Move parser functions to the header
- this way they can be inlined by the compiler if need be
This commit is contained in:
parent
7a3092b765
commit
3f4cde7488
3 changed files with 128 additions and 166 deletions
169
include/parser.h
169
include/parser.h
|
|
@ -1,17 +1,38 @@
|
|||
#ifndef __tree_sitter_parser_h__
|
||||
#define __tree_sitter_parser_h__
|
||||
|
||||
#ifdef __cplusplus
|
||||
extern "C" {
|
||||
#endif
|
||||
|
||||
#include "tree.h"
|
||||
#include "parse_config.h"
|
||||
#include <stdio.h>
|
||||
|
||||
#ifdef __cplusplus
|
||||
extern "C" {
|
||||
// #define TS_DEBUG_PARSE
|
||||
// #define TS_DEBUG_LEX
|
||||
|
||||
#ifdef TS_DEBUG_LEX
|
||||
#define DEBUG_LEX(...) fprintf(stderr, __VA_ARGS__)
|
||||
#else
|
||||
#define DEBUG_LEX(...)
|
||||
#endif
|
||||
|
||||
#ifdef TS_DEBUG_PARSE
|
||||
#define DEBUG_PARSE(...) fprintf(stderr, __VA_ARGS__)
|
||||
#else
|
||||
#define DEBUG_PARSE(...)
|
||||
#endif
|
||||
|
||||
static int INITIAL_STACK_SIZE = 100;
|
||||
|
||||
typedef int TSState;
|
||||
typedef struct TSStackEntry TSStackEntry;
|
||||
typedef struct TSParser {
|
||||
|
||||
typedef struct {
|
||||
TSState state;
|
||||
TSTree *node;
|
||||
} TSStackEntry;
|
||||
|
||||
typedef struct {
|
||||
const char *input;
|
||||
size_t position;
|
||||
TSTree *lookahead_node;
|
||||
|
|
@ -21,29 +42,96 @@ typedef struct TSParser {
|
|||
TSParseResult result;
|
||||
} TSParser;
|
||||
|
||||
TSParser TSParserMake(const char *input);
|
||||
void TSParserShift(TSParser *parser, TSState state);
|
||||
void TSParserReduce(TSParser *parser, TSSymbol symbol, int child_count);
|
||||
void TSParserLexError(TSParser *parser, size_t count, const char **expected_inputs);
|
||||
void TSParserError(TSParser *parser, size_t count, const char **expected_inputs);
|
||||
void TSParserAcceptInput(TSParser *parser);
|
||||
void TSParserAdvance(TSParser *parser, TSState lex_state);
|
||||
TSState TSParserParseState(const TSParser *parser);
|
||||
TSState TSParserLexState(const TSParser *parser);
|
||||
TSParseResult TSParserResult(TSParser *parser);
|
||||
void TSParserSetLexState(TSParser *parser, TSState state);
|
||||
char TSParserLookaheadChar(const TSParser *parser);
|
||||
long TSParserLookaheadSym(const TSParser *parser);
|
||||
void TSParserSetLookaheadSym(TSParser *parser, TSSymbol symbol);
|
||||
static TSParser TSParserMake(const char *input) {
|
||||
TSParser result = {
|
||||
.input = input,
|
||||
.position = 0,
|
||||
.lookahead_node = NULL,
|
||||
.lex_state = 0,
|
||||
.stack = calloc(INITIAL_STACK_SIZE, sizeof(TSStackEntry)),
|
||||
.stack_size = 0,
|
||||
.result = {
|
||||
.tree = NULL,
|
||||
.error = {
|
||||
.type = TSParseErrorTypeNone,
|
||||
.expected_inputs = NULL,
|
||||
.expected_input_count = 0
|
||||
},
|
||||
},
|
||||
};
|
||||
return result;
|
||||
}
|
||||
|
||||
static char TSParserLookaheadChar(const TSParser *parser) {
|
||||
return parser->input[parser->position];
|
||||
}
|
||||
|
||||
static long TSParserLookaheadSym(const TSParser *parser) {
|
||||
TSTree *node = parser->lookahead_node;
|
||||
return node ? node->value : -1;
|
||||
}
|
||||
|
||||
static TSState TSParserParseState(const TSParser *parser) {
|
||||
return parser->stack[parser->stack_size - 1].state;
|
||||
}
|
||||
|
||||
static void TSParserShift(TSParser *parser, TSState parse_state) {
|
||||
DEBUG_PARSE("shift %d \n", parse_state);
|
||||
TSStackEntry *entry = (parser->stack + parser->stack_size);
|
||||
entry->state = parse_state;
|
||||
entry->node = parser->lookahead_node;
|
||||
parser->lookahead_node = NULL;
|
||||
parser->stack_size++;
|
||||
}
|
||||
|
||||
static void TSParserReduce(TSParser *parser, TSSymbol symbol, int child_count) {
|
||||
parser->stack_size -= child_count;
|
||||
|
||||
TSTree **children = malloc(child_count * sizeof(TSTree *));
|
||||
for (int i = 0; i < child_count; i++) {
|
||||
children[i] = parser->stack[parser->stack_size + i].node;
|
||||
}
|
||||
|
||||
parser->lookahead_node = TSTreeMake(symbol, child_count, children);
|
||||
DEBUG_PARSE("reduce: %ld, state: %u \n", symbol, TSParserParseState(parser));
|
||||
}
|
||||
|
||||
static void TSParserError(TSParser *parser, size_t count, const char **expected_inputs) {
|
||||
TSParseError *error = &parser->result.error;
|
||||
error->type = TSParseErrorTypeSyntactic;
|
||||
error->expected_input_count = count;
|
||||
error->expected_inputs = expected_inputs;
|
||||
error->lookahead_sym = TSParserLookaheadSym(parser);
|
||||
}
|
||||
|
||||
static void TSParserLexError(TSParser *parser, size_t count, const char **expected_inputs) {
|
||||
TSParseError *error = &parser->result.error;
|
||||
error->type = TSParseErrorTypeLexical;
|
||||
error->expected_input_count = count;
|
||||
error->expected_inputs = expected_inputs;
|
||||
error->lookahead_sym = TSParserLookaheadSym(parser);
|
||||
}
|
||||
|
||||
static void TSParserAdvance(TSParser *parser, TSState lex_state) {
|
||||
DEBUG_LEX("character: '%c' \n", TSParserLookaheadChar(parser));
|
||||
parser->position++;
|
||||
parser->lex_state = lex_state;
|
||||
}
|
||||
|
||||
static void TSParserSetLookaheadSym(TSParser *parser, TSSymbol symbol) {
|
||||
DEBUG_LEX("token: %ld \n", symbol);
|
||||
parser->lookahead_node = TSTreeMake(symbol, 0, NULL);
|
||||
}
|
||||
|
||||
static void TSParserAcceptInput(TSParser *parser) {
|
||||
parser->result.tree = parser->stack[parser->stack_size - 1].node;
|
||||
}
|
||||
|
||||
#pragma mark - DSL
|
||||
|
||||
#define START_PARSER() \
|
||||
TSParser p = TSParserMake(input), *parser = &p; \
|
||||
TSParser p = TSParserMake(input), *parser = &p; \
|
||||
next_state:
|
||||
|
||||
#define LOOKAHEAD_SYM_NAME() \
|
||||
ts_symbol_names[LOOKAHEAD_SYM()]
|
||||
|
||||
#define START_LEXER() \
|
||||
if (LOOKAHEAD_SYM() >= 0) return; \
|
||||
|
|
@ -60,26 +148,26 @@ TSParserLookaheadChar(parser)
|
|||
TSParserParseState(parser)
|
||||
|
||||
#define LEX_STATE() \
|
||||
TSParserLexState(parser)
|
||||
parser->lex_state
|
||||
|
||||
#define SET_LEX_STATE(state_index) \
|
||||
{ parser->lex_state = state_index; ts_lex(parser); }
|
||||
|
||||
#define SHIFT(state) \
|
||||
{ TSParserShift(parser, state); goto next_state; }
|
||||
|
||||
#define SET_LEX_STATE(state_index) \
|
||||
{ TSParserSetLexState(parser, state_index); ts_lex(parser); }
|
||||
|
||||
#define ADVANCE(state_index) \
|
||||
{ TSParserAdvance(parser, state_index); goto next_state; }
|
||||
|
||||
#define REDUCE(symbol, child_count) \
|
||||
{ TSParserReduce(parser, symbol, child_count); goto next_state; }
|
||||
|
||||
#define ACCEPT_INPUT() \
|
||||
{ TSParserAcceptInput(parser); goto done; }
|
||||
|
||||
#define ACCEPT_TOKEN(symbol) \
|
||||
{ TSParserSetLookaheadSym(parser, symbol); goto done; }
|
||||
|
||||
#define REDUCE(symbol, child_count) \
|
||||
{ TSParserReduce(parser, symbol, child_count); goto next_state; }
|
||||
|
||||
#define PARSE_ERROR(count, inputs) \
|
||||
{ \
|
||||
static const char *expected_inputs[] = inputs; \
|
||||
|
|
@ -89,28 +177,27 @@ goto done; \
|
|||
|
||||
#define LEX_ERROR(count, inputs) \
|
||||
{ \
|
||||
static const char *expected_inputs[] = inputs; \
|
||||
TSParserLexError(parser, count, expected_inputs); \
|
||||
goto done; \
|
||||
static const char *expected_inputs[] = inputs; \
|
||||
TSParserLexError(parser, count, expected_inputs); \
|
||||
goto done; \
|
||||
}
|
||||
|
||||
#define LEX_PANIC() \
|
||||
printf("Lex error: unexpected state %ud", LEX_STATE());
|
||||
|
||||
#define PARSE_PANIC() \
|
||||
printf("Parse error: unexpected state %ud", PARSE_STATE());
|
||||
|
||||
#define EXPECT(...) __VA_ARGS__
|
||||
|
||||
#define FINISH_PARSER() \
|
||||
done: \
|
||||
return TSParserResult(parser);
|
||||
return parser->result;
|
||||
|
||||
#define FINISH_LEXER() \
|
||||
done:
|
||||
|
||||
#define LEX_PANIC() \
|
||||
printf("Lex error: unexpected state %ud", LEX_STATE());
|
||||
|
||||
#define PARSE_PANIC() \
|
||||
printf("Parse error: unexpected state %ud", PARSE_STATE());
|
||||
|
||||
#ifdef __cplusplus
|
||||
}
|
||||
#endif
|
||||
|
||||
#endif
|
||||
|
|
|
|||
|
|
@ -1,121 +0,0 @@
|
|||
#include "parser.h"
|
||||
#include <stdio.h>
|
||||
|
||||
#define TS_DEBUG_PARSE
|
||||
#define TS_DEBUG_LEX
|
||||
|
||||
#ifdef TS_DEBUG_LEX
|
||||
#define DEBUG_LEX(...) fprintf(stderr, __VA_ARGS__)
|
||||
#else
|
||||
#define DEBUG_LEX(...)
|
||||
#endif
|
||||
|
||||
#ifdef TS_DEBUG_PARSE
|
||||
#define DEBUG_PARSE(...) fprintf(stderr, __VA_ARGS__)
|
||||
#else
|
||||
#define DEBUG_PARSE(...)
|
||||
#endif
|
||||
|
||||
static int INITIAL_STACK_SIZE = 100;
|
||||
|
||||
struct TSStackEntry {
|
||||
TSState state;
|
||||
TSTree *node;
|
||||
};
|
||||
|
||||
TSParser TSParserMake(const char *input) {
|
||||
TSParser result = {
|
||||
.input = input,
|
||||
.position = 0,
|
||||
.lookahead_node = NULL,
|
||||
.lex_state = 0,
|
||||
.stack = calloc(INITIAL_STACK_SIZE, sizeof(TSStackEntry)),
|
||||
.stack_size = 0,
|
||||
.result = {
|
||||
.tree = NULL,
|
||||
.error = {
|
||||
.type = TSParseErrorTypeNone,
|
||||
.expected_inputs = NULL,
|
||||
.expected_input_count = 0
|
||||
},
|
||||
},
|
||||
};
|
||||
return result;
|
||||
}
|
||||
|
||||
void TSParserShift(TSParser *parser, TSState parse_state) {
|
||||
DEBUG_PARSE("shift %d \n", parse_state);
|
||||
TSStackEntry *entry = (parser->stack + parser->stack_size);
|
||||
entry->state = parse_state;
|
||||
entry->node = parser->lookahead_node;
|
||||
parser->lookahead_node = NULL;
|
||||
parser->stack_size++;
|
||||
}
|
||||
|
||||
void TSParserReduce(TSParser *parser, TSSymbol symbol, int child_count) {
|
||||
parser->stack_size -= child_count;
|
||||
|
||||
TSTree **children = malloc(child_count * sizeof(TSTree *));
|
||||
for (int i = 0; i < child_count; i++) {
|
||||
children[i] = parser->stack[parser->stack_size + i].node;
|
||||
}
|
||||
|
||||
parser->lookahead_node = TSTreeMake(symbol, child_count, children);
|
||||
DEBUG_PARSE("reduce: %ld, state: %u \n", symbol, TSParserParseState(parser));
|
||||
}
|
||||
|
||||
void TSParserError(TSParser *parser, size_t count, const char **expected_inputs) {
|
||||
TSParseError *error = &parser->result.error;
|
||||
error->type = TSParseErrorTypeSyntactic;
|
||||
error->expected_input_count = count;
|
||||
error->expected_inputs = expected_inputs;
|
||||
error->lookahead_sym = TSParserLookaheadSym(parser);
|
||||
}
|
||||
|
||||
void TSParserLexError(TSParser *parser, size_t count, const char **expected_inputs) {
|
||||
TSParseError *error = &parser->result.error;
|
||||
error->type = TSParseErrorTypeLexical;
|
||||
error->expected_input_count = count;
|
||||
error->expected_inputs = expected_inputs;
|
||||
error->lookahead_sym = TSParserLookaheadSym(parser);
|
||||
}
|
||||
|
||||
void TSParserAdvance(TSParser *parser, TSState lex_state) {
|
||||
DEBUG_LEX("character: '%c' \n", TSParserLookaheadChar(parser));
|
||||
parser->position++;
|
||||
parser->lex_state = lex_state;
|
||||
}
|
||||
|
||||
char TSParserLookaheadChar(const TSParser *parser) {
|
||||
return parser->input[parser->position];
|
||||
}
|
||||
|
||||
long TSParserLookaheadSym(const TSParser *parser) {
|
||||
TSTree *node = parser->lookahead_node;
|
||||
return node ? node->value : -1;
|
||||
}
|
||||
|
||||
void TSParserSetLookaheadSym(TSParser *parser, TSSymbol symbol) {
|
||||
DEBUG_LEX("token: %ld \n", symbol);
|
||||
parser->lookahead_node = TSTreeMake(symbol, 0, NULL);
|
||||
}
|
||||
|
||||
TSState TSParserParseState(const TSParser *parser) {
|
||||
return parser->stack[parser->stack_size - 1].state;
|
||||
}
|
||||
|
||||
TSState TSParserLexState(const TSParser *parser) {
|
||||
return parser->lex_state;
|
||||
}
|
||||
|
||||
void TSParserSetLexState(TSParser *parser, TSState lex_state) {
|
||||
parser->lex_state = lex_state;
|
||||
}
|
||||
|
||||
void TSParserAcceptInput(TSParser *parser) {
|
||||
parser->result.tree = parser->stack[parser->stack_size - 1].node;
|
||||
}
|
||||
|
||||
TSParseResult TSParserResult(TSParser *parser) {
|
||||
return parser->result;
|
||||
}
|
||||
|
|
@ -44,7 +44,6 @@
|
|||
12FD40D8185FEEDF0041A84E /* rules_spec.cpp in Sources */ = {isa = PBXBuildFile; fileRef = 121492EA181E200B008E9BDA /* rules_spec.cpp */; };
|
||||
12FD40D9185FEEDF0041A84E /* pattern_spec.cpp in Sources */ = {isa = PBXBuildFile; fileRef = 12D136A0183570F5005F3369 /* pattern_spec.cpp */; };
|
||||
12FD40DB185FEF0D0041A84E /* arithmetic_spec.cpp in Sources */ = {isa = PBXBuildFile; fileRef = 12FD40DA185FEF0D0041A84E /* arithmetic_spec.cpp */; };
|
||||
12FD40DD185FF12C0041A84E /* parser.c in Sources */ = {isa = PBXBuildFile; fileRef = 12FD40DC185FF12C0041A84E /* parser.c */; };
|
||||
12FD40DF1860064C0041A84E /* tree.c in Sources */ = {isa = PBXBuildFile; fileRef = 12FD40DE1860064C0041A84E /* tree.c */; };
|
||||
12FD40E718639B910041A84E /* visitor.cpp in Sources */ = {isa = PBXBuildFile; fileRef = 12FD40E618639B910041A84E /* visitor.cpp */; };
|
||||
12FD40E918641FB70041A84E /* rules.cpp in Sources */ = {isa = PBXBuildFile; fileRef = 12FD40E818641FB70041A84E /* rules.cpp */; };
|
||||
|
|
@ -145,7 +144,6 @@
|
|||
12FD40D1185EEB5E0041A84E /* runtime_specs */ = {isa = PBXFileReference; explicitFileType = "compiled.mach-o.executable"; includeInIndex = 0; path = runtime_specs; sourceTree = BUILT_PRODUCTS_DIR; };
|
||||
12FD40D4185FED9A0041A84E /* tree.h */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.c.h; path = tree.h; sourceTree = "<group>"; };
|
||||
12FD40DA185FEF0D0041A84E /* arithmetic_spec.cpp */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.cpp.cpp; path = arithmetic_spec.cpp; sourceTree = "<group>"; };
|
||||
12FD40DC185FF12C0041A84E /* parser.c */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.c; path = parser.c; sourceTree = "<group>"; };
|
||||
12FD40DE1860064C0041A84E /* tree.c */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.c; path = tree.c; sourceTree = "<group>"; };
|
||||
12FD40E41862B3530041A84E /* visitor.h */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.h; path = visitor.h; sourceTree = "<group>"; };
|
||||
12FD40E618639B910041A84E /* visitor.cpp */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.cpp.cpp; path = visitor.cpp; sourceTree = "<group>"; };
|
||||
|
|
@ -352,7 +350,6 @@
|
|||
12FD40AD185EE5440041A84E /* runtime */ = {
|
||||
isa = PBXGroup;
|
||||
children = (
|
||||
12FD40DC185FF12C0041A84E /* parser.c */,
|
||||
12FD40DE1860064C0041A84E /* tree.c */,
|
||||
12EDCF8C187C6282005A7A07 /* document.c */,
|
||||
12BC470318822A17005AC502 /* parse_config.cpp */,
|
||||
|
|
@ -511,7 +508,6 @@
|
|||
isa = PBXSourcesBuildPhase;
|
||||
buildActionMask = 2147483647;
|
||||
files = (
|
||||
12FD40DD185FF12C0041A84E /* parser.c in Sources */,
|
||||
12EDCF8A187B498C005A7A07 /* tree_spec.cpp in Sources */,
|
||||
12EDCF8D187C6282005A7A07 /* document.c in Sources */,
|
||||
12FD40DF1860064C0041A84E /* tree.c in Sources */,
|
||||
|
|
|
|||
Loading…
Add table
Add a link
Reference in a new issue