255 lines
8.1 KiB
Plaintext
255 lines
8.1 KiB
Plaintext
<% if @grammar.modulename %>
|
|
module <%= @grammar.modulename %>;
|
|
|
|
<% end %>
|
|
class <%= @classname %>
|
|
{
|
|
enum
|
|
{
|
|
<% @grammar.tokens.each_with_index do |token, index| %>
|
|
TOKEN_<%= token.c_name %> = <%= index %>,
|
|
<% end %>
|
|
TOKEN_EOF = <%= TOKEN_EOF %>,
|
|
TOKEN_DECODE_ERROR = <%= TOKEN_DECODE_ERROR %>,
|
|
TOKEN_DROP = <%= TOKEN_DROP %>,
|
|
TOKEN_NONE = <%= TOKEN_NONE %>,
|
|
}
|
|
|
|
static immutable string TokenNames[] = [
|
|
<% @grammar.tokens.each_with_index do |token, index| %>
|
|
"<%= token.name %>",
|
|
<% end %>
|
|
];
|
|
|
|
static class Decoder
|
|
{
|
|
enum
|
|
{
|
|
CODE_POINT_INVALID = 0xFFFFFFFE,
|
|
CODE_POINT_EOF = 0xFFFFFFFF,
|
|
}
|
|
|
|
struct DecodedCodePoint
|
|
{
|
|
uint code_point;
|
|
uint code_point_length;
|
|
}
|
|
|
|
static DecodedCodePoint decode_code_point(const(ubyte) * input, size_t input_length)
|
|
{
|
|
if (input_length == 0u)
|
|
{
|
|
return DecodedCodePoint(CODE_POINT_EOF, 0u);
|
|
}
|
|
ubyte c = *input;
|
|
uint code_point;
|
|
uint code_point_length;
|
|
if ((c & 0x80u) == 0u)
|
|
{
|
|
code_point = c;
|
|
code_point_length = 1u;
|
|
}
|
|
else
|
|
{
|
|
ubyte 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 DecodedCodePoint(CODE_POINT_INVALID, 0u);
|
|
}
|
|
if (input_length <= following_bytes)
|
|
{
|
|
return DecodedCodePoint(CODE_POINT_INVALID, 0u);
|
|
}
|
|
code_point_length = following_bytes + 1u;
|
|
while (following_bytes-- > 0u)
|
|
{
|
|
input++;
|
|
ubyte b = *input;
|
|
if ((b & 0xC0u) != 0u)
|
|
{
|
|
return DecodedCodePoint(CODE_POINT_INVALID, 0u);
|
|
}
|
|
code_point = (code_point << 6u) | b;
|
|
}
|
|
}
|
|
return DecodedCodePoint(code_point, code_point_length);
|
|
}
|
|
}
|
|
|
|
static class Lexer
|
|
{
|
|
private struct Transition
|
|
{
|
|
uint first;
|
|
uint last;
|
|
uint destination;
|
|
}
|
|
|
|
private struct State
|
|
{
|
|
uint transition_table_index;
|
|
uint n_transitions;
|
|
uint accepts;
|
|
}
|
|
|
|
<% transition_table, state_table = @lexer.build_tables %>
|
|
private static const Transition transitions[] = [
|
|
<% transition_table.each do |transition_table_entry| %>
|
|
Transition(<%= transition_table_entry[:first] %>u, <%= transition_table_entry[:last] %>u, <%= transition_table_entry[:destination] %>u),
|
|
<% end %>
|
|
];
|
|
|
|
private static const State states[] = [
|
|
<% state_table.each do |state_table_entry| %>
|
|
State(<%= state_table_entry[:transition_table_index] %>u, <%= state_table_entry[:n_transitions] %>u, <%= state_table_entry[:accepts] %>u),
|
|
<% end %>
|
|
];
|
|
|
|
struct LexedToken
|
|
{
|
|
size_t row;
|
|
size_t col;
|
|
size_t length;
|
|
uint token;
|
|
}
|
|
|
|
private const(ubyte) * m_input;
|
|
private size_t m_input_length;
|
|
private size_t m_input_position;
|
|
private size_t m_input_row;
|
|
private size_t m_input_col;
|
|
|
|
this(const(ubyte) * input, size_t input_length)
|
|
{
|
|
m_input = input;
|
|
m_input_length = input_length;
|
|
}
|
|
|
|
LexedToken lex_token()
|
|
{
|
|
for (;;)
|
|
{
|
|
LexedToken lt = attempt_lex_token();
|
|
if (lt.token != TOKEN_DROP)
|
|
{
|
|
return lt;
|
|
}
|
|
}
|
|
}
|
|
|
|
private LexedToken attempt_lex_token()
|
|
{
|
|
LexedToken lt = LexedToken(m_input_row, m_input_col, 0, TOKEN_NONE);
|
|
struct LexedTokenState
|
|
{
|
|
size_t length;
|
|
size_t delta_row;
|
|
size_t delta_col;
|
|
uint token;
|
|
}
|
|
LexedTokenState last_accepts_info;
|
|
last_accepts_info.token = TOKEN_NONE;
|
|
LexedTokenState attempt_info;
|
|
uint current_state;
|
|
for (;;)
|
|
{
|
|
auto decoded = Decoder.decode_code_point(&m_input[m_input_position + attempt_info.length], m_input_length - m_input_position - attempt_info.length);
|
|
if (decoded.code_point == Decoder.CODE_POINT_INVALID)
|
|
{
|
|
lt.token = TOKEN_DECODE_ERROR;
|
|
return lt;
|
|
}
|
|
bool lex_continue = false;
|
|
if (decoded.code_point != Decoder.CODE_POINT_EOF)
|
|
{
|
|
uint dest = transition(current_state, decoded.code_point);
|
|
if (dest != cast(uint)-1)
|
|
{
|
|
lex_continue = true;
|
|
attempt_info.length += decoded.code_point_length;
|
|
if (decoded.code_point == '\n')
|
|
{
|
|
attempt_info.delta_row++;
|
|
attempt_info.delta_col = 0u;
|
|
}
|
|
else
|
|
{
|
|
attempt_info.delta_col++;
|
|
}
|
|
current_state = dest;
|
|
if (states[current_state].accepts != TOKEN_NONE)
|
|
{
|
|
attempt_info.token = states[current_state].accepts;
|
|
last_accepts_info = attempt_info;
|
|
}
|
|
}
|
|
}
|
|
else if (attempt_info.length == 0u)
|
|
{
|
|
lt.token = TOKEN_EOF;
|
|
break;
|
|
}
|
|
if (!lex_continue)
|
|
{
|
|
if (last_accepts_info.token != TOKEN_NONE)
|
|
{
|
|
lt.token = last_accepts_info.token;
|
|
lt.length = last_accepts_info.length;
|
|
m_input_position += last_accepts_info.length;
|
|
m_input_row += last_accepts_info.delta_row;
|
|
if (last_accepts_info.delta_row != 0u)
|
|
{
|
|
m_input_col = last_accepts_info.delta_col;
|
|
}
|
|
else
|
|
{
|
|
m_input_col += last_accepts_info.delta_col;
|
|
}
|
|
}
|
|
break;
|
|
}
|
|
}
|
|
return lt;
|
|
}
|
|
|
|
private uint transition(uint current_state, uint code_point)
|
|
{
|
|
uint transition_table_index = states[current_state].transition_table_index;
|
|
for (uint i = 0u; i < states[current_state].n_transitions; i++)
|
|
{
|
|
if ((transitions[transition_table_index + i].first <= code_point) &&
|
|
(code_point <= transitions[transition_table_index + i].last))
|
|
{
|
|
return transitions[transition_table_index + i].destination;
|
|
}
|
|
}
|
|
return cast(uint)-1;
|
|
}
|
|
}
|
|
}
|