readtags: unescape input field (a.k.a {tagfile}) only if TAG_OUTPUT_M…

…ODE is u-ctags and TAG_OUTPUT_FILESEP is slash Close universal-ctags#3577. Close universal-ctags#3559 except documentation. Signed-off-by: Masatake YAMATO <yamato@redhat.com>
masatake · Dec 17, 2022 · 28d363d · 28d363d
1 parent 6cd0af4
commit 28d363d
Show file tree

Hide file tree

Showing 13 changed files with 198 additions and 10 deletions.
diff --git a/Tmain/readtags-ptags-escaping.d/pseudo-input-b_-u-ctags-mode.tags b/Tmain/readtags-ptags-escaping.d/pseudo-input-b_-u-ctags-mode.tags
@@ -0,0 +1,4 @@
+!_PTAG	value\	/comment/;"
+!_TAG_OUTPUT_MODE	u-ctags	//;"
+!_TAG_OUTPUT_FILESEP	slash	//
+regular_tag	/path/to/file	/^pattern$/;"	kind:
diff --git a/Tmain/readtags-ptags-escaping.d/pseudo-input-bb-u-ctags-mode.tags b/Tmain/readtags-ptags-escaping.d/pseudo-input-bb-u-ctags-mode.tags
@@ -0,0 +1,4 @@
+!_PTAG	value\\	/comment/;"
+!_TAG_OUTPUT_FILESEP	slash	//;"
+!_TAG_OUTPUT_MODE	u-ctags	//;"
+regular_tag	/path/to/file	/^pattern$/;"	kind:
diff --git a/Tmain/readtags-ptags-escaping.d/pseudo-input-bt-u-ctags-mode.tags b/Tmain/readtags-ptags-escaping.d/pseudo-input-bt-u-ctags-mode.tags
@@ -0,0 +1,4 @@
+!_PTAG	value\t	/comment/;"
+!_TAG_OUTPUT_MODE	u-ctags	//;"
+!_TAG_OUTPUT_FILESEP	slash	//
+regular_tag	/path/to/file	/^pattern$/;"	kind:
diff --git a/Tmain/readtags-ptags-escaping.d/regular-input-b_-u-ctags-mode.tags b/Tmain/readtags-ptags-escaping.d/regular-input-b_-u-ctags-mode.tags
@@ -0,0 +1,4 @@
+!_PTAG	value	/comment/;"
+!_TAG_OUTPUT_FILESEP	slash	//;"
+!_TAG_OUTPUT_MODE	u-ctags	//;"
+regular_tag	/path/to/file\	/^pattern$/;"	kind:
diff --git a/Tmain/readtags-ptags-escaping.d/regular-input-bb-u-ctags-mode.tags b/Tmain/readtags-ptags-escaping.d/regular-input-bb-u-ctags-mode.tags
@@ -0,0 +1,4 @@
+!_PTAG	value	/comment/;"
+!_TAG_OUTPUT_MODE	u-ctags	//;"
+!_TAG_OUTPUT_FILESEP	slash	//
+regular_tag	/path/to/file\\	/^pattern$/;"	kind:
diff --git a/Tmain/readtags-ptags-escaping.d/regular-input-bt-u-ctags-mode.tags b/Tmain/readtags-ptags-escaping.d/regular-input-bt-u-ctags-mode.tags
@@ -0,0 +1,5 @@
+!_PTAG	value	/comment/;"
+!_COMMENT	_	/intentionally remote semicolon double quotes char in the next line/;"
+!_TAG_OUTPUT_MODE	u-ctags	//
+!_TAG_OUTPUT_FILESEP	slash	//;"
+regular_tag	/path/to/file\t	/^pattern$/;"	kind:
diff --git a/Tmain/readtags-ptags-escaping.d/regular-input-bt-u-filesep-backslash.tags b/Tmain/readtags-ptags-escaping.d/regular-input-bt-u-filesep-backslash.tags
@@ -0,0 +1,5 @@
+!_PTAG	value	/comment/;"
+!_COMMENT	_	/intentionally remote semicolon double quotes char in the next line/;"
+!_TAG_OUTPUT_MODE	u-ctags	//
+!_TAG_OUTPUT_FILESEP	backslash	//;"
+regular_tag	/path/to/file\t	/^pattern$/;"	kind:
diff --git a/Tmain/readtags-ptags-escaping.d/regular-input-bt-u-no-filesep.tags b/Tmain/readtags-ptags-escaping.d/regular-input-bt-u-no-filesep.tags
@@ -0,0 +1,4 @@
+!_PTAG	value	/comment/;"
+!_COMMENT	_	/intentionally remote semicolon double quotes char in the next line/;"
+!_TAG_OUTPUT_MODE	u-ctags	//
+regular_tag	/path/to/file\t	/^pattern$/;"	kind:
diff --git a/Tmain/readtags-ptags-escaping.d/run.sh b/Tmain/readtags-ptags-escaping.d/run.sh
@@ -21,6 +21,9 @@ for t in INIT									\
 			 regular-input-b_.tags				\
 			 regular-input-bb.tags				\
 			 regular-input-bt.tags				\
+			 regular-input-b_-u-ctags-mode.tags	\
+			 regular-input-bb-u-ctags-mode.tags	\
+			 regular-input-bt-u-ctags-mode.tags	\
 			 regular-kind-b_.tags				\
 			 regular-kind-bb.tags				\
 			 pseudo-name-b_.tags				\
@@ -29,6 +32,11 @@ for t in INIT									\
 			 pseudo-input-b_.tags				\
 			 pseudo-input-bb.tags				\
 			 pseudo-input-bt.tags				\
+			 pseudo-input-b_-u-ctags-mode.tags	\
+			 pseudo-input-bb-u-ctags-mode.tags	\
+			 pseudo-input-bt-u-ctags-mode.tags	\
+			 regular-input-bt-u-filesep-backslash.tags \
+			 regular-input-bt-u-no-filesep.tags \
 		 ; do
 	case $t in
 		INIT)
@@ -49,3 +57,39 @@ for t in INIT									\
 			;;
 	esac
 done
+
+for t in INIT									\
+			 regular-input-b_.tags				\
+			 regular-input-bb.tags				\
+			 regular-input-bt.tags				\
+			 regular-input-b_-u-ctags-mode.tags	\
+			 regular-input-bb-u-ctags-mode.tags	\
+			 regular-input-bt-u-ctags-mode.tags	\
+			 pseudo-input-b_.tags				\
+			 pseudo-input-bb.tags				\
+			 pseudo-input-bt.tags				\
+			 pseudo-input-b_-u-ctags-mode.tags	\
+			 pseudo-input-bb-u-ctags-mode.tags	\
+			 pseudo-input-bt-u-ctags-mode.tags	\
+			 regular-input-bt-u-filesep-backslash.tags \
+			 regular-input-bt-u-no-filesep.tags \
+		 ; do
+	case $t in
+		INIT)
+			;;
+		regular-*)
+			echo "# <no unescaping> $t"
+			${V} ${READTAGS} -e -t $t -l || exit 1
+			echo
+			;;
+		pseudo-*)
+			echo "# <no unescaping> $t"
+			${V} ${READTAGS} -e -t $t -D || exit 1
+			echo
+			;;
+		*)
+			echo "INTERNAL BUG"
+			exit 1
+			;;
+	esac
+done
diff --git a/Tmain/readtags-ptags-escaping.d/stdout-expected.txt b/Tmain/readtags-ptags-escaping.d/stdout-expected.txt
@@ -8,13 +8,22 @@ regular_tag\\	/path/to/file	/^pattern$/
 regular_tag\t	/path/to/file	/^pattern$/
 
 # regular-input-b_.tags
-regular_tag	/path/to/file\\	/^pattern$/
+regular_tag	/path/to/file\	/^pattern$/
 
 # regular-input-bb.tags
-regular_tag	/path/to/file\\\\	/^pattern$/
+regular_tag	/path/to/file\\	/^pattern$/
 
 # regular-input-bt.tags
-regular_tag	/path/to/file\\t	/^pattern$/
+regular_tag	/path/to/file\t	/^pattern$/
+
+# regular-input-b_-u-ctags-mode.tags
+regular_tag	/path/to/file\\	/^pattern$/
+
+# regular-input-bb-u-ctags-mode.tags
+regular_tag	/path/to/file\\	/^pattern$/
+
+# regular-input-bt-u-ctags-mode.tags
+regular_tag	/path/to/file\t	/^pattern$/
 
 # regular-kind-b_.tags
 regular_tag	/path/to/file	/^pattern$/;"	kind:\\
@@ -32,11 +41,80 @@ regular_tag	/path/to/file	/^pattern$/;"	kind:\\
 !_PTAG\t	value	/comment/
 
 # pseudo-input-b_.tags
-!_PTAG	value\\	/comment/
+!_PTAG	value\	/comment/
 
 # pseudo-input-bb.tags
-!_PTAG	value\\\\	/comment/
+!_PTAG	value\\	/comment/
 
 # pseudo-input-bt.tags
-!_PTAG	value\\t	/comment/
+!_PTAG	value\t	/comment/
+
+# pseudo-input-b_-u-ctags-mode.tags
+!_PTAG	value\\	/comment/
+!_TAG_OUTPUT_MODE	u-ctags	//
+!_TAG_OUTPUT_FILESEP	slash	//
+
+# pseudo-input-bb-u-ctags-mode.tags
+!_PTAG	value\\	/comment/
+!_TAG_OUTPUT_FILESEP	slash	//
+!_TAG_OUTPUT_MODE	u-ctags	//
+
+# pseudo-input-bt-u-ctags-mode.tags
+!_PTAG	value\t	/comment/
+!_TAG_OUTPUT_MODE	u-ctags	//
+!_TAG_OUTPUT_FILESEP	slash	//
+
+# regular-input-bt-u-filesep-backslash.tags
+regular_tag	/path/to/file\t	/^pattern$/
+
+# regular-input-bt-u-no-filesep.tags
+regular_tag	/path/to/file\t	/^pattern$/
+
+# <no unescaping> regular-input-b_.tags
+regular_tag	/path/to/file\	/^pattern$/
+
+# <no unescaping> regular-input-bb.tags
+regular_tag	/path/to/file\\	/^pattern$/
+
+# <no unescaping> regular-input-bt.tags
+regular_tag	/path/to/file\t	/^pattern$/
+
+# <no unescaping> regular-input-b_-u-ctags-mode.tags
+regular_tag	/path/to/file\	/^pattern$/
+
+# <no unescaping> regular-input-bb-u-ctags-mode.tags
+regular_tag	/path/to/file\	/^pattern$/
+
+# <no unescaping> regular-input-bt-u-ctags-mode.tags
+regular_tag	/path/to/file		/^pattern$/
+
+# <no unescaping> pseudo-input-b_.tags
+!_PTAG	value\	/comment/
+
+# <no unescaping> pseudo-input-bb.tags
+!_PTAG	value\\	/comment/
+
+# <no unescaping> pseudo-input-bt.tags
+!_PTAG	value\t	/comment/
+
+# <no unescaping> pseudo-input-b_-u-ctags-mode.tags
+!_PTAG	value\	/comment/
+!_TAG_OUTPUT_MODE	u-ctags	//
+!_TAG_OUTPUT_FILESEP	slash	//
+
+# <no unescaping> pseudo-input-bb-u-ctags-mode.tags
+!_PTAG	value\	/comment/
+!_TAG_OUTPUT_FILESEP	slash	//
+!_TAG_OUTPUT_MODE	u-ctags	//
+
+# <no unescaping> pseudo-input-bt-u-ctags-mode.tags
+!_PTAG	value		/comment/
+!_TAG_OUTPUT_MODE	u-ctags	//
+!_TAG_OUTPUT_FILESEP	slash	//
+
+# <no unescaping> regular-input-bt-u-filesep-backslash.tags
+regular_tag	/path/to/file\t	/^pattern$/
+
+# <no unescaping> regular-input-bt-u-no-filesep.tags
+regular_tag	/path/to/file\t	/^pattern$/
 
diff --git a/extra-cmds/printtags.c b/extra-cmds/printtags.c
@@ -106,6 +106,7 @@ static void tagsPrintTag (const tagEntry *entry,
 						  int printingExtensionFields,
 						  int printingLineNumber,
 						  int printingWithEscaping,
+						  int printingFileInUniversalCtagsMode,
 						  int pseudoTag,
 						  int  (* print_str) (const char *, void *),
 						  int  (* print_char) (int, void *),
@@ -142,8 +143,15 @@ static void tagsPrintTag (const tagEntry *entry,
 					print_str, print_char, outfp);
 
 	print_char ('\t', outfp);
-	printValue  (entry->file, printingWithEscaping,
-				 print_str, print_char, outfp);
+	{
+		/* Universal Ctags on Windows converts \ , the file name separator
+		 * on the platform, to /; we can reliably expect \ characters in
+		 * entry->file is for escaping. So we can do unescape here.
+		 */
+		int esc = printingWithEscaping && printingFileInUniversalCtagsMode;
+		printValue  (entry->file, esc,
+					 print_str, print_char, outfp);
+	}
 	print_char ('\t', outfp);
 	print_str (entry->address.pattern, outfp);
 
@@ -201,6 +209,7 @@ extern int tagsPrint (const tagEntry *entry,
 				  opt->extensionFields,
 				  opt->lineNumber,
 				  opt->escaping,
+				  opt->outputMode == universalOutputMode,
 				  0,
 				  procs->printStr,
 				  procs->printChar,
@@ -218,6 +227,7 @@ extern int tagsPrintPseudoTag (const tagEntry *entry,
 				  opt->extensionFields,
 				  opt->lineNumber,
 				  opt->escaping,
+				  opt->outputMode == universalOutputMode,
 				  1,
 				  procs->printStr,
 				  procs->printChar,

diff --git a/extra-cmds/printtags.h b/extra-cmds/printtags.h
@@ -27,12 +27,16 @@ typedef struct {
 	int  (* printChar) (int, void *);
 } tagPrintProcs;
 
+typedef enum eTagPrintOutputMode {
+	exuberantOutputMode,
+	universalOutputMode
+} tagPrintOutputMode;
+
 typedef struct {
-	/* TODO: int version; */
 	int extensionFields;
 	int lineNumber;
 	int escaping;
-	/* version 1 */
+	tagPrintOutputMode outputMode;
 } tagPrintOptions;
 
 /*

diff --git a/extra-cmds/readtags-cmd.c b/extra-cmds/readtags-cmd.c
@@ -327,6 +327,20 @@ static tagFile *openTags (const char *const filePath, tagFileInfo *const info)
 	return tagsOpen (filePath, info);
 }
 
+static tagPrintOutputMode getTagOutputMode (tagFile *const file)
+{
+	tagEntry entry;
+
+	if ((tagsFindPseudoTag (file, &entry,
+						   "!_TAG_OUTPUT_MODE", TAG_FULLMATCH) == TagSuccess)
+		&& (strcmp(entry.file, "u-ctags") == 0)
+		&& (tagsFindPseudoTag (file, &entry,
+						   "!_TAG_OUTPUT_FILESEP", TAG_FULLMATCH) == TagSuccess)
+		&& (strcmp(entry.file, "slash") == 0))
+		return universalOutputMode;
+	return exuberantOutputMode;
+}
+
 static void findTag (const char *const name, readOptions *readOpts,
 					 tagPrintOptions *printOpts)
 {
@@ -343,6 +357,8 @@ static void findTag (const char *const name, readOptions *readOpts,
 		exit (1);
 	}
 
+	printOpts->outputMode = getTagOutputMode (file);
+
 	if (readOpts->sortOverride)
 	{
 		if (tagsSetSortType (file, readOpts->sortMethod) != TagSuccess)
@@ -391,6 +407,8 @@ static void listTags (int pseudoTags, tagPrintOptions *printOpts)
 		exit (1);
 	}
 
+	printOpts->outputMode = getTagOutputMode (file);
+
 	if (pseudoTags)
 	{
 		if (tagsFirstPseudoTag (file, &entry) == TagSuccess)