git apply: option to ignore whitespace differences

Introduce --ignore-whitespace option and corresponding config bool to
ignore whitespace differences while applying patches, akin to the
'patch' program.

'git am', 'git rebase' and the bash git completion are made aware of
this option.

Signed-off-by: Giuseppe Bilotta <giuseppe.bilotta@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>

Documentation/config.txt

@@ -461,6 +461,14 @@ it will be treated as a shell command.  For example, defining
 executed from the top-level directory of a repository, which may
 not necessarily be the current directory.
 
+apply.ignorewhitespace::
+	When set to 'change', tells 'git-apply' to ignore changes in
+	whitespace, in the same way as the '--ignore-space-change'
+	option.
+	When set to one of: no, none, never, false tells 'git-apply' to
+	respect all whitespace differences.
+	See linkgit:git-apply[1].
+
 apply.whitespace::
 	Tells 'git-apply' how to handle whitespaces, in the same way
 	as the '--whitespace' option. See linkgit:git-apply[1].

Documentation/git-am.txt

@@ -11,7 +11,7 @@ SYNOPSIS
 [verse]
 'git am' [--signoff] [--keep] [--utf8 | --no-utf8]
 	 [--3way] [--interactive] [--committer-date-is-author-date]
-	 [--ignore-date]
+	 [--ignore-date] [--ignore-space-change | --ignore-whitespace]
 	 [--whitespace=<option>] [-C<n>] [-p<n>] [--directory=<dir>]
 	 [--reject] [-q | --quiet]
 	 [<mbox> | <Maildir>...]
@@ -65,6 +65,9 @@ default.   You can use `--no-utf8` to override this.
 	it is supposed to apply to and we have those blobs
 	available locally.
 
+--ignore-date::
+--ignore-space-change::
+--ignore-whitespace::
 --whitespace=<option>::
 -C<n>::
 -p<n>::

Documentation/git-apply.txt

@@ -13,6 +13,7 @@ SYNOPSIS
 	  [--apply] [--no-add] [--build-fake-ancestor=<file>] [-R | --reverse]
 	  [--allow-binary-replacement | --binary] [--reject] [-z]
 	  [-pNUM] [-CNUM] [--inaccurate-eof] [--recount] [--cached]
+	  [--ignore-space-change | --ignore-whitespace ]
 	  [--whitespace=<nowarn|warn|fix|error|error-all>]
 	  [--exclude=PATH] [--include=PATH] [--directory=<root>]
 	  [--verbose] [<patch>...]
@@ -149,6 +150,14 @@ patch to each path is used.  A patch to a path that does not match any
 include/exclude pattern is used by default if there is no include pattern
 on the command line, and ignored if there is any include pattern.
 
+--ignore-space-change::
+--ignore-whitespace::
+	When applying a patch, ignore changes in whitespace in context
+	lines if necessary.
+	Context lines will preserve their whitespace, and they will not
+	undergo whitespace fixing regardless of the value of the
+	`--whitespace` option. New lines will still be fixed, though.
+
 --whitespace=<action>::
 	When applying a patch, detect a new or modified line that has
 	whitespace errors.  What are considered whitespace errors is
@@ -205,6 +214,10 @@ running `git apply --directory=modules/git-gui`.
 Configuration
 -------------
 
+apply.ignorewhitespace::
+	Set to 'change' if you want changes in whitespace to be ignored by default.
+	Set to one of: no, none, never, false if you want changes in
+	whitespace to be significant.
 apply.whitespace::
 	When no `--whitespace` flag is given from the command
 	line, this configuration item is used as the default.

Documentation/git-rebase.txt

@@ -268,8 +268,9 @@ OPTIONS
 	exit with the message "Current branch is up to date" in such a
 	situation.
 
+--ignore-whitespace::
 --whitespace=<option>::
-	This flag is passed to the 'git-apply' program
+	These flag are passed to the 'git-apply' program
 	(see linkgit:git-apply[1]) that applies the patch.
 	Incompatible with the --interactive option.
 

builtin-apply.c

@@ -61,6 +61,13 @@ static enum ws_error_action {
 static int whitespace_error;
 static int squelch_whitespace_errors = 5;
 static int applied_after_fixing_ws;
+
+static enum ws_ignore {
+	ignore_ws_none,
+	ignore_ws_change,
+} ws_ignore_action = ignore_ws_none;
+
+
 static const char *patch_input_file;
 static const char *root;
 static int root_len;
@@ -97,6 +104,21 @@ static void parse_whitespace_option(const char *option)
 	die("unrecognized whitespace option '%s'", option);
 }
 
+static void parse_ignorewhitespace_option(const char *option)
+{
+	if (!option || !strcmp(option, "no") ||
+	    !strcmp(option, "false") || !strcmp(option, "never") ||
+	    !strcmp(option, "none")) {
+		ws_ignore_action = ignore_ws_none;
+		return;
+	}
+	if (!strcmp(option, "change")) {
+		ws_ignore_action = ignore_ws_change;
+		return;
+	}
+	die("unrecognized whitespace ignore option '%s'", option);
+}
+
 static void set_default_whitespace_mode(const char *whitespace_option)
 {
 	if (!whitespace_option && !apply_default_whitespace)
@@ -214,6 +236,62 @@ static uint32_t hash_line(const char *cp, size_t len)
 	return h;
 }
 
+/*
+ * Compare lines s1 of length n1 and s2 of length n2, ignoring
+ * whitespace difference. Returns 1 if they match, 0 otherwise
+ */
+static int fuzzy_matchlines(const char *s1, size_t n1,
+			    const char *s2, size_t n2)
+{
+	const char *last1 = s1 + n1 - 1;
+	const char *last2 = s2 + n2 - 1;
+	int result = 0;
+
+	if (n1 < 0 || n2 < 0)
+		return 0;
+
+	/* ignore line endings */
+	while ((*last1 == '\r') || (*last1 == '\n'))
+		last1--;
+	while ((*last2 == '\r') || (*last2 == '\n'))
+		last2--;
+
+	/* skip leading whitespace */
+	while (isspace(*s1) && (s1 <= last1))
+		s1++;
+	while (isspace(*s2) && (s2 <= last2))
+		s2++;
+	/* early return if both lines are empty */
+	if ((s1 > last1) && (s2 > last2))
+		return 1;
+	while (!result) {
+		result = *s1++ - *s2++;
+		/*
+		 * Skip whitespace inside. We check for whitespace on
+		 * both buffers because we don't want "a b" to match
+		 * "ab"
+		 */
+		if (isspace(*s1) && isspace(*s2)) {
+			while (isspace(*s1) && s1 <= last1)
+				s1++;
+			while (isspace(*s2) && s2 <= last2)
+				s2++;
+		}
+		/*
+		 * If we reached the end on one side only,
+		 * lines don't match
+		 */
+		if (
+		    ((s2 > last2) && (s1 <= last1)) ||
+		    ((s1 > last1) && (s2 <= last2)))
+			return 0;
+		if ((s1 > last1) && (s2 > last2))
+			break;
+	}
+
+	return !result;
+}
+
 static void add_line_info(struct image *img, const char *bol, size_t len, unsigned flag)
 {
 	ALLOC_GROW(img->line_allocated, img->nr + 1, img->alloc);
@@ -1592,10 +1670,17 @@ static int read_old_data(struct stat *st, const char *path, struct strbuf *buf)
 	}
 }
 
+/*
+ * Update the preimage, and the common lines in postimage,
+ * from buffer buf of length len. If postlen is 0 the postimage
+ * is updated in place, otherwise it's updated on a new buffer
+ * of length postlen
+ */
+
 static void update_pre_post_images(struct image *preimage,
 				   struct image *postimage,
 				   char *buf,
-				   size_t len)
+				   size_t len, size_t postlen)
 {
 	int i, ctx;
 	char *new, *old, *fixed;
@@ -1614,11 +1699,19 @@ static void update_pre_post_images(struct image *preimage,
 	*preimage = fixed_preimage;
 
 	/*
-	 * Adjust the common context lines in postimage, in place.
-	 * This is possible because whitespace fixing does not make
-	 * the string grow.
+	 * Adjust the common context lines in postimage. This can be
+	 * done in-place when we are just doing whitespace fixing,
+	 * which does not make the string grow, but needs a new buffer
+	 * when ignoring whitespace causes the update, since in this case
+	 * we could have e.g. tabs converted to multiple spaces.
+	 * We trust the caller to tell us if the update can be done
+	 * in place (postlen==0) or not.
 	 */
-	new = old = postimage->buf;
+	old = postimage->buf;
+	if (postlen)
+		new = postimage->buf = xmalloc(postlen);
+	else
+		new = old;
 	fixed = preimage->buf;
 	for (i = ctx = 0; i < postimage->nr; i++) {
 		size_t len = postimage->line[i].len;
@@ -1693,12 +1786,58 @@ static int match_fragment(struct image *img,
 	    !memcmp(img->buf + try, preimage->buf, preimage->len))
 		return 1;
 
+	/*
+	 * No exact match. If we are ignoring whitespace, run a line-by-line
+	 * fuzzy matching. We collect all the line length information because
+	 * we need it to adjust whitespace if we match.
+	 */
+	if (ws_ignore_action == ignore_ws_change) {
+		size_t imgoff = 0;
+		size_t preoff = 0;
+		size_t postlen = postimage->len;
+		size_t imglen[preimage->nr];
+		for (i = 0; i < preimage->nr; i++) {
+			size_t prelen = preimage->line[i].len;
+
+			imglen[i] = img->line[try_lno+i].len;
+			if (!fuzzy_matchlines(
+				img->buf + try + imgoff, imglen[i],
+				preimage->buf + preoff, prelen))
+				return 0;
+			if (preimage->line[i].flag & LINE_COMMON)
+				postlen += imglen[i] - prelen;
+			imgoff += imglen[i];
+			preoff += prelen;
+		}
+
+		/*
+		 * Ok, the preimage matches with whitespace fuzz. Update it and
+		 * the common postimage lines to use the same whitespace as the
+		 * target. imgoff now holds the true length of the target that
+		 * matches the preimage, and we need to update the line lengths
+		 * of the preimage to match the target ones.
+		 */
+		fixed_buf = xmalloc(imgoff);
+		memcpy(fixed_buf, img->buf + try, imgoff);
+		for (i = 0; i < preimage->nr; i++)
+			preimage->line[i].len = imglen[i];
+
+		/*
+		 * Update the preimage buffer and the postimage context lines.
+		 */
+		update_pre_post_images(preimage, postimage,
+				fixed_buf, imgoff, postlen);
+		return 1;
+	}
+
 	if (ws_error_action != correct_ws_error)
 		return 0;
 
 	/*
 	 * The hunk does not apply byte-by-byte, but the hash says
-	 * it might with whitespace fuzz.
+	 * it might with whitespace fuzz. We haven't been asked to
+	 * ignore whitespace, we were asked to correct whitespace
+	 * errors, so let's try matching after whitespace correction.
 	 */
 	fixed_buf = xmalloc(preimage->len + 1);
 	buf = fixed_buf;
@@ -1750,7 +1889,7 @@ static int match_fragment(struct image *img,
 	 * hunk match.  Update the context lines in the postimage.
 	 */
 	update_pre_post_images(preimage, postimage,
-			       fixed_buf, buf - fixed_buf);
+			       fixed_buf, buf - fixed_buf, 0);
 	return 1;
 
  unmatch_exit:
@@ -3192,6 +3331,8 @@ static int git_apply_config(const char *var, const char *value, void *cb)
 {
 	if (!strcmp(var, "apply.whitespace"))
 		return git_config_string(&apply_default_whitespace, var, value);
+	else if (!strcmp(var, "apply.ignorewhitespace"))
+		return git_config_string(&apply_default_ignorewhitespace, var, value);
 	return git_default_config(var, value, cb);
 }
 
@@ -3228,6 +3369,16 @@ static int option_parse_z(const struct option *opt,
 	return 0;
 }
 
+static int option_parse_space_change(const struct option *opt,
+			  const char *arg, int unset)
+{
+	if (unset)
+		ws_ignore_action = ignore_ws_none;
+	else
+		ws_ignore_action = ignore_ws_change;
+	return 0;
+}
+
 static int option_parse_whitespace(const struct option *opt,
 				   const char *arg, int unset)
 {
@@ -3304,6 +3455,12 @@ int cmd_apply(int argc, const char **argv, const char *unused_prefix)
 		{ OPTION_CALLBACK, 0, "whitespace", &whitespace_option, "action",
 			"detect new or modified lines that have whitespace errors",
 			0, option_parse_whitespace },
+		{ OPTION_CALLBACK, 0, "ignore-space-change", NULL, NULL,
+			"ignore changes in whitespace when finding context",
+			PARSE_OPT_NOARG, option_parse_space_change },
+		{ OPTION_CALLBACK, 0, "ignore-whitespace", NULL, NULL,
+			"ignore changes in whitespace when finding context",
+			PARSE_OPT_NOARG, option_parse_space_change },
 		OPT_BOOLEAN('R', "reverse", &apply_in_reverse,
 			"apply the patch in reverse"),
 		OPT_BOOLEAN(0, "unidiff-zero", &unidiff_zero,
@@ -3328,6 +3485,8 @@ int cmd_apply(int argc, const char **argv, const char *unused_prefix)
 	git_config(git_apply_config, NULL);
 	if (apply_default_whitespace)
 		parse_whitespace_option(apply_default_whitespace);
+	if (apply_default_ignorewhitespace)
+		parse_ignorewhitespace_option(apply_default_ignorewhitespace);
 
 	argc = parse_options(argc, argv, prefix, builtin_apply_options,
 			apply_usage, 0);

cache.h

@@ -509,6 +509,7 @@ extern int log_all_ref_updates;
 extern int warn_ambiguous_refs;
 extern int shared_repository;
 extern const char *apply_default_whitespace;
+extern const char *apply_default_ignorewhitespace;
 extern int zlib_compression_level;
 extern int core_compression_level;
 extern int core_compression_seen;

contrib/completion/git-completion.bash

@@ -674,6 +674,7 @@ _git_am ()
 	--*)
 		__gitcomp "
 			--3way --committer-date-is-author-date --ignore-date
+			--ignore-whitespace --ignore-space-change
 			--interactive --keep --no-utf8 --signoff --utf8
 			--whitespace=
 			"
@@ -695,6 +696,7 @@ _git_apply ()
 			--stat --numstat --summary --check --index
 			--cached --index-info --reverse --reject --unidiff-zero
 			--apply --no-add --exclude=
+			--ignore-whitespace --ignore-space-change
 			--whitespace= --inaccurate-eof --verbose
 			"
 		return
@@ -1536,6 +1538,7 @@ _git_config ()
 	__gitcomp "
 		add.ignore-errors
 		alias.
+		apply.ignorewhitespace
 		apply.whitespace
 		branch.autosetupmerge
 		branch.autosetuprebase

environment.c

@@ -26,6 +26,7 @@ const char *git_commit_encoding;
 const char *git_log_output_encoding;
 int shared_repository = PERM_UMASK;
 const char *apply_default_whitespace;
+const char *apply_default_ignorewhitespace;
 int zlib_compression_level = Z_BEST_SPEED;
 int core_compression_level;
 int core_compression_seen;

git-am.sh

@@ -16,6 +16,8 @@ s,signoff       add a Signed-off-by line to the commit message
 u,utf8          recode into utf8 (default)
 k,keep          pass -k flag to git-mailinfo
 whitespace=     pass it through git-apply
+ignore-space-change pass it through git-apply
+ignore-whitespace pass it through git-apply
 directory=      pass it through git-apply
 C=              pass it through git-apply
 p=              pass it through git-apply
@@ -303,7 +305,7 @@ do
 		git_apply_opt="$git_apply_opt $(sq "$1$2")"; shift ;;
 	--patch-format)
 		shift ; patch_format="$1" ;;
-	--reject)
+	--reject|--ignore-whitespace|--ignore-space-change)
 		git_apply_opt="$git_apply_opt $1" ;;
 	--committer-date-is-author-date)
 		committer_date_is_author_date=t ;;