Re: [PATCH 01/17] mv: convert to using pathspec struct interface

On 12/07, Duy Nguyen wrote:

On Wed, Dec 7, 2016 at 4:51 AM, Brandon Williams [off-list ref] wrote:

quoted

Convert the 'internal_copy_pathspec()' function to use the pathspec
struct interface from using the deprecated 'get_pathspec()' interface.

In addition to this, fix a memory leak caused by only duplicating some
of the pathspec elements.  Instead always duplicate all of the the
pathspec elements as an intermediate step (with modificationed based on
the passed in flags).  This way the intermediate strings can then be
freed prior to duplicating the result of parse_pathspec (which contains
each of the elements with the prefix prepended).

Signed-off-by: Brandon Williams <redacted>
---
 builtin/mv.c | 45 ++++++++++++++++++++++++++++++++-------------
 1 file changed, 32 insertions(+), 13 deletions(-)

Stefan did something similar last year [1] but I couldn't find why it
did not get merged. Not sure if the reasons are still relevant or
not...

[1] http://public-inbox.org/git/%3C1438885632-26470-1-git-send-email-sbeller@google.com%3E/

quoted

diff --git a/builtin/mv.c b/builtin/mv.c
index 2f43877..4df4a12 100644
--- a/builtin/mv.c
+++ b/builtin/mv.c

@@ -4,6 +4,7 @@
  * Copyright (C) 2006 Johannes Schindelin
  */
 #include "builtin.h"
+#include "pathspec.h"
 #include "lockfile.h"
 #include "dir.h"
 #include "cache-tree.h"

@@ -25,25 +26,43 @@ static const char **internal_copy_pathspec(const char *prefix,
 {
        int i;
        const char **result;
+       struct pathspec ps;
        ALLOC_ARRAY(result, count + 1);
-       COPY_ARRAY(result, pathspec, count);
-       result[count] = NULL;
+
+       /* Create an intermediate copy of the pathspec based on the flags */
        for (i = 0; i < count; i++) {
-               int length = strlen(result[i]);
+               int length = strlen(pathspec[i]);
                int to_copy = length;
+               char *it;
                while (!(flags & KEEP_TRAILING_SLASH) &&
-                      to_copy > 0 && is_dir_sep(result[i][to_copy - 1]))
+                      to_copy > 0 && is_dir_sep(pathspec[i][to_copy - 1]))
                        to_copy--;
-               if (to_copy != length || flags & DUP_BASENAME) {
-                       char *it = xmemdupz(result[i], to_copy);
-                       if (flags & DUP_BASENAME) {
-                               result[i] = xstrdup(basename(it));
-                               free(it);
-                       } else
-                               result[i] = it;
-               }
+
+               it = xmemdupz(pathspec[i], to_copy);
+               if (flags & DUP_BASENAME) {
+                       result[i] = xstrdup(basename(it));
+                       free(it);
+               } else
+                       result[i] = it;
+       }
+       result[count] = NULL;
+
+       parse_pathspec(&ps,
+                      PATHSPEC_ALL_MAGIC &
+                      ~(PATHSPEC_FROMTOP | PATHSPEC_LITERAL),
+                      PATHSPEC_KEEP_ORDER | PATHSPEC_PREFER_CWD,
+                      prefix, result);
+       assert(count == ps.nr);
+
+       /* Copy the pathspec and free the old intermediate strings */
+       for (i = 0; i < count; i++) {
+               const char *match = xstrdup(ps.items[i].match);
+               free((char *) result[i]);
+               result[i] = match;

Sigh.. it looks so weird that we do all the parsing (in a _copy_
pathspec function) then remove struct pathspec and return the plain
string. I guess we can't do anything more until we rework cmd_mv code
to handle pathspec natively.

At the least I think we should rename this function to something else.
But if you have time I really wish we could kill this function. I
haven't stared at cmd_mv() long and hard, but it looks to me that we
combining two separate functionalities in the same function here.

If "mv" takes n arguments, then the first <n-1> arguments may be
pathspec, the last one is always a plain path. The "dest_path =
internal_copy_pathspec..." could be as simple as "dest_path =
prefix_path(argv[argc - 1])". the special treatment for this last
argument [1] can live here. Then, we can do parse_pathspec for the
<n-1> arguments in cmd_mv(). It's still far from perfect, because
cmd_mv can't handle pathspec properly, but it reduces the messy mess
in internal_copy_pathspec a bit, I hope.

[1] c57f628 (mv: let 'git mv file no-such-dir/' error out - 2013-12-03)

quoted

        }
-       return get_pathspec(prefix, result);
+
+       clear_pathspec(&ps);
+       return result;
 }

 static const char *add_slash(const char *path)
--
2.8.0.rc3.226.g39d4020

-- 
Duy

Actually, after looking at this a bit more it seems like we could
technically use prefix_path for both source and dest (based on how the
current code is structured) since the source's provied must all exist (as
in no wildcards are allowed).  We could drop using the pathspec struct
completely in addition to renaming the function (to what I'm still
unsure).  I agree that this code should probably be rewritten and made a
bit cleaner, I don't know if that fits in the scope of this series or
should be done as a followup patch.  If you think it fits here then I
can try and find some time to do the rework.

-- 
Brandon Williams