[PATCH 4/4] dir.c: don't exclude whole dir prematurely

["Nguyễn Thái Ngọc Duy" <pclouds@gmail.com>]

If there is a pattern "!foo/bar", this patch makes it not exclude
"foo" right away. This gives us a chance to examine "foo" and
re-include "foo/bar".

Helped-by: brian m. carlson <sandals@crustytoothpaste.net>
Helped-by: Micha Wiedenmann <mw-u2@gmx.de>
Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
---
 Documentation/gitignore.txt                 |  17 +++-
 dir.c                                       | 109 +++++++++++++++++++-
 t/t3001-ls-files-others-exclude.sh          |   7 +-
 t/t3007-ls-files-other-negative.sh (new +x) | 153 ++++++++++++++++++++++++++++
 4 files changed, 276 insertions(+), 10 deletions(-)
 create mode 100755 t/t3007-ls-files-other-negative.sh

diff --git a/Documentation/gitignore.txt b/Documentation/gitignore.txt
index 473623d..3ded6fd 100644
--- a/Documentation/gitignore.txt
+++ b/Documentation/gitignore.txt
@@ -82,12 +82,12 @@ PATTERN FORMAT
 
  - An optional prefix "`!`" which negates the pattern; any
    matching file excluded by a previous pattern will become
-   included again. It is not possible to re-include a file if a parent
-   directory of that file is excluded. Git doesn't list excluded
-   directories for performance reasons, so any patterns on contained
-   files have no effect, no matter where they are defined.
+   included again.
    Put a backslash ("`\`") in front of the first "`!`" for patterns
    that begin with a literal "`!`", for example, "`\!important!.txt`".
+   It is possible to re-include a file if a parent directory of that
+   file is excluded if certain conditions are met. See section NOTES
+   for detail.
 
  - If the pattern ends with a slash, it is removed for the
    purpose of the following description, but it would only find
@@ -141,6 +141,15 @@ not tracked by Git remain untracked.
 To stop tracking a file that is currently tracked, use
 'git rm --cached'.
 
+To re-include files or directories when their parent directory is
+excluded, the following conditions must be met:
+
+ - The rules to exclude a directory and re-include a subset back must
+   be in the same .gitignore file.
+
+ - The directory part in the re-include rules must be literal (i.e. no
+   wildcards)
+
 EXAMPLES
 --------
 
diff --git a/dir.c b/dir.c
index 8a9d8c0..552af23 100644
--- a/dir.c
+++ b/dir.c
@@ -930,6 +930,75 @@ static int match_sticky(struct exclude *exc, const char *pathname, int pathlen,
 	return 0;
 }
 
+static inline int different_decisions(const struct exclude *a,
+				      const struct exclude *b)
+{
+	return (a->flags & EXC_FLAG_NEGATIVE) != (b->flags & EXC_FLAG_NEGATIVE);
+}
+
+/*
+ * Return non-zero if pathname is a directory and an ancestor of the
+ * literal path in a pattern.
+ */
+static int match_directory_part(const char *pathname, int pathlen,
+				int *dtype, struct exclude *x)
+{
+	const char	*base	    = x->base;
+	int		 baselen    = x->baselen ? x->baselen - 1 : 0;
+	const char	*pattern    = x->pattern;
+	int		 prefix	    = x->nowildcardlen;
+	int		 patternlen = x->patternlen;
+
+	if (*dtype == DT_UNKNOWN)
+		*dtype = get_dtype(NULL, pathname, pathlen);
+	if (*dtype != DT_DIR)
+		return 0;
+
+	if (*pattern == '/') {
+		pattern++;
+		patternlen--;
+		prefix--;
+	}
+
+	if (baselen) {
+		if (((pathlen < baselen && base[pathlen] == '/') ||
+		     pathlen == baselen) &&
+		    !strncmp_icase(pathname, base, pathlen))
+			return 1;
+		pathname += baselen + 1;
+		pathlen  -= baselen + 1;
+	}
+
+
+	if (prefix &&
+	    (((pathlen < prefix && pattern[pathlen] == '/') ||
+	      pathlen == prefix) &&
+	     !strncmp_icase(pathname, pattern, pathlen)))
+		return 1;
+
+	return 0;
+}
+
+static struct exclude *should_descend(const char *pathname, int pathlen,
+				      int *dtype, struct exclude_list *el,
+				      struct exclude *exc)
+{
+	int i;
+
+	for (i = el->nr - 1; 0 <= i; i--) {
+		struct exclude *x = el->excludes[i];
+
+		if (x == exc)
+			break;
+
+		if (!(x->flags & EXC_FLAG_NODIR) &&
+		    different_decisions(x, exc) &&
+		    match_directory_part(pathname, pathlen, dtype, x))
+			return x;
+	}
+	return NULL;
+}
+
 /*
  * Scan the given exclude list in reverse to see whether pathname
  * should be ignored.  The first match (i.e. the last on the list), if
@@ -943,7 +1012,7 @@ static struct exclude *last_exclude_matching_from_list(const char *pathname,
 						       struct exclude_list *el)
 {
 	struct exclude *exc = NULL; /* undecided */
-	int i;
+	int i, maybe_descend = 0;
 
 	if (!el->nr)
 		return NULL;	/* undefined */
@@ -955,6 +1024,10 @@ static struct exclude *last_exclude_matching_from_list(const char *pathname,
 		const char *exclude = x->pattern;
 		int prefix = x->nowildcardlen;
 
+		if (!maybe_descend && i < el->nr - 1 &&
+		    different_decisions(x, el->excludes[i+1]))
+			maybe_descend = 1;
+
 		if (x->sticky_paths.nr) {
 			if (*dtype == DT_UNKNOWN)
 				*dtype = get_dtype(NULL, pathname, pathlen);
@@ -998,6 +1071,34 @@ static struct exclude *last_exclude_matching_from_list(const char *pathname,
 		return NULL;
 	}
 
+	/*
+	 * We have found a matching pattern "exc" that may exclude whole
+	 * directory. We also found that there may be a pattern that matches
+	 * something inside the directory and reincludes stuff.
+	 *
+	 * Go through the patterns again, find that pattern and double check.
+	 * If it's true, return "undecided" and keep descending in. "exc" is
+	 * marked sticky so that it continues to match inside the directory.
+	 */
+	if (!(exc->flags & EXC_FLAG_NEGATIVE) && maybe_descend) {
+		struct exclude *x;
+
+		if (*dtype == DT_UNKNOWN)
+			*dtype = get_dtype(NULL, pathname, pathlen);
+
+		if (*dtype == DT_DIR &&
+		    (x = should_descend(pathname, pathlen, dtype, el, exc))) {
+			add_sticky(exc, pathname, pathlen);
+			trace_printf_key(&trace_exclude,
+					 "exclude: %.*s vs %s at line %d => %s,"
+					 " forced open by %s at line %d => n/a\n",
+					 pathlen, pathname, exc->pattern, exc->srcpos,
+					 exc->flags & EXC_FLAG_NEGATIVE ? "no" : "yes",
+					 x->pattern, x->srcpos);
+			return NULL;
+		}
+	}
+
 	trace_printf_key(&trace_exclude, "exclude: %.*s vs %s at line %d => %s%s\n",
 			 pathlen, pathname, exc->pattern, exc->srcpos,
 			 exc->flags & EXC_FLAG_NEGATIVE ? "no" : "yes",
@@ -2097,6 +2198,12 @@ int read_directory(struct dir_struct *dir, const char *path, int len, const stru
 		return dir->nr;
 
 	/*
+	 * Stay on the safe side. if read_directory() has run once on
+	 * "dir", some sticky flag may have been left. Clear them all.
+	 */
+	clear_sticky(dir);
+
+	/*
 	 * exclude patterns are treated like positive ones in
 	 * create_simplify. Usually exclude patterns should be a
 	 * subset of positive ones, which has no impacts on
diff --git a/t/t3001-ls-files-others-exclude.sh b/t/t3001-ls-files-others-exclude.sh
index 3fc484e..d043078 100755
--- a/t/t3001-ls-files-others-exclude.sh
+++ b/t/t3001-ls-files-others-exclude.sh
@@ -175,13 +175,10 @@ test_expect_success 'negated exclude matches can override previous ones' '
 	grep "^a.1" output
 '
 
-test_expect_success 'excluded directory overrides content patterns' '
+test_expect_success 'excluded directory does not override content patterns' '
 
 	git ls-files --others --exclude="one" --exclude="!one/a.1" >output &&
-	if grep "^one/a.1" output
-	then
-		false
-	fi
+	grep "^one/a.1" output
 '
 
 test_expect_success 'negated directory doesn'\''t affect content patterns' '
diff --git a/t/t3007-ls-files-other-negative.sh b/t/t3007-ls-files-other-negative.sh
new file mode 100755
index 0000000..0797b86
--- /dev/null
+++ b/t/t3007-ls-files-other-negative.sh
@@ -0,0 +1,153 @@
+#!/bin/sh
+
+test_description='test re-include patterns'
+
+. ./test-lib.sh
+
+test_expect_success 'setup' '
+	mkdir -p fooo foo/bar tmp &&
+	touch abc foo/def foo/bar/ghi foo/bar/bar
+'
+
+test_expect_success 'no match, do not enter subdir and waste cycles' '
+	cat >.gitignore <<-\EOF &&
+	/tmp
+	/foo
+	!fooo/bar/bar
+	EOF
+	GIT_TRACE_EXCLUDE="$(pwd)/tmp/trace" git ls-files -o --exclude-standard >tmp/actual &&
+	! grep "enter .foo/.\$" tmp/trace &&
+	cat >tmp/expected <<-\EOF &&
+	.gitignore
+	abc
+	EOF
+	test_cmp tmp/expected tmp/actual
+'
+
+test_expect_success 'match, excluded by literal pathname pattern' '
+	cat >.gitignore <<-\EOF &&
+	/tmp
+	/fooo
+	/foo
+	!foo/bar/bar
+	EOF
+	cat >fooo/.gitignore <<-\EOF &&
+	!/*
+	EOF	git ls-files -o --exclude-standard >tmp/actual &&
+	cat >tmp/expected <<-\EOF &&
+	.gitignore
+	abc
+	foo/bar/bar
+	EOF
+	test_cmp tmp/expected tmp/actual
+'
+
+test_expect_success 'match, excluded by wildcard pathname pattern' '
+	cat >.gitignore <<-\EOF &&
+	/tmp
+	/fooo
+	/fo?
+	!foo/bar/bar
+	EOF
+	git ls-files -o --exclude-standard >tmp/actual &&
+	cat >tmp/expected <<-\EOF &&
+	.gitignore
+	abc
+	foo/bar/bar
+	EOF
+	test_cmp tmp/expected tmp/actual
+'
+
+test_expect_success 'match, excluded by literal basename pattern' '
+	cat >.gitignore <<-\EOF &&
+	/tmp
+	/fooo
+	foo
+	!foo/bar/bar
+	EOF
+	git ls-files -o --exclude-standard >tmp/actual &&
+	cat >tmp/expected <<-\EOF &&
+	.gitignore
+	abc
+	foo/bar/bar
+	EOF
+	test_cmp tmp/expected tmp/actual
+'
+
+test_expect_success 'match, excluded by wildcard basename pattern' '
+	cat >.gitignore <<-\EOF &&
+	/tmp
+	/fooo
+	fo?
+	!foo/bar/bar
+	EOF
+	git ls-files -o --exclude-standard >tmp/actual &&
+	cat >tmp/expected <<-\EOF &&
+	.gitignore
+	abc
+	foo/bar/bar
+	EOF
+	test_cmp tmp/expected tmp/actual
+'
+
+test_expect_success 'match, excluded by literal mustbedir, basename pattern' '
+	cat >.gitignore <<-\EOF &&
+	/tmp
+	/fooo
+	foo/
+	!foo/bar/bar
+	EOF
+	git ls-files -o --exclude-standard >tmp/actual &&
+	cat >tmp/expected <<-\EOF &&
+	.gitignore
+	abc
+	foo/bar/bar
+	EOF
+	test_cmp tmp/expected tmp/actual
+'
+
+test_expect_success 'match, excluded by literal mustbedir, pathname pattern' '
+	cat >.gitignore <<-\EOF &&
+	/tmp
+	/fooo
+	/foo/
+	!foo/bar/bar
+	EOF
+	git ls-files -o --exclude-standard >tmp/actual &&
+	cat >tmp/expected <<-\EOF &&
+	.gitignore
+	abc
+	foo/bar/bar
+	EOF
+	test_cmp tmp/expected tmp/actual
+'
+
+test_expect_success 'prepare for nested negatives' '
+	cat >.git/info/exclude <<-\EOF &&
+	/.gitignore
+	/tmp
+	/foo
+	/abc
+	EOF
+	git ls-files -o --exclude-standard >tmp/actual &&
+	test_must_be_empty tmp/actual &&
+	mkdir -p 1/2/3/4 &&
+	touch 1/f 1/2/f 1/2/3/f 1/2/3/4/f
+'
+
+test_expect_success 'match, literal pathname, nested negatives' '
+	cat >.gitignore <<-\EOF &&
+	/1
+	!1/2
+	1/2/3
+	!1/2/3/4
+	EOF
+	git ls-files -o --exclude-standard >tmp/actual &&
+	cat >tmp/expected <<-\EOF &&
+	1/2/3/4/f
+	1/2/f
+	EOF
+	test_cmp tmp/expected tmp/actual
+'
+
+test_done
-- 
2.7.0.377.g4cd97dd