change hard-coded perl path to variable
[platform/upstream/git.git] / tree-walk.c
index 3f54c02..4dc86c7 100644 (file)
@@ -3,6 +3,7 @@
 #include "unpack-trees.h"
 #include "dir.h"
 #include "tree.h"
+#include "pathspec.h"
 
 static const char *get_mode(const char *str, unsigned int *modep)
 {
@@ -36,7 +37,7 @@ static void decode_tree_entry(struct tree_desc *desc, const char *buf, unsigned
 
        /* Initialize the descriptor entry */
        desc->entry.path = path;
-       desc->entry.mode = mode;
+       desc->entry.mode = canon_mode(mode);
        desc->entry.sha1 = (const unsigned char *)(path + len);
 }
 
@@ -323,7 +324,6 @@ static inline int prune_traversal(struct name_entry *e,
 
 int traverse_trees(int n, struct tree_desc *t, struct traverse_info *info)
 {
-       int ret = 0;
        int error = 0;
        struct name_entry *entry = xmalloc(n*sizeof(*entry));
        int i;
@@ -341,6 +341,7 @@ int traverse_trees(int n, struct tree_desc *t, struct traverse_info *info)
                strbuf_setlen(&base, info->pathlen);
        }
        for (;;) {
+               int trees_used;
                unsigned long mask, dirmask;
                const char *first = NULL;
                int first_len = 0;
@@ -404,15 +405,14 @@ int traverse_trees(int n, struct tree_desc *t, struct traverse_info *info)
                if (interesting < 0)
                        break;
                if (interesting) {
-                       ret = info->fn(n, mask, dirmask, entry, info);
-                       if (ret < 0) {
-                               error = ret;
+                       trees_used = info->fn(n, mask, dirmask, entry, info);
+                       if (trees_used < 0) {
+                               error = trees_used;
                                if (!info->show_all_errors)
                                        break;
                        }
-                       mask &= ret;
+                       mask &= trees_used;
                }
-               ret = 0;
                for (i = 0; i < n; i++)
                        if (mask & (1ul << i))
                                update_extended_entry(tx + i, entry + i);
@@ -488,13 +488,25 @@ int get_tree_entry(const unsigned char *tree_sha1, const char *name, unsigned ch
        return retval;
 }
 
-static int match_entry(const struct name_entry *entry, int pathlen,
+static int match_entry(const struct pathspec_item *item,
+                      const struct name_entry *entry, int pathlen,
                       const char *match, int matchlen,
                       enum interesting *never_interesting)
 {
        int m = -1; /* signals that we haven't called strncmp() */
 
-       if (*never_interesting != entry_not_interesting) {
+       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) {
                /*
                 * We have not seen any match that sorts later
                 * than the current path.
@@ -531,7 +543,7 @@ static int match_entry(const struct name_entry *entry, int pathlen,
        if (matchlen > pathlen) {
                if (match[pathlen] != '/')
                        return 0;
-               if (!S_ISDIR(entry->mode))
+               if (!S_ISDIR(entry->mode) && !S_ISGITLINK(entry->mode))
                        return 0;
        }
 
@@ -540,7 +552,7 @@ static int match_entry(const struct name_entry *entry, int pathlen,
                 * we cheated and did not do strncmp(), so we do
                 * that here.
                 */
-               m = strncmp(match, entry->path, pathlen);
+               m = ps_strncmp(item, match, entry->path, pathlen);
 
        /*
         * If common part matched earlier then it is a hit,
@@ -548,15 +560,39 @@ static int match_entry(const struct name_entry *entry, int pathlen,
         * leading directory and is shorter than match.
         */
        if (!m)
+               /*
+                * 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.
+                */
                return 1;
 
        return 0;
 }
 
-static int match_dir_prefix(const char *base,
+/* :(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,
                            const char *match, int matchlen)
 {
-       if (strncmp(base, match, matchlen))
+       if (basecmp(item, base, match, matchlen))
                return 0;
 
        /*
@@ -573,22 +609,81 @@ static int match_dir_prefix(const char *base,
 }
 
 /*
+ * 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;
+                       return !basecmp(item, base, match, matchlen);
+               }
+
+               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
+                */
+               if (basecmp(item, base, match, baselen))
+                       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;
+}
+
+/*
  * 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)
+static enum interesting do_match(const struct name_entry *entry,
+                                struct strbuf *base, int base_offset,
+                                const struct pathspec *ps,
+                                int exclude)
 {
        int i;
        int pathlen, baselen = base->len - base_offset;
        enum interesting never_interesting = ps->has_wildcard ?
                entry_not_interesting : all_entries_not_interesting;
 
+       GUARD_PATHSPEC(ps,
+                      PATHSPEC_FROMTOP |
+                      PATHSPEC_MAXDEPTH |
+                      PATHSPEC_LITERAL |
+                      PATHSPEC_GLOB |
+                      PATHSPEC_ICASE |
+                      PATHSPEC_EXCLUDE);
+
        if (!ps->nr) {
-               if (!ps->recursive || ps->max_depth == -1)
+               if (!ps->recursive ||
+                   !(ps->magic & PATHSPEC_MAXDEPTH) ||
+                   ps->max_depth == -1)
                        return all_entries_interesting;
                return within_depth(base->buf + base_offset, baselen,
                                    !!S_ISDIR(entry->mode),
@@ -602,14 +697,20 @@ enum interesting tree_entry_interesting(const struct name_entry *entry,
                const struct pathspec_item *item = ps->items+i;
                const char *match = item->match;
                const char *base_str = base->buf + base_offset;
-               int matchlen = item->len;
+               int matchlen = item->len, matched = 0;
+
+               if ((!exclude &&   item->magic & PATHSPEC_EXCLUDE) ||
+                   ( exclude && !(item->magic & PATHSPEC_EXCLUDE)))
+                       continue;
 
                if (baselen >= matchlen) {
                        /* If it doesn't match, move along... */
-                       if (!match_dir_prefix(base_str, match, matchlen))
+                       if (!match_dir_prefix(item, base_str, match, matchlen))
                                goto match_wildcards;
 
-                       if (!ps->recursive || ps->max_depth == -1)
+                       if (!ps->recursive ||
+                           !(ps->magic & PATHSPEC_MAXDEPTH) ||
+                           ps->max_depth == -1)
                                return all_entries_interesting;
 
                        return within_depth(base_str + matchlen + 1,
@@ -620,14 +721,15 @@ enum interesting tree_entry_interesting(const struct name_entry *entry,
                }
 
                /* Either there must be no base, or the base must match. */
-               if (baselen == 0 || !strncmp(base_str, match, baselen)) {
-                       if (match_entry(entry, pathlen,
+               if (baselen == 0 || !basecmp(item, base_str, match, baselen)) {
+                       if (match_entry(item, entry, pathlen,
                                        match + baselen, matchlen - baselen,
                                        &never_interesting))
                                return entry_interesting;
 
-                       if (item->use_wildcard) {
-                               if (!fnmatch(match + baselen, entry->path, 0))
+                       if (item->nowildcard_len < item->len) {
+                               if (!git_fnmatch(item, match + baselen, entry->path,
+                                                item->nowildcard_len - baselen))
                                        return entry_interesting;
 
                                /*
@@ -642,17 +744,33 @@ enum interesting tree_entry_interesting(const struct name_entry *entry,
                }
 
 match_wildcards:
-               if (!item->use_wildcard)
+               if (item->nowildcard_len == item->len)
+                       continue;
+
+               if (item->nowildcard_len &&
+                   !match_wildcard_base(item, base_str, baselen, &matched))
                        continue;
 
                /*
                 * Concatenate base and entry->path into one and do
                 * fnmatch() on it.
+                *
+                * 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.
                 */
 
                strbuf_add(base, entry->path, pathlen);
 
-               if (!fnmatch(match, base->buf + base_offset, 0)) {
+               if (!git_fnmatch(item, match, base->buf + base_offset,
+                                item->nowildcard_len)) {
                        strbuf_setlen(base, base_offset + baselen);
                        return entry_interesting;
                }
@@ -670,3 +788,72 @@ match_wildcards:
        }
        return never_interesting; /* No matches */
 }
+
+/*
+ * 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 */
+}