X-Git-Url: https://git.m6w6.name/?p=m6w6%2Fext-psi;a=blobdiff_plain;f=src%2Ftoken.c;h=316c734384843e16e275face12f3c9c81cf382ad;hp=01c45f10e0b1932e5a5bbcdf5920c290c0fe6fab;hb=2559f07a178b681dc240e20153ff5abdff54cb34;hpb=9bcb1df0786a8193d65949c857baaba2f4296e84

diff --git a/src/token.c b/src/token.c
index 01c45f1..316c734 100644
--- a/src/token.c
+++ b/src/token.c
@@ -25,112 +25,134 @@
 
 #include "php_psi_stdinc.h"
 
+#include <ctype.h>
+
+#include <Zend/zend_smart_str.h>
+
 #include "token.h"
 #include "parser.h"
 
-size_t psi_token_alloc_size(size_t token_len, size_t fname_len) {
-	return sizeof(struct psi_token) + token_len + fname_len + 2;
-}
+#ifndef PSI_DEBUG_TOKEN_ALLOC
+#	define PSI_DEBUG_TOKEN_ALLOC 0
+#endif
 
-struct psi_token *psi_token_alloc(struct psi_parser *P) {
+struct psi_token *psi_token_init(token_t token_typ, const char *token_txt,
+		size_t token_len, unsigned col, unsigned line, zend_string *file)
+{
 	struct psi_token *T;
-	size_t token_len, fname_len;
-	token_t token_typ;
 
-	if (P->cur < P->tok) {
-		return NULL;
-	}
-
-	token_typ = P->num;
-	token_len = P->cur - P->tok;
-	fname_len = strlen(P->file.fn);
-
-	T = calloc(1, psi_token_alloc_size(token_len, fname_len));
+	T = calloc(1, sizeof(*T));
 	T->type = token_typ;
-	T->size = token_len;
-	T->text = &T->buf[0];
-	T->file = &T->buf[token_len + 1];
-	T->line = P->line;
-	T->col = P->col;
-
-	memcpy(T->text, P->tok, token_len);
-	memcpy(T->file, P->file.fn, fname_len);
-
+	T->col = col;
+	T->line = line;
+	T->file = zend_string_copy(file);
+	T->text = zend_string_init_interned(token_txt, token_len, 1);
+#if PSI_DEBUG_TOKEN_ALLOC
+	fprintf(stderr, "PSI: token_init %p\n", T);
+#endif
 	return T;
 }
 
-struct psi_token *psi_token_copy(struct psi_token *src) {
-	size_t strct_len = psi_token_alloc_size(src->size, strlen(src->file));
-	struct psi_token *ptr = malloc(strct_len);
+void psi_token_free(struct psi_token **token_ptr) {
+	if (*token_ptr) {
+		struct psi_token *token = *token_ptr;
+#if PSI_DEBUG_TOKEN_ALLOC
+		fprintf(stderr, "PSI: token_free %p\n", token);
+#endif
+		*token_ptr = NULL;
+		zend_string_release(token->text);
+		zend_string_release(token->file);
+		free(token);
+	}
+}
 
-	memcpy(ptr, src, strct_len);
+struct psi_token *psi_token_copy(struct psi_token *src) {
+	struct psi_token *ptr = malloc(sizeof(*ptr));
 
-	ptr->text = &ptr->buf[0];
-	ptr->file = &ptr->buf[ptr->size + 1];
+	*ptr = *src;
+#if PSI_DEBUG_TOKEN_ALLOC
+	fprintf(stderr, "PSI: token_copy %p <= %p\n", ptr, src);
+#endif
+	ptr->text = zend_string_copy(ptr->text);
+	ptr->file = zend_string_copy(ptr->file);
 
 	return ptr;
 }
 
-struct psi_token *psi_token_cat(unsigned argc, ...) {
+void psi_token_copy_ctor(struct psi_token **tok) {
+	*tok = psi_token_copy(*tok);
+}
+
+/* concatenate `argc` number of tokens separated by `sep` into a newly allocated token */
+struct psi_token *psi_token_cat(const char *sep, unsigned argc, ...) {
 	va_list argv;
 	unsigned i;
-	struct psi_token *T = NULL;
+	size_t sep_len = sep ? strlen(sep) : 0;
+	struct psi_token *T = malloc(sizeof(*T));
+	smart_str text = {0};
 
 	va_start(argv, argc);
-	for (i = 0; i < argc; ++i) {
-		struct psi_token *arg = va_arg(argv, struct psi_token *);
 
-		if (T) {
-			size_t token_len = T->size, fname_len = strlen(T->file);
-			struct psi_token *tmp = realloc(T, psi_token_alloc_size(T->size += arg->size + 1, fname_len));
+	*T = *(struct psi_token *) va_arg(argv, struct psi_token *);
+	T->type = PSI_T_NAME;
+	T->file = zend_string_copy(T->file);
 
-			if (tmp) {
-				T = tmp;
-			} else {
-				free(T);
-				va_end(argv);
-				return NULL;
-			}
+	smart_str_append_ex(&text, T->text, 1);
 
-			T->text = &T->buf[0];
-			T->file = &T->buf[T->size + 1];
-			T->buf[token_len] = ' ';
-			memmove(&T->buf[T->size + 1], &T->buf[token_len + 1], fname_len + 1);
-			memcpy(&T->buf[token_len + 1], arg->text, arg->size + 1);
-		} else {
-			T = psi_token_copy(arg);
-			T->type = PSI_T_NAME;
-		}
+	for (i = 1; i < argc; ++i) {
+		struct psi_token *arg = va_arg(argv, struct psi_token *);
+
+		smart_str_appendl_ex(&text, sep, sep_len, 1);
+		smart_str_append_ex(&text, arg->text, 1);
 	}
 	va_end(argv);
 
+	T->text = smart_str_extract(&text);
+
+#if PSI_DEBUG_TOKEN_ALLOC
+	fprintf(stderr, "PSI: token_cat  %p\n", T);
+#endif
 	return T;
 }
 
-struct psi_token *psi_token_append(struct psi_token *T, unsigned argc, ...) {
+/* append `argc` number of C strings separated by `sep` to token `T` */
+struct psi_token *psi_token_append(const char *sep, struct psi_token *T, unsigned argc, ...) {
 	va_list argv;
 	unsigned i;
+	size_t sep_len = sep ? strlen(sep) : 0;
+	smart_str text = {0};
+
+	smart_str_append_ex(&text, T->text, 1);
 
 	va_start(argv, argc);
 	for (i = 0; i < argc; ++i) {
 		char *str = va_arg(argv, char *);
-		size_t str_len = strlen(str), token_len = T->size, fname_len = strlen(T->file);
-
-		T = realloc(T, psi_token_alloc_size(T->size += str_len + 1, fname_len));
-		T->text = &T->buf[0];
-		T->file = &T->buf[T->size + 1];
-		T->buf[token_len] = ' ';
-		memmove(&T->buf[T->size + 1], &T->buf[token_len + 1], fname_len + 1);
-		memcpy(&T->buf[token_len + 1], str, str_len + 1);
+		size_t str_len = strlen(str);
+
+		if (sep_len && text.a) {
+			smart_str_appendl_ex(&text, sep, sep_len, 1);
+		}
+
+		smart_str_appendl_ex(&text, str, str_len, 1);
 	}
 	va_end(argv);
 
+	zend_string_release(T->text);
+	T->text = smart_str_extract(&text);
+
 	return T;
 }
 
 char *php_strtr(char *str, size_t len, char *str_from, char *str_to, size_t trlen);
 struct psi_token *psi_token_translit(struct psi_token *T, char *from, char *to) {
-	php_strtr(T->text, T->size, from, to, MIN(strlen(from), strlen(to)));
+	zend_string *tmp = zend_string_init(T->text->val, T->text->len, 1);
+
+	zend_string_release(T->text);
+	T->text = tmp;
+
+	php_strtr(T->text->val, T->text->len, from, to, MIN(strlen(from), strlen(to)));
+	zend_string_forget_hash_val(T->text);
+
 	return T;
 }
 
@@ -160,5 +182,57 @@ uint64_t psi_token_hash(struct psi_token *t, char *digest_buf) {
 	char loc_buf[48];
 
 	sprintf(loc_buf, "%u%u", t->line, t->col);
-	return psi_hash(digest_buf, t->file, loc_buf, (char *) NULL);
+	return psi_hash(digest_buf, t->file->val, loc_buf, (char *) NULL);
+}
+
+void psi_token_dump(int fd, struct psi_token *t)
+{
+	size_t i;
+
+	dprintf(fd, "TOKEN %p (%u) ", t, t->type);
+	if (t->type == PSI_T_EOF) {
+		dprintf(fd, "EOF");
+	} else {
+		dprintf(fd, "\"");
+		for (i = 0; i < t->text->len; ++i) {
+			switch (t->text->val[i]) {
+			case '\0':
+				dprintf(fd, "\\0");
+				break;
+			case '\a':
+				dprintf(fd, "\\a");
+				break;
+			case '\b':
+				dprintf(fd, "\\b");
+				break;
+			case '\f':
+				dprintf(fd, "\\f");
+				break;
+			case '\n':
+				dprintf(fd, "\\n");
+				break;
+			case '\r':
+				dprintf(fd, "\\r");
+				break;
+			case '\t':
+				dprintf(fd, "\\t");
+				break;
+			case '\v':
+				dprintf(fd, "\\v");
+				break;
+			case '"':
+				dprintf(fd, "\\\"");
+				break;
+			default:
+				if (isprint(t->text->val[i])) {
+					dprintf(fd, "%c", t->text->val[i]);
+				} else {
+					dprintf(fd, "\\x%02hhX", t->text->val[i]);
+				}
+				break;
+			}
+		}
+		dprintf(fd, "\"");
+	}
+	dprintf(fd, " at col %u in %s on line %u\n", t->col, t->file->val, t->line);
 }