,0 +1,19 @@
+import testparser;
+import std.stdio;
+
+int main()
+{
+    return 0;
+}
+
+unittest
+{
+    string input = "aba";
+    p_context_t context;
+    p_context_init(&context, input);
+    assert(p_parse(&context) == P_SUCCESS);
+
+    input = "abb";
+    p_context_init(&context, input);
+    assert(p_parse(&context) == P_SUCCESS);
+}

spec/test_parser_rule_from_multiple_states.c

@@ -0,0 +1,24 @@
+#include "testparser.h"
+#include <assert.h>
+#include <string.h>
+
+int main()
+{
+    char const * input = "a";
+    p_context_t context;
+    p_context_init(&context, (uint8_t const *)input, strlen(input));
+    assert(p_parse(&context) == P_UNEXPECTED_TOKEN);
+    assert(p_position(&context).row == 0);
+    assert(p_position(&context).col == 1);
+    assert(context.token == TOKEN___EOF);
+
+    input = "a b";
+    p_context_init(&context, (uint8_t const *)input, strlen(input));
+    assert(p_parse(&context) == P_SUCCESS);
+
+    input = "bb";
+    p_context_init(&context, (uint8_t const *)input, strlen(input));
+    assert(p_parse(&context) == P_SUCCESS);
+
+    return 0;
+}

spec/test_parser_rule_from_multiple_states.d

@@ -0,0 +1,25 @@
+import testparser;
+import std.stdio;
+
+int main()
+{
+    return 0;
+}
+
+unittest
+{
+    string input = "a";
+    p_context_t context;
+    p_context_init(&context, input);
+    assert(p_parse(&context) == P_UNEXPECTED_TOKEN);
+    assert(p_position(&context) == p_position_t(0, 1));
+    assert(context.token == TOKEN___EOF);
+
+    input = "a b";
+    p_context_init(&context, input);
+    assert(p_parse(&context) == P_SUCCESS);
+
+    input = "bb";
+    p_context_init(&context, input);
+    assert(p_parse(&context) == P_SUCCESS);
+}

spec/test_parser_rule_user_code.c

@@ -0,0 +1,13 @@
+#include "testparser.h"
+#include <assert.h>
+#include <string.h>
+
+int main()
+{
+    char const * input = "ab";
+    p_context_t context;
+    p_context_init(&context, (uint8_t const *)input, strlen(input));
+    assert(p_parse(&context) == P_SUCCESS);
+
+    return 0;
+}

spec/test_parser_rule_user_code.d

@@ -0,0 +1,15 @@
+import testparser;
+import std.stdio;
+
+int main()
+{
+    return 0;
+}
+
+unittest
+{
+    string input = "ab";
+    p_context_t context;
+    p_context_init(&context, input);
+    assert(p_parse(&context) == P_SUCCESS);
+}

spec/test_parsing_json.d

@@ -0,0 +1,56 @@
+import testparser;
+import std.stdio;
+import json_types;
+
+int main()
+{
+    return 0;
+}
+
+unittest
+{
+    string input = ``;
+    p_context_t context;
+    p_context_init(&context, input);
+    assert(p_parse(&context) == P_SUCCESS);
+
+    input = `{}`;
+    p_context_init(&context, input);
+    assert(p_parse(&context) == P_SUCCESS);
+    assert(cast(JSONObject)p_result(&context));
+
+    input = `[]`;
+    p_context_init(&context, input);
+    assert(p_parse(&context) == P_SUCCESS);
+    assert(cast(JSONArray)p_result(&context));
+
+    input = `-45.6`;
+    p_context_init(&context, input);
+    assert(p_parse(&context) == P_SUCCESS);
+    assert(cast(JSONNumber)p_result(&context));
+    assert((cast(JSONNumber)p_result(&context)).value == -45.6);
+
+    input = `2E-2`;
+    p_context_init(&context, input);
+    assert(p_parse(&context) == P_SUCCESS);
+    assert(cast(JSONNumber)p_result(&context));
+    assert((cast(JSONNumber)p_result(&context)).value == 0.02);
+
+    input = `{"hi":true}`;
+    p_context_init(&context, input);
+    assert(p_parse(&context) == P_SUCCESS);
+    assert(cast(JSONObject)p_result(&context));
+    JSONObject o = cast(JSONObject)p_result(&context);
+    assert(o.value["hi"]);
+    assert(cast(JSONTrue)o.value["hi"]);
+
+    input = `{"ff": false, "nn": null}`;
+    p_context_init(&context, input);
+    assert(p_parse(&context) == P_SUCCESS);
+    assert(cast(JSONObject)p_result(&context));
+    o = cast(JSONObject)p_result(&context);
+    assert(o.value["ff"]);
+    assert(cast(JSONFalse)o.value["ff"]);
+    assert(o.value["nn"]);
+    assert(cast(JSONNull)o.value["nn"]);
+}

spec/test_parsing_lists.c

@@ -0,0 +1,24 @@
+#include "testparser.h"
+#include <assert.h>
+#include <string.h>
+
+int main()
+{
+    char const * input = "a";
+    p_context_t context;
+    p_context_init(&context, (uint8_t const *)input, strlen(input));
+    assert(p_parse(&context) == P_SUCCESS);
+    assert(p_result(&context) == 1u);
+
+    input = "";
+    p_context_init(&context, (uint8_t const *)input, strlen(input));
+    assert(p_parse(&context) == P_SUCCESS);
+    assert(p_result(&context) == 0u);
+
+    input = "aaaaaaaaaaaaaaaa";
+    p_context_init(&context, (uint8_t const *)input, strlen(input));
+    assert(p_parse(&context) == P_SUCCESS);
+    assert(p_result(&context) == 16u);
+
+    return 0;
+}

spec/test_parsing_lists.d

@@ -0,0 +1,26 @@
+import testparser;
+import std.stdio;
+
+int main()
+{
+    return 0;
+}
+
+unittest
+{
+    string input = "a";
+    p_context_t context;
+    p_context_init(&context, input);
+    assert(p_parse(&context) == P_SUCCESS);
+    assert(p_result(&context) == 1u);
+
+    input = "";
+    p_context_init(&context, input);
+    assert(p_parse(&context) == P_SUCCESS);
+    assert(p_result(&context) == 0u);
+
+    input = "aaaaaaaaaaaaaaaa";
+    p_context_init(&context, input);
+    assert(p_parse(&context) == P_SUCCESS);
+    assert(p_result(&context) == 16u);
+}

spec/test_pattern.c

@@ -0,0 +1,20 @@
+#include "testparser.h"
+#include <stdio.h>
+#include <assert.h>
+#include <string.h>
+
+int main()
+{
+    char const * input = "abcdef";
+    p_context_t context;
+    p_context_init(&context, (uint8_t const *)input, strlen(input));
+    assert(p_parse(&context) == P_SUCCESS);
+    printf("pass1\n");
+
+    input = "defabcdef";
+    p_context_init(&context, (uint8_t const *)input, strlen(input));
+    assert(p_parse(&context) == P_SUCCESS);
+    printf("pass2\n");
+
+    return 0;
+}

spec/test_pattern.d

@@ -0,0 +1,21 @@
+import testparser;
+import std.stdio;
+
+int main()
+{
+    return 0;
+}
+
+unittest
+{
+    string input = "abcdef";
+    p_context_t context;
+    p_context_init(&context, input);
+    assert(p_parse(&context) == P_SUCCESS);
+    writeln("pass1");
+
+    input = "defabcdef";
+    p_context_init(&context, input);
+    assert(p_parse(&context) == P_SUCCESS);
+    writeln("pass2");
+}

spec/test_return_token_from_pattern.c

@@ -0,0 +1,13 @@
+#include "testparser.h"
+#include <assert.h>
+#include <string.h>
+
+int main()
+{
+    char const * input = "defghidef";
+    p_context_t context;
+    p_context_init(&context, (uint8_t const *)input, strlen(input));
+    assert(p_parse(&context) == P_SUCCESS);
+
+    return 0;
+}

spec/test_return_token_from_pattern.d

@@ -0,0 +1,15 @@
+import testparser;
+import std.stdio;
+
+int main()
+{
+    return 0;
+}
+
+unittest
+{
+    string input = "defghidef";
+    p_context_t context;
+    p_context_init(&context, input);
+    assert(p_parse(&context) == P_SUCCESS);
+}

spec/test_user_code.c

@@ -0,0 +1,20 @@
+#include "testparser.h"
+#include <assert.h>
+#include <stdio.h>
+#include <string.h>
+
+int main()
+{
+    char const * input = "abcdef";
+    p_context_t context;
+    p_context_init(&context, (uint8_t const *)input, strlen(input));
+    assert(p_parse(&context) == P_SUCCESS);
+    printf("pass1\n");
+
+    input = "abcabcdef";
+    p_context_init(&context, (uint8_t const *)input, strlen(input));
+    assert(p_parse(&context) == P_SUCCESS);
+    printf("pass2\n");
+
+    return 0;
+}

spec/test_user_code.d

@@ -0,0 +1,21 @@
+import testparser;
+import std.stdio;
+
+int main()
+{
+    return 0;
+}
+
+unittest
+{
+    string input = "abcdef";
+    p_context_t context;
+    p_context_init(&context, input);
+    assert(p_parse(&context) == P_SUCCESS);
+    writeln("pass1");
+
+    input = "abcabcdef";
+    p_context_init(&context, input);
+    assert(p_parse(&context) == P_SUCCESS);
+    writeln("pass2");
+}

spec/testutils.c

@@ -0,0 +1,12 @@
+#include <stdio.h>
+#include <assert.h>
+#include <stdbool.h>
+
+void assert_eq_size_t_i(size_t expected, size_t actual, char const * file, size_t line)
+{
+    if (expected != actual)
+    {
+        fprintf(stderr, "%s:%lu: expected %lu, got %lu\n", file, line, expected, actual);
+        assert(false);
+    }
+}

spec/testutils.d

@@ -0,0 +1,10 @@
+import std.stdio;
+
+void assert_eq(T)(T expected, T actual, string file = __FILE__, size_t line = __LINE__)
+{
+    if (expected != actual)
+    {
+        stderr.writeln(file, ":", line, ": expected ", expected, ", got ", actual);
+        assert(false);
+    }
+}

spec/testutils.h

@@ -0,0 +1,7 @@
+#pragma once
+
+void assert_eq_size_t_i(size_t expected, size_t actual, char const * file, size_t line);
+
+#define assert_eq(expected, actual) \
+    assert_eq_size_t_i(expected, actual, __FILE__, __LINE__)
+