ewah: support platforms that require aligned reads
[gitweb.git] / builtin / apply.c
index ace04c453be447b62b4907c45247f82ef4344033..ef32e4f62415956aa8e2366589d4cd6200d3bf37 100644 (file)
@@ -188,7 +188,6 @@ struct patch {
        int is_new, is_delete;  /* -1 = unknown, 0 = false, 1 = true */
        int rejected;
        unsigned ws_rule;
-       unsigned long deflate_origlen;
        int lines_added, lines_deleted;
        int score;
        unsigned int is_toplevel_relative:1;
@@ -474,7 +473,7 @@ static char *find_name_gnu(const char *line, const char *def, int p_value)
 
        /*
         * Proposed "new-style" GNU patch/diff format; see
-        * http://marc.theaimsgroup.com/?l=git&m=112927316408690&w=2
+        * http://marc.info/?l=git&m=112927316408690&w=2
         */
        if (unquote_c_style(&name, line, NULL)) {
                strbuf_release(&name);
@@ -723,7 +722,7 @@ static char *find_name(const char *line, char *def, int p_value, int terminate)
 
 static char *find_name_traditional(const char *line, char *def, int p_value)
 {
-       size_t len = strlen(line);
+       size_t len;
        size_t date_len;
 
        if (*line == '"') {
@@ -907,7 +906,7 @@ static void parse_traditional_patch(const char *first, const char *second, struc
                        patch->old_name = name;
                } else {
                        patch->old_name = name;
-                       patch->new_name = xstrdup(name);
+                       patch->new_name = null_strdup(name);
                }
        }
        if (!name)
@@ -1042,15 +1041,17 @@ static int gitdiff_renamedst(const char *line, struct patch *patch)
 
 static int gitdiff_similarity(const char *line, struct patch *patch)
 {
-       if ((patch->score = strtoul(line, NULL, 10)) == ULONG_MAX)
-               patch->score = 0;
+       unsigned long val = strtoul(line, NULL, 10);
+       if (val <= 100)
+               patch->score = val;
        return 0;
 }
 
 static int gitdiff_dissimilarity(const char *line, struct patch *patch)
 {
-       if ((patch->score = strtoul(line, NULL, 10)) == ULONG_MAX)
-               patch->score = 0;
+       unsigned long val = strtoul(line, NULL, 10);
+       if (val <= 100)
+               patch->score = val;
        return 0;
 }
 
@@ -1096,15 +1097,23 @@ static int gitdiff_unrecognized(const char *line, struct patch *patch)
        return -1;
 }
 
-static const char *stop_at_slash(const char *line, int llen)
+/*
+ * Skip p_value leading components from "line"; as we do not accept
+ * absolute paths, return NULL in that case.
+ */
+static const char *skip_tree_prefix(const char *line, int llen)
 {
-       int nslash = p_value;
+       int nslash;
        int i;
 
+       if (!p_value)
+               return (llen && line[0] == '/') ? NULL : line;
+
+       nslash = p_value;
        for (i = 0; i < llen; i++) {
                int ch = line[i];
                if (ch == '/' && --nslash <= 0)
-                       return &line[i];
+                       return (i == 0) ? NULL : &line[i + 1];
        }
        return NULL;
 }
@@ -1134,12 +1143,11 @@ static char *git_header_name(const char *line, int llen)
                if (unquote_c_style(&first, line, &second))
                        goto free_and_fail1;
 
-               /* advance to the first slash */
-               cp = stop_at_slash(first.buf, first.len);
-               /* we do not accept absolute paths */
-               if (!cp || cp == first.buf)
+               /* strip the a/b prefix including trailing slash */
+               cp = skip_tree_prefix(first.buf, first.len);
+               if (!cp)
                        goto free_and_fail1;
-               strbuf_remove(&first, 0, cp + 1 - first.buf);
+               strbuf_remove(&first, 0, cp - first.buf);
 
                /*
                 * second points at one past closing dq of name.
@@ -1153,22 +1161,21 @@ static char *git_header_name(const char *line, int llen)
                if (*second == '"') {
                        if (unquote_c_style(&sp, second, NULL))
                                goto free_and_fail1;
-                       cp = stop_at_slash(sp.buf, sp.len);
-                       if (!cp || cp == sp.buf)
+                       cp = skip_tree_prefix(sp.buf, sp.len);
+                       if (!cp)
                                goto free_and_fail1;
                        /* They must match, otherwise ignore */
-                       if (strcmp(cp + 1, first.buf))
+                       if (strcmp(cp, first.buf))
                                goto free_and_fail1;
                        strbuf_release(&sp);
                        return strbuf_detach(&first, NULL);
                }
 
                /* unquoted second */
-               cp = stop_at_slash(second, line + llen - second);
-               if (!cp || cp == second)
+               cp = skip_tree_prefix(second, line + llen - second);
+               if (!cp)
                        goto free_and_fail1;
-               cp++;
-               if (line + llen - cp != first.len + 1 ||
+               if (line + llen - cp != first.len ||
                    memcmp(first.buf, cp, first.len))
                        goto free_and_fail1;
                return strbuf_detach(&first, NULL);
@@ -1180,10 +1187,9 @@ static char *git_header_name(const char *line, int llen)
        }
 
        /* unquoted first name */
-       name = stop_at_slash(line, llen);
-       if (!name || name == line)
+       name = skip_tree_prefix(line, llen);
+       if (!name)
                return NULL;
-       name++;
 
        /*
         * since the first name is unquoted, a dq if exists must be
@@ -1197,10 +1203,9 @@ static char *git_header_name(const char *line, int llen)
                        if (unquote_c_style(&sp, second, NULL))
                                goto free_and_fail2;
 
-                       np = stop_at_slash(sp.buf, sp.len);
-                       if (!np || np == sp.buf)
+                       np = skip_tree_prefix(sp.buf, sp.len);
+                       if (!np)
                                goto free_and_fail2;
-                       np++;
 
                        len = sp.buf + sp.len - np;
                        if (len < second - name &&
@@ -1232,13 +1237,27 @@ static char *git_header_name(const char *line, int llen)
                case '\n':
                        return NULL;
                case '\t': case ' ':
-                       second = stop_at_slash(name + len, line_len - len);
+                       /*
+                        * Is this the separator between the preimage
+                        * and the postimage pathname?  Again, we are
+                        * only interested in the case where there is
+                        * no rename, as this is only to set def_name
+                        * and a rename patch has the names elsewhere
+                        * in an unambiguous form.
+                        */
+                       if (!name[len + 1])
+                               return NULL; /* no postimage name */
+                       second = skip_tree_prefix(name + len + 1,
+                                                 line_len - (len + 1));
                        if (!second)
                                return NULL;
-                       second++;
-                       if (second[len] == '\n' && !strncmp(name, second, len)) {
+                       /*
+                        * Does len bytes starting at "name" and "second"
+                        * (that are separated by one HT or SP we just
+                        * found) exactly match?
+                        */
+                       if (second[len] == '\n' && !strncmp(name, second, len))
                                return xmemdupz(name, len);
-                       }
                }
        }
 }
@@ -1902,7 +1921,7 @@ static int parse_binary(char *buffer, unsigned long size, struct patch *patch)
 }
 
 /*
- * Read the patch text in "buffer" taht extends for "size" bytes; stop
+ * Read the patch text in "buffer" that extends for "size" bytes; stop
  * reading after seeing a single patch (i.e. changes to a single file).
  * Create fragments (i.e. patch hunks) and hang them to the given patch.
  * Return the number of bytes consumed, so that the caller can call us
@@ -2078,7 +2097,7 @@ static void update_pre_post_images(struct image *preimage,
                                   char *buf,
                                   size_t len, size_t postlen)
 {
-       int i, ctx;
+       int i, ctx, reduced;
        char *new, *old, *fixed;
        struct image fixed_preimage;
 
@@ -2088,18 +2107,20 @@ static void update_pre_post_images(struct image *preimage,
         * free "oldlines".
         */
        prepare_image(&fixed_preimage, buf, len, 1);
-       assert(fixed_preimage.nr == preimage->nr);
-       for (i = 0; i < preimage->nr; i++)
+       assert(postlen
+              ? fixed_preimage.nr == preimage->nr
+              : fixed_preimage.nr <= preimage->nr);
+       for (i = 0; i < fixed_preimage.nr; i++)
                fixed_preimage.line[i].flag = preimage->line[i].flag;
        free(preimage->line_allocated);
        *preimage = fixed_preimage;
 
        /*
         * Adjust the common context lines in postimage. This can be
-        * done in-place when we are just doing whitespace fixing,
-        * which does not make the string grow, but needs a new buffer
-        * when ignoring whitespace causes the update, since in this case
-        * we could have e.g. tabs converted to multiple spaces.
+        * done in-place when we are shrinking it with whitespace
+        * fixing, but needs a new buffer when ignoring whitespace or
+        * expanding leading tabs to spaces.
+        *
         * We trust the caller to tell us if the update can be done
         * in place (postlen==0) or not.
         */
@@ -2109,7 +2130,8 @@ static void update_pre_post_images(struct image *preimage,
        else
                new = old;
        fixed = preimage->buf;
-       for (i = ctx = 0; i < postimage->nr; i++) {
+
+       for (i = reduced = ctx = 0; i < postimage->nr; i++) {
                size_t len = postimage->line[i].len;
                if (!(postimage->line[i].flag & LINE_COMMON)) {
                        /* an added line -- no counterparts in preimage */
@@ -2128,8 +2150,15 @@ static void update_pre_post_images(struct image *preimage,
                        fixed += preimage->line[ctx].len;
                        ctx++;
                }
-               if (preimage->nr <= ctx)
-                       die(_("oops"));
+
+               /*
+                * preimage is expected to run out, if the caller
+                * fixed addition of trailing blank lines.
+                */
+               if (preimage->nr <= ctx) {
+                       reduced++;
+                       continue;
+               }
 
                /* and copy it in, while fixing the line length */
                len = preimage->line[ctx].len;
@@ -2142,6 +2171,7 @@ static void update_pre_post_images(struct image *preimage,
 
        /* Fix the length of the whole thing */
        postimage->len = new - postimage->buf;
+       postimage->nr -= reduced;
 }
 
 static int match_fragment(struct image *img,
@@ -2155,7 +2185,7 @@ static int match_fragment(struct image *img,
        int i;
        char *fixed_buf, *buf, *orig, *target;
        struct strbuf fixed;
-       size_t fixed_len;
+       size_t fixed_len, postlen;
        int preimage_limit;
 
        if (preimage->nr + try_lno <= img->nr) {
@@ -2305,6 +2335,7 @@ static int match_fragment(struct image *img,
        strbuf_init(&fixed, preimage->len + 1);
        orig = preimage->buf;
        target = img->buf + try;
+       postlen = 0;
        for (i = 0; i < preimage_limit; i++) {
                size_t oldlen = preimage->line[i].len;
                size_t tgtlen = img->line[try_lno + i].len;
@@ -2332,6 +2363,7 @@ static int match_fragment(struct image *img,
                match = (tgtfix.len == fixed.len - fixstart &&
                         !memcmp(tgtfix.buf, fixed.buf + fixstart,
                                             fixed.len - fixstart));
+               postlen += tgtfix.len;
 
                strbuf_release(&tgtfix);
                if (!match)
@@ -2369,8 +2401,10 @@ static int match_fragment(struct image *img,
         * hunk match.  Update the context lines in the postimage.
         */
        fixed_buf = strbuf_detach(&fixed, &fixed_len);
+       if (postlen < postimage->len)
+               postlen = 0;
        update_pre_post_images(preimage, postimage,
-                              fixed_buf, fixed_len, 0);
+                              fixed_buf, fixed_len, postlen);
        return 1;
 
  unmatch_exit:
@@ -2965,7 +2999,7 @@ static int read_blob_object(struct strbuf *buf, const unsigned char *sha1, unsig
        return 0;
 }
 
-static int read_file_or_gitlink(struct cache_entry *ce, struct strbuf *buf)
+static int read_file_or_gitlink(const struct cache_entry *ce, struct strbuf *buf)
 {
        if (!ce)
                return 0;
@@ -2995,7 +3029,7 @@ static struct patch *in_fn_table(const char *name)
  *
  * The latter is needed to deal with a case where two paths A and B
  * are swapped by first renaming A to B and then renaming B to A;
- * moving A to B should not be prevented due to presense of B as we
+ * moving A to B should not be prevented due to presence of B as we
  * will remove it in a later patch.
  */
 #define PATH_TO_BE_DELETED ((struct patch *) -2)
@@ -3083,7 +3117,7 @@ static struct patch *previous_patch(struct patch *patch, int *gone)
        return previous;
 }
 
-static int verify_index_match(struct cache_entry *ce, struct stat *st)
+static int verify_index_match(const struct cache_entry *ce, struct stat *st)
 {
        if (S_ISGITLINK(ce->ce_mode)) {
                if (!S_ISDIR(st->st_mode))
@@ -3096,7 +3130,7 @@ static int verify_index_match(struct cache_entry *ce, struct stat *st)
 #define SUBMODULE_PATCH_WITHOUT_INDEX 1
 
 static int load_patch_target(struct strbuf *buf,
-                            struct cache_entry *ce,
+                            const struct cache_entry *ce,
                             struct stat *st,
                             const char *name,
                             unsigned expected_mode)
@@ -3126,7 +3160,8 @@ static int load_patch_target(struct strbuf *buf,
  * we read from the result of a previous diff.
  */
 static int load_preimage(struct image *image,
-                        struct patch *patch, struct stat *st, struct cache_entry *ce)
+                        struct patch *patch, struct stat *st,
+                        const struct cache_entry *ce)
 {
        struct strbuf buf = STRBUF_INIT;
        size_t len;
@@ -3239,7 +3274,7 @@ static int load_current(struct image *image, struct patch *patch)
 }
 
 static int try_threeway(struct image *image, struct patch *patch,
-                       struct stat *st, struct cache_entry *ce)
+                       struct stat *st, const struct cache_entry *ce)
 {
        unsigned char pre_sha1[20], post_sha1[20], our_sha1[20];
        struct strbuf buf = STRBUF_INIT;
@@ -3309,7 +3344,7 @@ static int try_threeway(struct image *image, struct patch *patch,
        return 0;
 }
 
-static int apply_data(struct patch *patch, struct stat *st, struct cache_entry *ce)
+static int apply_data(struct patch *patch, struct stat *st, const struct cache_entry *ce)
 {
        struct image image;
 
@@ -3479,7 +3514,7 @@ static int check_patch(struct patch *patch)
         *
         * A patch to swap-rename between A and B would first rename A
         * to B and then rename B to A.  While applying the first one,
-        * the presense of B should not stop A from getting renamed to
+        * the presence of B should not stop A from getting renamed to
         * B; ask to_be_deleted() about the later rename.  Removal of
         * B and rename from A to B is handled the same way by asking
         * was_deleted().
@@ -3491,7 +3526,7 @@ static int check_patch(struct patch *patch)
                ok_if_exists = 0;
 
        if (new_name &&
-           ((0 < patch->is_new) | (0 < patch->is_rename) | patch->is_copy)) {
+           ((0 < patch->is_new) || patch->is_rename || patch->is_copy)) {
                int err = check_to_create(new_name, ok_if_exists);
 
                if (err && threeway) {
@@ -3570,6 +3605,40 @@ static int get_current_sha1(const char *path, unsigned char *sha1)
        return 0;
 }
 
+static int preimage_sha1_in_gitlink_patch(struct patch *p, unsigned char sha1[20])
+{
+       /*
+        * A usable gitlink patch has only one fragment (hunk) that looks like:
+        * @@ -1 +1 @@
+        * -Subproject commit <old sha1>
+        * +Subproject commit <new sha1>
+        * or
+        * @@ -1 +0,0 @@
+        * -Subproject commit <old sha1>
+        * for a removal patch.
+        */
+       struct fragment *hunk = p->fragments;
+       static const char heading[] = "-Subproject commit ";
+       char *preimage;
+
+       if (/* does the patch have only one hunk? */
+           hunk && !hunk->next &&
+           /* is its preimage one line? */
+           hunk->oldpos == 1 && hunk->oldlines == 1 &&
+           /* does preimage begin with the heading? */
+           (preimage = memchr(hunk->patch, '\n', hunk->size)) != NULL &&
+           !prefixcmp(++preimage, heading) &&
+           /* does it record full SHA-1? */
+           !get_sha1_hex(preimage + sizeof(heading) - 1, sha1) &&
+           preimage[sizeof(heading) + 40 - 1] == '\n' &&
+           /* does the abbreviated name on the index line agree with it? */
+           !prefixcmp(preimage + sizeof(heading) - 1, p->old_sha1_prefix))
+               return 0; /* it all looks fine */
+
+       /* we may have full object name on the index line */
+       return get_sha1_hex(p->old_sha1_prefix, sha1);
+}
+
 /* Build an index that contains the just the files needed for a 3way merge */
 static void build_fake_ancestor(struct patch *list, const char *filename)
 {
@@ -3581,7 +3650,6 @@ static void build_fake_ancestor(struct patch *list, const char *filename)
         * worth showing the new sha1 prefix, but until then...
         */
        for (patch = list; patch; patch = patch->next) {
-               const unsigned char *sha1_ptr;
                unsigned char sha1[20];
                struct cache_entry *ce;
                const char *name;
@@ -3589,20 +3657,25 @@ static void build_fake_ancestor(struct patch *list, const char *filename)
                name = patch->old_name ? patch->old_name : patch->new_name;
                if (0 < patch->is_new)
                        continue;
-               else if (get_sha1(patch->old_sha1_prefix, sha1))
-                       /* git diff has no index line for mode/type changes */
-                       if (!patch->lines_added && !patch->lines_deleted) {
-                               if (get_current_sha1(patch->old_name, sha1))
-                                       die("mode change for %s, which is not "
-                                               "in current HEAD", name);
-                               sha1_ptr = sha1;
-                       } else
-                               die("sha1 information is lacking or useless "
-                                       "(%s).", name);
-               else
-                       sha1_ptr = sha1;
 
-               ce = make_cache_entry(patch->old_mode, sha1_ptr, name, 0, 0);
+               if (S_ISGITLINK(patch->old_mode)) {
+                       if (!preimage_sha1_in_gitlink_patch(patch, sha1))
+                               ; /* ok, the textual part looks sane */
+                       else
+                               die("sha1 information is lacking or useless for submoule %s",
+                                   name);
+               } else if (!get_sha1_blob(patch->old_sha1_prefix, sha1)) {
+                       ; /* ok */
+               } else if (!patch->lines_added && !patch->lines_deleted) {
+                       /* mode-only change: update the current */
+                       if (get_current_sha1(patch->old_name, sha1))
+                               die("mode change for %s, which is not "
+                                   "in current HEAD", name);
+               } else
+                       die("sha1 information is lacking or useless "
+                           "(%s).", name);
+
+               ce = make_cache_entry(patch->old_mode, sha1, name, 0, 0);
                if (!ce)
                        die(_("make_cache_entry failed for path '%s'"), name);
                if (add_index_entry(&result, ce, ADD_CACHE_OK_TO_ADD))
@@ -3769,12 +3842,13 @@ static void add_index_file(const char *path, unsigned mode, void *buf, unsigned
        ce = xcalloc(1, ce_size);
        memcpy(ce->name, path, namelen);
        ce->ce_mode = create_ce_mode(mode);
-       ce->ce_flags = namelen;
+       ce->ce_flags = create_ce_flags(0);
+       ce->ce_namelen = namelen;
        if (S_ISGITLINK(mode)) {
                const char *s = buf;
 
                if (get_sha1_hex(s + strlen("Subproject commit "), ce->sha1))
-                       die(_("corrupt patch for subproject %s"), path);
+                       die(_("corrupt patch for submodule %s"), path);
        } else {
                if (!cached) {
                        if (lstat(path, &st) < 0)
@@ -3890,7 +3964,8 @@ static void add_conflicted_stages_file(struct patch *patch)
                ce = xcalloc(1, ce_size);
                memcpy(ce->name, patch->new_name, namelen);
                ce->ce_mode = create_ce_mode(mode);
-               ce->ce_flags = create_ce_flags(namelen, stage);
+               ce->ce_flags = create_ce_flags(stage);
+               ce->ce_namelen = namelen;
                hashcpy(ce->sha1, patch->threeway_stage[stage - 1]);
                if (add_cache_entry(ce, ADD_CACHE_OK_TO_ADD) < 0)
                        die(_("unable to add cache entry for %s"), patch->new_name);
@@ -4288,23 +4363,23 @@ int cmd_apply(int argc, const char **argv, const char *prefix_)
                { OPTION_CALLBACK, 'p', NULL, NULL, N_("num"),
                        N_("remove <num> leading slashes from traditional diff paths"),
                        0, option_parse_p },
-               OPT_BOOLEAN(0, "no-add", &no_add,
+               OPT_BOOL(0, "no-add", &no_add,
                        N_("ignore additions made by the patch")),
-               OPT_BOOLEAN(0, "stat", &diffstat,
+               OPT_BOOL(0, "stat", &diffstat,
                        N_("instead of applying the patch, output diffstat for the input")),
                OPT_NOOP_NOARG(0, "allow-binary-replacement"),
                OPT_NOOP_NOARG(0, "binary"),
-               OPT_BOOLEAN(0, "numstat", &numstat,
-                       N_("shows number of added and deleted lines in decimal notation")),
-               OPT_BOOLEAN(0, "summary", &summary,
+               OPT_BOOL(0, "numstat", &numstat,
+                       N_("show number of added and deleted lines in decimal notation")),
+               OPT_BOOL(0, "summary", &summary,
                        N_("instead of applying the patch, output a summary for the input")),
-               OPT_BOOLEAN(0, "check", &check,
+               OPT_BOOL(0, "check", &check,
                        N_("instead of applying the patch, see if the patch is applicable")),
-               OPT_BOOLEAN(0, "index", &check_index,
+               OPT_BOOL(0, "index", &check_index,
                        N_("make sure the patch is applicable to the current index")),
-               OPT_BOOLEAN(0, "cached", &cached,
+               OPT_BOOL(0, "cached", &cached,
                        N_("apply a patch without touching the working tree")),
-               OPT_BOOLEAN(0, "apply", &force_apply,
+               OPT_BOOL(0, "apply", &force_apply,
                        N_("also apply the patch (use with --stat/--summary/--check)")),
                OPT_BOOL('3', "3way", &threeway,
                         N_( "attempt three-way merge if a patch does not apply")),
@@ -4324,13 +4399,13 @@ int cmd_apply(int argc, const char **argv, const char *prefix_)
                { OPTION_CALLBACK, 0, "ignore-whitespace", NULL, NULL,
                        N_("ignore changes in whitespace when finding context"),
                        PARSE_OPT_NOARG, option_parse_space_change },
-               OPT_BOOLEAN('R', "reverse", &apply_in_reverse,
+               OPT_BOOL('R', "reverse", &apply_in_reverse,
                        N_("apply the patch in reverse")),
-               OPT_BOOLEAN(0, "unidiff-zero", &unidiff_zero,
+               OPT_BOOL(0, "unidiff-zero", &unidiff_zero,
                        N_("don't expect at least one line of context")),
-               OPT_BOOLEAN(0, "reject", &apply_with_reject,
+               OPT_BOOL(0, "reject", &apply_with_reject,
                        N_("leave the rejected hunks in corresponding *.rej files")),
-               OPT_BOOLEAN(0, "allow-overlap", &allow_overlap,
+               OPT_BOOL(0, "allow-overlap", &allow_overlap,
                        N_("allow overlapping hunks")),
                OPT__VERBOSE(&apply_verbosely, N_("be verbose")),
                OPT_BIT(0, "inaccurate-eof", &options,