391 lines
11 KiB
C
391 lines
11 KiB
C
/* splc0.c - SPL compiler CLI (stage 1, 引导用)
|
||
*
|
||
* splc0 --dump tokens|ast|sema|ir|vm <file> dump 前端产物
|
||
* splc0 <in> <out> 编译
|
||
*/
|
||
#define __SCC_LOG_IMPL_IMPORT_SRC__
|
||
#include "../stage0/include/utils.h"
|
||
|
||
#include <stdio.h>
|
||
#include <stdlib.h>
|
||
#include <string.h>
|
||
|
||
#include "spl_ast.h"
|
||
#include "spl_ast2ir.h"
|
||
#include "spl_diag.h"
|
||
#include "spl_ir2vm.h"
|
||
#include "spl_lexer.h"
|
||
#include "spl_parser.h"
|
||
#include "spl_sema.h"
|
||
#include "spl_tok.h"
|
||
|
||
typedef enum {
|
||
DUMP_NONE,
|
||
DUMP_TOKEN,
|
||
DUMP_AST,
|
||
DUMP_SEMA,
|
||
DUMP_IR,
|
||
DUMP_VM,
|
||
} dump_t;
|
||
|
||
typedef struct {
|
||
spl_ast_t *ast;
|
||
char *abs_path;
|
||
char *src;
|
||
} spl_driver_module_t;
|
||
typedef VEC(spl_driver_module_t) spl_driver_module_vec_t;
|
||
|
||
typedef struct {
|
||
spl_ast_t main_ast;
|
||
spl_tok_vec_t main_toks;
|
||
char *main_path;
|
||
|
||
spl_driver_module_vec_t modules;
|
||
MAP(const char *, usize) path_to_idx;
|
||
|
||
spl_diag_t *diag;
|
||
} spl_driver_t;
|
||
|
||
static char *read_file(const char *path, long *out_len) {
|
||
FILE *f = fopen(path, "rb");
|
||
if (!f) {
|
||
fprintf(stderr, "path `%s` ", path);
|
||
perror("fopen");
|
||
return NULL;
|
||
}
|
||
fseek(f, 0, SEEK_END);
|
||
long len = ftell(f);
|
||
fseek(f, 0, SEEK_SET);
|
||
char *buf = malloc((size_t)len + 1);
|
||
if (!buf) {
|
||
fclose(f);
|
||
return NULL;
|
||
}
|
||
size_t nread = fread(buf, 1, (size_t)len, f);
|
||
fclose(f);
|
||
buf[nread] = '\0';
|
||
if (out_len)
|
||
*out_len = (long)nread;
|
||
return buf;
|
||
}
|
||
|
||
/* 注意：按值接收，内部 vec_free 释放与调用方共享的 data 指针。
|
||
调用后原 vec 视为已消费，不要再使用。 */
|
||
static void dump_tokens(spl_tok_vec_t toks) {
|
||
printf("tokens got (%zu)\n", toks.size);
|
||
for (usize i = 0; i < toks.size; i++) {
|
||
const spl_tok_t *t = &toks.data[i];
|
||
printf("[%s] %.*s (%zu:%zu)\n", spl_tok_type_name(t->type), (int)t->len, t->lexeme, t->line,
|
||
t->col);
|
||
}
|
||
vec_free(toks);
|
||
}
|
||
|
||
/* 把 import_path 拼接到 importing_file 所在目录。
|
||
from_fname 为空时按相对 cwd 处理。绝对路径直接透传。 */
|
||
static void resolve_relative(const char *from, const char *path, char *out, usize cap) {
|
||
if (!from || !path) {
|
||
if (cap)
|
||
out[0] = 0;
|
||
return;
|
||
}
|
||
if (path[0] == '/' || path[0] == '\\' || (path[0] && path[1] == ':')) {
|
||
snprintf(out, cap, "%s", path);
|
||
return;
|
||
}
|
||
const char *last_sep = NULL;
|
||
for (const char *p = from; *p; ++p)
|
||
if (*p == '/' || *p == '\\')
|
||
last_sep = p;
|
||
if (!last_sep) {
|
||
snprintf(out, cap, "%s", path);
|
||
return;
|
||
}
|
||
usize dir_len = (usize)(last_sep - from) + 1;
|
||
snprintf(out, cap, "%.*s%s", (int)dir_len, from, path);
|
||
}
|
||
|
||
static spl_ast_t *driver_load(void *user, const char *from_fname, const char *path,
|
||
spl_diag_t *diag) {
|
||
spl_driver_t *d = user;
|
||
char abs[1024];
|
||
resolve_relative(from_fname, path, abs, sizeof abs);
|
||
|
||
usize idx;
|
||
if (map_get(d->path_to_idx, abs, &idx)) {
|
||
spl_ast_t *a = vec_at(d->modules, idx).ast;
|
||
if (!a) {
|
||
spl_diag_error(diag, SPL_DIAG_KIND_CIRCULAR_DEP, NULL, "circular import: `%s`", abs);
|
||
return NULL;
|
||
}
|
||
return a;
|
||
}
|
||
|
||
/* 先登记 ast=NULL 占位：若后续解析过程中又有 @import 指回本路径，
|
||
会命中这里，判定为环。 */
|
||
char *key = strdup(abs);
|
||
idx = vec_size(d->modules);
|
||
spl_driver_module_t placeholder = {.ast = NULL, .abs_path = key, .src = NULL};
|
||
vec_push(d->modules, placeholder);
|
||
map_put(d->path_to_idx, key, idx);
|
||
|
||
long len = 0;
|
||
char *src = read_file(abs, &len);
|
||
if (!src) {
|
||
spl_diag_error(diag, SPL_DIAG_KIND_LEX_UNEXPECTED, NULL, "@import: cannot open `%s`", abs);
|
||
return NULL;
|
||
}
|
||
(void)len;
|
||
|
||
/* key 同时用作 lexer 的 fname：token.fname 与 map key 共享一份 strdup，
|
||
减少一次分配；cleanup 时只 free module.abs_path */
|
||
spl_tok_vec_t toks = spl_lex(src, key);
|
||
spl_ast_t *ast = malloc(sizeof *ast);
|
||
spl_ast_init(ast, &toks);
|
||
spl_parser_t parser;
|
||
spl_parser_init(&parser);
|
||
parser.diag = diag;
|
||
spl_ast_parse(&parser, ast);
|
||
spl_parser_drop(&parser);
|
||
|
||
vec_at(d->modules, idx).ast = ast;
|
||
vec_at(d->modules, idx).src = src;
|
||
return ast;
|
||
}
|
||
|
||
static int compile_spl(const char *src, const char *fname, const char *outpath, int gen_debug,
|
||
dump_t dump) {
|
||
(void)gen_debug;
|
||
|
||
spl_diag_t diag;
|
||
spl_diag_init(&diag, stderr);
|
||
|
||
spl_driver_t drv = {0};
|
||
drv.diag = &diag;
|
||
drv.main_path = strdup(fname);
|
||
drv.main_toks = spl_lex(src, drv.main_path);
|
||
spl_ast_init(&drv.main_ast, &drv.main_toks);
|
||
map_init(drv.path_to_idx, MAP_HASH_STR, MAP_CMP_STR);
|
||
vec_init(drv.modules);
|
||
|
||
int rc = 0;
|
||
|
||
/* dump_tokens 消费 drv.main_toks（共享 data），之后 main_ast.input 悬空，
|
||
所以这条分支必须在 spl_ast_parse 之前、且不能再 drop main_ast。 */
|
||
if (dump == DUMP_TOKEN) {
|
||
dump_tokens(drv.main_toks);
|
||
free(drv.main_path);
|
||
map_free(drv.path_to_idx);
|
||
spl_diag_drop(&diag);
|
||
return 0;
|
||
}
|
||
|
||
spl_attr_t attr;
|
||
spl_lang_type_t lang_type;
|
||
spl_layout_target_t layout_target;
|
||
spl_layout_t layout;
|
||
spl_scope_t scope;
|
||
spl_sema_t sema;
|
||
spl_ast2ir_t a2ir;
|
||
spl_ir_builder_t ir_builder;
|
||
spl_ir2vm_t ir2vm;
|
||
spl_ast_ptr_vec_t all_asts;
|
||
|
||
int attr_ok = 0, lang_type_ok = 0, layout_ok = 0, scope_ok = 0;
|
||
int sema_ok = 0, a2ir_ok = 0, irb_ok = 0, ir2vm_ok = 0;
|
||
|
||
spl_diag_set_stage(&diag, SPL_STAGE_PARSE);
|
||
spl_parser_t parser;
|
||
spl_parser_init(&parser);
|
||
parser.diag = &diag;
|
||
spl_ast_parse(&parser, &drv.main_ast);
|
||
spl_parser_drop(&parser);
|
||
if (spl_diag_has_error(&diag) || spl_diag_has_fatal(&diag)) {
|
||
spl_diag_summary(&diag);
|
||
rc = 1;
|
||
goto cleanup;
|
||
}
|
||
if (dump == DUMP_AST) {
|
||
spl_ast_dump(&drv.main_ast, drv.main_ast.root);
|
||
goto cleanup;
|
||
}
|
||
|
||
spl_attr_init(&attr);
|
||
attr_ok = 1;
|
||
spl_lang_type_init(&lang_type);
|
||
lang_type_ok = 1;
|
||
|
||
/* 目标描述符：剥离目标机器细节，layout 是 size/align/offset 的唯一来源。 */
|
||
memset(&layout_target, 0, sizeof layout_target);
|
||
layout_target.endian = SPL_LITTLE_ENDIAN;
|
||
layout_target.ptr_size = sizeof(void *);
|
||
layout_target.ptr_align = sizeof(void *);
|
||
spl_layout_init(&layout, &layout_target);
|
||
layout_ok = 1;
|
||
|
||
spl_scope_init(&scope);
|
||
scope_ok = 1;
|
||
|
||
spl_sema_init(&sema, &drv.main_ast, &lang_type, &layout, &scope, &attr);
|
||
sema_ok = 1;
|
||
spl_module_loader_t loader = {.load = driver_load, .user = &drv};
|
||
sema.loader = &loader;
|
||
spl_diag_set_stage(&diag, SPL_STAGE_SEMA);
|
||
sema.diag = &diag;
|
||
spl_sema_run(&sema);
|
||
if (spl_diag_has_error(&diag) || spl_diag_has_fatal(&diag)) {
|
||
spl_diag_summary(&diag);
|
||
rc = 1;
|
||
goto cleanup;
|
||
}
|
||
if (dump == DUMP_SEMA) {
|
||
spl_sema_dump(&sema);
|
||
spl_ast_dump(&drv.main_ast, drv.main_ast.root);
|
||
goto cleanup;
|
||
}
|
||
|
||
/* 收集所有 ast：主文件在前，被 import 的文件按加载顺序在后。
|
||
ast2ir 内部会 vec_free(all_asts)。 */
|
||
vec_init(all_asts);
|
||
vec_push(all_asts, &drv.main_ast);
|
||
for (usize i = 0; i < vec_size(drv.modules); ++i)
|
||
vec_push(all_asts, vec_at(drv.modules, i).ast);
|
||
|
||
spl_ir_builder_init(&ir_builder);
|
||
irb_ok = 1;
|
||
spl_ast2ir_init(&a2ir, &ir_builder, &layout, &sema);
|
||
a2ir_ok = 1;
|
||
spl_diag_set_stage(&diag, SPL_STAGE_AST2IR);
|
||
a2ir.diag = &diag;
|
||
spl_ast2ir_run(&a2ir, all_asts);
|
||
if (spl_diag_has_error(&diag) || spl_diag_has_fatal(&diag)) {
|
||
spl_diag_summary(&diag);
|
||
rc = 1;
|
||
goto cleanup;
|
||
}
|
||
if (dump == DUMP_IR) {
|
||
spl_ir_dump(&ir_builder.ir, &lang_type.ir, &attr);
|
||
goto cleanup;
|
||
}
|
||
|
||
/* 类型已全部就绪：解析 layout，供 ir2vm 的 size/align/offset 使用。 */
|
||
spl_layout_resolve(&layout, &lang_type.ir);
|
||
|
||
spl_ir2vm_init(&ir2vm, &a2ir.ir->ir, &lang_type.ir, &layout, &attr);
|
||
ir2vm_ok = 1;
|
||
spl_diag_set_stage(&diag, SPL_STAGE_IR2VM);
|
||
ir2vm.diag = &diag;
|
||
rc = spl_ir2vm_run(&ir2vm);
|
||
if (spl_diag_has_error(&diag) || spl_diag_has_fatal(&diag)) {
|
||
spl_diag_summary(&diag);
|
||
goto cleanup;
|
||
}
|
||
if (dump == DUMP_VM) {
|
||
spl_prog_dump(&ir2vm.prog);
|
||
goto cleanup;
|
||
}
|
||
|
||
spl_prog_store_to_file(outpath, &ir2vm.prog);
|
||
rc = 0;
|
||
|
||
cleanup:
|
||
if (ir2vm_ok)
|
||
spl_ir2vm_drop(&ir2vm);
|
||
if (a2ir_ok)
|
||
spl_ast2ir_drop(&a2ir);
|
||
if (irb_ok)
|
||
spl_ir_builder_drop(&ir_builder);
|
||
if (sema_ok)
|
||
spl_sema_drop(&sema);
|
||
if (scope_ok)
|
||
spl_scope_drop(&scope);
|
||
if (layout_ok)
|
||
spl_layout_drop(&layout);
|
||
if (lang_type_ok)
|
||
spl_lang_type_drop(&lang_type);
|
||
if (attr_ok)
|
||
spl_attr_drop(&attr);
|
||
|
||
spl_ast_drop(&drv.main_ast);
|
||
for (usize i = 0; i < vec_size(drv.modules); ++i) {
|
||
spl_ast_t *a = vec_at(drv.modules, i).ast;
|
||
if (a) {
|
||
spl_ast_drop(a);
|
||
free(a);
|
||
}
|
||
free(vec_at(drv.modules, i).abs_path);
|
||
free(vec_at(drv.modules, i).src);
|
||
}
|
||
vec_free(drv.modules);
|
||
/* map 的 key 是 module 里 strdup 出的 abs_path，已随 module 释放 */
|
||
map_free(drv.path_to_idx);
|
||
free(drv.main_path);
|
||
spl_diag_drop(&diag);
|
||
return rc;
|
||
}
|
||
|
||
int main(int argc, char **argv) {
|
||
if (argc < 2) {
|
||
LOG_FATAL("Usage: splc0 [--dump <flags>] <in> [out]");
|
||
return 1;
|
||
}
|
||
if (strcmp(argv[1], "--help") == 0 || strcmp(argv[1], "-h") == 0) {
|
||
LOG_INFO("splc0 <in> <out> compile (.spl -> .sir, stage B)");
|
||
LOG_INFO("splc0 --dump <flags> <file> dump: lex,ast,sema,ir,vm,none");
|
||
return 0;
|
||
}
|
||
int argi = 1;
|
||
if (argi >= argc) {
|
||
LOG_FATAL("Usage: splc0 [--dump <flags>] <in> [out]");
|
||
return 1;
|
||
}
|
||
|
||
dump_t dump = DUMP_NONE;
|
||
if (strcmp(argv[argi], "--dump") == 0) {
|
||
if (argc < argi + 3) {
|
||
LOG_INFO("splc0: --dump need <flags> <file>");
|
||
return 1;
|
||
}
|
||
|
||
const char *flags = argv[argi + 1];
|
||
if (strcmp(flags, "lex") == 0) {
|
||
dump = DUMP_TOKEN;
|
||
} else if (strcmp(flags, "ast") == 0) {
|
||
dump = DUMP_AST;
|
||
} else if (strcmp(flags, "sema") == 0) {
|
||
dump = DUMP_SEMA;
|
||
} else if (strcmp(flags, "ir") == 0) {
|
||
dump = DUMP_IR;
|
||
} else if (strcmp(flags, "vm") == 0) {
|
||
dump = DUMP_VM;
|
||
} else if (strcmp(flags, "none") == 0) {
|
||
dump = DUMP_NONE;
|
||
} else {
|
||
LOG_FATAL("splc0: --dump takes one of lex|ast|sema|ir|vm|none, got `%s`", flags);
|
||
return 1;
|
||
}
|
||
argi += 2;
|
||
}
|
||
/* splc0 [-g] <in> <out> */
|
||
if (argc < argi + 2) {
|
||
LOG_FATAL("Usage: splc0 <in> <out> [-g] or splc0 --dump <flags> <file>\n");
|
||
return 1;
|
||
}
|
||
int gen_debug = 0;
|
||
if (strcmp(argv[argi], "-g") == 0) {
|
||
gen_debug = 1;
|
||
argi++;
|
||
}
|
||
if (argc < argi + 2) {
|
||
LOG_FATAL("Usage: splc0 <in> <out> [-g] or splc0 --dump <flags> <file>\n");
|
||
return 1;
|
||
}
|
||
long len;
|
||
char *src = read_file(argv[argi], &len);
|
||
if (!src)
|
||
return 1;
|
||
int rc = compile_spl(src, argv[argi], argv[argi + 1], gen_debug, dump);
|
||
free(src);
|
||
return rc;
|
||
}
|