X-Git-Url: https://git.m6w6.name/?p=m6w6%2Fext-psi;a=blobdiff_plain;f=src%2Fcpp.c;h=5ab649247d17a4a2f468d7411a825e2c9333541c;hp=8fcd5cf986fc3e0f9bf4d9c2e07c1d4fdd1fd4ab;hb=c9384515a81cb64d345b299908b2852f51bb8e6e;hpb=6509a2053456d0e63b6f383b757289d3016ed1a5 diff --git a/src/cpp.c b/src/cpp.c index 8fcd5cf..5ab6492 100644 --- a/src/cpp.c +++ b/src/cpp.c @@ -25,176 +25,91 @@ #include "php_psi_stdinc.h" +#include + #include "cpp.h" #include "parser.h" -static inline bool psi_cpp_level_skipped(struct psi_cpp_data *cpp) -{ - return cpp->skip == cpp->level; -} -static inline void psi_cpp_level_skip(struct psi_cpp_data *cpp) -{ - assert(!cpp->skip); - cpp->skip = cpp->level; -} -static inline void psi_cpp_level_unskip(struct psi_cpp_data *cpp) +#define PSI_CPP_SEARCH +#define PSI_CPP_PREDEF +#include "php_psi_cpp.h" + +#include "php_psi.h" + +static void free_cpp_def(zval *p) { - if (psi_cpp_level_skipped(cpp)) { - cpp->skip = 0; + if (Z_TYPE_P(p) == IS_PTR) { + psi_cpp_macro_decl_free((void *) &Z_PTR_P(p)); } } -static inline bool psi_cpp_level_masked(struct psi_cpp_data *cpp) + +struct psi_cpp *psi_cpp_init(struct psi_parser *P) { - return cpp->seen & (1 << cpp->level); + struct psi_cpp *cpp = calloc(1, sizeof(*cpp)); + + cpp->parser = P; + zend_hash_init(&cpp->defs, 0, NULL, free_cpp_def, 1); + zend_hash_init(&cpp->once, 0, NULL, NULL, 1); + + return cpp; } -static inline void psi_cpp_level_mask(struct psi_cpp_data *cpp) + +bool psi_cpp_load_defaults(struct psi_cpp *cpp) { - assert(!psi_cpp_level_masked(cpp)); - cpp->seen |= (1 << cpp->level); + struct psi_parser_input *predef; + + if ((predef = psi_parser_open_string(cpp->parser, psi_cpp_predef, sizeof(psi_cpp_predef) - 1))) { + bool parsed = psi_parser_parse(cpp->parser, predef); + free(predef); + return parsed; + } + + return false; } -static inline void psi_cpp_level_unmask(struct psi_cpp_data *cpp) + +#if PSI_CPP_DEBUG +static int dump_def(zval *p) { - cpp->seen &= ~(1 << cpp->level); + struct psi_cpp_macro_decl *decl = Z_PTR_P(p); + + if (decl) { + dprintf(2, "#define "); + psi_cpp_macro_decl_dump(2, decl); + dprintf(2, "\n"); + } + return ZEND_HASH_APPLY_KEEP; } +#endif -static void psi_cpp_eval(struct psi_data *D, struct psi_cpp_data *cpp) +void psi_cpp_free(struct psi_cpp **cpp_ptr) { - assert(cpp->exp); - - PSI_DEBUG_PRINT(D, "PSI: CPP EVAL < %s (level=%u, skip=%u)\n", - cpp->exp->token->text, cpp->level, cpp->skip); + if (*cpp_ptr) { + struct psi_cpp *cpp = *cpp_ptr; #if PSI_CPP_DEBUG - psi_cpp_exp_dump(2, cpp->exp); + fprintf(stderr, "PSI: CPP decls:\n"); + zend_hash_apply(&cpp->defs, dump_def); #endif - - switch (cpp->exp->type) { - case PSI_T_ERROR: - if (!cpp->skip) { - D->error(D, cpp->exp->token, PSI_ERROR, "%s", - cpp->exp->data.tok->text); - } - break; - case PSI_T_WARNING: - if (!cpp->skip) { - D->error(D, cpp->exp->token, PSI_WARNING, "%s", - cpp->exp->data.tok->text); - } - break; - case PSI_T_UNDEF: - if (!cpp->skip) { - psi_cpp_undef(cpp, cpp->exp->data.tok); - } - break; - case PSI_T_DEFINE: - if (!cpp->skip) { - psi_cpp_define(cpp, cpp->exp->data.decl); - /* FIXME: copy */ - cpp->exp->data.decl = NULL; - } - break; - case PSI_T_IFDEF: - ++cpp->level; - if (!cpp->skip) { - if (psi_cpp_defined(cpp, cpp->exp->data.tok)) { - psi_cpp_level_mask(cpp); - } else { - psi_cpp_level_skip(cpp); - } - } - break; - case PSI_T_IFNDEF: - ++cpp->level; - if (!cpp->skip) { - if (psi_cpp_defined(cpp, cpp->exp->data.tok)) { - psi_cpp_level_skip(cpp); - } else { - psi_cpp_level_mask(cpp); - } - } - break; - case PSI_T_IF: - ++cpp->level; - if (!cpp->skip) { - if (psi_cpp_if(cpp->exp, &cpp->defs, D)) { - psi_cpp_level_mask(cpp); - } else { - psi_cpp_level_skip(cpp); - } - } - break; - case PSI_T_ENDIF: - if (!cpp->level) { - D->error(D, cpp->exp->token, PSI_WARNING, "Ingoring lone #endif"); - } else { - psi_cpp_level_unskip(cpp); - psi_cpp_level_unmask(cpp); - --cpp->level; - } - break; - case PSI_T_ELSE: - /* FIXME: catch "else" after "else" */ - if (!cpp->level) { - D->error(D, cpp->exp->token, PSI_WARNING, "Ingoring lone #else"); - } else if (psi_cpp_level_skipped(cpp) && !psi_cpp_level_masked(cpp)) { - /* - * if skip is set on this level and the level has - * not been masked yet, then unskip and mask this level - */ - psi_cpp_level_unskip(cpp); - psi_cpp_level_mask(cpp); - } else if (!cpp->skip && psi_cpp_level_masked(cpp)) { - /* - * previous block masked this level - */ - psi_cpp_level_skip(cpp); - } else { - assert(cpp->skip < cpp->level); - } - break; - case PSI_T_ELIF: - if (!cpp->level) { - D->error(D, cpp->exp->token, PSI_WARNING, "Ingoring lone #elif"); - } else if (psi_cpp_level_skipped(cpp) && !psi_cpp_level_masked(cpp)) { - /* - * if skip is set on this level and the level has - * not been masked yet, then unskip and mask this - * level, if the condition evals truthy - */ - if (psi_cpp_if(cpp->exp, &cpp->defs, D)) { - psi_cpp_level_unskip(cpp); - psi_cpp_level_mask(cpp); - } - } else if (!cpp->skip && psi_cpp_level_masked(cpp)) { - /* - * previous block masked this level - */ - psi_cpp_level_skip(cpp); - } else { - assert(cpp->skip < cpp->level); - } - break; - default: - assert(0); - break; + *cpp_ptr = NULL; + zend_hash_destroy(&cpp->defs); + zend_hash_destroy(&cpp->once); + free(cpp); } - - PSI_DEBUG_PRINT(D, "PSI: CPP EVAL > %s (level=%u, skip=%u)\n", - cpp->exp->token->text, cpp->level, cpp->skip); - - psi_cpp_exp_free(&cpp->exp); } -static bool psi_cpp_stage1(struct psi_parser *P, struct psi_cpp_data *cpp) +static bool psi_cpp_stage1(struct psi_cpp *cpp) { bool name = false, define = false, hash = false, eol = true, esc = false, ws = false; + PSI_DEBUG_PRINT(cpp->parser, "PSI: CPP %s\n", "stage1"); + psi_cpp_tokiter_reset(cpp); while (psi_cpp_tokiter_valid(cpp)) { struct psi_token *token = psi_cpp_tokiter_current(cpp); - /* strip comments */ - if (token->type == PSI_T_COMMENT) { + /* strip comments and attributes */ + if (token->type == PSI_T_COMMENT + || token->type == PSI_T_CPP_ATTRIBUTE) { psi_cpp_tokiter_del_cur(cpp, true); continue; } @@ -219,6 +134,9 @@ static bool psi_cpp_stage1(struct psi_parser *P, struct psi_cpp_data *cpp) */ if (token->type == PSI_T_WHITESPACE) { + if (name) { + name = false; + } ws = true; psi_cpp_tokiter_del_cur(cpp, true); continue; @@ -272,148 +190,160 @@ static bool psi_cpp_stage1(struct psi_parser *P, struct psi_cpp_data *cpp) return true; } -static bool psi_cpp_stage2(struct psi_parser *P, struct psi_cpp_data *cpp) +static bool psi_cpp_stage2(struct psi_cpp *cpp) { - do { - bool is_eol = true, do_cpp = false, do_expansion = true, skip_paren = false, skip_all = false; + struct psi_plist *parser_tokens = psi_plist_init((psi_plist_dtor) psi_token_free); + bool is_eol = true, do_cpp = false, do_expansion = true, skip_paren = false, skip_all = false; - psi_cpp_tokiter_reset(cpp); + PSI_DEBUG_PRINT(cpp->parser, "PSI: CPP %s\n", "stage2"); - while (psi_cpp_tokiter_valid(cpp)) { - struct psi_token *current = psi_cpp_tokiter_current(cpp); + psi_cpp_tokiter_reset(cpp); + while (psi_cpp_tokiter_valid(cpp)) { + struct psi_token *current = psi_cpp_tokiter_current(cpp); - if (current->type == PSI_T_HASH) { - if (is_eol) { - do_cpp = true; - is_eol = false; - } - } else if (current->type == PSI_T_EOL) { + if (current->type == PSI_T_HASH) { + if (is_eol) { + do_cpp = true; + is_eol = false; + } + } else if (current->type == PSI_T_EOL) { #if PSI_CPP_DEBUG - fprintf(stderr, "PSI: CPP do_expansion=true, PSI_T_EOL\n"); + fprintf(stderr, "PSI: CPP do_expansion=true, PSI_T_EOL\n"); #endif - is_eol = true; - skip_all = false; - do_expansion = true; - if (!do_cpp) { - psi_cpp_tokiter_del_cur(cpp, true); - continue; - } - } else { - is_eol = false; + is_eol = true; + skip_all = false; + do_expansion = true; + if (!do_cpp) { + psi_cpp_tokiter_del_cur(cpp, true); + continue; + } + } else { + is_eol = false; - if (do_cpp) { - switch (current->type) { - case PSI_T_DEFINE: + if (do_cpp) { + switch (current->type) { + case PSI_T_DEFINE: #if PSI_CPP_DEBUG - fprintf(stderr, "PSI: CPP do_expansion=false, PSI_T_DEFINE, skip_all\n"); + fprintf(stderr, "PSI: CPP do_expansion=false, PSI_T_DEFINE, skip_all\n"); #endif - do_expansion = false; - skip_all = true; - break; - case PSI_T_DEFINED: - skip_paren = true; - /* no break */ - case PSI_T_IFDEF: - case PSI_T_IFNDEF: - case PSI_T_UNDEF: + do_expansion = false; + skip_all = true; + break; + case PSI_T_DEFINED: + skip_paren = true; + /* no break */ + case PSI_T_IFDEF: + case PSI_T_IFNDEF: + case PSI_T_UNDEF: #if PSI_CPP_DEBUG - fprintf(stderr, "PSI: CPP do_expansion=false, PSI_T_{IF{,N},UN}DEF\n"); + fprintf(stderr, "PSI: CPP do_expansion=false, PSI_T_{IF{,N},UN}DEF\n"); #endif - do_expansion = false; - break; - case PSI_T_LPAREN: - if (!skip_all) { - if (skip_paren) { - skip_paren = false; - } else { - do_expansion = true; + do_expansion = false; + break; + case PSI_T_LPAREN: + + if (!skip_all) { + if (skip_paren) { + skip_paren = false; + } else { + do_expansion = true; #if PSI_CPP_DEBUG - fprintf(stderr, "PSI: CPP do_expansion=true, PSI_T_LPAREN, !skip_all, !skip_paren\n"); + fprintf(stderr, "PSI: CPP do_expansion=true, PSI_T_LPAREN, !skip_all, !skip_paren\n"); #endif - } } - break; - case PSI_T_NAME: - break; - default: - do_expansion = !skip_all; + } + break; + case PSI_T_NAME: + break; + default: + do_expansion = !skip_all; #if PSI_CPP_DEBUG - fprintf(stderr, "PSI CPP do_expansion=%s, <- !skip_all\n", do_expansion?"true":"false"); + fprintf(stderr, "PSI: CPP do_expansion=%s, <- !skip_all\n", do_expansion?"true":"false"); #endif - } } } + } - if (cpp->skip) { - /* FIXME: del_range */ - if (!do_cpp) { + if (cpp->skip) { + /* FIXME: del_range */ + if (!do_cpp) { #if PSI_CPP_DEBUG - fprintf(stderr, "PSI: CPP skip "); - psi_token_dump(2, current); + fprintf(stderr, "PSI: CPP skip "); + psi_token_dump(2, current); #endif - psi_cpp_tokiter_del_cur(cpp, true); - continue; - } + psi_cpp_tokiter_del_cur(cpp, true); + continue; } + } - if (do_expansion && current->type == PSI_T_NAME && psi_cpp_tokiter_defined(cpp)) { - bool expanded = false; + if (do_expansion && current->type == PSI_T_NAME && psi_cpp_tokiter_defined(cpp)) { + bool expanded = false; - while (psi_cpp_tokiter_expand(cpp)) { - expanded = true; - } - if (expanded) { - continue; - } + while (psi_cpp_tokiter_expand(cpp)) { + expanded = true; } + if (expanded) { + continue; + } + } - if (do_cpp) { - if (is_eol) { - do_cpp = false; - skip_all = false; - } + if (do_cpp) { + parser_tokens = psi_plist_add(parser_tokens, ¤t); - if (P->flags & PSI_DEBUG) { - fprintf(stderr, "PSI> Parse (%zu) ", psi_cpp_tokiter_index(cpp)); - psi_token_dump(2, current); - } + if (is_eol) { + size_t processed = 0; + bool parsed = psi_parser_process(cpp->parser, parser_tokens, &processed); - psi_parser_proc_parse(P->proc, current->type, current, P); - psi_cpp_tokiter_del_cur(cpp, false); + /* EOL */ + psi_plist_pop(parser_tokens, NULL); + psi_plist_clean(parser_tokens); + do_cpp = false; - if (is_eol) { - psi_parser_proc_parse(P->proc, 0, NULL, P); - psi_cpp_eval(PSI_DATA(P), cpp); + if (!parsed) { + psi_plist_free(parser_tokens); + return false; } + } else { + /* leave EOLs in the input stream, else we might end up + * with a hash not preceded with a new line after include */ + psi_cpp_tokiter_del_cur(cpp, false); + } -#if PSI_CPP_DEBUG - psi_cpp_tokiter_dump(2, cpp); +#if PSI_CPP_DEBUG > 1 + psi_cpp_tokiter_dump(2, cpp); #endif - continue; - } - - psi_cpp_tokiter_next(cpp); + continue; } - } while (cpp->expanded); + + psi_cpp_tokiter_next(cpp); + } + + psi_plist_free(parser_tokens); return true; } -bool psi_cpp_preprocess(struct psi_parser *P, struct psi_cpp_data *cpp) +bool psi_cpp_process(struct psi_cpp *cpp, struct psi_plist **tokens) { - if (!psi_cpp_stage1(P, cpp)) { - return false; + bool parsed = false; + struct psi_cpp temp = *cpp; + + cpp->tokens = *tokens; + if (psi_cpp_stage1(cpp) && psi_cpp_stage2(cpp)) { + parsed = true; } + *tokens = cpp->tokens; - if (!psi_cpp_stage2(P, cpp)) { - return false; + if (temp.tokens) { + cpp->tokens = temp.tokens; + cpp->index = temp.index; } - return true; + return parsed; } -bool psi_cpp_defined(struct psi_cpp_data *cpp, struct psi_token *tok) +bool psi_cpp_defined(struct psi_cpp *cpp, struct psi_token *tok) { bool defined; @@ -425,29 +355,174 @@ bool psi_cpp_defined(struct psi_cpp_data *cpp, struct psi_token *tok) #if PSI_CPP_DEBUG fprintf(stderr, "PSI: CPP defined -> %s ", defined ? "true" : "false"); + if (defined) { + struct psi_cpp_macro_decl *macro = zend_hash_str_find_ptr(&cpp->defs, tok->text, tok->size); + fprintf(stderr, " @ %s:%u ", macro->token->file, macro->token->line); + } psi_token_dump(2, tok); #endif return defined; } -void psi_cpp_define(struct psi_cpp_data *cpp, struct psi_cpp_macro_decl *decl) +void psi_cpp_define(struct psi_cpp *cpp, struct psi_cpp_macro_decl *decl) { - zend_hash_str_add_ptr(&cpp->defs, decl->token->text, decl->token->size, decl); + struct psi_cpp_macro_decl *old = zend_hash_str_find_ptr(&cpp->defs, decl->token->text, decl->token->size); + + if (old && !psi_cpp_macro_decl_equal(old, decl)) { + cpp->parser->error(PSI_DATA(cpp->parser), decl->token, PSI_WARNING, + "'%s' redefined", decl->token->text); + cpp->parser->error(PSI_DATA(cpp->parser), old->token, PSI_WARNING, + "'%s' previously defined", old->token->text); + } +#if PSI_CPP_DEBUG + if (decl->exp) { + fprintf(stderr, "MACRO: num_exp: ", decl->token->text); + } else if (decl->tokens) { + fprintf(stderr, "MACRO: decl : ", decl->token->text); + } + psi_cpp_macro_decl_dump(2, decl); + fprintf(stderr, "\n"); +#endif + zend_hash_str_update_ptr(&cpp->defs, decl->token->text, decl->token->size, decl); } -bool psi_cpp_undef(struct psi_cpp_data *cpp, struct psi_token *tok) +bool psi_cpp_undef(struct psi_cpp *cpp, struct psi_token *tok) { return SUCCESS == zend_hash_str_del(&cpp->defs, tok->text, tok->size); } -bool psi_cpp_if(struct psi_cpp_exp *exp, HashTable *defs, struct psi_data *D) +bool psi_cpp_if(struct psi_cpp *cpp, struct psi_cpp_exp *exp) { - if (!psi_num_exp_validate(D, exp->data.num, NULL, NULL, NULL, NULL, NULL)) { + struct psi_validate_scope scope = {0}; + + scope.defs = &cpp->defs; + if (!psi_num_exp_validate(PSI_DATA(cpp->parser), exp->data.num, &scope)) { return false; } - if (!psi_long_num_exp(exp->data.num, NULL, defs)) { + if (!psi_num_exp_get_long(exp->data.num, NULL, &cpp->defs)) { return false; } return true; } + +static inline bool try_include(struct psi_cpp *cpp, const char *path, bool *parsed) +{ + struct psi_parser_input *include; + + PSI_DEBUG_PRINT(cpp->parser, "PSI: CPP include trying %s\n", path); + + include = psi_parser_open_file(cpp->parser, path, false); + if (include) { + struct psi_plist *tokens; + + PSI_DEBUG_PRINT(cpp->parser, "PSI: CPP include scanning %s\n", path); + + tokens = psi_parser_scan(cpp->parser, include); + if (tokens) { + *parsed = psi_cpp_process(cpp, &tokens); + + if (*parsed) { + size_t num_tokens = psi_plist_count(tokens); + + ++cpp->expanded; + psi_cpp_tokiter_ins_range(cpp, cpp->index, + num_tokens, psi_plist_eles(tokens)); + /* skip already processed tokens */ + cpp->index += num_tokens; + free(tokens); + } else { + psi_plist_free(tokens); + } + } + free(include); + + zend_hash_str_add_empty_element(&cpp->once, path, strlen(path)); + return true; + } + return false; +} + +static inline void include_path(const struct psi_token *file, char **path) +{ + if (*file->text == '/') { + *path = file->text; + } else { + char *dir; + size_t len; + + strncpy(*path, file->file, PATH_MAX); + + dir = dirname(*path); + len = strlen(dir); + + assert(len + file->size + 1 < PATH_MAX); + + memmove(*path, dir, len); + (*path)[len] = '/'; + memcpy(&(*path)[len + 1], file->text, file->size + 1); + } +} + +bool psi_cpp_include(struct psi_cpp *cpp, const struct psi_token *file, unsigned flags) +{ + bool parsed = false; + int f_len = strlen(file->text); + + if (!(flags & PSI_CPP_INCLUDE_NEXT) || *file->text == '/') { + /* first try as is, full or relative path */ + char temp[PATH_MAX], *path = temp; + + include_path(file, &path); + + if ((flags & PSI_CPP_INCLUDE_ONCE) && zend_hash_str_exists(&cpp->once, path, f_len)) { + return true; + } + if (try_include(cpp, path, &parsed)) { + /* found */ + return parsed; + } + } + + /* look through search paths */ + if (*file->text != '/') { + char path[PATH_MAX]; + const char *sep; + int p_len; + + if ((flags & PSI_CPP_INCLUDE_NEXT) && cpp->search) { + if ((sep = strchr(cpp->search, ':'))) { + cpp->search = sep + 1; + } else { + /* point to end of string */ + cpp->search += strlen(cpp->search); + } + } + + if (!(flags & PSI_CPP_INCLUDE_NEXT) || !cpp->search) { + cpp->search = PSI_G(search_path); + } + + do { + int d_len; + + sep = strchr(cpp->search, ':'); + d_len = sep ? sep - cpp->search : strlen(cpp->search); + + if (PATH_MAX > (p_len = snprintf(path, PATH_MAX, "%.*s/%.*s", d_len, cpp->search, f_len, file->text))) { + if ((flags & PSI_CPP_INCLUDE_ONCE) && zend_hash_str_exists(&cpp->once, path, p_len)) { + return true; + } + if (try_include(cpp, path, &parsed)) { + break; + } + } + + if (sep) { + cpp->search = sep + 1; + } + } while (sep); + } + + return parsed; +}