X-Git-Url: https://git.m6w6.name/?p=m6w6%2Fext-psi;a=blobdiff_plain;f=src%2Fcpp.c;h=49cf041b6e2b4606d8877cbe30a1f68f5ce330ea;hp=5ab649247d17a4a2f468d7411a825e2c9333541c;hb=fb8f7887c289ae74d6e8dd85d55ae09e6796e890;hpb=c9384515a81cb64d345b299908b2852f51bb8e6e

diff --git a/src/cpp.c b/src/cpp.c
index 5ab6492..49cf041 100644
--- a/src/cpp.c
+++ b/src/cpp.c
@@ -32,64 +32,93 @@
 
 #define PSI_CPP_SEARCH
 #define PSI_CPP_PREDEF
-#include "php_psi_cpp.h"
+#include "php_psi_predef.h"
 
 #include "php_psi.h"
 
-static void free_cpp_def(zval *p)
+HashTable psi_cpp_defaults;
+
+PHP_MINIT_FUNCTION(psi_cpp)
 {
-	if (Z_TYPE_P(p) == IS_PTR) {
-		psi_cpp_macro_decl_free((void *) &Z_PTR_P(p));
+	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;
 	}
-}
 
-struct psi_cpp *psi_cpp_init(struct psi_parser *P)
-{
-	struct psi_cpp *cpp = calloc(1, sizeof(*cpp));
+	if (!(predef = psi_parser_open_string(&parser, psi_cpp_predef, sizeof(psi_cpp_predef) - 1))) {
+		psi_parser_dtor(&parser);
+		return FAILURE;
+	}
 
-	cpp->parser = P;
-	zend_hash_init(&cpp->defs, 0, NULL, free_cpp_def, 1);
-	zend_hash_init(&cpp->once, 0, NULL, NULL, 1);
+	if (!psi_parser_parse(&parser, predef)) {
+		psi_parser_input_free(&predef);
+		psi_parser_dtor(&parser);
+		return FAILURE;
+	}
+	psi_parser_input_free(&predef);
 
-	return cpp;
+	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;
 }
 
-bool psi_cpp_load_defaults(struct psi_cpp *cpp)
+PHP_MSHUTDOWN_FUNCTION(psi_cpp)
 {
-	struct psi_parser_input *predef;
+	struct psi_cpp_macro_decl *macro;
 
-	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;
+	ZEND_HASH_FOREACH_PTR(&psi_cpp_defaults, macro)
+	{
+		psi_cpp_macro_decl_free(&macro);
 	}
+	ZEND_HASH_FOREACH_END();
 
-	return false;
+	zend_hash_destroy(&psi_cpp_defaults);
+
+	return SUCCESS;
 }
 
-#if PSI_CPP_DEBUG
-static int dump_def(zval *p)
+static void free_cpp_def(zval *p)
 {
-	struct psi_cpp_macro_decl *decl = Z_PTR_P(p);
+	if (Z_TYPE_P(p) == IS_PTR) {
+		struct psi_cpp_macro_decl *macro = Z_PTR_P(p);
 
-	if (decl) {
-		dprintf(2, "#define ");
-		psi_cpp_macro_decl_dump(2, decl);
-		dprintf(2, "\n");
+		if (!zend_hash_exists(&psi_cpp_defaults, macro->token->text)) {
+			psi_cpp_macro_decl_free(&macro);
+		}
 	}
-	return ZEND_HASH_APPLY_KEEP;
 }
-#endif
+
+struct psi_cpp *psi_cpp_init(struct psi_parser *P)
+{
+	struct psi_cpp *cpp = pecalloc(1, sizeof(*cpp), 1);
+
+	cpp->parser = P;
+	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;
+}
+
+static char *include_flavor[] = {
+		"include",
+		"include next",
+		"include once"
+};
 
 void psi_cpp_free(struct psi_cpp **cpp_ptr)
 {
 	if (*cpp_ptr) {
 		struct psi_cpp *cpp = *cpp_ptr;
 
-#if PSI_CPP_DEBUG
-		fprintf(stderr, "PSI: CPP decls:\n");
-		zend_hash_apply(&cpp->defs, dump_def);
-#endif
 		*cpp_ptr = NULL;
 		zend_hash_destroy(&cpp->defs);
 		zend_hash_destroy(&cpp->once);
@@ -117,8 +146,8 @@ 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;
 			}
@@ -172,8 +201,9 @@ 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 = psi_string_init_interned("\xA0", 1, 1);
+					psi_cpp_tokiter_add(cpp, no_ws);
 					continue;
 				}
 			}
@@ -184,6 +214,7 @@ static bool psi_cpp_stage1(struct psi_cpp *cpp)
 		}
 
 		ws = false;
+		psi_cpp_tokiter_add_cur(cpp);
 		psi_cpp_tokiter_next(cpp);
 	}
 
@@ -208,7 +239,7 @@ static bool psi_cpp_stage2(struct psi_cpp *cpp)
 			}
 		} 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;
@@ -224,7 +255,7 @@ static bool psi_cpp_stage2(struct psi_cpp *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;
@@ -236,7 +267,7 @@ static bool psi_cpp_stage2(struct psi_cpp *cpp)
 				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;
@@ -248,7 +279,7 @@ static bool psi_cpp_stage2(struct psi_cpp *cpp)
 						} 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
 						}
 					}
@@ -258,18 +289,17 @@ static bool psi_cpp_stage2(struct psi_cpp *cpp)
 				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 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;
@@ -310,12 +340,13 @@ static bool psi_cpp_stage2(struct psi_cpp *cpp)
 			}
 
 #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_add_cur(cpp);
 		psi_cpp_tokiter_next(cpp);
 	}
 
@@ -327,18 +358,31 @@ 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;
 }
@@ -348,18 +392,21 @@ 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_DEBUG_PRINT(cpp->parser, "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);
+		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_token_dump(2, tok);
+	PSI_DEBUG_DUMP(cpp->parser, psi_token_dump, tok);
 #endif
 
 	return defined;
@@ -367,126 +414,134 @@ 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);
 	}
 #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_DEBUG_PRINT(cpp->parser, "PSI: CPP MACRO num_exp -> ");
+	} else {
+		PSI_DEBUG_PRINT(cpp->parser, "PSI: CPP MACRO decl    -> ");
 	}
-	psi_cpp_macro_decl_dump(2, decl);
-	fprintf(stderr, "\n");
+	PSI_DEBUG_DUMP(cpp->parser, psi_cpp_macro_decl_dump, decl);
+	PSI_DEBUG_PRINT(cpp->parser, "\n");
 #endif
-	zend_hash_str_update_ptr(&cpp->defs, decl->token->text, decl->token->size, decl);
+	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)
 {
 	struct psi_validate_scope scope = {0};
 
-	scope.defs = &cpp->defs;
+	scope.cpp = cpp;
 	if (!psi_num_exp_validate(PSI_DATA(cpp->parser), exp->data.num, &scope)) {
 		return false;
 	}
-	if (!psi_num_exp_get_long(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) {
-				size_t num_tokens = psi_plist_count(tokens);
+	tokens = psi_parser_scan(cpp->parser, include);
+	psi_parser_input_free(&include);
 
-				++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);
+	if (!tokens) {
+		return false;
+	}
 
-		zend_hash_str_add_empty_element(&cpp->once, path, strlen(path));
-		return true;
+	parsed = psi_cpp_process(cpp, &tokens);
+	if (!parsed) {
+		psi_plist_free(tokens);
+		return false;
 	}
-	return false;
+
+	psi_cpp_tokiter_add_range(cpp, psi_plist_count(tokens), psi_plist_eles(tokens));
+	free(tokens);
+
+	++cpp->expanded;
+	return true;
 }
 
-static inline void include_path(const struct psi_token *file, char **path)
+#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)
 {
-	if (*file->text == '/') {
-		*path = file->text;
-	} else {
-		char *dir;
-		size_t len;
+	char temp[PATH_MAX];
 
-		strncpy(*path, file->file, PATH_MAX);
+	if (!path) {
+		path = temp;
+	}
 
-		dir = dirname(*path);
-		len = strlen(dir);
+	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;
 
-		assert(len + file->size + 1 < PATH_MAX);
+			strncpy(path, file->file->val, PATH_MAX);
 
-		memmove(*path, dir, len);
-		(*path)[len] = '/';
-		memcpy(&(*path)[len + 1], file->text, file->size + 1);
-	}
-}
+			dir = dirname(path);
+			len = strlen(dir);
 
-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;
+			assert(len + file->text->len + 1 < PATH_MAX);
 
-		include_path(file, &path);
+			memmove(path, dir, len);
+			path[len] = '/';
+			memcpy(&(path)[len + 1], file->text->val, file->text->len + 1);
+		}
 
-		if ((flags & PSI_CPP_INCLUDE_ONCE) && zend_hash_str_exists(&cpp->once, path, f_len)) {
+		PSI_DEBUG_PRINT(cpp->parser, "PSI: CPP %s trying %s\n",
+				include_flavor[flags], path);
+		if (0 == eaccess(path, R_OK)) {
 			return true;
 		}
-		if (try_include(cpp, path, &parsed)) {
-			/* found */
-			return parsed;
-		}
 	}
 
 	/* look through search paths */
-	if (*file->text != '/') {
-		char path[PATH_MAX];
+	if (file->text->val[0] != '/') {
 		const char *sep;
 		int p_len;
 
@@ -499,7 +554,7 @@ bool psi_cpp_include(struct psi_cpp *cpp, const struct psi_token *file, unsigned
 			}
 		}
 
-		if (!(flags & PSI_CPP_INCLUDE_NEXT) || !cpp->search) {
+		if (!(flags & PSI_CPP_INCLUDE_NEXT)) {
 			cpp->search = PSI_G(search_path);
 		}
 
@@ -509,13 +564,12 @@ bool psi_cpp_include(struct psi_cpp *cpp, const struct psi_token *file, unsigned
 			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 (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) {
@@ -524,5 +578,5 @@ bool psi_cpp_include(struct psi_cpp *cpp, const struct psi_token *file, unsigned
 		} while (sep);
 	}
 
-	return parsed;
+	return false;
 }