2006-03-30 10:55:43 +04:00
|
|
|
#include "cache.h"
|
|
|
|
#include "tree-walk.h"
|
2010-08-11 12:38:07 +04:00
|
|
|
#include "unpack-trees.h"
|
2010-12-15 18:02:44 +03:00
|
|
|
#include "dir.h"
|
2006-04-02 16:44:09 +04:00
|
|
|
#include "tree.h"
|
2013-07-14 12:35:25 +04:00
|
|
|
#include "pathspec.h"
|
2006-03-30 10:55:43 +04:00
|
|
|
|
2007-03-21 20:09:56 +03:00
|
|
|
static const char *get_mode(const char *str, unsigned int *modep)
|
|
|
|
{
|
|
|
|
unsigned char c;
|
|
|
|
unsigned int mode = 0;
|
|
|
|
|
2008-01-06 20:21:10 +03:00
|
|
|
if (*str == ' ')
|
|
|
|
return NULL;
|
|
|
|
|
2007-03-21 20:09:56 +03:00
|
|
|
while ((c = *str++) != ' ') {
|
|
|
|
if (c < '0' || c > '7')
|
|
|
|
return NULL;
|
|
|
|
mode = (mode << 3) + (c - '0');
|
|
|
|
}
|
|
|
|
*modep = mode;
|
|
|
|
return str;
|
|
|
|
}
|
|
|
|
|
2016-09-27 23:59:51 +03:00
|
|
|
static int decode_tree_entry(struct tree_desc *desc, const char *buf, unsigned long size, struct strbuf *err)
|
2007-03-21 20:09:56 +03:00
|
|
|
{
|
|
|
|
const char *path;
|
|
|
|
unsigned int mode, len;
|
|
|
|
|
2016-09-27 23:59:51 +03:00
|
|
|
if (size < 23 || buf[size - 21]) {
|
|
|
|
strbuf_addstr(err, _("too-short tree object"));
|
|
|
|
return -1;
|
|
|
|
}
|
2008-01-06 20:21:10 +03:00
|
|
|
|
2007-03-21 20:09:56 +03:00
|
|
|
path = get_mode(buf, &mode);
|
2016-09-27 23:59:51 +03:00
|
|
|
if (!path) {
|
|
|
|
strbuf_addstr(err, _("malformed mode in tree entry"));
|
|
|
|
return -1;
|
|
|
|
}
|
|
|
|
if (!*path) {
|
|
|
|
strbuf_addstr(err, _("empty filename in tree entry"));
|
|
|
|
return -1;
|
|
|
|
}
|
2007-03-21 20:09:56 +03:00
|
|
|
len = strlen(path) + 1;
|
|
|
|
|
|
|
|
/* Initialize the descriptor entry */
|
|
|
|
desc->entry.path = path;
|
tree-walk: finally switch over tree descriptors to contain a pre-parsed entry
This continues 4651ece8 (Switch over tree descriptors to contain a
pre-parsed entry) and moves the only rest computational part
mode = canon_mode(mode)
from tree_entry_extract() to tree entry decode phase - to
decode_tree_entry().
The reason to do it, is that canon_mode() is at least 2 conditional
jumps for regular files, and that could be noticeable should canon_mode()
be invoked several times.
That does not matter for current Git codebase, where typical tree
traversal is
while (t->size) {
sha1 = tree_entry_extract(t, &path, &mode);
...
update_tree_entry(t);
}
i.e. we do t -> sha1,path.mode "extraction" only once per entry. In such
cases, it does not matter performance-wise, where that mode
canonicalization is done - either once in tree_entry_extract(), or once
in decode_tree_entry() called by update_tree_entry() - it is
approximately the same.
But for future code, which could need to work with several tree_desc's
in parallel, it could be handy to operate on tree_desc descriptors, and
do "extracts" only when needed, or at all, access only relevant part of
it through structure fields directly.
And for such situations, having canon_mode() be done once in decode
phase is better - we won't need to pay the performance price of 2 extra
conditional jumps on every t->mode access.
So let's move mode canonicalization to decode_tree_entry(). That was the
final bit. Now after tree entry is decoded, it is fully ready and could
be accessed either directly via field, or through tree_entry_extract()
which this time got really "totally trivial".
Signed-off-by: Kirill Smelkov <kirr@mns.spb.ru>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2014-02-06 15:36:31 +04:00
|
|
|
desc->entry.mode = canon_mode(mode);
|
2016-04-18 02:10:39 +03:00
|
|
|
desc->entry.oid = (const struct object_id *)(path + len);
|
2016-09-27 23:59:51 +03:00
|
|
|
|
|
|
|
return 0;
|
2007-03-21 20:09:56 +03:00
|
|
|
}
|
|
|
|
|
2016-09-27 23:59:51 +03:00
|
|
|
static int init_tree_desc_internal(struct tree_desc *desc, const void *buffer, unsigned long size, struct strbuf *err)
|
2007-03-21 20:08:25 +03:00
|
|
|
{
|
|
|
|
desc->buffer = buffer;
|
|
|
|
desc->size = size;
|
2007-03-21 20:09:56 +03:00
|
|
|
if (size)
|
2016-09-27 23:59:51 +03:00
|
|
|
return decode_tree_entry(desc, buffer, size, err);
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
void init_tree_desc(struct tree_desc *desc, const void *buffer, unsigned long size)
|
|
|
|
{
|
|
|
|
struct strbuf err = STRBUF_INIT;
|
|
|
|
if (init_tree_desc_internal(desc, buffer, size, &err))
|
|
|
|
die("%s", err.buf);
|
|
|
|
strbuf_release(&err);
|
|
|
|
}
|
|
|
|
|
|
|
|
int init_tree_desc_gently(struct tree_desc *desc, const void *buffer, unsigned long size)
|
|
|
|
{
|
|
|
|
struct strbuf err = STRBUF_INIT;
|
|
|
|
int result = init_tree_desc_internal(desc, buffer, size, &err);
|
|
|
|
if (result)
|
|
|
|
error("%s", err.buf);
|
|
|
|
strbuf_release(&err);
|
|
|
|
return result;
|
2007-03-21 20:08:25 +03:00
|
|
|
}
|
|
|
|
|
2017-08-12 11:32:59 +03:00
|
|
|
void *fill_tree_descriptor(struct tree_desc *desc, const struct object_id *oid)
|
2006-03-30 10:55:43 +04:00
|
|
|
{
|
|
|
|
unsigned long size = 0;
|
|
|
|
void *buf = NULL;
|
|
|
|
|
2017-08-12 11:32:59 +03:00
|
|
|
if (oid) {
|
|
|
|
buf = read_object_with_reference(oid->hash, tree_type, &size,
|
|
|
|
NULL);
|
2006-03-30 10:55:43 +04:00
|
|
|
if (!buf)
|
2017-08-12 11:32:59 +03:00
|
|
|
die("unable to read tree %s", oid_to_hex(oid));
|
2006-03-30 10:55:43 +04:00
|
|
|
}
|
2007-03-21 20:08:25 +03:00
|
|
|
init_tree_desc(desc, buf, size);
|
2006-03-30 10:55:43 +04:00
|
|
|
return buf;
|
|
|
|
}
|
|
|
|
|
|
|
|
static void entry_clear(struct name_entry *a)
|
|
|
|
{
|
|
|
|
memset(a, 0, sizeof(*a));
|
|
|
|
}
|
|
|
|
|
|
|
|
static void entry_extract(struct tree_desc *t, struct name_entry *a)
|
|
|
|
{
|
2007-03-21 20:09:56 +03:00
|
|
|
*a = t->entry;
|
2006-03-30 10:55:43 +04:00
|
|
|
}
|
|
|
|
|
2016-09-27 23:59:51 +03:00
|
|
|
static int update_tree_entry_internal(struct tree_desc *desc, struct strbuf *err)
|
2006-03-30 10:55:43 +04:00
|
|
|
{
|
2007-03-21 20:08:25 +03:00
|
|
|
const void *buf = desc->buffer;
|
2016-04-18 02:10:39 +03:00
|
|
|
const unsigned char *end = desc->entry.oid->hash + 20;
|
2006-03-30 10:55:43 +04:00
|
|
|
unsigned long size = desc->size;
|
2007-03-21 20:09:56 +03:00
|
|
|
unsigned long len = end - (const unsigned char *)buf;
|
2006-03-30 10:55:43 +04:00
|
|
|
|
|
|
|
if (size < len)
|
2016-09-27 23:59:50 +03:00
|
|
|
die(_("too-short tree file"));
|
2007-03-21 20:09:56 +03:00
|
|
|
buf = end;
|
|
|
|
size -= len;
|
|
|
|
desc->buffer = buf;
|
|
|
|
desc->size = size;
|
|
|
|
if (size)
|
2016-09-27 23:59:51 +03:00
|
|
|
return decode_tree_entry(desc, buf, size, err);
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
void update_tree_entry(struct tree_desc *desc)
|
|
|
|
{
|
|
|
|
struct strbuf err = STRBUF_INIT;
|
|
|
|
if (update_tree_entry_internal(desc, &err))
|
|
|
|
die("%s", err.buf);
|
|
|
|
strbuf_release(&err);
|
|
|
|
}
|
|
|
|
|
|
|
|
int update_tree_entry_gently(struct tree_desc *desc)
|
|
|
|
{
|
|
|
|
struct strbuf err = STRBUF_INIT;
|
|
|
|
if (update_tree_entry_internal(desc, &err)) {
|
|
|
|
error("%s", err.buf);
|
|
|
|
strbuf_release(&err);
|
|
|
|
/* Stop processing this tree after error */
|
|
|
|
desc->size = 0;
|
|
|
|
return -1;
|
|
|
|
}
|
|
|
|
strbuf_release(&err);
|
|
|
|
return 0;
|
2006-03-30 10:55:43 +04:00
|
|
|
}
|
|
|
|
|
tree_entry(): new tree-walking helper function
This adds a "tree_entry()" function that combines the common operation of
doing a "tree_entry_extract()" + "update_tree_entry()".
It also has a simplified calling convention, designed for simple loops
that traverse over a whole tree: the arguments are pointers to the tree
descriptor and a name_entry structure to fill in, and it returns a boolean
"true" if there was an entry left to be gotten in the tree.
This allows tree traversal with
struct tree_desc desc;
struct name_entry entry;
desc.buf = tree->buffer;
desc.size = tree->size;
while (tree_entry(&desc, &entry) {
... use "entry.{path, sha1, mode, pathlen}" ...
}
which is not only shorter than writing it out in full, it's hopefully less
error prone too.
[ It's actually a tad faster too - we don't need to recalculate the entry
pathlength in both extract and update, but need to do it only once.
Also, some callers can avoid doing a "strlen()" on the result, since
it's returned as part of the name_entry structure.
However, by now we're talking just 1% speedup on "git-rev-list --objects
--all", and we're definitely at the point where tree walking is no
longer the issue any more. ]
NOTE! Not everybody wants to use this new helper function, since some of
the tree walkers very much on purpose do the descriptor update separately
from the entry extraction. So the "extract + update" sequence still
remains as the core sequence, this is just a simplified interface.
We should probably add a silly two-line inline helper function for
initializing the descriptor from the "struct tree" too, just to cut down
on the noise from that common "desc" initializer.
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Signed-off-by: Junio C Hamano <junkio@cox.net>
2006-05-30 20:45:45 +04:00
|
|
|
int tree_entry(struct tree_desc *desc, struct name_entry *entry)
|
|
|
|
{
|
2007-03-21 20:09:56 +03:00
|
|
|
if (!desc->size)
|
tree_entry(): new tree-walking helper function
This adds a "tree_entry()" function that combines the common operation of
doing a "tree_entry_extract()" + "update_tree_entry()".
It also has a simplified calling convention, designed for simple loops
that traverse over a whole tree: the arguments are pointers to the tree
descriptor and a name_entry structure to fill in, and it returns a boolean
"true" if there was an entry left to be gotten in the tree.
This allows tree traversal with
struct tree_desc desc;
struct name_entry entry;
desc.buf = tree->buffer;
desc.size = tree->size;
while (tree_entry(&desc, &entry) {
... use "entry.{path, sha1, mode, pathlen}" ...
}
which is not only shorter than writing it out in full, it's hopefully less
error prone too.
[ It's actually a tad faster too - we don't need to recalculate the entry
pathlength in both extract and update, but need to do it only once.
Also, some callers can avoid doing a "strlen()" on the result, since
it's returned as part of the name_entry structure.
However, by now we're talking just 1% speedup on "git-rev-list --objects
--all", and we're definitely at the point where tree walking is no
longer the issue any more. ]
NOTE! Not everybody wants to use this new helper function, since some of
the tree walkers very much on purpose do the descriptor update separately
from the entry extraction. So the "extract + update" sequence still
remains as the core sequence, this is just a simplified interface.
We should probably add a silly two-line inline helper function for
initializing the descriptor from the "struct tree" too, just to cut down
on the noise from that common "desc" initializer.
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Signed-off-by: Junio C Hamano <junkio@cox.net>
2006-05-30 20:45:45 +04:00
|
|
|
return 0;
|
|
|
|
|
2007-03-21 20:09:56 +03:00
|
|
|
*entry = desc->entry;
|
|
|
|
update_tree_entry(desc);
|
tree_entry(): new tree-walking helper function
This adds a "tree_entry()" function that combines the common operation of
doing a "tree_entry_extract()" + "update_tree_entry()".
It also has a simplified calling convention, designed for simple loops
that traverse over a whole tree: the arguments are pointers to the tree
descriptor and a name_entry structure to fill in, and it returns a boolean
"true" if there was an entry left to be gotten in the tree.
This allows tree traversal with
struct tree_desc desc;
struct name_entry entry;
desc.buf = tree->buffer;
desc.size = tree->size;
while (tree_entry(&desc, &entry) {
... use "entry.{path, sha1, mode, pathlen}" ...
}
which is not only shorter than writing it out in full, it's hopefully less
error prone too.
[ It's actually a tad faster too - we don't need to recalculate the entry
pathlength in both extract and update, but need to do it only once.
Also, some callers can avoid doing a "strlen()" on the result, since
it's returned as part of the name_entry structure.
However, by now we're talking just 1% speedup on "git-rev-list --objects
--all", and we're definitely at the point where tree walking is no
longer the issue any more. ]
NOTE! Not everybody wants to use this new helper function, since some of
the tree walkers very much on purpose do the descriptor update separately
from the entry extraction. So the "extract + update" sequence still
remains as the core sequence, this is just a simplified interface.
We should probably add a silly two-line inline helper function for
initializing the descriptor from the "struct tree" too, just to cut down
on the noise from that common "desc" initializer.
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Signed-off-by: Junio C Hamano <junkio@cox.net>
2006-05-30 20:45:45 +04:00
|
|
|
return 1;
|
|
|
|
}
|
|
|
|
|
2016-09-27 23:59:51 +03:00
|
|
|
int tree_entry_gently(struct tree_desc *desc, struct name_entry *entry)
|
|
|
|
{
|
|
|
|
if (!desc->size)
|
|
|
|
return 0;
|
|
|
|
|
|
|
|
*entry = desc->entry;
|
|
|
|
if (update_tree_entry_gently(desc))
|
|
|
|
return 0;
|
|
|
|
return 1;
|
|
|
|
}
|
|
|
|
|
2008-03-06 05:59:29 +03:00
|
|
|
void setup_traverse_info(struct traverse_info *info, const char *base)
|
|
|
|
{
|
|
|
|
int pathlen = strlen(base);
|
2008-03-07 02:44:48 +03:00
|
|
|
static struct traverse_info dummy;
|
2008-03-06 05:59:29 +03:00
|
|
|
|
|
|
|
memset(info, 0, sizeof(*info));
|
|
|
|
if (pathlen && base[pathlen-1] == '/')
|
|
|
|
pathlen--;
|
|
|
|
info->pathlen = pathlen ? pathlen + 1 : 0;
|
|
|
|
info->name.path = base;
|
2016-04-18 02:10:39 +03:00
|
|
|
info->name.oid = (void *)(base + pathlen + 1);
|
2008-03-07 02:44:48 +03:00
|
|
|
if (pathlen)
|
|
|
|
info->prev = &dummy;
|
2008-03-06 05:59:29 +03:00
|
|
|
}
|
|
|
|
|
|
|
|
char *make_traverse_path(char *path, const struct traverse_info *info, const struct name_entry *n)
|
|
|
|
{
|
2011-10-24 10:36:09 +04:00
|
|
|
int len = tree_entry_len(n);
|
2008-03-06 05:59:29 +03:00
|
|
|
int pathlen = info->pathlen;
|
|
|
|
|
|
|
|
path[pathlen + len] = 0;
|
|
|
|
for (;;) {
|
|
|
|
memcpy(path + pathlen, n->path, len);
|
|
|
|
if (!pathlen)
|
|
|
|
break;
|
|
|
|
path[--pathlen] = '/';
|
|
|
|
n = &info->name;
|
2011-10-24 10:36:09 +04:00
|
|
|
len = tree_entry_len(n);
|
2008-03-06 05:59:29 +03:00
|
|
|
info = info->prev;
|
|
|
|
pathlen -= len;
|
|
|
|
}
|
|
|
|
return path;
|
|
|
|
}
|
|
|
|
|
traverse_trees(): handle D/F conflict case sanely
traverse_trees() is supposed to call its callback with all the matching
entries from the given trees. The current algorithm keeps a pointer to
each of the tree being traversed, and feeds the entry with the earliest
name to the callback.
This breaks down if the trees being traversed looks like this:
A B
t-1 t
t-2 u
t/a v
When we are currently looking at an entry "t-1" in tree A, and tree B has
returned "t", feeding "t" from the B and not feeding anything from A, only
because "t-1" sorts later than "t", will miss an entry for a subtree "t"
behind the current entry in tree A.
This introduces extended_entry_extract() helper function that gives what
name is expected from the tree, and implements a mechanism to look-ahead
in the tree object using it, to make sure such a case is handled sanely.
Traversal in tree A in the above example will first return "t" to match
that of B, and then the next request for an entry to A then returns "t-1".
This roughly corresponds to what Linus's "prepare for one-entry lookahead"
wanted to do, but because this does implement look ahead, t6035 and one more
test in t1012 reveal that the approach would not work without adjusting the
side that walks the index in unpack_trees() as well.
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2009-09-20 01:07:14 +04:00
|
|
|
struct tree_desc_skip {
|
|
|
|
struct tree_desc_skip *prev;
|
|
|
|
const void *ptr;
|
|
|
|
};
|
|
|
|
|
|
|
|
struct tree_desc_x {
|
|
|
|
struct tree_desc d;
|
|
|
|
struct tree_desc_skip *skip;
|
|
|
|
};
|
|
|
|
|
|
|
|
static int check_entry_match(const char *a, int a_len, const char *b, int b_len)
|
|
|
|
{
|
|
|
|
/*
|
|
|
|
* The caller wants to pick *a* from a tree or nothing.
|
|
|
|
* We are looking at *b* in a tree.
|
|
|
|
*
|
|
|
|
* (0) If a and b are the same name, we are trivially happy.
|
|
|
|
*
|
|
|
|
* There are three possibilities where *a* could be hiding
|
|
|
|
* behind *b*.
|
|
|
|
*
|
|
|
|
* (1) *a* == "t", *b* == "ab" i.e. *b* sorts earlier than *a* no
|
|
|
|
* matter what.
|
|
|
|
* (2) *a* == "t", *b* == "t-2" and "t" is a subtree in the tree;
|
|
|
|
* (3) *a* == "t-2", *b* == "t" and "t-2" is a blob in the tree.
|
|
|
|
*
|
|
|
|
* Otherwise we know *a* won't appear in the tree without
|
|
|
|
* scanning further.
|
|
|
|
*/
|
|
|
|
|
|
|
|
int cmp = name_compare(a, a_len, b, b_len);
|
|
|
|
|
|
|
|
/* Most common case first -- reading sync'd trees */
|
|
|
|
if (!cmp)
|
|
|
|
return cmp;
|
|
|
|
|
|
|
|
if (0 < cmp) {
|
|
|
|
/* a comes after b; it does not matter if it is case (3)
|
|
|
|
if (b_len < a_len && !memcmp(a, b, b_len) && a[b_len] < '/')
|
|
|
|
return 1;
|
|
|
|
*/
|
|
|
|
return 1; /* keep looking */
|
|
|
|
}
|
|
|
|
|
|
|
|
/* b comes after a; are we looking at case (2)? */
|
|
|
|
if (a_len < b_len && !memcmp(a, b, a_len) && b[a_len] < '/')
|
|
|
|
return 1; /* keep looking */
|
|
|
|
|
|
|
|
return -1; /* a cannot appear in the tree */
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* From the extended tree_desc, extract the first name entry, while
|
|
|
|
* paying attention to the candidate "first" name. Most importantly,
|
|
|
|
* when looking for an entry, if there are entries that sorts earlier
|
|
|
|
* in the tree object representation than that name, skip them and
|
|
|
|
* process the named entry first. We will remember that we haven't
|
|
|
|
* processed the first entry yet, and in the later call skip the
|
|
|
|
* entry we processed early when update_extended_entry() is called.
|
|
|
|
*
|
|
|
|
* E.g. if the underlying tree object has these entries:
|
|
|
|
*
|
|
|
|
* blob "t-1"
|
|
|
|
* blob "t-2"
|
|
|
|
* tree "t"
|
|
|
|
* blob "t=1"
|
|
|
|
*
|
|
|
|
* and the "first" asks for "t", remember that we still need to
|
|
|
|
* process "t-1" and "t-2" but extract "t". After processing the
|
|
|
|
* entry "t" from this call, the caller will let us know by calling
|
|
|
|
* update_extended_entry() that we can remember "t" has been processed
|
|
|
|
* already.
|
|
|
|
*/
|
|
|
|
|
|
|
|
static void extended_entry_extract(struct tree_desc_x *t,
|
|
|
|
struct name_entry *a,
|
|
|
|
const char *first,
|
|
|
|
int first_len)
|
|
|
|
{
|
|
|
|
const char *path;
|
|
|
|
int len;
|
|
|
|
struct tree_desc probe;
|
|
|
|
struct tree_desc_skip *skip;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Extract the first entry from the tree_desc, but skip the
|
|
|
|
* ones that we already returned in earlier rounds.
|
|
|
|
*/
|
|
|
|
while (1) {
|
|
|
|
if (!t->d.size) {
|
|
|
|
entry_clear(a);
|
|
|
|
break; /* not found */
|
|
|
|
}
|
|
|
|
entry_extract(&t->d, a);
|
|
|
|
for (skip = t->skip; skip; skip = skip->prev)
|
|
|
|
if (a->path == skip->ptr)
|
|
|
|
break; /* found */
|
|
|
|
if (!skip)
|
|
|
|
break;
|
|
|
|
/* We have processed this entry already. */
|
|
|
|
update_tree_entry(&t->d);
|
|
|
|
}
|
|
|
|
|
|
|
|
if (!first || !a->path)
|
|
|
|
return;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* The caller wants "first" from this tree, or nothing.
|
|
|
|
*/
|
|
|
|
path = a->path;
|
2011-10-24 10:36:09 +04:00
|
|
|
len = tree_entry_len(a);
|
traverse_trees(): handle D/F conflict case sanely
traverse_trees() is supposed to call its callback with all the matching
entries from the given trees. The current algorithm keeps a pointer to
each of the tree being traversed, and feeds the entry with the earliest
name to the callback.
This breaks down if the trees being traversed looks like this:
A B
t-1 t
t-2 u
t/a v
When we are currently looking at an entry "t-1" in tree A, and tree B has
returned "t", feeding "t" from the B and not feeding anything from A, only
because "t-1" sorts later than "t", will miss an entry for a subtree "t"
behind the current entry in tree A.
This introduces extended_entry_extract() helper function that gives what
name is expected from the tree, and implements a mechanism to look-ahead
in the tree object using it, to make sure such a case is handled sanely.
Traversal in tree A in the above example will first return "t" to match
that of B, and then the next request for an entry to A then returns "t-1".
This roughly corresponds to what Linus's "prepare for one-entry lookahead"
wanted to do, but because this does implement look ahead, t6035 and one more
test in t1012 reveal that the approach would not work without adjusting the
side that walks the index in unpack_trees() as well.
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2009-09-20 01:07:14 +04:00
|
|
|
switch (check_entry_match(first, first_len, path, len)) {
|
|
|
|
case -1:
|
|
|
|
entry_clear(a);
|
|
|
|
case 0:
|
|
|
|
return;
|
|
|
|
default:
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* We need to look-ahead -- we suspect that a subtree whose
|
|
|
|
* name is "first" may be hiding behind the current entry "path".
|
|
|
|
*/
|
|
|
|
probe = t->d;
|
|
|
|
while (probe.size) {
|
|
|
|
entry_extract(&probe, a);
|
|
|
|
path = a->path;
|
2011-10-24 10:36:09 +04:00
|
|
|
len = tree_entry_len(a);
|
traverse_trees(): handle D/F conflict case sanely
traverse_trees() is supposed to call its callback with all the matching
entries from the given trees. The current algorithm keeps a pointer to
each of the tree being traversed, and feeds the entry with the earliest
name to the callback.
This breaks down if the trees being traversed looks like this:
A B
t-1 t
t-2 u
t/a v
When we are currently looking at an entry "t-1" in tree A, and tree B has
returned "t", feeding "t" from the B and not feeding anything from A, only
because "t-1" sorts later than "t", will miss an entry for a subtree "t"
behind the current entry in tree A.
This introduces extended_entry_extract() helper function that gives what
name is expected from the tree, and implements a mechanism to look-ahead
in the tree object using it, to make sure such a case is handled sanely.
Traversal in tree A in the above example will first return "t" to match
that of B, and then the next request for an entry to A then returns "t-1".
This roughly corresponds to what Linus's "prepare for one-entry lookahead"
wanted to do, but because this does implement look ahead, t6035 and one more
test in t1012 reveal that the approach would not work without adjusting the
side that walks the index in unpack_trees() as well.
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2009-09-20 01:07:14 +04:00
|
|
|
switch (check_entry_match(first, first_len, path, len)) {
|
|
|
|
case -1:
|
|
|
|
entry_clear(a);
|
|
|
|
case 0:
|
|
|
|
return;
|
|
|
|
default:
|
|
|
|
update_tree_entry(&probe);
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
/* keep looking */
|
|
|
|
}
|
|
|
|
entry_clear(a);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void update_extended_entry(struct tree_desc_x *t, struct name_entry *a)
|
|
|
|
{
|
|
|
|
if (t->d.entry.path == a->path) {
|
|
|
|
update_tree_entry(&t->d);
|
|
|
|
} else {
|
|
|
|
/* we have returned this entry early */
|
|
|
|
struct tree_desc_skip *skip = xmalloc(sizeof(*skip));
|
|
|
|
skip->ptr = a->path;
|
|
|
|
skip->prev = t->skip;
|
|
|
|
t->skip = skip;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
static void free_extended_entry(struct tree_desc_x *t)
|
|
|
|
{
|
|
|
|
struct tree_desc_skip *p, *s;
|
|
|
|
|
|
|
|
for (s = t->skip; s; s = p) {
|
|
|
|
p = s->prev;
|
|
|
|
free(s);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2011-08-29 23:26:05 +04:00
|
|
|
static inline int prune_traversal(struct name_entry *e,
|
|
|
|
struct traverse_info *info,
|
|
|
|
struct strbuf *base,
|
|
|
|
int still_interesting)
|
|
|
|
{
|
|
|
|
if (!info->pathspec || still_interesting == 2)
|
|
|
|
return 2;
|
|
|
|
if (still_interesting < 0)
|
|
|
|
return still_interesting;
|
|
|
|
return tree_entry_interesting(e, base, 0, info->pathspec);
|
|
|
|
}
|
|
|
|
|
2008-03-06 06:44:06 +03:00
|
|
|
int traverse_trees(int n, struct tree_desc *t, struct traverse_info *info)
|
2006-03-30 10:55:43 +04:00
|
|
|
{
|
2010-08-11 12:38:07 +04:00
|
|
|
int error = 0;
|
2006-03-30 10:55:43 +04:00
|
|
|
struct name_entry *entry = xmalloc(n*sizeof(*entry));
|
traverse_trees(): handle D/F conflict case sanely
traverse_trees() is supposed to call its callback with all the matching
entries from the given trees. The current algorithm keeps a pointer to
each of the tree being traversed, and feeds the entry with the earliest
name to the callback.
This breaks down if the trees being traversed looks like this:
A B
t-1 t
t-2 u
t/a v
When we are currently looking at an entry "t-1" in tree A, and tree B has
returned "t", feeding "t" from the B and not feeding anything from A, only
because "t-1" sorts later than "t", will miss an entry for a subtree "t"
behind the current entry in tree A.
This introduces extended_entry_extract() helper function that gives what
name is expected from the tree, and implements a mechanism to look-ahead
in the tree object using it, to make sure such a case is handled sanely.
Traversal in tree A in the above example will first return "t" to match
that of B, and then the next request for an entry to A then returns "t-1".
This roughly corresponds to what Linus's "prepare for one-entry lookahead"
wanted to do, but because this does implement look ahead, t6035 and one more
test in t1012 reveal that the approach would not work without adjusting the
side that walks the index in unpack_trees() as well.
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2009-09-20 01:07:14 +04:00
|
|
|
int i;
|
|
|
|
struct tree_desc_x *tx = xcalloc(n, sizeof(*tx));
|
2011-08-29 23:26:05 +04:00
|
|
|
struct strbuf base = STRBUF_INIT;
|
|
|
|
int interesting = 1;
|
2015-12-22 01:34:20 +03:00
|
|
|
char *traverse_path;
|
traverse_trees(): handle D/F conflict case sanely
traverse_trees() is supposed to call its callback with all the matching
entries from the given trees. The current algorithm keeps a pointer to
each of the tree being traversed, and feeds the entry with the earliest
name to the callback.
This breaks down if the trees being traversed looks like this:
A B
t-1 t
t-2 u
t/a v
When we are currently looking at an entry "t-1" in tree A, and tree B has
returned "t", feeding "t" from the B and not feeding anything from A, only
because "t-1" sorts later than "t", will miss an entry for a subtree "t"
behind the current entry in tree A.
This introduces extended_entry_extract() helper function that gives what
name is expected from the tree, and implements a mechanism to look-ahead
in the tree object using it, to make sure such a case is handled sanely.
Traversal in tree A in the above example will first return "t" to match
that of B, and then the next request for an entry to A then returns "t-1".
This roughly corresponds to what Linus's "prepare for one-entry lookahead"
wanted to do, but because this does implement look ahead, t6035 and one more
test in t1012 reveal that the approach would not work without adjusting the
side that walks the index in unpack_trees() as well.
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2009-09-20 01:07:14 +04:00
|
|
|
|
|
|
|
for (i = 0; i < n; i++)
|
|
|
|
tx[i].d = t[i];
|
2006-03-30 10:55:43 +04:00
|
|
|
|
2011-08-29 23:26:05 +04:00
|
|
|
if (info->prev) {
|
|
|
|
strbuf_grow(&base, info->pathlen);
|
|
|
|
make_traverse_path(base.buf, info->prev, &info->name);
|
|
|
|
base.buf[info->pathlen-1] = '/';
|
|
|
|
strbuf_setlen(&base, info->pathlen);
|
2015-12-22 01:34:20 +03:00
|
|
|
traverse_path = xstrndup(base.buf, info->pathlen);
|
|
|
|
} else {
|
|
|
|
traverse_path = xstrndup(info->name.path, info->pathlen);
|
2011-08-29 23:26:05 +04:00
|
|
|
}
|
2015-12-22 01:34:20 +03:00
|
|
|
info->traverse_path = traverse_path;
|
2006-03-30 10:55:43 +04:00
|
|
|
for (;;) {
|
2013-07-20 00:26:32 +04:00
|
|
|
int trees_used;
|
traverse_trees(): handle D/F conflict case sanely
traverse_trees() is supposed to call its callback with all the matching
entries from the given trees. The current algorithm keeps a pointer to
each of the tree being traversed, and feeds the entry with the earliest
name to the callback.
This breaks down if the trees being traversed looks like this:
A B
t-1 t
t-2 u
t/a v
When we are currently looking at an entry "t-1" in tree A, and tree B has
returned "t", feeding "t" from the B and not feeding anything from A, only
because "t-1" sorts later than "t", will miss an entry for a subtree "t"
behind the current entry in tree A.
This introduces extended_entry_extract() helper function that gives what
name is expected from the tree, and implements a mechanism to look-ahead
in the tree object using it, to make sure such a case is handled sanely.
Traversal in tree A in the above example will first return "t" to match
that of B, and then the next request for an entry to A then returns "t-1".
This roughly corresponds to what Linus's "prepare for one-entry lookahead"
wanted to do, but because this does implement look ahead, t6035 and one more
test in t1012 reveal that the approach would not work without adjusting the
side that walks the index in unpack_trees() as well.
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2009-09-20 01:07:14 +04:00
|
|
|
unsigned long mask, dirmask;
|
|
|
|
const char *first = NULL;
|
|
|
|
int first_len = 0;
|
2011-09-11 23:39:32 +04:00
|
|
|
struct name_entry *e = NULL;
|
traverse_trees(): handle D/F conflict case sanely
traverse_trees() is supposed to call its callback with all the matching
entries from the given trees. The current algorithm keeps a pointer to
each of the tree being traversed, and feeds the entry with the earliest
name to the callback.
This breaks down if the trees being traversed looks like this:
A B
t-1 t
t-2 u
t/a v
When we are currently looking at an entry "t-1" in tree A, and tree B has
returned "t", feeding "t" from the B and not feeding anything from A, only
because "t-1" sorts later than "t", will miss an entry for a subtree "t"
behind the current entry in tree A.
This introduces extended_entry_extract() helper function that gives what
name is expected from the tree, and implements a mechanism to look-ahead
in the tree object using it, to make sure such a case is handled sanely.
Traversal in tree A in the above example will first return "t" to match
that of B, and then the next request for an entry to A then returns "t-1".
This roughly corresponds to what Linus's "prepare for one-entry lookahead"
wanted to do, but because this does implement look ahead, t6035 and one more
test in t1012 reveal that the approach would not work without adjusting the
side that walks the index in unpack_trees() as well.
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2009-09-20 01:07:14 +04:00
|
|
|
int len;
|
|
|
|
|
|
|
|
for (i = 0; i < n; i++) {
|
|
|
|
e = entry + i;
|
|
|
|
extended_entry_extract(tx + i, e, NULL, 0);
|
|
|
|
}
|
2006-03-30 10:55:43 +04:00
|
|
|
|
traverse_trees(): handle D/F conflict case sanely
traverse_trees() is supposed to call its callback with all the matching
entries from the given trees. The current algorithm keeps a pointer to
each of the tree being traversed, and feeds the entry with the earliest
name to the callback.
This breaks down if the trees being traversed looks like this:
A B
t-1 t
t-2 u
t/a v
When we are currently looking at an entry "t-1" in tree A, and tree B has
returned "t", feeding "t" from the B and not feeding anything from A, only
because "t-1" sorts later than "t", will miss an entry for a subtree "t"
behind the current entry in tree A.
This introduces extended_entry_extract() helper function that gives what
name is expected from the tree, and implements a mechanism to look-ahead
in the tree object using it, to make sure such a case is handled sanely.
Traversal in tree A in the above example will first return "t" to match
that of B, and then the next request for an entry to A then returns "t-1".
This roughly corresponds to what Linus's "prepare for one-entry lookahead"
wanted to do, but because this does implement look ahead, t6035 and one more
test in t1012 reveal that the approach would not work without adjusting the
side that walks the index in unpack_trees() as well.
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2009-09-20 01:07:14 +04:00
|
|
|
/*
|
|
|
|
* A tree may have "t-2" at the current location even
|
|
|
|
* though it may have "t" that is a subtree behind it,
|
|
|
|
* and another tree may return "t". We want to grab
|
|
|
|
* all "t" from all trees to match in such a case.
|
|
|
|
*/
|
2006-03-30 10:55:43 +04:00
|
|
|
for (i = 0; i < n; i++) {
|
traverse_trees(): handle D/F conflict case sanely
traverse_trees() is supposed to call its callback with all the matching
entries from the given trees. The current algorithm keeps a pointer to
each of the tree being traversed, and feeds the entry with the earliest
name to the callback.
This breaks down if the trees being traversed looks like this:
A B
t-1 t
t-2 u
t/a v
When we are currently looking at an entry "t-1" in tree A, and tree B has
returned "t", feeding "t" from the B and not feeding anything from A, only
because "t-1" sorts later than "t", will miss an entry for a subtree "t"
behind the current entry in tree A.
This introduces extended_entry_extract() helper function that gives what
name is expected from the tree, and implements a mechanism to look-ahead
in the tree object using it, to make sure such a case is handled sanely.
Traversal in tree A in the above example will first return "t" to match
that of B, and then the next request for an entry to A then returns "t-1".
This roughly corresponds to what Linus's "prepare for one-entry lookahead"
wanted to do, but because this does implement look ahead, t6035 and one more
test in t1012 reveal that the approach would not work without adjusting the
side that walks the index in unpack_trees() as well.
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2009-09-20 01:07:14 +04:00
|
|
|
e = entry + i;
|
|
|
|
if (!e->path)
|
2006-03-30 10:55:43 +04:00
|
|
|
continue;
|
2011-10-24 10:36:09 +04:00
|
|
|
len = tree_entry_len(e);
|
traverse_trees(): handle D/F conflict case sanely
traverse_trees() is supposed to call its callback with all the matching
entries from the given trees. The current algorithm keeps a pointer to
each of the tree being traversed, and feeds the entry with the earliest
name to the callback.
This breaks down if the trees being traversed looks like this:
A B
t-1 t
t-2 u
t/a v
When we are currently looking at an entry "t-1" in tree A, and tree B has
returned "t", feeding "t" from the B and not feeding anything from A, only
because "t-1" sorts later than "t", will miss an entry for a subtree "t"
behind the current entry in tree A.
This introduces extended_entry_extract() helper function that gives what
name is expected from the tree, and implements a mechanism to look-ahead
in the tree object using it, to make sure such a case is handled sanely.
Traversal in tree A in the above example will first return "t" to match
that of B, and then the next request for an entry to A then returns "t-1".
This roughly corresponds to what Linus's "prepare for one-entry lookahead"
wanted to do, but because this does implement look ahead, t6035 and one more
test in t1012 reveal that the approach would not work without adjusting the
side that walks the index in unpack_trees() as well.
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2009-09-20 01:07:14 +04:00
|
|
|
if (!first) {
|
|
|
|
first = e->path;
|
|
|
|
first_len = len;
|
|
|
|
continue;
|
|
|
|
}
|
|
|
|
if (name_compare(e->path, len, first, first_len) < 0) {
|
|
|
|
first = e->path;
|
|
|
|
first_len = len;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
if (first) {
|
|
|
|
for (i = 0; i < n; i++) {
|
|
|
|
e = entry + i;
|
|
|
|
extended_entry_extract(tx + i, e, first, first_len);
|
|
|
|
/* Cull the ones that are not the earliest */
|
|
|
|
if (!e->path)
|
2006-03-30 10:55:43 +04:00
|
|
|
continue;
|
2011-10-24 10:36:09 +04:00
|
|
|
len = tree_entry_len(e);
|
traverse_trees(): handle D/F conflict case sanely
traverse_trees() is supposed to call its callback with all the matching
entries from the given trees. The current algorithm keeps a pointer to
each of the tree being traversed, and feeds the entry with the earliest
name to the callback.
This breaks down if the trees being traversed looks like this:
A B
t-1 t
t-2 u
t/a v
When we are currently looking at an entry "t-1" in tree A, and tree B has
returned "t", feeding "t" from the B and not feeding anything from A, only
because "t-1" sorts later than "t", will miss an entry for a subtree "t"
behind the current entry in tree A.
This introduces extended_entry_extract() helper function that gives what
name is expected from the tree, and implements a mechanism to look-ahead
in the tree object using it, to make sure such a case is handled sanely.
Traversal in tree A in the above example will first return "t" to match
that of B, and then the next request for an entry to A then returns "t-1".
This roughly corresponds to what Linus's "prepare for one-entry lookahead"
wanted to do, but because this does implement look ahead, t6035 and one more
test in t1012 reveal that the approach would not work without adjusting the
side that walks the index in unpack_trees() as well.
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2009-09-20 01:07:14 +04:00
|
|
|
if (name_compare(e->path, len, first, first_len))
|
|
|
|
entry_clear(e);
|
2006-03-30 10:55:43 +04:00
|
|
|
}
|
traverse_trees(): handle D/F conflict case sanely
traverse_trees() is supposed to call its callback with all the matching
entries from the given trees. The current algorithm keeps a pointer to
each of the tree being traversed, and feeds the entry with the earliest
name to the callback.
This breaks down if the trees being traversed looks like this:
A B
t-1 t
t-2 u
t/a v
When we are currently looking at an entry "t-1" in tree A, and tree B has
returned "t", feeding "t" from the B and not feeding anything from A, only
because "t-1" sorts later than "t", will miss an entry for a subtree "t"
behind the current entry in tree A.
This introduces extended_entry_extract() helper function that gives what
name is expected from the tree, and implements a mechanism to look-ahead
in the tree object using it, to make sure such a case is handled sanely.
Traversal in tree A in the above example will first return "t" to match
that of B, and then the next request for an entry to A then returns "t-1".
This roughly corresponds to what Linus's "prepare for one-entry lookahead"
wanted to do, but because this does implement look ahead, t6035 and one more
test in t1012 reveal that the approach would not work without adjusting the
side that walks the index in unpack_trees() as well.
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2009-09-20 01:07:14 +04:00
|
|
|
}
|
|
|
|
|
|
|
|
/* Now we have in entry[i] the earliest name from the trees */
|
|
|
|
mask = 0;
|
|
|
|
dirmask = 0;
|
|
|
|
for (i = 0; i < n; i++) {
|
|
|
|
if (!entry[i].path)
|
|
|
|
continue;
|
2006-03-30 10:55:43 +04:00
|
|
|
mask |= 1ul << i;
|
2008-03-06 07:06:18 +03:00
|
|
|
if (S_ISDIR(entry[i].mode))
|
|
|
|
dirmask |= 1ul << i;
|
2011-08-29 23:26:05 +04:00
|
|
|
e = &entry[i];
|
2006-03-30 10:55:43 +04:00
|
|
|
}
|
|
|
|
if (!mask)
|
|
|
|
break;
|
2011-08-29 23:26:05 +04:00
|
|
|
interesting = prune_traversal(e, info, &base, interesting);
|
|
|
|
if (interesting < 0)
|
|
|
|
break;
|
|
|
|
if (interesting) {
|
2013-07-20 00:26:32 +04:00
|
|
|
trees_used = info->fn(n, mask, dirmask, entry, info);
|
|
|
|
if (trees_used < 0) {
|
|
|
|
error = trees_used;
|
2011-08-29 23:26:05 +04:00
|
|
|
if (!info->show_all_errors)
|
|
|
|
break;
|
|
|
|
}
|
2013-07-20 00:26:32 +04:00
|
|
|
mask &= trees_used;
|
2010-08-11 12:38:07 +04:00
|
|
|
}
|
traverse_trees(): handle D/F conflict case sanely
traverse_trees() is supposed to call its callback with all the matching
entries from the given trees. The current algorithm keeps a pointer to
each of the tree being traversed, and feeds the entry with the earliest
name to the callback.
This breaks down if the trees being traversed looks like this:
A B
t-1 t
t-2 u
t/a v
When we are currently looking at an entry "t-1" in tree A, and tree B has
returned "t", feeding "t" from the B and not feeding anything from A, only
because "t-1" sorts later than "t", will miss an entry for a subtree "t"
behind the current entry in tree A.
This introduces extended_entry_extract() helper function that gives what
name is expected from the tree, and implements a mechanism to look-ahead
in the tree object using it, to make sure such a case is handled sanely.
Traversal in tree A in the above example will first return "t" to match
that of B, and then the next request for an entry to A then returns "t-1".
This roughly corresponds to what Linus's "prepare for one-entry lookahead"
wanted to do, but because this does implement look ahead, t6035 and one more
test in t1012 reveal that the approach would not work without adjusting the
side that walks the index in unpack_trees() as well.
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2009-09-20 01:07:14 +04:00
|
|
|
for (i = 0; i < n; i++)
|
2008-03-06 06:44:06 +03:00
|
|
|
if (mask & (1ul << i))
|
traverse_trees(): handle D/F conflict case sanely
traverse_trees() is supposed to call its callback with all the matching
entries from the given trees. The current algorithm keeps a pointer to
each of the tree being traversed, and feeds the entry with the earliest
name to the callback.
This breaks down if the trees being traversed looks like this:
A B
t-1 t
t-2 u
t/a v
When we are currently looking at an entry "t-1" in tree A, and tree B has
returned "t", feeding "t" from the B and not feeding anything from A, only
because "t-1" sorts later than "t", will miss an entry for a subtree "t"
behind the current entry in tree A.
This introduces extended_entry_extract() helper function that gives what
name is expected from the tree, and implements a mechanism to look-ahead
in the tree object using it, to make sure such a case is handled sanely.
Traversal in tree A in the above example will first return "t" to match
that of B, and then the next request for an entry to A then returns "t-1".
This roughly corresponds to what Linus's "prepare for one-entry lookahead"
wanted to do, but because this does implement look ahead, t6035 and one more
test in t1012 reveal that the approach would not work without adjusting the
side that walks the index in unpack_trees() as well.
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2009-09-20 01:07:14 +04:00
|
|
|
update_extended_entry(tx + i, entry + i);
|
2006-03-30 10:55:43 +04:00
|
|
|
}
|
|
|
|
free(entry);
|
traverse_trees(): handle D/F conflict case sanely
traverse_trees() is supposed to call its callback with all the matching
entries from the given trees. The current algorithm keeps a pointer to
each of the tree being traversed, and feeds the entry with the earliest
name to the callback.
This breaks down if the trees being traversed looks like this:
A B
t-1 t
t-2 u
t/a v
When we are currently looking at an entry "t-1" in tree A, and tree B has
returned "t", feeding "t" from the B and not feeding anything from A, only
because "t-1" sorts later than "t", will miss an entry for a subtree "t"
behind the current entry in tree A.
This introduces extended_entry_extract() helper function that gives what
name is expected from the tree, and implements a mechanism to look-ahead
in the tree object using it, to make sure such a case is handled sanely.
Traversal in tree A in the above example will first return "t" to match
that of B, and then the next request for an entry to A then returns "t-1".
This roughly corresponds to what Linus's "prepare for one-entry lookahead"
wanted to do, but because this does implement look ahead, t6035 and one more
test in t1012 reveal that the approach would not work without adjusting the
side that walks the index in unpack_trees() as well.
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2009-09-20 01:07:14 +04:00
|
|
|
for (i = 0; i < n; i++)
|
|
|
|
free_extended_entry(tx + i);
|
|
|
|
free(tx);
|
2015-12-22 01:34:20 +03:00
|
|
|
free(traverse_path);
|
|
|
|
info->traverse_path = NULL;
|
2011-08-29 23:26:05 +04:00
|
|
|
strbuf_release(&base);
|
2010-08-11 12:38:07 +04:00
|
|
|
return error;
|
2006-03-30 10:55:43 +04:00
|
|
|
}
|
|
|
|
|
2015-05-20 20:03:38 +03:00
|
|
|
struct dir_state {
|
|
|
|
void *tree;
|
|
|
|
unsigned long size;
|
|
|
|
unsigned char sha1[20];
|
|
|
|
};
|
|
|
|
|
2006-04-20 01:05:47 +04:00
|
|
|
static int find_tree_entry(struct tree_desc *t, const char *name, unsigned char *result, unsigned *mode)
|
|
|
|
{
|
|
|
|
int namelen = strlen(name);
|
|
|
|
while (t->size) {
|
|
|
|
const char *entry;
|
2016-04-18 02:10:40 +03:00
|
|
|
const struct object_id *oid;
|
2006-04-20 01:05:47 +04:00
|
|
|
int entrylen, cmp;
|
|
|
|
|
2016-04-18 02:10:40 +03:00
|
|
|
oid = tree_entry_extract(t, &entry, mode);
|
2011-10-24 10:36:09 +04:00
|
|
|
entrylen = tree_entry_len(&t->entry);
|
2006-04-20 01:05:47 +04:00
|
|
|
update_tree_entry(t);
|
|
|
|
if (entrylen > namelen)
|
|
|
|
continue;
|
|
|
|
cmp = memcmp(name, entry, entrylen);
|
|
|
|
if (cmp > 0)
|
|
|
|
continue;
|
|
|
|
if (cmp < 0)
|
|
|
|
break;
|
|
|
|
if (entrylen == namelen) {
|
2016-04-18 02:10:40 +03:00
|
|
|
hashcpy(result, oid->hash);
|
2006-04-20 01:05:47 +04:00
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
if (name[entrylen] != '/')
|
|
|
|
continue;
|
|
|
|
if (!S_ISDIR(*mode))
|
|
|
|
break;
|
|
|
|
if (++entrylen == namelen) {
|
2016-04-18 02:10:40 +03:00
|
|
|
hashcpy(result, oid->hash);
|
2006-04-20 01:05:47 +04:00
|
|
|
return 0;
|
|
|
|
}
|
2016-04-18 02:10:40 +03:00
|
|
|
return get_tree_entry(oid->hash, name + entrylen, result, mode);
|
2006-04-20 01:05:47 +04:00
|
|
|
}
|
|
|
|
return -1;
|
|
|
|
}
|
|
|
|
|
|
|
|
int get_tree_entry(const unsigned char *tree_sha1, const char *name, unsigned char *sha1, unsigned *mode)
|
|
|
|
{
|
|
|
|
int retval;
|
|
|
|
void *tree;
|
2007-03-21 20:08:25 +03:00
|
|
|
unsigned long size;
|
2007-01-09 19:11:47 +03:00
|
|
|
unsigned char root[20];
|
2006-04-20 01:05:47 +04:00
|
|
|
|
2007-03-21 20:08:25 +03:00
|
|
|
tree = read_object_with_reference(tree_sha1, tree_type, &size, root);
|
2006-04-20 01:05:47 +04:00
|
|
|
if (!tree)
|
|
|
|
return -1;
|
2007-01-09 19:11:47 +03:00
|
|
|
|
|
|
|
if (name[0] == '\0') {
|
|
|
|
hashcpy(sha1, root);
|
2010-02-14 12:56:46 +03:00
|
|
|
free(tree);
|
2007-01-09 19:11:47 +03:00
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2011-10-27 22:18:40 +04:00
|
|
|
if (!size) {
|
|
|
|
retval = -1;
|
|
|
|
} else {
|
|
|
|
struct tree_desc t;
|
|
|
|
init_tree_desc(&t, tree, size);
|
|
|
|
retval = find_tree_entry(&t, name, sha1, mode);
|
|
|
|
}
|
2006-04-20 01:05:47 +04:00
|
|
|
free(tree);
|
|
|
|
return retval;
|
|
|
|
}
|
2010-12-15 18:02:40 +03:00
|
|
|
|
2015-05-20 20:03:38 +03:00
|
|
|
/*
|
|
|
|
* This is Linux's built-in max for the number of symlinks to follow.
|
|
|
|
* That limit, of course, does not affect git, but it's a reasonable
|
|
|
|
* choice.
|
|
|
|
*/
|
|
|
|
#define GET_TREE_ENTRY_FOLLOW_SYMLINKS_MAX_LINKS 40
|
|
|
|
|
|
|
|
/**
|
|
|
|
* Find a tree entry by following symlinks in tree_sha (which is
|
|
|
|
* assumed to be the root of the repository). In the event that a
|
|
|
|
* symlink points outside the repository (e.g. a link to /foo or a
|
|
|
|
* root-level link to ../foo), the portion of the link which is
|
|
|
|
* outside the repository will be returned in result_path, and *mode
|
|
|
|
* will be set to 0. It is assumed that result_path is uninitialized.
|
|
|
|
* If there are no symlinks, or the end result of the symlink chain
|
|
|
|
* points to an object inside the repository, result will be filled in
|
|
|
|
* with the sha1 of the found object, and *mode will hold the mode of
|
|
|
|
* the object.
|
|
|
|
*
|
|
|
|
* See the code for enum follow_symlink_result for a description of
|
|
|
|
* the return values.
|
|
|
|
*/
|
|
|
|
enum follow_symlinks_result get_tree_entry_follow_symlinks(unsigned char *tree_sha1, const char *name, unsigned char *result, struct strbuf *result_path, unsigned *mode)
|
|
|
|
{
|
|
|
|
int retval = MISSING_OBJECT;
|
|
|
|
struct dir_state *parents = NULL;
|
|
|
|
size_t parents_alloc = 0;
|
2017-09-21 19:49:38 +03:00
|
|
|
size_t i, parents_nr = 0;
|
2015-05-20 20:03:38 +03:00
|
|
|
unsigned char current_tree_sha1[20];
|
|
|
|
struct strbuf namebuf = STRBUF_INIT;
|
|
|
|
struct tree_desc t;
|
|
|
|
int follows_remaining = GET_TREE_ENTRY_FOLLOW_SYMLINKS_MAX_LINKS;
|
|
|
|
|
|
|
|
init_tree_desc(&t, NULL, 0UL);
|
|
|
|
strbuf_addstr(&namebuf, name);
|
|
|
|
hashcpy(current_tree_sha1, tree_sha1);
|
|
|
|
|
|
|
|
while (1) {
|
|
|
|
int find_result;
|
|
|
|
char *first_slash;
|
|
|
|
char *remainder = NULL;
|
|
|
|
|
|
|
|
if (!t.buffer) {
|
|
|
|
void *tree;
|
|
|
|
unsigned char root[20];
|
|
|
|
unsigned long size;
|
|
|
|
tree = read_object_with_reference(current_tree_sha1,
|
|
|
|
tree_type, &size,
|
|
|
|
root);
|
|
|
|
if (!tree)
|
|
|
|
goto done;
|
|
|
|
|
|
|
|
ALLOC_GROW(parents, parents_nr + 1, parents_alloc);
|
|
|
|
parents[parents_nr].tree = tree;
|
|
|
|
parents[parents_nr].size = size;
|
|
|
|
hashcpy(parents[parents_nr].sha1, root);
|
|
|
|
parents_nr++;
|
|
|
|
|
|
|
|
if (namebuf.buf[0] == '\0') {
|
|
|
|
hashcpy(result, root);
|
|
|
|
retval = FOUND;
|
|
|
|
goto done;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (!size)
|
|
|
|
goto done;
|
|
|
|
|
|
|
|
/* descend */
|
|
|
|
init_tree_desc(&t, tree, size);
|
|
|
|
}
|
|
|
|
|
|
|
|
/* Handle symlinks to e.g. a//b by removing leading slashes */
|
|
|
|
while (namebuf.buf[0] == '/') {
|
|
|
|
strbuf_remove(&namebuf, 0, 1);
|
|
|
|
}
|
|
|
|
|
|
|
|
/* Split namebuf into a first component and a remainder */
|
|
|
|
if ((first_slash = strchr(namebuf.buf, '/'))) {
|
|
|
|
*first_slash = 0;
|
|
|
|
remainder = first_slash + 1;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (!strcmp(namebuf.buf, "..")) {
|
|
|
|
struct dir_state *parent;
|
|
|
|
/*
|
|
|
|
* We could end up with .. in the namebuf if it
|
|
|
|
* appears in a symlink.
|
|
|
|
*/
|
|
|
|
|
|
|
|
if (parents_nr == 1) {
|
|
|
|
if (remainder)
|
|
|
|
*first_slash = '/';
|
|
|
|
strbuf_add(result_path, namebuf.buf,
|
|
|
|
namebuf.len);
|
|
|
|
*mode = 0;
|
|
|
|
retval = FOUND;
|
|
|
|
goto done;
|
|
|
|
}
|
|
|
|
parent = &parents[parents_nr - 1];
|
|
|
|
free(parent->tree);
|
|
|
|
parents_nr--;
|
|
|
|
parent = &parents[parents_nr - 1];
|
|
|
|
init_tree_desc(&t, parent->tree, parent->size);
|
|
|
|
strbuf_remove(&namebuf, 0, remainder ? 3 : 2);
|
|
|
|
continue;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* We could end up here via a symlink to dir/.. */
|
|
|
|
if (namebuf.buf[0] == '\0') {
|
|
|
|
hashcpy(result, parents[parents_nr - 1].sha1);
|
|
|
|
retval = FOUND;
|
|
|
|
goto done;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* Look up the first (or only) path component in the tree. */
|
|
|
|
find_result = find_tree_entry(&t, namebuf.buf,
|
|
|
|
current_tree_sha1, mode);
|
|
|
|
if (find_result) {
|
|
|
|
goto done;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (S_ISDIR(*mode)) {
|
|
|
|
if (!remainder) {
|
|
|
|
hashcpy(result, current_tree_sha1);
|
|
|
|
retval = FOUND;
|
|
|
|
goto done;
|
|
|
|
}
|
|
|
|
/* Descend the tree */
|
|
|
|
t.buffer = NULL;
|
|
|
|
strbuf_remove(&namebuf, 0,
|
|
|
|
1 + first_slash - namebuf.buf);
|
|
|
|
} else if (S_ISREG(*mode)) {
|
|
|
|
if (!remainder) {
|
|
|
|
hashcpy(result, current_tree_sha1);
|
|
|
|
retval = FOUND;
|
|
|
|
} else {
|
|
|
|
retval = NOT_DIR;
|
|
|
|
}
|
|
|
|
goto done;
|
|
|
|
} else if (S_ISLNK(*mode)) {
|
|
|
|
/* Follow a symlink */
|
|
|
|
unsigned long link_len;
|
|
|
|
size_t len;
|
|
|
|
char *contents, *contents_start;
|
|
|
|
struct dir_state *parent;
|
|
|
|
enum object_type type;
|
|
|
|
|
|
|
|
if (follows_remaining-- == 0) {
|
|
|
|
/* Too many symlinks followed */
|
|
|
|
retval = SYMLINK_LOOP;
|
|
|
|
goto done;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* At this point, we have followed at a least
|
|
|
|
* one symlink, so on error we need to report this.
|
|
|
|
*/
|
|
|
|
retval = DANGLING_SYMLINK;
|
|
|
|
|
|
|
|
contents = read_sha1_file(current_tree_sha1, &type,
|
|
|
|
&link_len);
|
|
|
|
|
|
|
|
if (!contents)
|
|
|
|
goto done;
|
|
|
|
|
|
|
|
if (contents[0] == '/') {
|
|
|
|
strbuf_addstr(result_path, contents);
|
|
|
|
free(contents);
|
|
|
|
*mode = 0;
|
|
|
|
retval = FOUND;
|
|
|
|
goto done;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (remainder)
|
|
|
|
len = first_slash - namebuf.buf;
|
|
|
|
else
|
|
|
|
len = namebuf.len;
|
|
|
|
|
|
|
|
contents_start = contents;
|
|
|
|
|
|
|
|
parent = &parents[parents_nr - 1];
|
|
|
|
init_tree_desc(&t, parent->tree, parent->size);
|
|
|
|
strbuf_splice(&namebuf, 0, len,
|
|
|
|
contents_start, link_len);
|
|
|
|
if (remainder)
|
|
|
|
namebuf.buf[link_len] = '/';
|
|
|
|
free(contents);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
done:
|
|
|
|
for (i = 0; i < parents_nr; i++)
|
|
|
|
free(parents[i].tree);
|
|
|
|
free(parents);
|
|
|
|
|
|
|
|
strbuf_release(&namebuf);
|
|
|
|
return retval;
|
|
|
|
}
|
|
|
|
|
2013-07-14 12:36:09 +04:00
|
|
|
static int match_entry(const struct pathspec_item *item,
|
|
|
|
const struct name_entry *entry, int pathlen,
|
2010-12-15 18:02:43 +03:00
|
|
|
const char *match, int matchlen,
|
2012-10-19 21:14:42 +04:00
|
|
|
enum interesting *never_interesting)
|
2010-12-15 18:02:43 +03:00
|
|
|
{
|
|
|
|
int m = -1; /* signals that we haven't called strncmp() */
|
|
|
|
|
2013-07-14 12:36:09 +04:00
|
|
|
if (item->magic & PATHSPEC_ICASE)
|
|
|
|
/*
|
|
|
|
* "Never interesting" trick requires exact
|
|
|
|
* matching. We could do something clever with inexact
|
|
|
|
* matching, but it's trickier (and not to forget that
|
|
|
|
* strcasecmp is locale-dependent, at least in
|
|
|
|
* glibc). Just disable it for now. It can't be worse
|
|
|
|
* than the wildcard's codepath of '[Tt][Hi][Is][Ss]'
|
|
|
|
* pattern.
|
|
|
|
*/
|
|
|
|
*never_interesting = entry_not_interesting;
|
|
|
|
else if (*never_interesting != entry_not_interesting) {
|
2010-12-15 18:02:43 +03:00
|
|
|
/*
|
|
|
|
* We have not seen any match that sorts later
|
|
|
|
* than the current path.
|
|
|
|
*/
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Does match sort strictly earlier than path
|
|
|
|
* with their common parts?
|
|
|
|
*/
|
|
|
|
m = strncmp(match, entry->path,
|
|
|
|
(matchlen < pathlen) ? matchlen : pathlen);
|
|
|
|
if (m < 0)
|
|
|
|
return 0;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* If we come here even once, that means there is at
|
|
|
|
* least one pathspec that would sort equal to or
|
|
|
|
* later than the path we are currently looking at.
|
|
|
|
* In other words, if we have never reached this point
|
|
|
|
* after iterating all pathspecs, it means all
|
|
|
|
* pathspecs are either outside of base, or inside the
|
|
|
|
* base but sorts strictly earlier than the current
|
|
|
|
* one. In either case, they will never match the
|
|
|
|
* subsequent entries. In such a case, we initialized
|
|
|
|
* the variable to -1 and that is what will be
|
|
|
|
* returned, allowing the caller to terminate early.
|
|
|
|
*/
|
2012-10-19 21:14:42 +04:00
|
|
|
*never_interesting = entry_not_interesting;
|
2010-12-15 18:02:43 +03:00
|
|
|
}
|
|
|
|
|
|
|
|
if (pathlen > matchlen)
|
|
|
|
return 0;
|
|
|
|
|
|
|
|
if (matchlen > pathlen) {
|
|
|
|
if (match[pathlen] != '/')
|
|
|
|
return 0;
|
2014-01-23 17:22:05 +04:00
|
|
|
if (!S_ISDIR(entry->mode) && !S_ISGITLINK(entry->mode))
|
2010-12-15 18:02:43 +03:00
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (m == -1)
|
|
|
|
/*
|
|
|
|
* we cheated and did not do strncmp(), so we do
|
|
|
|
* that here.
|
|
|
|
*/
|
2013-07-14 12:36:09 +04:00
|
|
|
m = ps_strncmp(item, match, entry->path, pathlen);
|
2010-12-15 18:02:43 +03:00
|
|
|
|
|
|
|
/*
|
|
|
|
* If common part matched earlier then it is a hit,
|
|
|
|
* because we rejected the case where path is not a
|
|
|
|
* leading directory and is shorter than match.
|
|
|
|
*/
|
|
|
|
if (!m)
|
2013-07-14 12:36:09 +04:00
|
|
|
/*
|
|
|
|
* match_entry does not check if the prefix part is
|
|
|
|
* matched case-sensitively. If the entry is a
|
|
|
|
* directory and part of prefix, it'll be rematched
|
|
|
|
* eventually by basecmp with special treatment for
|
|
|
|
* the prefix.
|
|
|
|
*/
|
2010-12-15 18:02:43 +03:00
|
|
|
return 1;
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2013-07-14 12:36:09 +04:00
|
|
|
/* :(icase)-aware string compare */
|
|
|
|
static int basecmp(const struct pathspec_item *item,
|
|
|
|
const char *base, const char *match, int len)
|
|
|
|
{
|
|
|
|
if (item->magic & PATHSPEC_ICASE) {
|
|
|
|
int ret, n = len > item->prefix ? item->prefix : len;
|
|
|
|
ret = strncmp(base, match, n);
|
|
|
|
if (ret)
|
|
|
|
return ret;
|
|
|
|
base += n;
|
|
|
|
match += n;
|
|
|
|
len -= n;
|
|
|
|
}
|
|
|
|
return ps_strncmp(item, base, match, len);
|
|
|
|
}
|
|
|
|
|
|
|
|
static int match_dir_prefix(const struct pathspec_item *item,
|
|
|
|
const char *base,
|
2010-12-15 18:02:43 +03:00
|
|
|
const char *match, int matchlen)
|
|
|
|
{
|
2013-07-14 12:36:09 +04:00
|
|
|
if (basecmp(item, base, match, matchlen))
|
2010-12-15 18:02:43 +03:00
|
|
|
return 0;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* If the base is a subdirectory of a path which
|
|
|
|
* was specified, all of them are interesting.
|
|
|
|
*/
|
|
|
|
if (!matchlen ||
|
|
|
|
base[matchlen] == '/' ||
|
|
|
|
match[matchlen - 1] == '/')
|
|
|
|
return 1;
|
|
|
|
|
|
|
|
/* Just a random prefix match */
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2012-11-24 08:33:51 +04:00
|
|
|
/*
|
|
|
|
* Perform matching on the leading non-wildcard part of
|
|
|
|
* pathspec. item->nowildcard_len must be greater than zero. Return
|
|
|
|
* non-zero if base is matched.
|
|
|
|
*/
|
|
|
|
static int match_wildcard_base(const struct pathspec_item *item,
|
|
|
|
const char *base, int baselen,
|
|
|
|
int *matched)
|
|
|
|
{
|
|
|
|
const char *match = item->match;
|
|
|
|
/* the wildcard part is not considered in this function */
|
|
|
|
int matchlen = item->nowildcard_len;
|
|
|
|
|
|
|
|
if (baselen) {
|
|
|
|
int dirlen;
|
|
|
|
/*
|
|
|
|
* Return early if base is longer than the
|
|
|
|
* non-wildcard part but it does not match.
|
|
|
|
*/
|
|
|
|
if (baselen >= matchlen) {
|
|
|
|
*matched = matchlen;
|
2013-07-14 12:36:09 +04:00
|
|
|
return !basecmp(item, base, match, matchlen);
|
2012-11-24 08:33:51 +04:00
|
|
|
}
|
|
|
|
|
|
|
|
dirlen = matchlen;
|
|
|
|
while (dirlen && match[dirlen - 1] != '/')
|
|
|
|
dirlen--;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Return early if base is shorter than the
|
|
|
|
* non-wildcard part but it does not match. Note that
|
|
|
|
* base ends with '/' so we are sure it really matches
|
|
|
|
* directory
|
|
|
|
*/
|
2013-07-14 12:36:09 +04:00
|
|
|
if (basecmp(item, base, match, baselen))
|
2012-11-24 08:33:51 +04:00
|
|
|
return 0;
|
|
|
|
*matched = baselen;
|
|
|
|
} else
|
|
|
|
*matched = 0;
|
|
|
|
/*
|
|
|
|
* we could have checked entry against the non-wildcard part
|
|
|
|
* that is not in base and does similar never_interesting
|
|
|
|
* optimization as in match_entry. For now just be happy with
|
|
|
|
* base comparison.
|
|
|
|
*/
|
|
|
|
return entry_interesting;
|
|
|
|
}
|
|
|
|
|
2010-12-15 18:02:40 +03:00
|
|
|
/*
|
|
|
|
* Is a tree entry interesting given the pathspec we have?
|
|
|
|
*
|
2010-12-17 15:45:33 +03:00
|
|
|
* Pre-condition: either baselen == base_offset (i.e. empty path)
|
|
|
|
* or base[baselen-1] == '/' (i.e. with trailing slash).
|
2010-12-15 18:02:40 +03:00
|
|
|
*/
|
2013-12-06 11:30:48 +04:00
|
|
|
static enum interesting do_match(const struct name_entry *entry,
|
|
|
|
struct strbuf *base, int base_offset,
|
|
|
|
const struct pathspec *ps,
|
|
|
|
int exclude)
|
2010-12-15 18:02:40 +03:00
|
|
|
{
|
|
|
|
int i;
|
2010-12-17 15:45:33 +03:00
|
|
|
int pathlen, baselen = base->len - base_offset;
|
2012-10-19 21:14:42 +04:00
|
|
|
enum interesting never_interesting = ps->has_wildcard ?
|
2011-10-24 10:36:10 +04:00
|
|
|
entry_not_interesting : all_entries_not_interesting;
|
2010-12-15 18:02:40 +03:00
|
|
|
|
2013-07-14 12:36:06 +04:00
|
|
|
GUARD_PATHSPEC(ps,
|
|
|
|
PATHSPEC_FROMTOP |
|
|
|
|
PATHSPEC_MAXDEPTH |
|
2013-07-14 12:36:08 +04:00
|
|
|
PATHSPEC_LITERAL |
|
2013-07-14 12:36:09 +04:00
|
|
|
PATHSPEC_GLOB |
|
2013-12-06 11:30:48 +04:00
|
|
|
PATHSPEC_ICASE |
|
|
|
|
PATHSPEC_EXCLUDE);
|
2013-07-14 12:35:36 +04:00
|
|
|
|
2010-12-15 18:02:44 +03:00
|
|
|
if (!ps->nr) {
|
2013-07-14 12:35:32 +04:00
|
|
|
if (!ps->recursive ||
|
|
|
|
!(ps->magic & PATHSPEC_MAXDEPTH) ||
|
|
|
|
ps->max_depth == -1)
|
2011-10-24 10:36:10 +04:00
|
|
|
return all_entries_interesting;
|
|
|
|
return within_depth(base->buf + base_offset, baselen,
|
|
|
|
!!S_ISDIR(entry->mode),
|
|
|
|
ps->max_depth) ?
|
|
|
|
entry_interesting : entry_not_interesting;
|
2010-12-15 18:02:44 +03:00
|
|
|
}
|
2010-12-15 18:02:40 +03:00
|
|
|
|
2011-10-24 10:36:09 +04:00
|
|
|
pathlen = tree_entry_len(entry);
|
2010-12-15 18:02:40 +03:00
|
|
|
|
2010-12-17 15:45:33 +03:00
|
|
|
for (i = ps->nr - 1; i >= 0; i--) {
|
2010-12-15 18:02:40 +03:00
|
|
|
const struct pathspec_item *item = ps->items+i;
|
|
|
|
const char *match = item->match;
|
2010-12-17 15:45:33 +03:00
|
|
|
const char *base_str = base->buf + base_offset;
|
2012-11-24 08:33:51 +04:00
|
|
|
int matchlen = item->len, matched = 0;
|
2010-12-15 18:02:40 +03:00
|
|
|
|
2013-12-06 11:30:48 +04:00
|
|
|
if ((!exclude && item->magic & PATHSPEC_EXCLUDE) ||
|
|
|
|
( exclude && !(item->magic & PATHSPEC_EXCLUDE)))
|
|
|
|
continue;
|
|
|
|
|
2010-12-15 18:02:40 +03:00
|
|
|
if (baselen >= matchlen) {
|
|
|
|
/* If it doesn't match, move along... */
|
2013-07-14 12:36:09 +04:00
|
|
|
if (!match_dir_prefix(item, base_str, match, matchlen))
|
2010-12-15 18:02:46 +03:00
|
|
|
goto match_wildcards;
|
2010-12-15 18:02:44 +03:00
|
|
|
|
2013-07-14 12:35:32 +04:00
|
|
|
if (!ps->recursive ||
|
|
|
|
!(ps->magic & PATHSPEC_MAXDEPTH) ||
|
|
|
|
ps->max_depth == -1)
|
2011-10-24 10:36:10 +04:00
|
|
|
return all_entries_interesting;
|
2010-12-15 18:02:44 +03:00
|
|
|
|
2011-10-24 10:36:10 +04:00
|
|
|
return within_depth(base_str + matchlen + 1,
|
|
|
|
baselen - matchlen - 1,
|
|
|
|
!!S_ISDIR(entry->mode),
|
|
|
|
ps->max_depth) ?
|
|
|
|
entry_interesting : entry_not_interesting;
|
2010-12-15 18:02:40 +03:00
|
|
|
}
|
|
|
|
|
tree-walk: micro-optimization in tree_entry_interesting
In the case of a wide breadth top-level tree (~2400 entries, all trees
in this case), we can see a noticeable cost in the profiler calling
strncmp() here. Most of the time we are at the base level of the
repository, so base is "" and baselen == 0, which means we will always
test true. Break out this one tiny case so we can short circuit the
strncmp() call.
Test cases are as follows. packages.git is the Arch Linux git-svn clone
of the packages repository which has the characteristics above.
Commands:
[1] packages.git, /usr/bin/time git log >/dev/null
[2] packages.git, /usr/bin/time git log -- autogen/trunk pacman/trunk wget/trunk >/dev/null
[3] linux.git, /usr/bin/time git log >/dev/null
[4] linux.git, /usr/bin/time git log -- drivers/ata drivers/uio tools >/dev/null
Results:
before after %faster
[1] 2.56 2.55 0.4%
[2] 51.82 48.66 6.5%
[3] 5.58 5.61 -0.5%
[4] 1.55 1.51 0.2%
The takeaway here is this doesn't matter in many operations, but it does
for a certain style of repository and operation where it nets a 6.5%
measured improvement. The other changes are likely not significant by
reasonable statistics methods.
Note: the measured improvement when originally submitted was ~11% (43 to
38 secs) for operation [2]. At the time, the repository had 117220
commits; it now has 137537 commits.
Signed-off-by: Dan McGee <dpmcgee@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2011-09-09 06:02:46 +04:00
|
|
|
/* Either there must be no base, or the base must match. */
|
2013-07-14 12:36:09 +04:00
|
|
|
if (baselen == 0 || !basecmp(item, base_str, match, baselen)) {
|
|
|
|
if (match_entry(item, entry, pathlen,
|
2010-12-15 18:02:43 +03:00
|
|
|
match + baselen, matchlen - baselen,
|
|
|
|
&never_interesting))
|
2011-10-24 10:36:10 +04:00
|
|
|
return entry_interesting;
|
2010-12-15 18:02:47 +03:00
|
|
|
|
2012-11-18 13:13:06 +04:00
|
|
|
if (item->nowildcard_len < item->len) {
|
2013-07-14 12:36:08 +04:00
|
|
|
if (!git_fnmatch(item, match + baselen, entry->path,
|
2012-11-24 08:33:50 +04:00
|
|
|
item->nowildcard_len - baselen))
|
2011-10-24 10:36:10 +04:00
|
|
|
return entry_interesting;
|
2010-12-15 18:02:47 +03:00
|
|
|
|
|
|
|
/*
|
|
|
|
* Match all directories. We'll try to
|
|
|
|
* match files later on.
|
|
|
|
*/
|
|
|
|
if (ps->recursive && S_ISDIR(entry->mode))
|
2011-10-24 10:36:10 +04:00
|
|
|
return entry_interesting;
|
2016-12-16 22:03:21 +03:00
|
|
|
|
|
|
|
/*
|
|
|
|
* When matching against submodules with
|
|
|
|
* wildcard characters, ensure that the entry
|
|
|
|
* at least matches up to the first wild
|
|
|
|
* character. More accurate matching can then
|
|
|
|
* be performed in the submodule itself.
|
|
|
|
*/
|
|
|
|
if (ps->recursive && S_ISGITLINK(entry->mode) &&
|
|
|
|
!ps_strncmp(item, match + baselen,
|
|
|
|
entry->path,
|
|
|
|
item->nowildcard_len - baselen))
|
|
|
|
return entry_interesting;
|
2010-12-15 18:02:47 +03:00
|
|
|
}
|
|
|
|
|
|
|
|
continue;
|
2010-12-15 18:02:40 +03:00
|
|
|
}
|
2010-12-15 18:02:46 +03:00
|
|
|
|
|
|
|
match_wildcards:
|
2012-11-18 13:13:06 +04:00
|
|
|
if (item->nowildcard_len == item->len)
|
2010-12-15 18:02:46 +03:00
|
|
|
continue;
|
|
|
|
|
2012-11-24 08:33:51 +04:00
|
|
|
if (item->nowildcard_len &&
|
|
|
|
!match_wildcard_base(item, base_str, baselen, &matched))
|
2014-01-26 02:06:46 +04:00
|
|
|
continue;
|
2012-11-24 08:33:51 +04:00
|
|
|
|
2010-12-15 18:02:46 +03:00
|
|
|
/*
|
|
|
|
* Concatenate base and entry->path into one and do
|
|
|
|
* fnmatch() on it.
|
2012-11-24 08:33:51 +04:00
|
|
|
*
|
|
|
|
* While we could avoid concatenation in certain cases
|
|
|
|
* [1], which saves a memcpy and potentially a
|
|
|
|
* realloc, it turns out not worth it. Measurement on
|
|
|
|
* linux-2.6 does not show any clear improvements,
|
|
|
|
* partly because of the nowildcard_len optimization
|
|
|
|
* in git_fnmatch(). Avoid micro-optimizations here.
|
|
|
|
*
|
|
|
|
* [1] if match_wildcard_base() says the base
|
|
|
|
* directory is already matched, we only need to match
|
|
|
|
* the rest, which is shorter so _in theory_ faster.
|
2010-12-15 18:02:46 +03:00
|
|
|
*/
|
|
|
|
|
|
|
|
strbuf_add(base, entry->path, pathlen);
|
|
|
|
|
2013-07-14 12:36:08 +04:00
|
|
|
if (!git_fnmatch(item, match, base->buf + base_offset,
|
2012-11-24 08:33:50 +04:00
|
|
|
item->nowildcard_len)) {
|
2010-12-17 15:45:33 +03:00
|
|
|
strbuf_setlen(base, base_offset + baselen);
|
2011-10-24 10:36:10 +04:00
|
|
|
return entry_interesting;
|
2010-12-15 18:02:46 +03:00
|
|
|
}
|
2016-12-16 22:03:21 +03:00
|
|
|
|
|
|
|
/*
|
|
|
|
* When matching against submodules with
|
|
|
|
* wildcard characters, ensure that the entry
|
|
|
|
* at least matches up to the first wild
|
|
|
|
* character. More accurate matching can then
|
|
|
|
* be performed in the submodule itself.
|
|
|
|
*/
|
|
|
|
if (ps->recursive && S_ISGITLINK(entry->mode) &&
|
|
|
|
!ps_strncmp(item, match, base->buf + base_offset,
|
|
|
|
item->nowildcard_len)) {
|
|
|
|
strbuf_setlen(base, base_offset + baselen);
|
|
|
|
return entry_interesting;
|
|
|
|
}
|
|
|
|
|
2010-12-17 15:45:33 +03:00
|
|
|
strbuf_setlen(base, base_offset + baselen);
|
2010-12-15 18:02:46 +03:00
|
|
|
|
|
|
|
/*
|
|
|
|
* Match all directories. We'll try to match files
|
|
|
|
* later on.
|
2012-01-14 13:23:22 +04:00
|
|
|
* max_depth is ignored but we may consider support it
|
|
|
|
* in future, see
|
2017-05-08 04:38:59 +03:00
|
|
|
* https://public-inbox.org/git/7vmxo5l2g4.fsf@alter.siamese.dyndns.org/
|
2010-12-15 18:02:46 +03:00
|
|
|
*/
|
|
|
|
if (ps->recursive && S_ISDIR(entry->mode))
|
2011-10-24 10:36:10 +04:00
|
|
|
return entry_interesting;
|
2010-12-15 18:02:40 +03:00
|
|
|
}
|
|
|
|
return never_interesting; /* No matches */
|
|
|
|
}
|
2013-12-06 11:30:48 +04:00
|
|
|
|
|
|
|
/*
|
|
|
|
* Is a tree entry interesting given the pathspec we have?
|
|
|
|
*
|
|
|
|
* Pre-condition: either baselen == base_offset (i.e. empty path)
|
|
|
|
* or base[baselen-1] == '/' (i.e. with trailing slash).
|
|
|
|
*/
|
|
|
|
enum interesting tree_entry_interesting(const struct name_entry *entry,
|
|
|
|
struct strbuf *base, int base_offset,
|
|
|
|
const struct pathspec *ps)
|
|
|
|
{
|
|
|
|
enum interesting positive, negative;
|
|
|
|
positive = do_match(entry, base, base_offset, ps, 0);
|
|
|
|
|
|
|
|
/*
|
|
|
|
* case | entry | positive | negative | result
|
|
|
|
* -----+-------+----------+----------+-------
|
|
|
|
* 1 | file | -1 | -1..2 | -1
|
|
|
|
* 2 | file | 0 | -1..2 | 0
|
|
|
|
* 3 | file | 1 | -1 | 1
|
|
|
|
* 4 | file | 1 | 0 | 1
|
|
|
|
* 5 | file | 1 | 1 | 0
|
|
|
|
* 6 | file | 1 | 2 | 0
|
|
|
|
* 7 | file | 2 | -1 | 2
|
|
|
|
* 8 | file | 2 | 0 | 2
|
|
|
|
* 9 | file | 2 | 1 | 0
|
|
|
|
* 10 | file | 2 | 2 | -1
|
|
|
|
* -----+-------+----------+----------+-------
|
|
|
|
* 11 | dir | -1 | -1..2 | -1
|
|
|
|
* 12 | dir | 0 | -1..2 | 0
|
|
|
|
* 13 | dir | 1 | -1 | 1
|
|
|
|
* 14 | dir | 1 | 0 | 1
|
|
|
|
* 15 | dir | 1 | 1 | 1 (*)
|
|
|
|
* 16 | dir | 1 | 2 | 0
|
|
|
|
* 17 | dir | 2 | -1 | 2
|
|
|
|
* 18 | dir | 2 | 0 | 2
|
|
|
|
* 19 | dir | 2 | 1 | 1 (*)
|
|
|
|
* 20 | dir | 2 | 2 | -1
|
|
|
|
*
|
|
|
|
* (*) An exclude pattern interested in a directory does not
|
|
|
|
* necessarily mean it will exclude all of the directory. In
|
|
|
|
* wildcard case, it can't decide until looking at individual
|
|
|
|
* files inside. So don't write such directories off yet.
|
|
|
|
*/
|
|
|
|
|
|
|
|
if (!(ps->magic & PATHSPEC_EXCLUDE) ||
|
|
|
|
positive <= entry_not_interesting) /* #1, #2, #11, #12 */
|
|
|
|
return positive;
|
|
|
|
|
|
|
|
negative = do_match(entry, base, base_offset, ps, 1);
|
|
|
|
|
|
|
|
/* #3, #4, #7, #8, #13, #14, #17, #18 */
|
|
|
|
if (negative <= entry_not_interesting)
|
|
|
|
return positive;
|
|
|
|
|
|
|
|
/* #15, #19 */
|
|
|
|
if (S_ISDIR(entry->mode) &&
|
|
|
|
positive >= entry_interesting &&
|
|
|
|
negative == entry_interesting)
|
|
|
|
return entry_interesting;
|
|
|
|
|
|
|
|
if ((positive == entry_interesting &&
|
|
|
|
negative >= entry_interesting) || /* #5, #6, #16 */
|
|
|
|
(positive == all_entries_interesting &&
|
|
|
|
negative == entry_interesting)) /* #9 */
|
|
|
|
return entry_not_interesting;
|
|
|
|
|
|
|
|
return all_entries_not_interesting; /* #10, #20 */
|
|
|
|
}
|