Merge pull request #47 from masatake/unescaping-input-field

Unescaping input field only when !_TAG_OUTPUT_MODE is "u-ctags"

See universal-ctags/ctags#3559 (comment) especially universal-ctags/ctags#3559 (comment) .

Author: masatake

.gitignore

@@ -41,3 +41,6 @@ tests/test-api-tagsSetSortType
 tests/test-fix-large-tags
 tests/test-fix-null-deref
 tests/test-fix-unescaping
+tests/test-fix-unescaping-input-fields
+tests/test-fix-unescaping-input-fields-exuberant
+tests/test-fix-unescaping-input-fields-no-mode

NEWS.md

@@ -1,3 +1,10 @@
+# Version ???
+
+- read input fields, values at the second column in a tag file, with
+  unescaping if !_TAG_OUTPUT_MODE in the tag file is "u-ctags".
+
+- LT_VERSION ?:?:?
+
 # Version 0.2.1
 
 - use "m" mode flag of fopen only when compiling with glibc 2.3 or higher.

readtags.c

@@ -42,9 +42,12 @@ typedef off_t rt_off_t;
 /* Information about current tag file */
 struct sTagFile {
 		/* has the file been opened and this structure initialized? */
-	short initialized;
+	unsigned char initialized;
 		/* format of tag file */
-	short format;
+	unsigned char format;
+		/* 1 "u-ctags" is set to !_TAG_OUTPUT_MODE pseudo tag
+		 * in the tags file. */
+	unsigned char inputUCtagsMode;
 		/* how is the tag file sorted? */
 	tagSortType sortMethod;
 		/* pointer to file structure */
@@ -519,26 +522,15 @@ static unsigned int countContinuousBackslashesBackward(const char *from,
 	return counter;
 }
 
-static tagResult parseTagLine (tagFile *file, tagEntry *const entry, int *err)
+/* When unescaping, the input string becomes shorter.
+ * e.g. \t occupies two bytes on the tag file.
+ * It is converted to 0x9 and occupies one byte.
+ * memmove called here for shortening the line
+ * buffer. */
+static char *unescapeInPlace (char *q, char **tab, size_t *p_len)
 {
-	int i;
-	char *p = file->line.buffer;
-	size_t p_len = strlen (p);
-	char *tab = strchr (p, TAB);
+	char *p = q;
 
-	memset(entry, 0, sizeof(*entry));
-
-	entry->name = p;
-	if (tab != NULL)
-	{
-		*tab = '\0';
-	}
-
-	/* When unescaping, the input string becomes shorter.
-	 * e.g. \t occupies two bytes on the tag file.
-	 * It is converted to 0x9 and occupies one byte.
-	 * memmove called here for shortening the line
-	 * buffer. */
 	while (*p != '\0')
 	{
 		const char *next = p;
@@ -547,21 +539,50 @@ static tagResult parseTagLine (tagFile *file, tagEntry *const entry, int *err)
 
 		*p = (char) ch;
 		p++;
-		p_len -= skip;
+		*p_len -= skip;
 		if (skip > 1)
 		{
 			/* + 1 is for moving the area including the last '\0'. */
-			memmove (p, next, p_len + 1);
-			if (tab)
-				tab -= skip - 1;
+			memmove (p, next, *p_len + 1);
+			if (*tab)
+				*tab -= skip - 1;
 		}
 	}
 
+	return p;
+}
+
+static tagResult parseTagLine (tagFile *file, tagEntry *const entry, int *err)
+{
+	int i;
+	char *p = file->line.buffer;
+	size_t p_len = strlen (p);
+	char *tab = strchr (p, TAB);
+
+	memset(entry, 0, sizeof(*entry));
+
+	entry->name = p;
+	if (tab != NULL)
+	{
+		*tab = '\0';
+	}
+
+	p = unescapeInPlace (p, &tab, &p_len);
+
 	if (tab != NULL)
 	{
 		p = tab + 1;
 		entry->file = p;
 		tab = strchr (p, TAB);
+		if (file->inputUCtagsMode)
+		{
+			if (tab != NULL)
+			{
+				*tab = '\0';
+			}
+			p = unescapeInPlace (p, &tab, &p_len);
+		}
+
 		if (tab != NULL)
 		{
 			int fieldsPresent;
@@ -717,7 +738,7 @@ static tagResult readPseudoTags (tagFile *const file, tagFileInfo *const info)
 					err = TagErrnoUnexpectedFormat;
 					break;
 				}
-				file->format = (short) m;
+				file->format = (unsigned char) m;
 			}
 			else if (strcmp (key, "TAG_PROGRAM_AUTHOR") == 0)
 			{
@@ -755,6 +776,11 @@ static tagResult readPseudoTags (tagFile *const file, tagFileInfo *const info)
 					break;
 				}
 			}
+			else if (strcmp (key, "TAG_OUTPUT_MODE") == 0)
+			{
+				if (strcmp (value, "u-ctags") == 0)
+					file->inputUCtagsMode = 1;
+			}
 
 			info->file.format     = file->format;
 			info->file.sort       = file->sortMethod;

tests/Makefile.am

@@ -11,6 +11,9 @@ TESTS = \
 	test-fix-unescaping \
 	test-fix-null-deref \
 	test-fix-large-tags \
+	test-fix-unescaping-input-fields \
+	test-fix-unescaping-input-fields-exuberant \
+	test-fix-unescaping-input-fields-no-mode \
 	\
 	$(NULL)
 
@@ -27,10 +30,13 @@ check_PROGRAMS = \
 	test-fix-unescaping \
 	test-fix-null-deref \
 	test-fix-large-tags \
+	test-fix-unescaping-input-fields \
+	test-fix-unescaping-input-fields-exuberant \
+	test-fix-unescaping-input-fields-no-mode \
 	\
 	$(NULL)
 
-EXTRA_DIST =
+EXTRA_DIST = test-fields.h
 
 AM_CPPFLAGS = -I $(top_srcdir) -DTAG_NO_COMPAT_SORT_TYPE
 AM_CFLAGS = $(GCOV_CFLAGS)
@@ -90,3 +96,15 @@ EXTRA_DIST += null-deref.tags
 
 test_fix_large_tags = test-fix-large-tags.c
 test_fix_large_tags_DEPENDENCIES = $(DEPS)
+
+test_fix_unescaping_input_fields = test-fix-unescaping-input-fields.c
+test_fix_unescaping_input_fields_DEPENDENCIES = $(DEPS)
+EXTRA_DIST += unescaping-input-fields.tags
+
+test_fix_unescaping_input_fields_exuberant = test-fix-unescaping-input-fields-exuberant.c
+test_fix_unescaping_input_fields_exuberant_DEPENDENCIES = $(DEPS)
+EXTRA_DIST += unescaping-input-fields-exuberant.tags
+
+test_fix_unescaping_input_fields_no_mode = test-fix-unescaping-input-fields-no-mode.c
+test_fix_unescaping_input_fields_no_mode_DEPENDENCIES = $(DEPS)
+EXTRA_DIST += unescaping-input-fields-no-mode.tags

tests/test-fields.h

@@ -0,0 +1,64 @@
+/*
+*   Copyright (c) 2022, Masatake YAMATO
+*
+*   This source code is released into the public domain.
+*
+*   Testing the fix for handling unescaping
+*/
+
+#ifndef TEST_FIELDS_H
+#define TEST_FIELDS_H
+
+#define NEXT()										\
+	r = tagsNext (t, &e);							\
+	if (r != TagSuccess)							\
+	{												\
+		fprintf (stderr, "error in tagsNext\n");	\
+		return 1;									\
+	}												\
+	do {} while (0)
+
+#define CHECK(EXP,FIELD)												\
+	if (strcmp (EXP, e.FIELD) != 0)										\
+	{																	\
+		fprintf (stderr, "unexpected " #FIELD "(expected: %s, actual: %s) in tagsFirst\n", \
+				 EXP, e.FIELD);											\
+		return 1;														\
+	} \
+	do {} while (0)
+
+#define CHECK_X(FIELD,EXP)												\
+	{																	\
+		unsigned short i;												\
+		for (i = 0; i < e.fields.count; i++)							\
+		{																\
+			if (strcmp (e.fields.list [i].key, FIELD) == 0)				\
+			{															\
+				if (strcmp(e.fields.list [i].value, EXP) == 0)			\
+					break;												\
+				else													\
+				{														\
+					fprintf (stderr, "unexpected " #FIELD "(expected: %s, actual: %s) in tagsFirst\n", \
+							 EXP, e.fields.list [i].value);				\
+					return 1;											\
+				}														\
+			}															\
+		}																\
+		if (i >= e.fields.count)										\
+		{																\
+			fprintf (stderr, "unexpected " #FIELD " field is not found in tagsFirst (count: %u)\n", \
+					 e.fields.count);									\
+			return 1;													\
+		}																\
+	}
+
+#define CHECK3(NAME,FILE,PAT)					\
+	CHECK ((NAME), name);						\
+	CHECK ((FILE), file);						\
+	CHECK ((PAT),  address.pattern)
+
+#define NEXT_CHECK3(NAME,FILE,PAT)				\
+	NEXT ();									\
+	CHECK3 (NAME,FILE,PAT)
+
+#endif	/* !TEST_FIELDS_H */

tests/test-fix-unescaping-input-fields-exuberant.c

@@ -0,0 +1,85 @@
+/*
+*   Copyright (c) 2022, Masatake YAMATO
+*
+*   This source code is released into the public domain.
+*
+*   Testing the fix for handling unescaping
+*/
+
+#include "readtags.h"
+
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <unistd.h>
+
+#include "test-fields.h"
+
+
+int
+main (void)
+{
+	char *srcdir = getenv ("srcdir");
+	if (srcdir)
+	{
+		if (chdir (srcdir) == -1)
+		{
+			perror ("chdir");
+			return 99;
+		}
+	}
+
+	tagFile *t;
+	tagFileInfo info;
+
+	const char *tags0 = "./unescaping-input-fields-exuberant.tags";
+	t = tagsOpen (tags0, &info);
+	if (t == NULL
+		|| info.status.opened == 0)
+	{
+		fprintf (stderr, "unexpected result (t: %p, opened: %d)\n",
+				 t, info.status.opened);
+		return 1;
+	}
+	fprintf (stderr, "ok\n");
+
+	tagEntry e;
+	tagResult r;
+
+	r = tagsFirst (t, &e);
+	if (r != TagSuccess)
+	{
+		fprintf (stderr, "error in tagsFirst\n");
+		return 1;
+	}
+
+	CHECK3 ("tab0", "\\tabc", "/^tab0$/");
+	NEXT_CHECK3 ("tab1", "a\\tbc", "/^tab1$/");
+	NEXT_CHECK3 ("tab2", "ab\\tc", "/^tab2$/");
+	NEXT_CHECK3 ("tab3", "abc\\t", "/^tab3$/");
+	NEXT_CHECK3 ("tab4", "\\\\abc", "/^tab4$/");
+	NEXT_CHECK3 ("tab5", "a\\\\bc", "/^tab5$/");
+	NEXT_CHECK3 ("tab6", "ab\\\\c", "/^tab6$/");
+	NEXT_CHECK3 ("tab7", "abc\\\\", "/^tab7$/");
+	NEXT_CHECK3 ("tab8", "\\nabc", "/^tab8$/");
+	NEXT_CHECK3 ("tab9", "a\\nbc", "/^tab9$/");
+	NEXT_CHECK3 ("taba", "ab\\nc", "/^taba$/");
+	NEXT_CHECK3 ("tabb", "abc\\n", "/^tabb$/");
+	NEXT_CHECK3 ("tabc", "\\n\\\\abc", "/^tabc$/");
+	NEXT_CHECK3 ("tabd", "\\\\\\nabc", "/^tabd$/");
+	NEXT_CHECK3 ("tabe", "a\\n\\\\bc", "/^tabe$/");
+	NEXT_CHECK3 ("tabf", "a\\\\\\nbc", "/^tabf$/");
+	NEXT_CHECK3 ("tabg", "abc\\n\\\\", "/^tabg$/");
+	NEXT_CHECK3 ("tabh", "abc\\\\\\n", "/^tabh$/");
+	NEXT_CHECK3 ("tabi", "\\t\\\\\\n",  "/^tabi$/");
+	NEXT_CHECK3 ("tabj", "ab\\t\\\\\\nc",  "/^tabj$/");
+
+	r = tagsClose(t);
+	if (r != TagSuccess)
+	{
+		fprintf (stderr, "error in tagsClose\n");
+		return 1;
+	}
+
+	return 0;
+}