X-Git-Url: https://git.m6w6.name/?p=m6w6%2Fext-psi;a=blobdiff_plain;f=src%2Fcpp.c;h=4982c7842337b8a4d8bcc26780b8af95b73e9cbd;hp=2cf4094e93efbd5ef9303c67e23ce75c97fefecc;hb=a7ac1c0a3c855321f21682c127a4b707de33a303;hpb=b1d875d13e5a8f7600d3df6b44c75e8613cfe509 diff --git a/src/cpp.c b/src/cpp.c index 2cf4094..4982c78 100644 --- a/src/cpp.c +++ b/src/cpp.c @@ -25,6 +25,8 @@ #include "php_psi_stdinc.h" +#include + #include "cpp.h" #include "parser.h" @@ -32,59 +34,109 @@ #define PSI_CPP_PREDEF #include "php_psi_cpp.h" +#include "php_psi.h" + +HashTable psi_cpp_defaults; + +PHP_MINIT_FUNCTION(psi_cpp) +{ + struct psi_parser parser; + struct psi_parser_input *predef; + + PSI_G(search_path) = pemalloc(strlen(PSI_G(directory)) + strlen(psi_cpp_search) + 1 + 1, 1); + sprintf(PSI_G(search_path), "%s:%s", PSI_G(directory), psi_cpp_search); + + if (!psi_parser_init(&parser, psi_error_wrapper, PSI_SILENT)) { + return FAILURE; + } + + if (!(predef = psi_parser_open_string(&parser, psi_cpp_predef, sizeof(psi_cpp_predef) - 1))) { + psi_parser_dtor(&parser); + return FAILURE; + } + + if (!psi_parser_parse(&parser, predef)) { + psi_parser_input_free(&predef); + psi_parser_dtor(&parser); + return FAILURE; + } + psi_parser_input_free(&predef); + + zend_hash_init(&psi_cpp_defaults, 0, NULL, NULL, 1); + zend_hash_copy(&psi_cpp_defaults, &parser.preproc->defs, NULL); + + psi_parser_dtor(&parser); + + return SUCCESS; +} + +PHP_MSHUTDOWN_FUNCTION(psi_cpp) +{ + struct psi_cpp_macro_decl *macro; + + ZEND_HASH_FOREACH_PTR(&psi_cpp_defaults, macro) + { + psi_cpp_macro_decl_free(¯o); + } + ZEND_HASH_FOREACH_END(); + + zend_hash_destroy(&psi_cpp_defaults); + + return SUCCESS; +} + static void free_cpp_def(zval *p) { if (Z_TYPE_P(p) == IS_PTR) { - psi_cpp_macro_decl_free((void *) &Z_PTR_P(p)); + struct psi_cpp_macro_decl *macro = Z_PTR_P(p); + + if (!zend_hash_exists(&psi_cpp_defaults, macro->token->text)) { + psi_cpp_macro_decl_free(¯o); + } } } struct psi_cpp *psi_cpp_init(struct psi_parser *P) { - struct psi_cpp *cpp = calloc(1, sizeof(*cpp)); + struct psi_cpp *cpp = pecalloc(1, sizeof(*cpp), 1); cpp->parser = P; - zend_hash_init(&cpp->defs, 0, NULL, free_cpp_def, 1); zend_hash_init(&cpp->once, 0, NULL, NULL, 1); + zend_hash_init(&cpp->defs, 0, NULL, free_cpp_def, 1); + zend_hash_copy(&cpp->defs, &psi_cpp_defaults, NULL); return cpp; } -bool psi_cpp_load_defaults(struct psi_cpp *cpp) -{ - 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 char *include_flavor[] = { + "include", + "include next", + "include once" +}; +#if PSI_CPP_DEBUG > 1 static int dump_def(zval *p) { struct psi_cpp_macro_decl *decl = Z_PTR_P(p); if (decl) { - dprintf(2, "#define "); + dprintf(2, "PSI: CPP decl -> #define "); psi_cpp_macro_decl_dump(2, decl); dprintf(2, "\n"); } return ZEND_HASH_APPLY_KEEP; } +#endif void psi_cpp_free(struct psi_cpp **cpp_ptr) { if (*cpp_ptr) { struct psi_cpp *cpp = *cpp_ptr; +#if PSI_CPP_DEBUG > 1 + zend_hash_apply(&cpp->defs, dump_def); +#endif *cpp_ptr = NULL; - if (cpp->parser->flags & PSI_DEBUG) { - fprintf(stderr, "PSI: CPP decls:\n"); - zend_hash_apply(&cpp->defs, dump_def); - } zend_hash_destroy(&cpp->defs); zend_hash_destroy(&cpp->once); free(cpp); @@ -95,12 +147,15 @@ 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 and attributes */ - if (token->type == PSI_T_COMMENT || token->type == PSI_T_CPP_ATTRIBUTE) { + if (token->type == PSI_T_COMMENT + || token->type == PSI_T_CPP_ATTRIBUTE) { psi_cpp_tokiter_del_cur(cpp, true); continue; } @@ -108,8 +163,8 @@ static bool psi_cpp_stage1(struct psi_cpp *cpp) /* line continuations */ if (token->type == PSI_T_EOL) { if (esc) { - psi_cpp_tokiter_del_range(cpp, psi_cpp_tokiter_index(cpp) - 1, 2, true); - psi_cpp_tokiter_prev(cpp); + psi_cpp_tokiter_del_prev(cpp, true); + psi_cpp_tokiter_del_cur(cpp, true); esc = false; continue; } @@ -125,6 +180,9 @@ static bool psi_cpp_stage1(struct psi_cpp *cpp) */ if (token->type == PSI_T_WHITESPACE) { + if (name) { + name = false; + } ws = true; psi_cpp_tokiter_del_cur(cpp, true); continue; @@ -160,8 +218,9 @@ static bool psi_cpp_stage1(struct psi_cpp *cpp) struct psi_token *no_ws = psi_token_copy(token); no_ws->type = PSI_T_NO_WHITESPACE; - no_ws->text[0] = '\xA0'; - psi_cpp_tokiter_ins_cur(cpp, no_ws); + zend_string_release(no_ws->text); + no_ws->text = zend_string_init_interned("\xA0", 1, 1); + psi_cpp_tokiter_add(cpp, no_ws); continue; } } @@ -172,6 +231,7 @@ static bool psi_cpp_stage1(struct psi_cpp *cpp) } ws = false; + psi_cpp_tokiter_add_cur(cpp); psi_cpp_tokiter_next(cpp); } @@ -181,133 +241,131 @@ static bool psi_cpp_stage1(struct psi_cpp *cpp) static bool psi_cpp_stage2(struct psi_cpp *cpp) { 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; - do { - 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"); + PSI_DEBUG_PRINT(cpp->parser, "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"); + PSI_DEBUG_PRINT(cpp->parser, "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"); + PSI_DEBUG_PRINT(cpp->parser, "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"); + PSI_DEBUG_PRINT(cpp->parser, "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"); + PSI_DEBUG_PRINT(cpp->parser, "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) { + if (!do_cpp) { #if PSI_CPP_DEBUG - fprintf(stderr, "PSI: CPP skip "); - psi_token_dump(2, current); + PSI_DEBUG_PRINT(cpp->parser, "PSI: CPP skip "); + PSI_DEBUG_DUMP(cpp->parser, psi_token_dump, 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) { - parser_tokens = psi_plist_add(parser_tokens, ¤t); + if (do_cpp) { + parser_tokens = psi_plist_add(parser_tokens, ¤t); - if (is_eol) { - size_t processed = 0; - bool parsed = psi_parser_process(cpp->parser, parser_tokens, &processed); + if (is_eol) { + size_t processed = 0; + bool parsed = psi_parser_process(cpp->parser, parser_tokens, &processed); - /* EOL */ - psi_plist_pop(parser_tokens, NULL); - psi_plist_clean(parser_tokens); - do_cpp = false; + /* EOL */ + psi_plist_pop(parser_tokens, NULL); + psi_plist_clean(parser_tokens); + do_cpp = false; - 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 (!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 > 1 - psi_cpp_tokiter_dump(2, cpp); + PSI_DEBUG_DUMP(cpp->parser, psi_cpp_tokiter_dump, cpp); #endif - continue; - } - - psi_cpp_tokiter_next(cpp); + continue; } - } while (cpp->expanded); + + psi_cpp_tokiter_add_cur(cpp); + psi_cpp_tokiter_next(cpp); + } psi_plist_free(parser_tokens); @@ -317,18 +375,31 @@ static bool psi_cpp_stage2(struct psi_cpp *cpp) bool psi_cpp_process(struct psi_cpp *cpp, struct psi_plist **tokens) { bool parsed = false; - struct psi_cpp temp = *cpp; + struct psi_cpp temp = *cpp; cpp->level = temp.level; + + cpp->tokens.iter = *tokens; + cpp->tokens.next = NULL; - cpp->tokens = *tokens; if (psi_cpp_stage1(cpp) && psi_cpp_stage2(cpp)) { parsed = true; } - *tokens = cpp->tokens; - if (temp.tokens) { - cpp->tokens = temp.tokens; - cpp->index = temp.index; + if (cpp->tokens.next) { + free(cpp->tokens.iter); + cpp->tokens.iter = cpp->tokens.next; + cpp->tokens.next = NULL; + } + + *tokens = cpp->tokens.iter; + + if (temp.tokens.iter) { + cpp->tokens.iter = temp.tokens.iter; + cpp->tokens.next = temp.tokens.next; } + cpp->index = temp.index; + cpp->skip = temp.skip; + cpp->level = temp.level; + cpp->seen = temp.seen; return parsed; } @@ -338,14 +409,21 @@ bool psi_cpp_defined(struct psi_cpp *cpp, struct psi_token *tok) bool defined; if (tok->type == PSI_T_NAME) { - defined = zend_hash_str_exists(&cpp->defs, tok->text, tok->size); + defined = zend_hash_exists(&cpp->defs, tok->text) + || psi_builtin_exists(tok->text); } else { defined = false; } #if PSI_CPP_DEBUG - fprintf(stderr, "PSI: CPP defined -> %s ", defined ? "true" : "false"); - psi_token_dump(2, tok); + PSI_DEBUG_PRINT(cpp->parser, "PSI: CPP defined -> %s ", defined ? "true" : "false"); + if (defined) { + struct psi_cpp_macro_decl *macro = zend_hash_find_ptr(&cpp->defs, tok->text); + if (macro) { + PSI_DEBUG_PRINT(cpp->parser, " @ %s:%u ", macro->token->file->val, macro->token->line); + } + } + PSI_DEBUG_DUMP(cpp->parser, psi_token_dump, tok); #endif return defined; @@ -353,81 +431,136 @@ bool psi_cpp_defined(struct psi_cpp *cpp, struct psi_token *tok) void psi_cpp_define(struct psi_cpp *cpp, struct psi_cpp_macro_decl *decl) { - struct psi_cpp_macro_decl *old = zend_hash_str_find_ptr(&cpp->defs, decl->token->text, decl->token->size); + struct psi_cpp_macro_decl *old = zend_hash_find_ptr(&cpp->defs, decl->token->text); 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); + "'%s' redefined", decl->token->text->val); cpp->parser->error(PSI_DATA(cpp->parser), old->token, PSI_WARNING, - "'%s' previously defined", old->token->text); + "'%s' previously defined", old->token->text->val); } - zend_hash_str_update_ptr(&cpp->defs, decl->token->text, decl->token->size, decl); +#if PSI_CPP_DEBUG + if (decl->exp) { + PSI_DEBUG_PRINT(cpp->parser, "PSI: CPP MACRO num_exp -> "); + } else { + PSI_DEBUG_PRINT(cpp->parser, "PSI: CPP MACRO decl -> "); + } + PSI_DEBUG_DUMP(cpp->parser, psi_cpp_macro_decl_dump, decl); + PSI_DEBUG_PRINT(cpp->parser, "\n"); +#endif + zend_hash_update_ptr(&cpp->defs, decl->token->text, decl); } bool psi_cpp_undef(struct psi_cpp *cpp, struct psi_token *tok) { - return SUCCESS == zend_hash_str_del(&cpp->defs, tok->text, tok->size); + return SUCCESS == zend_hash_del(&cpp->defs, tok->text); } bool psi_cpp_if(struct psi_cpp *cpp, struct psi_cpp_exp *exp) { - if (!psi_num_exp_validate(PSI_DATA(cpp->parser), exp->data.num, NULL, NULL, NULL, NULL, NULL)) { + struct psi_validate_scope scope = {0}; + + scope.cpp = cpp; + if (!psi_num_exp_validate(PSI_DATA(cpp->parser), exp->data.num, &scope)) { return false; } - if (!psi_long_num_exp(exp->data.num, NULL, &cpp->defs)) { + if (!psi_num_exp_get_long(exp->data.num, NULL, cpp)) { return false; } return true; } -static inline bool try_include(struct psi_cpp *cpp, const char *path, bool *parsed) +bool psi_cpp_include(struct psi_cpp *cpp, const struct psi_token *file, unsigned flags) { + bool parsed = false; + char path[PATH_MAX]; + struct psi_plist *tokens; struct psi_parser_input *include; - PSI_DEBUG_PRINT(cpp->parser, "PSI: CPP include trying %s\n", path); + if (!psi_cpp_has_include(cpp, file, flags, path)) { + return false; + } + + if (flags & PSI_CPP_INCLUDE_ONCE) { + if (zend_hash_str_exists(&cpp->once, path, strlen(path))) { + return true; + } + } + + PSI_DEBUG_PRINT(cpp->parser, "PSI: CPP %s opening %s\n", + include_flavor[flags], path); include = psi_parser_open_file(cpp->parser, path, false); - if (include) { - struct psi_plist *tokens; + if (!include) { + return false; + } - PSI_DEBUG_PRINT(cpp->parser, "PSI: CPP include scanning %s\n", path); + zend_hash_str_add_empty_element(&cpp->once, path, strlen(path)); - tokens = psi_parser_scan(cpp->parser, include); - if (tokens) { - *parsed = psi_cpp_process(cpp, &tokens); + PSI_DEBUG_PRINT(cpp->parser, "PSI: CPP include scanning %s\n", path); - if (*parsed) { - ++cpp->expanded; - psi_cpp_tokiter_ins_range(cpp, cpp->index, - psi_plist_count(tokens), psi_plist_eles(tokens)); - free(tokens); - } else { - psi_plist_free(tokens); - } - } - free(include); + tokens = psi_parser_scan(cpp->parser, include); + psi_parser_input_free(&include); - zend_hash_str_add_empty_element(&cpp->once, path, strlen(path)); - return true; + if (!tokens) { + return false; } - return false; + + parsed = psi_cpp_process(cpp, &tokens); + if (!parsed) { + psi_plist_free(tokens); + return false; + } + + psi_cpp_tokiter_add_range(cpp, psi_plist_count(tokens), psi_plist_eles(tokens)); + free(tokens); + + ++cpp->expanded; + return true; } -bool psi_cpp_include(struct psi_cpp *cpp, const char *file, unsigned flags) +#ifndef HAVE_EACCESS +# define eaccess access +#endif +bool psi_cpp_has_include(struct psi_cpp *cpp, const struct psi_token *file, unsigned flags, char *path) { - char path[PATH_MAX]; - bool parsed = false; - int p_len, f_len = strlen(file) - 2; + char temp[PATH_MAX]; - if (file[1] == '/') { - if (PATH_MAX > (p_len = snprintf(path, PATH_MAX, "%.*s", f_len, file + 1))) { - if ((flags & PSI_CPP_INCLUDE_ONCE) && zend_hash_str_exists(&cpp->once, path, p_len)) { - return true; - } - return try_include(cpp, path, &parsed) && parsed; + if (!path) { + path = temp; + } + + if (file->type == PSI_T_QUOTED_STRING && (!(flags & PSI_CPP_INCLUDE_NEXT) || file->text->val[0] == '/')) { + /* first try as is, full or relative path */ + if (file->text->val[0] == '/') { + path = file->text->val; + } else { + char *dir; + size_t len; + + strncpy(path, file->file->val, PATH_MAX); + + dir = dirname(path); + len = strlen(dir); + + assert(len + file->text->len + 1 < PATH_MAX); + + memmove(path, dir, len); + path[len] = '/'; + memcpy(&(path)[len + 1], file->text->val, file->text->len + 1); } - } else { + + PSI_DEBUG_PRINT(cpp->parser, "PSI: CPP %s trying %s\n", + include_flavor[flags], path); + if (0 == eaccess(path, R_OK)) { + return true; + } + } + + /* look through search paths */ + if (file->text->val[0] != '/') { const char *sep; + int p_len; if ((flags & PSI_CPP_INCLUDE_NEXT) && cpp->search) { if ((sep = strchr(cpp->search, ':'))) { @@ -438,8 +571,8 @@ bool psi_cpp_include(struct psi_cpp *cpp, const char *file, unsigned flags) } } - if (!(flags & PSI_CPP_INCLUDE_NEXT) || !cpp->search) { - cpp->search = &psi_cpp_search[0]; + if (!(flags & PSI_CPP_INCLUDE_NEXT)) { + cpp->search = PSI_G(search_path); } do { @@ -448,17 +581,19 @@ bool psi_cpp_include(struct psi_cpp *cpp, const char *file, unsigned flags) 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 + 1))) { - 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 (PATH_MAX > (p_len = snprintf(path, PATH_MAX, "%.*s/%.*s", d_len, cpp->search, (int) file->text->len, file->text->val))) { + PSI_DEBUG_PRINT(cpp->parser, "PSI: CPP %s trying %s\n", + include_flavor[flags], path); + if (0 == eaccess(path, R_OK)) { + return true; + } + } + + if (sep) { + cpp->search = sep + 1; } - cpp->search = sep + 1; } while (sep); } - return parsed; + return false; }