wip

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

Gemfile.lock

@@ -2,26 +2,26 @@ GEM
   remote: https://rubygems.org/
   specs:
     diff-lcs (1.5.0)
-    psych (5.0.1)
+    psych (5.1.0)
       stringio
     rake (13.0.6)
     rdoc (6.5.0)
       psych (>= 4.0.0)
-    redcarpet (3.5.1)
-    rspec (3.11.0)
-      rspec-core (~> 3.11.0)
-      rspec-expectations (~> 3.11.0)
-      rspec-mocks (~> 3.11.0)
-    rspec-core (3.11.0)
-      rspec-support (~> 3.11.0)
-    rspec-expectations (3.11.0)
+    redcarpet (3.6.0)
+    rspec (3.12.0)
+      rspec-core (~> 3.12.0)
+      rspec-expectations (~> 3.12.0)
+      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-mocks (3.11.1)
+      rspec-support (~> 3.12.0)
+    rspec-mocks (3.12.6)
       diff-lcs (>= 1.2.0, < 2.0)
-      rspec-support (~> 3.11.0)
-    rspec-support (3.11.0)
-    stringio (3.0.4)
+      rspec-support (~> 3.12.0)
+    rspec-support (3.12.1)
+    stringio (3.0.7)
     syntax (1.2.2)
 
 PLATFORMS

assets/parser.c.erb

@@ -0,0 +1,883 @@
+#include "<TBD>.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;
+    lexer_match_info_t attempt_match;
+    *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;
+    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. */
+alias reduce_id_t = <%= get_type_for(@parser.reduce_table.size) %>;
+
+/**
+ * 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.
+ */
+alias symbol_id_t = <%= get_type_for(@parser.rule_sets.map(&:last).map(&:id).max) %>;
+
+/** Parser state ID type. */
+alias parser_state_id_t = <%= get_type_for(@parser.state_table.size) %>;
+
+/** Parser rule ID type. */
+alias rule_id_t = <%= get_type_for(@grammar.rules.size) %>;
+
+/** Parser shift ID type. */
+alias shift_id_t = <%= get_type_for(@parser.shift_table.size) %>;
+
+/** 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 immutable shift_t[] parser_shift_table = [
+<%   @parser.shift_table.each do |shift| %>
+    shift_t(<%= shift[:symbol_id] %>u, <%= shift[:state_id] %>u),
+<%   end %>
+];
+
+/** Parser reduce table. */
+static immutable reduce_t[] parser_reduce_table = [
+<%   @parser.reduce_table.each do |reduce| %>
+    reduce_t(<%= reduce[:token_id] %>u, <%= reduce[:rule_id] %>u, <%= reduce[:rule_set_id] %>u, <%= reduce[:n_states] %>u),
+<%   end %>
+];
+
+/** Parser state table. */
+static immutable parser_state_t[] parser_state_table = [
+<%   @parser.state_table.each do |state| %>
+    parser_state_t(<%= 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.
+ */
+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));
+}
+
+/**
+ * Push a new state_value to the state_values stack.
+ *
+ * @param stack
+ *   state_values stack structure.
+ */
+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;
+}
+
+/**
+ * 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_value_t[] statevalues, uint32_t n_states)
+{
+    <%= @grammar.prefix %>value_t _pvalue;
+
+    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_value_t[] statevalues = new state_value_t[](1);
+    size_t reduced_rule_set = INVALID_ID;
+    <%= @grammar.prefix %>value_t reduced_parser_value;
+    for (;;)
+    {
+        if (token == INVALID_TOKEN_ID)
+        {
+            size_t lexer_result = <%= @grammar.prefix %>lex(context, &token_info);
+            if (lexer_result != P_SUCCESS)
+            {
+                return lexer_result;
+            }
+            token = token_info.token;
+        }
+        size_t shift_state = INVALID_ID;
+        if (reduced_rule_set != INVALID_ID)
+        {
+            shift_state = check_shift(statevalues[$-1].state_id, reduced_rule_set);
+        }
+        if (shift_state == INVALID_ID)
+        {
+            shift_state = check_shift(statevalues[$-1].state_id, token);
+            if ((shift_state != INVALID_ID) && (token == TOKEN___EOF))
+            {
+                /* Successful parse. */
+                context.parse_result = statevalues[$-1].pvalue;
+                return P_SUCCESS;
+            }
+        }
+        if (shift_state != INVALID_ID)
+        {
+            /* We have something to shift. */
+            statevalues ~= state_value_t(shift_state);
+            if (reduced_rule_set == INVALID_ID)
+            {
+                /* We shifted a token, mark it consumed. */
+                token = INVALID_TOKEN_ID;
+                statevalues[$-1].pvalue = token_info.pvalue;
+            }
+            else
+            {
+                /* We shifted a RuleSet. */
+                statevalues[$-1].pvalue = reduced_parser_value;
+                <%= @grammar.prefix %>value_t new_parse_result;
+                reduced_parser_value = new_parse_result;
+                reduced_rule_set = INVALID_ID;
+            }
+            continue;
+        }
+
+        size_t reduce_index = check_reduce(statevalues[$-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;
+            statevalues.length -= 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;
+        return P_UNEXPECTED_TOKEN;
+    }
+}
+
+/**
+ * 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 +1,108 @@
+/**
+ * @file
+ *
+ * This file is generated by Propane.
+ */
+
+#ifndef PROPANE_PARSER_H
+#define PROPANE_PARSER_H
+
+#include <stdint.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 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;
+
+#endif

doc/user_guide.md

@@ -49,6 +49,60 @@ detailed information about the parser states and transitions.
 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:

lib/propane/generator.rb

@@ -183,7 +183,7 @@ class Propane
     #   Expanded user code block.
     def expand_code(code, parser, rule, pattern)
       code = code.gsub(/\$token\(([$\w]+)\)/) do |match|
-        "Token(TOKEN_#{Token.code_name($1)})"
+        "TOKEN_#{Token.code_name($1)}"
       end
       if parser
         code = code.gsub(/\$\$/) do |match|
@@ -195,7 +195,7 @@ class Propane
         end
       else
         code = code.gsub(/\$\$/) do |match|
-          "result.pvalue.v_#{pattern.ptypename}"
+          "out_token_info.pvalue.v_#{pattern.ptypename}"
         end
         code = code.gsub(/\$mode\(([a-zA-Z_][a-zA-Z_0-9]*)\)/) do |match|
           mode_name = $1
@@ -203,7 +203,7 @@ class Propane
           unless mode_id
             raise Error.new("Lexer mode '#{mode_name}' not found")
           end
-          "m_mode = #{mode_id}u"
+          "context.mode = #{mode_id}u"
         end
       end
       code
@@ -220,6 +220,23 @@ class Propane
       [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
+        "ubyte"
+      elsif max <= 0xFFFF
+        "ushort"
+      else
+        "uint"
+      end
+    end
+
   end
 
 end

lib/propane/grammar.rb

@@ -12,6 +12,7 @@ class Propane
     attr_reader :tokens
     attr_reader :code_blocks
     attr_reader :ptypes
+    attr_reader :prefix
 
     def initialize(input)
       @patterns = []
@@ -23,6 +24,7 @@ class Propane
       @mode = nil
       @input = input.gsub("\r\n", "\n")
       @ptypes = {"default" => "void *"}
+      @prefix = "p_"
       parse_grammar!
     end
 
@@ -30,6 +32,10 @@ class Propane
       @ptypes["default"]
     end
 
+    def invalid_token_id
+      @tokens.size
+    end
+
     private
 
     def parse_grammar!
@@ -51,6 +57,7 @@ class Propane
       elsif parse_drop_statement!
       elsif parse_rule_statement!
       elsif parse_code_block_statement!
+      elsif parse_prefix_statement!
       else
         if @input.size > 25
           @input = @input.slice(0..20) + "..."
@@ -148,7 +155,7 @@ class Propane
         end
         consume!(/\s+/)
         consume!(/;/, "expected `;'")
-        @patterns << Pattern.new(pattern: pattern, line_number: @line_number, drop: true, mode: @mode)
+        @patterns << Pattern.new(pattern: pattern, line_number: @line_number, mode: @mode)
         @mode = nil
         true
       end
@@ -191,6 +198,13 @@ class Propane
       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 = ""

lib/propane/lexer.rb

@@ -1,55 +1,13 @@
 class Propane
   class Lexer
 
+    attr_reader :state_table
+    attr_reader :transition_table
+    attr_reader :mode_table
+
     def initialize(grammar)
       @grammar = grammar
-    end
-
-    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
-      transition_table = []
-      state_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|
-          drop = state.accepts && state.accepts.drop?
-          token =
-            if state.accepts && state.accepts.token
-              state.accepts.token.id
-            end
-          code_id =
-            if state.accepts && state.accepts.code_id
-              state.accepts.code_id
-            end
-          state_table << {
-            transition_table_index: transition_table.size,
-            n_transitions: state.transitions.size,
-            drop: drop,
-            token: token,
-            code_id: code_id,
-          }
-          state.transitions.each do |transition|
-            transition_table << {
-              first: transition.code_point_range.first,
-              last: transition.code_point_range.last,
-              destination: states[transition.destination] + state_table_offset,
-            }
-          end
-        end
-      end
-      [transition_table, state_table, mode_table]
+      build_tables!
     end
 
     # Get ID for a mode.
@@ -65,5 +23,46 @@ class Propane
       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|
+          token = state.accepts && state.accepts.token && state.accepts.token.id
+          code_id = state.accepts && state.accepts.code_id && state.accepts.code_id
+          @state_table << {
+            transition_table_index: @transition_table.size,
+            n_transitions: state.transitions.size,
+            accepts: !!state.accepts,
+            token: token,
+            code_id: code_id,
+          }
+          state.transitions.each do |transition|
+            @transition_table << {
+              first: transition.code_point_range.first,
+              last: transition.code_point_range.last,
+              destination: states[transition.destination] + state_table_offset,
+            }
+          end
+        end
+      end
+    end
+
   end
 end

lib/propane/parser.rb

@@ -2,6 +2,11 @@ class Propane
 
   class Parser
 
+    attr_reader :state_table
+    attr_reader :shift_table
+    attr_reader :reduce_table
+    attr_reader :rule_sets
+
     def initialize(grammar, rule_sets, log)
       @grammar = grammar
       @rule_sets = rule_sets
@@ -33,12 +38,15 @@ class Propane
 
       build_reduce_actions!
       write_log!
+      build_tables!
     end
 
-    def build_tables
-      shift_table = []
-      state_table = []
-      reduce_table = []
+    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 =
@@ -48,14 +56,14 @@ class Propane
               item_set.following_item_set[following_symbol].id
             end
           {
-            token_id: following_symbol.id,
+            symbol_id: following_symbol.id,
             state_id: state_id,
           }
         end
         reduce_entries =
           case ra = item_set.reduce_actions
           when Rule
-            [{token_id: @grammar.tokens.size, rule_id: ra.id,
+            [{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|
@@ -65,20 +73,17 @@ class Propane
           else
             []
           end
-        state_table << {
-          shift_index: shift_table.size,
+        @state_table << {
+          shift_index: @shift_table.size,
           n_shifts: shift_entries.size,
-          reduce_index: reduce_table.size,
+          reduce_index: @reduce_table.size,
           n_reduces: reduce_entries.size,
         }
-        shift_table += shift_entries
-        reduce_table += reduce_entries
+        @shift_table += shift_entries
+        @reduce_table += reduce_entries
       end
-      [state_table, shift_table, reduce_table]
     end
 
-    private
-
     def process_item_set(item_set)
       item_set.following_symbols.each do |following_symbol|
         unless following_symbol.name == "$EOF"
@@ -242,7 +247,7 @@ class Propane
         @log.puts "  Reduce actions:"
         case item_set.reduce_actions
         when Rule
-          @log.puts "    * => #{item_set.reduce_actions.id} (#{item_set.reduce_actions.name})"
+          @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}"

lib/propane/pattern.rb

@@ -40,8 +40,6 @@ class Propane
     #   Optional parameters.
     # @option options [String, nil] :code
     #   Code block to execute when the pattern is matched.
-    # @option options [Boolean] :drop
-    #   Whether this is a drop pattern.
     # @option options [String, nil] :pattern
     #   Pattern.
     # @option options [Token, nil] :token
@@ -52,7 +50,6 @@ class Propane
     #   Lexer mode for this pattern.
     def initialize(options)
       @code = options[:code]
-      @drop = options[:drop]
       @pattern = options[:pattern]
       @token = options[:token]
       @line_number = options[:line_number]
@@ -63,14 +60,6 @@ class Propane
       @nfa = regex.nfa
     end
 
-    # Whether the pattern is a drop pattern.
-    #
-    # @return [Boolean]
-    #   Whether the pattern is a drop pattern.
-    def drop?
-      @drop
-    end
-
   end
 
 end

lib/propane/rule_set.rb

@@ -4,7 +4,7 @@ class Propane
 
     # @return [Integer]
     #   ID of the RuleSet.
-    attr_accessor :id
+    attr_reader :id
 
     # @return [String]
     #   Name of the RuleSet.

spec/propane/grammar_spec.rb

@@ -34,6 +34,7 @@ EOF
       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
@@ -111,8 +112,11 @@ token code2 <<
 >>
 
 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

spec/propane_spec.rb

@@ -4,12 +4,14 @@ require "open3"
 Results = Struct.new(:stdout, :stderr, :status)
 
 describe Propane do
-  def write_grammar(grammar)
-    File.write("spec/run/testparser.propane", grammar)
+  def write_grammar(grammar, options = {})
+    options[:name] ||= ""
+    File.write("spec/run/testparser#{options[:name]}.propane", grammar)
   end
 
   def build_parser(options = {})
-    command = %w[./propane.sh spec/run/testparser.propane spec/run/testparser.d --log spec/run/testparser.log]
+    options[:name] ||= ""
+    command = %W[./propane.sh spec/run/testparser#{options[:name]}.propane spec/run/testparser#{options[:name]}.d --log spec/run/testparser#{options[:name]}.log]
     if (options[:capture])
       stdout, stderr, status = Open3.capture3(*command)
       Results.new(stdout, stderr, status)
@@ -19,8 +21,13 @@ describe Propane do
     end
   end
 
-  def compile(*test_files)
-    result = system(*%w[ldc2 --unittest -of spec/run/testparser spec/run/testparser.d -Ispec], *test_files)
+  def compile(test_files, options = {})
+    test_files = Array(test_files)
+    options[:parsers] ||= [""]
+    parsers = options[:parsers].map do |name|
+      "spec/run/testparser#{name}.d"
+    end
+    result = system(*%w[ldc2 --unittest -of spec/run/testparser -Ispec], *parsers, *test_files, "spec/testutils.d")
     expect(result).to be_truthy
   end
 
@@ -62,7 +69,7 @@ describe Propane do
     FileUtils.mkdir_p("spec/run")
   end
 
-  it "generates a D lexer" do
+  it "generates a lexer" do
     write_grammar <<EOF
 token int /\\d+/;
 token plus /\\+/;
@@ -75,7 +82,30 @@ Foo -> plus <<
 >>
 EOF
     build_parser
-    compile("spec/test_d_lexer.d")
+    compile("spec/test_lexer.d")
+    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
+    write_grammar <<EOF
+ptype int;
+token int /\\d+/ <<
+  int v;
+  foreach (c; match)
+  {
+    v *= 10;
+    v += (c - '0');
+  }
+  $$ = v;
+>>
+Start -> int <<
+  $$ = $1;
+>>
+EOF
+    build_parser
+    compile("spec/test_lexer_unknown_character.d")
     results = run
     expect(results.stderr).to eq ""
     expect(results.status).to eq 0
@@ -97,6 +127,65 @@ EOF
     build_parser
   end
 
+  it "generates a parser that does basic math - user guide example" do
+    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
+    build_parser
+    compile("spec/test_basic_math_grammar.d")
+    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/;
@@ -117,7 +206,7 @@ R1 -> a b;
 R2 -> a b;
 EOF
     build_parser
-    compile("spec/test_d_parser_identical_rules_lookahead.d")
+    compile("spec/test_parser_identical_rules_lookahead.d")
     results = run
     expect(results.status).to eq 0
   end
@@ -132,13 +221,16 @@ Start -> b R1;
 R1 -> b;
 EOF
     build_parser
-    compile("spec/test_d_parser_rule_from_multiple_states.d")
+    compile("spec/test_parser_rule_from_multiple_states.d")
     results = run
     expect(results.status).to eq 0
   end
 
   it "executes user code when matching lexer token" do
     write_grammar <<EOF
+<<
+import std.stdio;
+>>
 token abc <<
   writeln("abc!");
 >>
@@ -162,6 +254,9 @@ EOF
 
   it "supports a pattern statement" do
     write_grammar <<EOF
+<<
+import std.stdio;
+>>
 token abc;
 /def/ <<
   writeln("def!");
@@ -183,6 +278,9 @@ EOF
 
   it "supports returning tokens from pattern code blocks" do
     write_grammar <<EOF
+<<
+import std.stdio;
+>>
 token abc;
 /def/ <<
   writeln("def!");
@@ -206,6 +304,9 @@ EOF
 
   it "supports lexer modes" do
     write_grammar <<EOF
+<<
+import std.stdio;
+>>
 token abc;
 token def;
 tokenid string;
@@ -239,6 +340,9 @@ EOF
 
   it "executes user code associated with a parser rule" do
     write_grammar <<EOF
+<<
+import std.stdio;
+>>
 token a;
 token b;
 Start -> A B <<
@@ -304,6 +408,9 @@ EOF
 
   it "provides matched text to user code blocks" do
     write_grammar <<EOF
+<<
+import std.stdio;
+>>
 token id /[a-zA-Z_][a-zA-Z0-9_]*/ <<
   writeln("Matched token is ", match);
 >>
@@ -336,9 +443,46 @@ EOF
     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
+    compile("spec/test_error_positions.d")
+    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
-    compile("spec/test_parsing_json.d", "spec/json_types.d")
+    compile(["spec/test_parsing_json.d", "spec/json_types.d"])
+  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")
+    write_grammar(<<EOF, name: "myp2")
+prefix myp2_;
+token b;
+token c;
+Start -> b c b;
+EOF
+    build_parser(name: "myp2")
+    compile("spec/test_multiple_parsers.d", parsers: %w[myp1 myp2])
+    results = run
+    expect(results.stderr).to eq ""
+    expect(results.status).to eq 0
   end
 end

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

@@ -1,52 +0,0 @@
-import testparser;
-import std.stdio;
-
-int main()
-{
-    return 0;
-}
-
-unittest
-{
-    alias Result = Testparser.Decoder.Result;
-    Result result;
-
-    result = Testparser.Decoder.decode_code_point("5");
-    assert(result == Result.success('5', 1u));
-
-    result = Testparser.Decoder.decode_code_point("");
-    assert(result == Result.eof());
-
-    result = Testparser.Decoder.decode_code_point("\xC2\xA9");
-    assert(result == Result.success(0xA9u, 2u));
-
-    result = Testparser.Decoder.decode_code_point("\xf0\x9f\xa7\xa1");
-    assert(result == Result.success(0x1F9E1, 4u));
-
-    result = Testparser.Decoder.decode_code_point("\xf0\x9f\x27");
-    assert(result == Result.decode_error());
-
-    result = Testparser.Decoder.decode_code_point("\xf0\x9f\xa7\xFF");
-    assert(result == Result.decode_error());
-
-    result = Testparser.Decoder.decode_code_point("\xfe");
-    assert(result == Result.decode_error());
-}
-
-unittest
-{
-    alias Result = Testparser.Lexer.Result;
-    string input = "5 + 4 * \n677 + 567";
-    Testparser.Lexer lexer = new Testparser.Lexer(input);
-    assert(lexer.lex_token() == Result(Result.Type.TOKEN, 0, 0, 1, Testparser.TOKEN_int));
-    assert(lexer.lex_token() == Result(Result.Type.TOKEN, 0, 2, 1, Testparser.TOKEN_plus));
-    assert(lexer.lex_token() == Result(Result.Type.TOKEN, 0, 4, 1, Testparser.TOKEN_int));
-    assert(lexer.lex_token() == Result(Result.Type.TOKEN, 0, 6, 1, Testparser.TOKEN_times));
-    assert(lexer.lex_token() == Result(Result.Type.TOKEN, 1, 0, 3, Testparser.TOKEN_int));
-    assert(lexer.lex_token() == Result(Result.Type.TOKEN, 1, 4, 1, Testparser.TOKEN_plus));
-    assert(lexer.lex_token() == Result(Result.Type.TOKEN, 1, 6, 3, Testparser.TOKEN_int));
-    assert(lexer.lex_token() == Result(Result.Type.TOKEN, 1, 9, 0, Testparser.TOKEN___EOF));
-
-    lexer = new Testparser.Lexer("");
-    assert(lexer.lex_token() == Result(Result.Type.TOKEN, 0, 0, 0, Testparser.TOKEN___EOF));
-}

spec/test_d_parser_identical_rules_lookahead.d

@@ -1,18 +0,0 @@
-import testparser;
-import std.stdio;
-
-int main()
-{
-    return 0;
-}
-
-unittest
-{
-    string input = "aba";
-    auto parser = new Testparser.Parser(input);
-    assert(parser.parse() == true);
-
-    input = "abb";
-    parser = new Testparser.Parser(input);
-    assert(parser.parse() == true);
-}

spec/test_d_parser_rule_from_multiple_states.d

@@ -1,22 +0,0 @@
-import testparser;
-import std.stdio;
-
-int main()
-{
-    return 0;
-}
-
-unittest
-{
-    string input = "a";
-    auto parser = new Testparser.Parser(input);
-    assert(parser.parse() == false);
-
-    input = "a b";
-    parser = new Testparser.Parser(input);
-    assert(parser.parse() == true);
-
-    input = "bb";
-    parser = new Testparser.Parser(input);
-    assert(parser.parse() == true);
-}

spec/test_error_positions.d

@@ -0,0 +1,38 @@
+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);
+    writeln(p_position(&context));
+    assert(p_position(&context) == p_position_t(5, 4));
+}

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.d

@@ -9,7 +9,8 @@ int main()
 unittest
 {
     string input = `identifier_123`;
-    auto parser = new Testparser.Parser(input);
-    assert(parser.parse() == true);
+    p_context_t context;
+    p_context_init(&context, input);
+    assert(p_parse(&context) == P_SUCCESS);
     writeln("pass1");
 }

spec/test_lexer_modes.d

@@ -9,12 +9,13 @@ int main()
 unittest
 {
     string input = `abc "a string" def`;
-    auto parser = new Testparser.Parser(input);
-    assert(parser.parse() == true);
+    p_context_t context;
+    p_context_init(&context, input);
+    assert(p_parse(&context) == P_SUCCESS);
     writeln("pass1");
 
     input = `abc "abc def" def`;
-    parser = new Testparser.Parser(input);
-    assert(parser.parse() == true);
+    p_context_init(&context, input);
+    assert(p_parse(&context) == P_SUCCESS);
     writeln("pass2");
 }

spec/test_lexer_result_value.d

@@ -9,12 +9,13 @@ int main()
 unittest
 {
     string input = `x`;
-    auto parser = new Testparser.Parser(input);
-    assert(parser.parse() == true);
-    assert(parser.result == 1u);
+    p_context_t context;
+    p_context_init(&context, input);
+    assert(p_parse(&context) == P_SUCCESS);
+    assert(p_result(&context) == 1u);
 
     input = `fabulous`;
-    parser = new Testparser.Parser(input);
-    assert(parser.parse() == true);
-    assert(parser.result == 8u);
+    p_context_init(&context, input);
+    assert(p_parse(&context) == P_SUCCESS);
+    assert(p_result(&context) == 8u);
 }

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.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.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.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.d

@@ -9,6 +9,7 @@ int main()
 unittest
 {
     string input = "ab";
-    auto parser = new Testparser.Parser(input);
-    assert(parser.parse() == true);
+    p_context_t context;
+    p_context_init(&context, input);
+    assert(p_parse(&context) == P_SUCCESS);
 }

spec/test_parsing_json.d

@@ -10,44 +10,45 @@ int main()
 unittest
 {
     string input = ``;
-    auto parser = new Testparser.Parser(input);
-    assert(parser.parse());
+    p_context_t context;
+    p_context_init(&context, input);
+    assert(p_parse(&context) == P_SUCCESS);
 
     input = `{}`;
-    parser = new Testparser.Parser(input);
-    assert(parser.parse());
-    assert(cast(JSONObject)parser.result);
+    p_context_init(&context, input);
+    assert(p_parse(&context) == P_SUCCESS);
+    assert(cast(JSONObject)p_result(&context));
 
     input = `[]`;
-    parser = new Testparser.Parser(input);
-    assert(parser.parse());
-    assert(cast(JSONArray)parser.result);
+    p_context_init(&context, input);
+    assert(p_parse(&context) == P_SUCCESS);
+    assert(cast(JSONArray)p_result(&context));
 
     input = `-45.6`;
-    parser = new Testparser.Parser(input);
-    assert(parser.parse());
-    assert(cast(JSONNumber)parser.result);
-    assert((cast(JSONNumber)parser.result).value == -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`;
-    parser = new Testparser.Parser(input);
-    assert(parser.parse());
-    assert(cast(JSONNumber)parser.result);
-    assert((cast(JSONNumber)parser.result).value == 0.02);
+    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}`;
-    parser = new Testparser.Parser(input);
-    assert(parser.parse());
-    assert(cast(JSONObject)parser.result);
-    JSONObject o = cast(JSONObject)parser.result;
+    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}`;
-    parser = new Testparser.Parser(input);
-    assert(parser.parse());
-    assert(cast(JSONObject)parser.result);
-    o = cast(JSONObject)parser.result;
+    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"]);

spec/test_parsing_lists.d

@@ -9,17 +9,18 @@ int main()
 unittest
 {
     string input = "a";
-    auto parser = new Testparser.Parser(input);
-    assert(parser.parse() == true);
-    assert(parser.result == 1u);
+    p_context_t context;
+    p_context_init(&context, input);
+    assert(p_parse(&context) == P_SUCCESS);
+    assert(p_result(&context) == 1u);
 
     input = "";
-    parser = new Testparser.Parser(input);
-    assert(parser.parse() == true);
-    assert(parser.result == 0u);
+    p_context_init(&context, input);
+    assert(p_parse(&context) == P_SUCCESS);
+    assert(p_result(&context) == 0u);
 
     input = "aaaaaaaaaaaaaaaa";
-    parser = new Testparser.Parser(input);
-    assert(parser.parse() == true);
-    assert(parser.result == 16u);
+    p_context_init(&context, input);
+    assert(p_parse(&context) == P_SUCCESS);
+    assert(p_result(&context) == 16u);
 }

spec/test_pattern.d

@@ -9,12 +9,13 @@ int main()
 unittest
 {
     string input = "abcdef";
-    auto parser = new Testparser.Parser(input);
-    assert(parser.parse() == true);
+    p_context_t context;
+    p_context_init(&context, input);
+    assert(p_parse(&context) == P_SUCCESS);
     writeln("pass1");
 
     input = "defabcdef";
-    parser = new Testparser.Parser(input);
-    assert(parser.parse() == true);
+    p_context_init(&context, input);
+    assert(p_parse(&context) == P_SUCCESS);
     writeln("pass2");
 }

spec/test_return_token_from_pattern.d

@@ -9,6 +9,7 @@ int main()
 unittest
 {
     string input = "defghidef";
-    auto parser = new Testparser.Parser(input);
-    assert(parser.parse() == true);
+    p_context_t context;
+    p_context_init(&context, input);
+    assert(p_parse(&context) == P_SUCCESS);
 }

spec/test_user_code.d

@@ -9,12 +9,13 @@ int main()
 unittest
 {
     string input = "abcdef";
-    auto parser = new Testparser.Parser(input);
-    assert(parser.parse() == true);
+    p_context_t context;
+    p_context_init(&context, input);
+    assert(p_parse(&context) == P_SUCCESS);
     writeln("pass1");
 
     input = "abcabcdef";
-    parser = new Testparser.Parser(input);
-    assert(parser.parse() == true);
+    p_context_init(&context, input);
+    assert(p_parse(&context) == P_SUCCESS);
     writeln("pass2");
 }

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);
+    }
+}