fast-export: make extra_refs global
[gitweb.git] / commit.c
index 8b84ebf2b70561dc4b3bd4a13afa4fcfcc5c5aea..a575564a15c1c340b36d10c90da8fa343692264d 100644 (file)
--- a/commit.c
+++ b/commit.c
@@ -469,14 +469,23 @@ static void clear_commit_marks_1(struct commit_list **plist,
        }
 }
 
-void clear_commit_marks(struct commit *commit, unsigned int mark)
+void clear_commit_marks_many(int nr, struct commit **commit, unsigned int mark)
 {
        struct commit_list *list = NULL;
-       commit_list_insert(commit, &list);
+
+       while (nr--) {
+               commit_list_insert(*commit, &list);
+               commit++;
+       }
        while (list)
                clear_commit_marks_1(&list, pop_commit(&list), mark);
 }
 
+void clear_commit_marks(struct commit *commit, unsigned int mark)
+{
+       clear_commit_marks_many(1, &commit, mark);
+}
+
 void clear_commit_marks_for_object_array(struct object_array *a, unsigned mark)
 {
        struct object *object;
@@ -510,7 +519,69 @@ struct commit *pop_commit(struct commit_list **stack)
 /* count number of children that have not been emitted */
 define_commit_slab(indegree_slab, int);
 
-static int compare_commits_by_commit_date(const void *a_, const void *b_, void *unused)
+/* record author-date for each commit object */
+define_commit_slab(author_date_slab, unsigned long);
+
+static void record_author_date(struct author_date_slab *author_date,
+                              struct commit *commit)
+{
+       const char *buf, *line_end;
+       char *buffer = NULL;
+       struct ident_split ident;
+       char *date_end;
+       unsigned long date;
+
+       if (!commit->buffer) {
+               unsigned long size;
+               enum object_type type;
+               buffer = read_sha1_file(commit->object.sha1, &type, &size);
+               if (!buffer)
+                       return;
+       }
+
+       for (buf = commit->buffer ? commit->buffer : buffer;
+            buf;
+            buf = line_end + 1) {
+               line_end = strchrnul(buf, '\n');
+               if (prefixcmp(buf, "author ")) {
+                       if (!line_end[0] || line_end[1] == '\n')
+                               return; /* end of header */
+                       continue;
+               }
+               if (split_ident_line(&ident,
+                                    buf + strlen("author "),
+                                    line_end - (buf + strlen("author "))) ||
+                   !ident.date_begin || !ident.date_end)
+                       goto fail_exit; /* malformed "author" line */
+               break;
+       }
+
+       date = strtoul(ident.date_begin, &date_end, 10);
+       if (date_end != ident.date_end)
+               goto fail_exit; /* malformed date */
+       *(author_date_slab_at(author_date, commit)) = date;
+
+fail_exit:
+       free(buffer);
+}
+
+static int compare_commits_by_author_date(const void *a_, const void *b_,
+                                         void *cb_data)
+{
+       const struct commit *a = a_, *b = b_;
+       struct author_date_slab *author_date = cb_data;
+       unsigned long a_date = *(author_date_slab_at(author_date, a));
+       unsigned long b_date = *(author_date_slab_at(author_date, b));
+
+       /* newer commits with larger date first */
+       if (a_date < b_date)
+               return 1;
+       else if (a_date > b_date)
+               return -1;
+       return 0;
+}
+
+int compare_commits_by_commit_date(const void *a_, const void *b_, void *unused)
 {
        const struct commit *a = a_, *b = b_;
        /* newer commits with larger date first */
@@ -531,6 +602,7 @@ void sort_in_topological_order(struct commit_list **list, enum rev_sort_order so
        struct indegree_slab indegree;
        struct prio_queue queue;
        struct commit *commit;
+       struct author_date_slab author_date;
 
        if (!orig)
                return;
@@ -538,6 +610,7 @@ void sort_in_topological_order(struct commit_list **list, enum rev_sort_order so
 
        init_indegree_slab(&indegree);
        memset(&queue, '\0', sizeof(queue));
+
        switch (sort_order) {
        default: /* REV_SORT_IN_GRAPH_ORDER */
                queue.compare = NULL;
@@ -545,12 +618,20 @@ void sort_in_topological_order(struct commit_list **list, enum rev_sort_order so
        case REV_SORT_BY_COMMIT_DATE:
                queue.compare = compare_commits_by_commit_date;
                break;
+       case REV_SORT_BY_AUTHOR_DATE:
+               init_author_date_slab(&author_date);
+               queue.compare = compare_commits_by_author_date;
+               queue.cb_data = &author_date;
+               break;
        }
 
        /* Mark them and clear the indegree */
        for (next = orig; next; next = next->next) {
                struct commit *commit = next->item;
                *(indegree_slab_at(&indegree, commit)) = 1;
+               /* also record the author dates, if needed */
+               if (sort_order == REV_SORT_BY_AUTHOR_DATE)
+                       record_author_date(&author_date, commit);
        }
 
        /* update the indegree */
@@ -621,6 +702,8 @@ void sort_in_topological_order(struct commit_list **list, enum rev_sort_order so
 
        clear_indegree_slab(&indegree);
        clear_prio_queue(&queue);
+       if (sort_order == REV_SORT_BY_AUTHOR_DATE)
+               clear_author_date_slab(&author_date);
 }
 
 /* merge-base stuff */
@@ -833,8 +916,7 @@ struct commit_list *get_merge_bases_many(struct commit *one,
        if (!result || !result->next) {
                if (cleanup) {
                        clear_commit_marks(one, all_flags);
-                       for (i = 0; i < n; i++)
-                               clear_commit_marks(twos[i], all_flags);
+                       clear_commit_marks_many(n, twos, all_flags);
                }
                return result;
        }
@@ -852,8 +934,7 @@ struct commit_list *get_merge_bases_many(struct commit *one,
        free_commit_list(result);
 
        clear_commit_marks(one, all_flags);
-       for (i = 0; i < n; i++)
-               clear_commit_marks(twos[i], all_flags);
+       clear_commit_marks_many(n, twos, all_flags);
 
        cnt = remove_redundant(rslt, cnt);
        result = NULL;
@@ -888,25 +969,36 @@ int is_descendant_of(struct commit *commit, struct commit_list *with_commit)
 }
 
 /*
- * Is "commit" an ancestor of (i.e. reachable from) the "reference"?
+ * Is "commit" an ancestor of one of the "references"?
  */
-int in_merge_bases(struct commit *commit, struct commit *reference)
+int in_merge_bases_many(struct commit *commit, int nr_reference, struct commit **reference)
 {
        struct commit_list *bases;
-       int ret = 0;
+       int ret = 0, i;
 
-       if (parse_commit(commit) || parse_commit(reference))
+       if (parse_commit(commit))
                return ret;
+       for (i = 0; i < nr_reference; i++)
+               if (parse_commit(reference[i]))
+                       return ret;
 
-       bases = paint_down_to_common(commit, 1, &reference);
+       bases = paint_down_to_common(commit, nr_reference, reference);
        if (commit->object.flags & PARENT2)
                ret = 1;
        clear_commit_marks(commit, all_flags);
-       clear_commit_marks(reference, all_flags);
+       clear_commit_marks_many(nr_reference, reference, all_flags);
        free_commit_list(bases);
        return ret;
 }
 
+/*
+ * Is "commit" an ancestor of (i.e. reachable from) the "reference"?
+ */
+int in_merge_bases(struct commit *commit, struct commit *reference)
+{
+       return in_merge_bases_many(commit, 1, &reference);
+}
+
 struct commit_list *reduce_heads(struct commit_list *heads)
 {
        struct commit_list *p;
@@ -1059,6 +1151,76 @@ static void handle_signed_tag(struct commit *parent, struct commit_extra_header
        free(buf);
 }
 
+static struct {
+       char result;
+       const char *check;
+} sigcheck_gpg_status[] = {
+       { 'G', "\n[GNUPG:] GOODSIG " },
+       { 'B', "\n[GNUPG:] BADSIG " },
+       { 'U', "\n[GNUPG:] TRUST_NEVER" },
+       { 'U', "\n[GNUPG:] TRUST_UNDEFINED" },
+};
+
+static void parse_gpg_output(struct signature_check *sigc)
+{
+       const char *buf = sigc->gpg_status;
+       int i;
+
+       /* Iterate over all search strings */
+       for (i = 0; i < ARRAY_SIZE(sigcheck_gpg_status); i++) {
+               const char *found, *next;
+
+               if (!prefixcmp(buf, sigcheck_gpg_status[i].check + 1)) {
+                       /* At the very beginning of the buffer */
+                       found = buf + strlen(sigcheck_gpg_status[i].check + 1);
+               } else {
+                       found = strstr(buf, sigcheck_gpg_status[i].check);
+                       if (!found)
+                               continue;
+                       found += strlen(sigcheck_gpg_status[i].check);
+               }
+               sigc->result = sigcheck_gpg_status[i].result;
+               /* The trust messages are not followed by key/signer information */
+               if (sigc->result != 'U') {
+                       sigc->key = xmemdupz(found, 16);
+                       found += 17;
+                       next = strchrnul(found, '\n');
+                       sigc->signer = xmemdupz(found, next - found);
+               }
+       }
+}
+
+void check_commit_signature(const struct commit* commit, struct signature_check *sigc)
+{
+       struct strbuf payload = STRBUF_INIT;
+       struct strbuf signature = STRBUF_INIT;
+       struct strbuf gpg_output = STRBUF_INIT;
+       struct strbuf gpg_status = STRBUF_INIT;
+       int status;
+
+       sigc->result = 'N';
+
+       if (parse_signed_commit(commit->object.sha1,
+                               &payload, &signature) <= 0)
+               goto out;
+       status = verify_signed_buffer(payload.buf, payload.len,
+                                     signature.buf, signature.len,
+                                     &gpg_output, &gpg_status);
+       if (status && !gpg_output.len)
+               goto out;
+       sigc->gpg_output = strbuf_detach(&gpg_output, NULL);
+       sigc->gpg_status = strbuf_detach(&gpg_status, NULL);
+       parse_gpg_output(sigc);
+
+ out:
+       strbuf_release(&gpg_status);
+       strbuf_release(&gpg_output);
+       strbuf_release(&payload);
+       strbuf_release(&signature);
+}
+
+
+
 void append_merge_tag_headers(struct commit_list *parents,
                              struct commit_extra_header ***tail)
 {
@@ -1188,10 +1350,15 @@ int commit_tree(const struct strbuf *msg, unsigned char *tree,
 static int find_invalid_utf8(const char *buf, int len)
 {
        int offset = 0;
+       static const unsigned int max_codepoint[] = {
+               0x7f, 0x7ff, 0xffff, 0x10ffff
+       };
 
        while (len) {
                unsigned char c = *buf++;
                int bytes, bad_offset;
+               unsigned int codepoint;
+               unsigned int min_val, max_val;
 
                len--;
                offset++;
@@ -1212,24 +1379,48 @@ static int find_invalid_utf8(const char *buf, int len)
                        bytes++;
                }
 
-               /* Must be between 1 and 5 more bytes */
-               if (bytes < 1 || bytes > 5)
+               /*
+                * Must be between 1 and 3 more bytes.  Longer sequences result in
+                * codepoints beyond U+10FFFF, which are guaranteed never to exist.
+                */
+               if (bytes < 1 || 3 < bytes)
                        return bad_offset;
 
                /* Do we *have* that many bytes? */
                if (len < bytes)
                        return bad_offset;
 
+               /*
+                * Place the encoded bits at the bottom of the value and compute the
+                * valid range.
+                */
+               codepoint = (c & 0x7f) >> bytes;
+               min_val = max_codepoint[bytes-1] + 1;
+               max_val = max_codepoint[bytes];
+
                offset += bytes;
                len -= bytes;
 
                /* And verify that they are good continuation bytes */
                do {
+                       codepoint <<= 6;
+                       codepoint |= *buf & 0x3f;
                        if ((*buf++ & 0xc0) != 0x80)
                                return bad_offset;
                } while (--bytes);
 
-               /* We could/should check the value and length here too */
+               /* Reject codepoints that are out of range for the sequence length. */
+               if (codepoint < min_val || codepoint > max_val)
+                       return bad_offset;
+               /* Surrogates are only for UTF-16 and cannot be encoded in UTF-8. */
+               if ((codepoint & 0x1ff800) == 0xd800)
+                       return bad_offset;
+               /* U+xxFFFE and U+xxFFFF are guaranteed non-characters. */
+               if ((codepoint & 0xfffe) == 0xfffe)
+                       return bad_offset;
+               /* So are anything in the range U+FDD0..U+FDEF. */
+               if (codepoint >= 0xfdd0 && codepoint <= 0xfdef)
+                       return bad_offset;
        }
        return -1;
 }
@@ -1239,9 +1430,6 @@ static int find_invalid_utf8(const char *buf, int len)
  *
  * If it isn't, it assumes any non-utf8 characters are Latin1,
  * and does the conversion.
- *
- * Fixme: we should probably also disallow overlong forms and
- * invalid characters. But we don't do that currently.
  */
 static int verify_utf8(struct strbuf *buf)
 {