zzy/scc
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

feat(pproc): 实现C语言预处理器功能并重构项目依赖

Browse Source 
- 新增预处理器库(pproc)，替代原有的pprocessor模块
- 实现完整的宏定义解析功能，支持对象宏和函数宏
- 添加条件编译指令处理(#if、#ifdef、#ifndef、#else、#elif、#endif)
- 实现宏展开机制，包括嵌套宏和递归宏处理
- 添加宏定义测试用例，覆盖基本功能和复杂场景
- 在cbuild.toml中更新依赖配置，移除parser、ast、ast2ir、ir等未完成模块
- 新增lexer工具函数用于token流处理
- 添加宏定义表管理功能，支持宏的创建、查找、删除操作
- 实现宏参数解析和替换列表处理
This commit is contained in:
zzy
2026-02-17 22:47:25 +08:00
parent 681a15cb44
commit 2de5ae59f5
13 changed files with 1083 additions and 84 deletions
Download Patch File Download Diff File Expand all files Collapse all files

187
src/main.c
View File

@@ -1,41 +1,22 @@
#include <argparse.h>
#include <lexer.h>
#include <parser.h>
#include <pprocessor.h>
#include <scc_lexer.h>
#include <scc_pproc.h>
#include <ast_dump.h>
#include <ir_dump.h>
#include <scc_ast2ir.h>
// #include <scc_parser.h>
// #include <ast_dump.h>
// #include <ir_dump.h>
// #include <scc_ast2ir.h>
#include <stdio.h>
static scc_probe_stream_t *from_file_stream(FILE *fp) {
if (fseek(fp, 0, SEEK_END) != 0) {
perror("fseek failed");
return NULL;
}
usize fsize = ftell(fp);
if (fseek(fp, 0, SEEK_SET)) {
perror("fseek failed");
return NULL;
}
char *buffer = (char *)scc_malloc(fsize);
scc_memset(buffer, 0, fsize);
usize read_ret = fread(buffer, 1, fsize, fp);
fclose(fp);
scc_probe_stream_t *stream =
scc_mem_probe_stream_alloc(buffer, read_ret, true);
return stream;
}
typedef struct {
const char *input_file;
const char *output_file;
int verbose;
cbool dump_ast;
cbool dump_ir;
cbool emit_lex;
cbool emit_pp;
cbool emit_ast;
cbool emit_ir;
} scc_config_t;
static void setup_argparse(scc_argparse_t *argparse, scc_config_t *config,
@@ -46,16 +27,20 @@ static void setup_argparse(scc_argparse_t *argparse, scc_config_t *config,
SCC_HINT_OUTPUT_FILE,
SCC_HINT_INPUT_FILE,
SCC_HINT_VERBOSE,
SCC_HINT_EMIT_LEX,
SCC_HINT_EMIT_PP,
SCC_HINT_EMIT_AST,
SCC_HINT_EMIT_IR,
};
static const char *scc_hints_en[] = {
[SCC_HINT_PROG_NAME] = "scc",
[SCC_HINT_DESCRIPTION] = "A simple C compiler",
[SCC_HINT_OUTPUT_FILE] = "Output file",
[SCC_HINT_INPUT_FILE] = "Input source file",
[SCC_HINT_VERBOSE] = "Increase verbosity (can be used multiple times)",
[SCC_HINT_EMIT_LEX] = "Generate lexer sources tokens and exit",
[SCC_HINT_EMIT_PP] = "Generate preprocessed tokens and exit",
[SCC_HINT_EMIT_AST] = "Generate AST and exit",
[SCC_HINT_EMIT_IR] = "Generate IR and exit",
};
@@ -65,8 +50,10 @@ static void setup_argparse(scc_argparse_t *argparse, scc_config_t *config,
[SCC_HINT_OUTPUT_FILE] = "输出文件",
[SCC_HINT_INPUT_FILE] = "输入源文件",
[SCC_HINT_VERBOSE] = "增加详细输出(可多次使用)",
[SCC_HINT_EMIT_AST] = "生成 AST 并退出",
[SCC_HINT_EMIT_IR] = "生成 IR 并退出",
[SCC_HINT_EMIT_LEX] = "生成`源代码的词法单元`并退出",
[SCC_HINT_EMIT_PP] = "生成`预处理后的词法单元`并退出",
[SCC_HINT_EMIT_AST] = "生成`抽象语法树`并退出",
[SCC_HINT_EMIT_IR] = "生成`中间代码`并退出",
};
const char **scc_hints;
@@ -103,22 +90,35 @@ static void setup_argparse(scc_argparse_t *argparse, scc_config_t *config,
// -v, --verbose (计数)
scc_argparse_opt_t opt_verbose;
scc_argparse_opt_init(&opt_verbose, 'v', "verbose",
scc_argparse_opt_init(&opt_verbose, 'V', "verbose",
scc_hints[SCC_HINT_VERBOSE]);
scc_argparse_spec_setup_count(&opt_verbose.spec, &(config->verbose));
scc_argparse_cmd_add_opt(root, &opt_verbose);
// -T, --ast
// --emit-lex
scc_argparse_opt_t opt_lex;
scc_argparse_opt_init(&opt_lex, 0, "emit-lex",
scc_hints[SCC_HINT_EMIT_LEX]);
scc_argparse_spec_setup_bool(&opt_lex.spec, &(config->emit_lex));
scc_argparse_cmd_add_opt(root, &opt_lex);
// --emit-pp
scc_argparse_opt_t opt_pp;
scc_argparse_opt_init(&opt_pp, 0, "emit-pp", scc_hints[SCC_HINT_EMIT_PP]);
scc_argparse_spec_setup_bool(&opt_pp.spec, &(config->emit_pp));
scc_argparse_cmd_add_opt(root, &opt_pp);
// -T, --emit-ast
scc_argparse_opt_t opt_ast;
scc_argparse_opt_init(&opt_ast, 'T', "emit-ast",
scc_hints[SCC_HINT_EMIT_AST]);
scc_argparse_spec_setup_bool(&opt_ast.spec, &(config->dump_ast));
scc_argparse_spec_setup_bool(&opt_ast.spec, &(config->emit_ast));
scc_argparse_cmd_add_opt(root, &opt_ast);
// -R, --ir
// -R, --emit-ir
scc_argparse_opt_t opt_ir;
scc_argparse_opt_init(&opt_ir, 'R', "emit-ir", scc_hints[SCC_HINT_EMIT_IR]);
scc_argparse_spec_setup_bool(&opt_ir.spec, &(config->dump_ir));
scc_argparse_spec_setup_bool(&opt_ir.spec, &(config->emit_ir));
scc_argparse_cmd_add_opt(root, &opt_ir);
}
@@ -127,17 +127,42 @@ static void setup_argparse(scc_argparse_t *argparse, scc_config_t *config,
#include <windows.h>
#endif
static void print_ring(scc_lexer_tok_ring_t *ring, int verbose) {
scc_lexer_tok_t tok = {0};
int ret = 0;
while (1) {
scc_ring_next_consume(*ring, tok, ret);
if (ret == false || tok.type == SCC_TOK_EOF) {
break;
}
if (verbose == 0) {
scc_printf("%s ", scc_get_tok_name(tok.type));
} else if (verbose >= 1) {
scc_printf("token [%-8s] `%s` at %s:%d:%d\n",
scc_get_tok_name(tok.type),
scc_cstring_as_cstr(&tok.lexeme), tok.loc.name,
tok.loc.line, tok.loc.col);
}
}
}
int main(int argc, const char **argv, const char **envp) {
#ifdef _WIN32
SetConsoleOutputCP(CP_UTF8);
SetConsoleCP(CP_UTF8);
#endif
setbuf(stdout, NULL);
scc_config_t config = {
.input_file = NULL,
#ifdef _WIN32
.output_file = "a.exe",
#else
.output_file = "a.out",
#endif
.verbose = 0,
.dump_ast = false,
.dump_ir = false,
.emit_ast = false,
.emit_ir = false,
};
scc_argparse_t argparse;
setup_argparse(&argparse, &config, SCC_ARGPARSE_LANG_ZH);
@@ -148,51 +173,57 @@ int main(int argc, const char **argv, const char **envp) {
}
scc_argparse_drop(&argparse);
setbuf(stdout, NULL);
FILE *fp = fopen(config.input_file, "r");
if (!fp) {
perror("fopen");
scc_argparse_drop(&argparse);
return 1;
scc_sstream_t sstream;
if (scc_sstream_init(&sstream, config.input_file, 1024)) {
return 0;
}
scc_lexer_t lexer;
scc_lexer_init(&lexer, scc_sstream_to_ring(&sstream));
if (config.emit_lex) {
scc_lexer_tok_ring_t *tok_ring = scc_lexer_to_ring(&lexer, 8, false);
print_ring(tok_ring, config.verbose);
return 0;
}
scc_pproc_t pproc;
scc_probe_stream_t *source_code_stream = from_file_stream(fp);
// scc_probe_stream_t *pprocessed_code_stream =
// scc_pproc_init(&pproc, source_code_stream);
scc_lexer_t lexer;
scc_lexer_init(&lexer, source_code_stream);
scc_lexer_stream_t lexer_stream;
scc_lexer_to_stream(&lexer, &lexer_stream, false);
scc_parser_t parser;
scc_parser_init(&parser, &lexer_stream, null);
scc_ast_translation_unit_t *translation_unit =
scc_parse_translation_unit(&parser);
if (config.dump_ast) {
scc_tree_dump_ctx_t tree_dump;
scc_tree_dump_ctx_init(&tree_dump, true);
scc_ast_dump_node(&tree_dump, (scc_ast_node_t *)translation_unit);
scc_tree_dump_ctx_drop(&tree_dump);
return 0;
scc_pproc_init(&pproc, scc_lexer_to_ring(&lexer, 8, false));
if (config.emit_pp) {
scc_lexer_tok_ring_t *tok_ring = scc_pproc_to_ring(&pproc, 8);
print_ring(tok_ring, config.verbose);
}
scc_ir_builder_t ir_builder;
scc_ast2ir(translation_unit, &ir_builder);
scc_pproc_drop(&pproc);
scc_lexer_drop(&lexer);
scc_sstream_drop(&sstream);
if (config.dump_ir) {
scc_ir_dump_ctx_t ir_dump_ctx;
scc_tree_dump_ctx_t tree_dump; // 仅为 ir dump 辅助
scc_tree_dump_ctx_init(&tree_dump, true);
scc_ir_dump_ctx_init(&ir_dump_ctx, &tree_dump, &ir_builder.cprog,
&ir_builder.ctx);
// scc_ir_dump_cprog(&ir_dump_ctx);
scc_ir_dump_cprog_linear(&ir_dump_ctx);
scc_tree_dump_ctx_drop(&tree_dump);
return 0;
}
// scc_parser_t parser;
// scc_parser_init(&parser, &lexer_stream, null);
// scc_ast_translation_unit_t *translation_unit =
// scc_parse_translation_unit(&parser);
// if (config.emit_ast) {
// scc_tree_dump_ctx_t tree_dump;
// scc_tree_dump_ctx_init(&tree_dump, true);
// scc_ast_dump_node(&tree_dump, (scc_ast_node_t *)translation_unit);
// scc_tree_dump_ctx_drop(&tree_dump);
// return 0;
// }
// scc_ir_builder_t ir_builder;
// scc_ast2ir(translation_unit, &ir_builder);
// if (config.emit_ir) {
// scc_ir_dump_ctx_t ir_dump_ctx;
// scc_tree_dump_ctx_t tree_dump; // 仅为 ir dump 辅助
// scc_tree_dump_ctx_init(&tree_dump, true);
// scc_ir_dump_ctx_init(&ir_dump_ctx, &tree_dump, &ir_builder.cprog,
// &ir_builder.ctx);
// // scc_ir_dump_cprog(&ir_dump_ctx);
// scc_ir_dump_cprog_linear(&ir_dump_ctx);
// scc_tree_dump_ctx_drop(&tree_dump);
// return 0;
// }
scc_printf("output exe at %s", config.output_file);
return 0;