X-Git-Url: https://git.m6w6.name/?p=m6w6%2Fext-psi;a=blobdiff_plain;f=src%2Fcpp.c;h=5ab649247d17a4a2f468d7411a825e2c9333541c;hp=0c64246098ae8ba42844bea3e52170a8365fa329;hb=c9384515a81cb64d345b299908b2852f51bb8e6e;hpb=42f44eb5bf4ecd36e26e051fada79d861d0f92d2

diff --git a/src/cpp.c b/src/cpp.c
index 0c64246..5ab6492 100644
--- a/src/cpp.c
+++ b/src/cpp.c
@@ -25,6 +25,8 @@
 
 #include "php_psi_stdinc.h"
 
+#include <libgen.h>
+
 #include "cpp.h"
 #include "parser.h"
 
@@ -32,6 +34,8 @@
 #define PSI_CPP_PREDEF
 #include "php_psi_cpp.h"
 
+#include "php_psi.h"
+
 static void free_cpp_def(zval *p)
 {
 	if (Z_TYPE_P(p) == IS_PTR) {
@@ -44,8 +48,8 @@ struct psi_cpp *psi_cpp_init(struct psi_parser *P)
 	struct psi_cpp *cpp = calloc(1, sizeof(*cpp));
 
 	cpp->parser = P;
-	ALLOC_HASHTABLE(cpp->defs);
-	zend_hash_init(cpp->defs, 0, NULL, free_cpp_def, 1);
+	zend_hash_init(&cpp->defs, 0, NULL, free_cpp_def, 1);
+	zend_hash_init(&cpp->once, 0, NULL, NULL, 1);
 
 	return cpp;
 }
@@ -63,6 +67,7 @@ bool psi_cpp_load_defaults(struct psi_cpp *cpp)
 	return false;
 }
 
+#if PSI_CPP_DEBUG
 static int dump_def(zval *p)
 {
 	struct psi_cpp_macro_decl *decl = Z_PTR_P(p);
@@ -74,19 +79,20 @@ static int dump_def(zval *p)
 	}
 	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
+		fprintf(stderr, "PSI: CPP decls:\n");
+		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);
-		FREE_HASHTABLE(cpp->defs);
+		zend_hash_destroy(&cpp->defs);
+		zend_hash_destroy(&cpp->once);
 		free(cpp);
 	}
 }
@@ -95,12 +101,15 @@ 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 */
-		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;
 		}
@@ -125,6 +134,9 @@ 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;
@@ -181,126 +193,131 @@ 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");
+			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) {
-				parser_tokens = psi_plist_add(parser_tokens, &current);
-				psi_cpp_tokiter_del_cur(cpp, false);
+		if (do_cpp) {
+			parser_tokens = psi_plist_add(parser_tokens, &current);
 
-				if (is_eol) {
-					size_t processed = 0;
+			if (is_eol) {
+				size_t processed = 0;
+				bool parsed = psi_parser_process(cpp->parser, parser_tokens, &processed);
 
-					if (!psi_parser_process(cpp->parser, parser_tokens, &processed)) {
-						psi_plist_free(parser_tokens);
-						return false;
-					}
-					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 PSI_CPP_DEBUG > 1
-				psi_cpp_tokiter_dump(2, cpp);
+			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);
 
@@ -309,17 +326,21 @@ 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;
 
-	temp.tokens = *tokens;
+	cpp->tokens = *tokens;
+	if (psi_cpp_stage1(cpp) && psi_cpp_stage2(cpp)) {
+		parsed = true;
+	}
+	*tokens = cpp->tokens;
 
-	if (psi_cpp_stage1(&temp) && psi_cpp_stage2(&temp)) {
-		*tokens = temp.tokens;
-		return true;
+	if (temp.tokens) {
+		cpp->tokens = temp.tokens;
+		cpp->index = temp.index;
 	}
 
-	*tokens = temp.tokens;
-	return false;
+	return parsed;
 }
 
 bool psi_cpp_defined(struct psi_cpp *cpp, struct psi_token *tok)
@@ -327,13 +348,17 @@ 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_str_exists(&cpp->defs, tok->text, tok->size);
 	} else {
 		defined = false;
 	}
 
 #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
 
@@ -342,7 +367,7 @@ 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_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,
@@ -350,20 +375,32 @@ void psi_cpp_define(struct psi_cpp *cpp, struct psi_cpp_macro_decl *decl)
 		cpp->parser->error(PSI_DATA(cpp->parser), old->token, PSI_WARNING,
 				"'%s' previously defined", old->token->text);
 	}
-	zend_hash_str_update_ptr(cpp->defs, decl->token->text, decl->token->size, decl);
+#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 *cpp, struct psi_token *tok)
 {
-	return SUCCESS == zend_hash_str_del(cpp->defs, tok->text, tok->size);
+	return SUCCESS == zend_hash_str_del(&cpp->defs, tok->text, tok->size);
 }
 
 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.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, cpp->defs)) {
+	if (!psi_num_exp_get_long(exp->data.num, NULL, &cpp->defs)) {
 		return false;
 	}
 	return true;
@@ -383,38 +420,87 @@ static inline bool try_include(struct psi_cpp *cpp, const char *path, bool *pars
 
 		tokens = psi_parser_scan(cpp->parser, include);
 		if (tokens) {
-			if ((*parsed = psi_cpp_process(cpp, &tokens))) {
-				psi_cpp_tokiter_ins_range(cpp, psi_cpp_tokiter_index(cpp),
-						psi_plist_count(tokens), psi_plist_eles(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);
 			}
-			psi_plist_free(tokens);
 		}
 		free(include);
+
+		zend_hash_str_add_empty_element(&cpp->once, path, strlen(path));
 		return true;
 	}
 	return false;
 }
 
-bool psi_cpp_include(struct psi_cpp *cpp, const char *file, unsigned flags)
+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)
 {
-	char path[PATH_MAX];
 	bool parsed = false;
-	int f_len = strlen(file) - 2;
+	int f_len = strlen(file->text);
 
-	if (file[1] == '/' && PATH_MAX > snprintf(path, PATH_MAX, "%.*s", f_len, file + 1)) {
-		return try_include(cpp, path, &parsed) && parsed;
-	} else {
+	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 {
-				cpp->search += strlen(cpp->search); /* point to end of string */
+				/* point to end of string */
+				cpp->search += strlen(cpp->search);
 			}
 		}
+
 		if (!(flags & PSI_CPP_INCLUDE_NEXT) || !cpp->search) {
-			cpp->search = &psi_cpp_search[0];
+			cpp->search = PSI_G(search_path);
 		}
 
 		do {
@@ -423,12 +509,18 @@ 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 > snprintf(path, PATH_MAX, "%.*s/%.*s", d_len, cpp->search, f_len, file + 1)) {
+			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;
 				}
 			}
-			cpp->search = sep + 1;
+
+			if (sep) {
+				cpp->search = sep + 1;
+			}
 		} while (sep);
 	}