Implement normalize_absolute_path

normalize_absolute_path removes several oddities form absolute paths, giving nice clean paths like "/dir/sub1/sub2". Also add a test case for this utility, based on a new test program (in the style of test-sha1).

---
Junio thought it would be good to do normalization of paths in the GIT_CEILING_DIRECTORIES patch. This function does the normalization.

The closest thing I could find to this in the existing code base was sanitary_path_copy, but it does not remove trailing slashes.

If this test suite takes too long to run, I'm fine with commenting most of it out.

 .gitignore            |    1 +
 Makefile              |    2 +-
 cache.h               |    1 +
 path.c                |   50 +++++++++++++++++++++++++++++++++++++++++++++++++
 t/t0060-path-utils.sh |   40 +++++++++++++++++++++++++++++++++++++++
 test-path-utils.c     |   13 ++++++++++++
 6 files changed, 106 insertions(+), 1 deletions(-)
 create mode 100755 t/t0060-path-utils.sh
 create mode 100644 test-path-utils.c

index 4ff2fec..c54c473 100644
@@ -150,6 +150,7 @@ test-dump-cache-tree
 test-genrandom
 test-match-trees
 test-parse-options
+test-path-utils
 test-sha1
 common-cmds.h
 *.tar.gz

index 649ee56..d76c16b 100644
@@ -1182,7 +1182,7 @@ endif
 
 ### Testing rules
 
-TEST_PROGRAMS = test-chmtime$X test-genrandom$X test-date$X test-delta$X test-sha1$X test-match-trees$X test-absolute-path$X test-parse-options$X
+TEST_PROGRAMS = test-chmtime$X test-genrandom$X test-date$X test-delta$X test-sha1$X test-match-trees$X test-absolute-path$X test-parse-options$X test-path-utils$X
 
 all:: $(TEST_PROGRAMS)
 

index 9cee9a5..5fd1e5b 100644
@@ -514,6 +514,7 @@ static inline int is_absolute_path(const char *path)
 	return path[0] == '/';
 }
 const char *make_absolute_path(const char *path);
+int normalize_absolute_path(char *buf, const char *path);
 
 /* Read and unpack a sha1 file into memory, write memory to a sha1 file */
 extern int sha1_object_info(const unsigned char *, unsigned long *);

new file mode 100755
index 0000000..9076b3b
@@ -0,0 +1,40 @@
+#!/bin/sh
+#
+# Copyright (c) 2008 David Reiss
+#
+
+test_description='Test various path utilities'
+
+. ./test-lib.sh
+
+norm_abs() {
+	test_expect_success "normalize absolute" \
+	"test \$(test-path-utils normalize_absolute_path '$1') = '$2'"
+}
+
+norm_abs "" /
+norm_abs / /
+norm_abs // /
+norm_abs /// /
+norm_abs /. /
+norm_abs /./ /
+norm_abs /./.. /
+norm_abs /../. /
+norm_abs /./../.// /
+norm_abs /dir/.. /
+norm_abs /dir/sub/../.. /
+norm_abs /dir /dir
+norm_abs /dir// /dir
+norm_abs /./dir /dir
+norm_abs /dir/. /dir
+norm_abs /dir///./ /dir
+norm_abs /dir//sub/.. /dir
+norm_abs /dir/sub/../ /dir
+norm_abs //dir/sub/../. /dir
+norm_abs /dir/s1/../s2/ /dir/s2
+norm_abs /d1/s1///s2/..//../s3/ /d1/s3
+norm_abs /d1/s1//../s2/../../d2 /d2
+norm_abs /d1/.../d2 /d1/.../d2
+norm_abs /d1/..././../d2 /d1/d2
+
+test_done

new file mode 100644
index 0000000..1bd4321
@@ -0,0 +1,13 @@
+#include "cache.h"
+
+int main(int argc, char **argv)
+{
+	if (argc == 3 && !strcmp(argv[1], "normalize_absolute_path")) {
+		char *buf = xmalloc(strlen(argv[2])+1);
+		int rv = normalize_absolute_path(buf, argv[2]);
+		assert(strlen(buf) == rv);
+		puts(buf);
+	}
+
+	return 0;
+}

1.5.4

David Reiss, Tue, May 20, 2008 08:48:54 +0200:

Heh...
http://thread.gmane.org/gmane.comp.version-control.git/68786/focus=68812

The test of the function has these cases:

static void check(const char *cwd, const char *path, const char *good);

int main(int argc, char **argv)

{
	/* 1 */ check("/onelevel", "aa", "onelevel/aa");
	/* 2 */ check("/", "..", "");
	/* 3 */ check("/", "../..", "");
	/* 4 */ check("/one", "aa/../bb", "one/bb");
	/* 5 */ check("/one/two", "aa//bb", "one/two/aa/bb");
	/* 6 */ check("", "/aa//bb", "aa/bb");
	/* 7 */ check("/one/two", "", "one/two");
	/* 8 */ check("/one/two", "aa/..bb/x/../cc/", "one/two/aa/..bb/cc");
	/* 9 */ check("/one/two", "aa/x/././cc////", "one/two/aa/x/cc");
	/* 10 */ check("/one/two", "../../../../aa", "aa");
	/* 11 */ check("one/", "../one/two", "one/two");
	/* 12 */ check("", "../../two", "two");
	/* 13 */ check("a/b/c", "../../two", "a/two");
	/* 14 */ check("a/b/", "../two", "a/two");
	/* 15 */ check("///", "../two", "two");
	return 0;
}

static void check(const char *cwd, const char *path, const char *good)

{
	static int n = 0;
	printf("%-2d: %10s$ cd %s", ++n, cwd, path);
	char *t = pathexpand(cwd, path);
	if ( strcmp(t, good) )
		printf(" ____________________failed(%s)\n", t);
	else
		printf(" \033[32m%s\033[0m\n", t);
	free(t);
}

This is close enough to what I want that I could definitely use it if it were in the main tree. I read through most of that thread, but I couldn't figure out what ever happened to your patch. (I couldn't find a clear rejection, but I also don't see it in master.)

--David

David Reiss, Wed, May 21, 2008 10:34:17 +0200:

Me too. OTOH, it was not a clear proposal for inclusion.

Hi,

I wonder why my advice to look at setup.c (I think I did not mention the function name "sanitary_path_copy()" in particular, because I figured you would spot it right away) went unheeded.

So I think I will not even bother to look at your patches either.

Ciao,
Dscho

I wonder why my comment in the patch went unheeded:

Honestly, I found it quite daunting to try to track down all of the indirect callers of sanitary_path_copy and ensure that they would continue to work if I changed the behavior of a function that does not document what it is supposed to do. Frankly, it would be easier for me to just keep rebasing my patches and apply them to my local builds.

--David