]> git.ipfire.org Git - thirdparty/git.git/commitdiff
Merge branch 'dd/sparse-fixes'
authorJunio C Hamano <gitster@pobox.com>
Fri, 1 May 2020 20:39:56 +0000 (13:39 -0700)
committerJunio C Hamano <gitster@pobox.com>
Fri, 1 May 2020 20:39:56 +0000 (13:39 -0700)
Compilation fix.

* dd/sparse-fixes:
  progress.c: silence cgcc suggestion about internal linkage
  graph.c: limit linkage of internal variable
  compat/regex: move stdlib.h up in inclusion chain
  test-parse-pathspec-file.c: s/0/NULL/ for pointer type

1  2 
fmt-merge-msg.c

diff --cc fmt-merge-msg.c
index 08022ed2e493df124e6347700d930b228acf1905,0000000000000000000000000000000000000000..72d32bd73b11ddf7291f5e80b7e6afb6df5fdd33
mode 100644,000000..100644
--- /dev/null
@@@ -1,656 -1,0 +1,656 @@@
-               struct signature_check sigc = { 0 };
 +#include "config.h"
 +#include "refs.h"
 +#include "object-store.h"
 +#include "diff.h"
 +#include "revision.h"
 +#include "tag.h"
 +#include "string-list.h"
 +#include "branch.h"
 +#include "fmt-merge-msg.h"
 +#include "commit-reach.h"
 +
 +static int use_branch_desc;
 +
 +int fmt_merge_msg_config(const char *key, const char *value, void *cb)
 +{
 +      if (!strcmp(key, "merge.log") || !strcmp(key, "merge.summary")) {
 +              int is_bool;
 +              merge_log_config = git_config_bool_or_int(key, value, &is_bool);
 +              if (!is_bool && merge_log_config < 0)
 +                      return error("%s: negative length %s", key, value);
 +              if (is_bool && merge_log_config)
 +                      merge_log_config = DEFAULT_MERGE_LOG_LEN;
 +      } else if (!strcmp(key, "merge.branchdesc")) {
 +              use_branch_desc = git_config_bool(key, value);
 +      } else {
 +              return git_default_config(key, value, cb);
 +      }
 +      return 0;
 +}
 +
 +/* merge data per repository where the merged tips came from */
 +struct src_data {
 +      struct string_list branch, tag, r_branch, generic;
 +      int head_status;
 +};
 +
 +struct origin_data {
 +      struct object_id oid;
 +      unsigned is_local_branch:1;
 +};
 +
 +static void init_src_data(struct src_data *data)
 +{
 +      data->branch.strdup_strings = 1;
 +      data->tag.strdup_strings = 1;
 +      data->r_branch.strdup_strings = 1;
 +      data->generic.strdup_strings = 1;
 +}
 +
 +static struct string_list srcs = STRING_LIST_INIT_DUP;
 +static struct string_list origins = STRING_LIST_INIT_DUP;
 +
 +struct merge_parents {
 +      int alloc, nr;
 +      struct merge_parent {
 +              struct object_id given;
 +              struct object_id commit;
 +              unsigned char used;
 +      } *item;
 +};
 +
 +/*
 + * I know, I know, this is inefficient, but you won't be pulling and merging
 + * hundreds of heads at a time anyway.
 + */
 +static struct merge_parent *find_merge_parent(struct merge_parents *table,
 +                                            struct object_id *given,
 +                                            struct object_id *commit)
 +{
 +      int i;
 +      for (i = 0; i < table->nr; i++) {
 +              if (given && !oideq(&table->item[i].given, given))
 +                      continue;
 +              if (commit && !oideq(&table->item[i].commit, commit))
 +                      continue;
 +              return &table->item[i];
 +      }
 +      return NULL;
 +}
 +
 +static void add_merge_parent(struct merge_parents *table,
 +                           struct object_id *given,
 +                           struct object_id *commit)
 +{
 +      if (table->nr && find_merge_parent(table, given, commit))
 +              return;
 +      ALLOC_GROW(table->item, table->nr + 1, table->alloc);
 +      oidcpy(&table->item[table->nr].given, given);
 +      oidcpy(&table->item[table->nr].commit, commit);
 +      table->item[table->nr].used = 0;
 +      table->nr++;
 +}
 +
 +static int handle_line(char *line, struct merge_parents *merge_parents)
 +{
 +      int i, len = strlen(line);
 +      struct origin_data *origin_data;
 +      char *src;
 +      const char *origin, *tag_name;
 +      struct src_data *src_data;
 +      struct string_list_item *item;
 +      int pulling_head = 0;
 +      struct object_id oid;
 +      const unsigned hexsz = the_hash_algo->hexsz;
 +
 +      if (len < hexsz + 3 || line[hexsz] != '\t')
 +              return 1;
 +
 +      if (starts_with(line + hexsz + 1, "not-for-merge"))
 +              return 0;
 +
 +      if (line[hexsz + 1] != '\t')
 +              return 2;
 +
 +      i = get_oid_hex(line, &oid);
 +      if (i)
 +              return 3;
 +
 +      if (!find_merge_parent(merge_parents, &oid, NULL))
 +              return 0; /* subsumed by other parents */
 +
 +      origin_data = xcalloc(1, sizeof(struct origin_data));
 +      oidcpy(&origin_data->oid, &oid);
 +
 +      if (line[len - 1] == '\n')
 +              line[len - 1] = 0;
 +      line += hexsz + 2;
 +
 +      /*
 +       * At this point, line points at the beginning of comment e.g.
 +       * "branch 'frotz' of git://that/repository.git".
 +       * Find the repository name and point it with src.
 +       */
 +      src = strstr(line, " of ");
 +      if (src) {
 +              *src = 0;
 +              src += 4;
 +              pulling_head = 0;
 +      } else {
 +              src = line;
 +              pulling_head = 1;
 +      }
 +
 +      item = unsorted_string_list_lookup(&srcs, src);
 +      if (!item) {
 +              item = string_list_append(&srcs, src);
 +              item->util = xcalloc(1, sizeof(struct src_data));
 +              init_src_data(item->util);
 +      }
 +      src_data = item->util;
 +
 +      if (pulling_head) {
 +              origin = src;
 +              src_data->head_status |= 1;
 +      } else if (skip_prefix(line, "branch ", &origin)) {
 +              origin_data->is_local_branch = 1;
 +              string_list_append(&src_data->branch, origin);
 +              src_data->head_status |= 2;
 +      } else if (skip_prefix(line, "tag ", &tag_name)) {
 +              origin = line;
 +              string_list_append(&src_data->tag, tag_name);
 +              src_data->head_status |= 2;
 +      } else if (skip_prefix(line, "remote-tracking branch ", &origin)) {
 +              string_list_append(&src_data->r_branch, origin);
 +              src_data->head_status |= 2;
 +      } else {
 +              origin = src;
 +              string_list_append(&src_data->generic, line);
 +              src_data->head_status |= 2;
 +      }
 +
 +      if (!strcmp(".", src) || !strcmp(src, origin)) {
 +              int len = strlen(origin);
 +              if (origin[0] == '\'' && origin[len - 1] == '\'')
 +                      origin = xmemdupz(origin + 1, len - 2);
 +      } else
 +              origin = xstrfmt("%s of %s", origin, src);
 +      if (strcmp(".", src))
 +              origin_data->is_local_branch = 0;
 +      string_list_append(&origins, origin)->util = origin_data;
 +      return 0;
 +}
 +
 +static void print_joined(const char *singular, const char *plural,
 +              struct string_list *list, struct strbuf *out)
 +{
 +      if (list->nr == 0)
 +              return;
 +      if (list->nr == 1) {
 +              strbuf_addf(out, "%s%s", singular, list->items[0].string);
 +      } else {
 +              int i;
 +              strbuf_addstr(out, plural);
 +              for (i = 0; i < list->nr - 1; i++)
 +                      strbuf_addf(out, "%s%s", i > 0 ? ", " : "",
 +                                  list->items[i].string);
 +              strbuf_addf(out, " and %s", list->items[list->nr - 1].string);
 +      }
 +}
 +
 +static void add_branch_desc(struct strbuf *out, const char *name)
 +{
 +      struct strbuf desc = STRBUF_INIT;
 +
 +      if (!read_branch_desc(&desc, name)) {
 +              const char *bp = desc.buf;
 +              while (*bp) {
 +                      const char *ep = strchrnul(bp, '\n');
 +                      if (*ep)
 +                              ep++;
 +                      strbuf_addf(out, "  : %.*s", (int)(ep - bp), bp);
 +                      bp = ep;
 +              }
 +              strbuf_complete_line(out);
 +      }
 +      strbuf_release(&desc);
 +}
 +
 +#define util_as_integral(elem) ((intptr_t)((elem)->util))
 +
 +static void record_person_from_buf(int which, struct string_list *people,
 +                                 const char *buffer)
 +{
 +      char *name_buf, *name, *name_end;
 +      struct string_list_item *elem;
 +      const char *field;
 +
 +      field = (which == 'a') ? "\nauthor " : "\ncommitter ";
 +      name = strstr(buffer, field);
 +      if (!name)
 +              return;
 +      name += strlen(field);
 +      name_end = strchrnul(name, '<');
 +      if (*name_end)
 +              name_end--;
 +      while (isspace(*name_end) && name <= name_end)
 +              name_end--;
 +      if (name_end < name)
 +              return;
 +      name_buf = xmemdupz(name, name_end - name + 1);
 +
 +      elem = string_list_lookup(people, name_buf);
 +      if (!elem) {
 +              elem = string_list_insert(people, name_buf);
 +              elem->util = (void *)0;
 +      }
 +      elem->util = (void*)(util_as_integral(elem) + 1);
 +      free(name_buf);
 +}
 +
 +
 +static void record_person(int which, struct string_list *people,
 +                        struct commit *commit)
 +{
 +      const char *buffer = get_commit_buffer(commit, NULL);
 +      record_person_from_buf(which, people, buffer);
 +      unuse_commit_buffer(commit, buffer);
 +}
 +
 +static int cmp_string_list_util_as_integral(const void *a_, const void *b_)
 +{
 +      const struct string_list_item *a = a_, *b = b_;
 +      return util_as_integral(b) - util_as_integral(a);
 +}
 +
 +static void add_people_count(struct strbuf *out, struct string_list *people)
 +{
 +      if (people->nr == 1)
 +              strbuf_addstr(out, people->items[0].string);
 +      else if (people->nr == 2)
 +              strbuf_addf(out, "%s (%d) and %s (%d)",
 +                          people->items[0].string,
 +                          (int)util_as_integral(&people->items[0]),
 +                          people->items[1].string,
 +                          (int)util_as_integral(&people->items[1]));
 +      else if (people->nr)
 +              strbuf_addf(out, "%s (%d) and others",
 +                          people->items[0].string,
 +                          (int)util_as_integral(&people->items[0]));
 +}
 +
 +static void credit_people(struct strbuf *out,
 +                        struct string_list *them,
 +                        int kind)
 +{
 +      const char *label;
 +      const char *me;
 +
 +      if (kind == 'a') {
 +              label = "By";
 +              me = git_author_info(IDENT_NO_DATE);
 +      } else {
 +              label = "Via";
 +              me = git_committer_info(IDENT_NO_DATE);
 +      }
 +
 +      if (!them->nr ||
 +          (them->nr == 1 &&
 +           me &&
 +           skip_prefix(me, them->items->string, &me) &&
 +           starts_with(me, " <")))
 +              return;
 +      strbuf_addf(out, "\n%c %s ", comment_line_char, label);
 +      add_people_count(out, them);
 +}
 +
 +static void add_people_info(struct strbuf *out,
 +                          struct string_list *authors,
 +                          struct string_list *committers)
 +{
 +      QSORT(authors->items, authors->nr,
 +            cmp_string_list_util_as_integral);
 +      QSORT(committers->items, committers->nr,
 +            cmp_string_list_util_as_integral);
 +
 +      credit_people(out, authors, 'a');
 +      credit_people(out, committers, 'c');
 +}
 +
 +static void shortlog(const char *name,
 +                   struct origin_data *origin_data,
 +                   struct commit *head,
 +                   struct rev_info *rev,
 +                   struct fmt_merge_msg_opts *opts,
 +                   struct strbuf *out)
 +{
 +      int i, count = 0;
 +      struct commit *commit;
 +      struct object *branch;
 +      struct string_list subjects = STRING_LIST_INIT_DUP;
 +      struct string_list authors = STRING_LIST_INIT_DUP;
 +      struct string_list committers = STRING_LIST_INIT_DUP;
 +      int flags = UNINTERESTING | TREESAME | SEEN | SHOWN | ADDED;
 +      struct strbuf sb = STRBUF_INIT;
 +      const struct object_id *oid = &origin_data->oid;
 +      int limit = opts->shortlog_len;
 +
 +      branch = deref_tag(the_repository, parse_object(the_repository, oid),
 +                         oid_to_hex(oid),
 +                         the_hash_algo->hexsz);
 +      if (!branch || branch->type != OBJ_COMMIT)
 +              return;
 +
 +      setup_revisions(0, NULL, rev, NULL);
 +      add_pending_object(rev, branch, name);
 +      add_pending_object(rev, &head->object, "^HEAD");
 +      head->object.flags |= UNINTERESTING;
 +      if (prepare_revision_walk(rev))
 +              die("revision walk setup failed");
 +      while ((commit = get_revision(rev)) != NULL) {
 +              struct pretty_print_context ctx = {0};
 +
 +              if (commit->parents && commit->parents->next) {
 +                      /* do not list a merge but count committer */
 +                      if (opts->credit_people)
 +                              record_person('c', &committers, commit);
 +                      continue;
 +              }
 +              if (!count && opts->credit_people)
 +                      /* the 'tip' committer */
 +                      record_person('c', &committers, commit);
 +              if (opts->credit_people)
 +                      record_person('a', &authors, commit);
 +              count++;
 +              if (subjects.nr > limit)
 +                      continue;
 +
 +              format_commit_message(commit, "%s", &sb, &ctx);
 +              strbuf_ltrim(&sb);
 +
 +              if (!sb.len)
 +                      string_list_append(&subjects,
 +                                         oid_to_hex(&commit->object.oid));
 +              else
 +                      string_list_append_nodup(&subjects,
 +                                               strbuf_detach(&sb, NULL));
 +      }
 +
 +      if (opts->credit_people)
 +              add_people_info(out, &authors, &committers);
 +      if (count > limit)
 +              strbuf_addf(out, "\n* %s: (%d commits)\n", name, count);
 +      else
 +              strbuf_addf(out, "\n* %s:\n", name);
 +
 +      if (origin_data->is_local_branch && use_branch_desc)
 +              add_branch_desc(out, name);
 +
 +      for (i = 0; i < subjects.nr; i++)
 +              if (i >= limit)
 +                      strbuf_addstr(out, "  ...\n");
 +              else
 +                      strbuf_addf(out, "  %s\n", subjects.items[i].string);
 +
 +      clear_commit_marks((struct commit *)branch, flags);
 +      clear_commit_marks(head, flags);
 +      free_commit_list(rev->commits);
 +      rev->commits = NULL;
 +      rev->pending.nr = 0;
 +
 +      string_list_clear(&authors, 0);
 +      string_list_clear(&committers, 0);
 +      string_list_clear(&subjects, 0);
 +}
 +
 +static void fmt_merge_msg_title(struct strbuf *out,
 +                              const char *current_branch)
 +{
 +      int i = 0;
 +      char *sep = "";
 +
 +      strbuf_addstr(out, "Merge ");
 +      for (i = 0; i < srcs.nr; i++) {
 +              struct src_data *src_data = srcs.items[i].util;
 +              const char *subsep = "";
 +
 +              strbuf_addstr(out, sep);
 +              sep = "; ";
 +
 +              if (src_data->head_status == 1) {
 +                      strbuf_addstr(out, srcs.items[i].string);
 +                      continue;
 +              }
 +              if (src_data->head_status == 3) {
 +                      subsep = ", ";
 +                      strbuf_addstr(out, "HEAD");
 +              }
 +              if (src_data->branch.nr) {
 +                      strbuf_addstr(out, subsep);
 +                      subsep = ", ";
 +                      print_joined("branch ", "branches ", &src_data->branch,
 +                                      out);
 +              }
 +              if (src_data->r_branch.nr) {
 +                      strbuf_addstr(out, subsep);
 +                      subsep = ", ";
 +                      print_joined("remote-tracking branch ", "remote-tracking branches ",
 +                                      &src_data->r_branch, out);
 +              }
 +              if (src_data->tag.nr) {
 +                      strbuf_addstr(out, subsep);
 +                      subsep = ", ";
 +                      print_joined("tag ", "tags ", &src_data->tag, out);
 +              }
 +              if (src_data->generic.nr) {
 +                      strbuf_addstr(out, subsep);
 +                      print_joined("commit ", "commits ", &src_data->generic,
 +                                      out);
 +              }
 +              if (strcmp(".", srcs.items[i].string))
 +                      strbuf_addf(out, " of %s", srcs.items[i].string);
 +      }
 +
 +      if (!strcmp("master", current_branch))
 +              strbuf_addch(out, '\n');
 +      else
 +              strbuf_addf(out, " into %s\n", current_branch);
 +}
 +
 +static void fmt_tag_signature(struct strbuf *tagbuf,
 +                            struct strbuf *sig,
 +                            const char *buf,
 +                            unsigned long len)
 +{
 +      const char *tag_body = strstr(buf, "\n\n");
 +      if (tag_body) {
 +              tag_body += 2;
 +              strbuf_add(tagbuf, tag_body, buf + len - tag_body);
 +      }
 +      strbuf_complete_line(tagbuf);
 +      if (sig->len) {
 +              strbuf_addch(tagbuf, '\n');
 +              strbuf_add_commented_lines(tagbuf, sig->buf, sig->len);
 +      }
 +}
 +
 +static void fmt_merge_msg_sigs(struct strbuf *out)
 +{
 +      int i, tag_number = 0, first_tag = 0;
 +      struct strbuf tagbuf = STRBUF_INIT;
 +
 +      for (i = 0; i < origins.nr; i++) {
 +              struct object_id *oid = origins.items[i].util;
 +              enum object_type type;
 +              unsigned long size, len;
 +              char *buf = read_object_file(oid, &type, &size);
++              struct signature_check sigc = { NULL };
 +              struct strbuf sig = STRBUF_INIT;
 +
 +              if (!buf || type != OBJ_TAG)
 +                      goto next;
 +              len = parse_signature(buf, size);
 +
 +              if (size == len)
 +                      ; /* merely annotated */
 +              else if (check_signature(buf, len, buf + len, size - len, &sigc) &&
 +                      !sigc.gpg_output)
 +                      strbuf_addstr(&sig, "gpg verification failed.\n");
 +              else
 +                      strbuf_addstr(&sig, sigc.gpg_output);
 +              signature_check_clear(&sigc);
 +
 +              if (!tag_number++) {
 +                      fmt_tag_signature(&tagbuf, &sig, buf, len);
 +                      first_tag = i;
 +              } else {
 +                      if (tag_number == 2) {
 +                              struct strbuf tagline = STRBUF_INIT;
 +                              strbuf_addch(&tagline, '\n');
 +                              strbuf_add_commented_lines(&tagline,
 +                                              origins.items[first_tag].string,
 +                                              strlen(origins.items[first_tag].string));
 +                              strbuf_insert(&tagbuf, 0, tagline.buf,
 +                                            tagline.len);
 +                              strbuf_release(&tagline);
 +                      }
 +                      strbuf_addch(&tagbuf, '\n');
 +                      strbuf_add_commented_lines(&tagbuf,
 +                                      origins.items[i].string,
 +                                      strlen(origins.items[i].string));
 +                      fmt_tag_signature(&tagbuf, &sig, buf, len);
 +              }
 +              strbuf_release(&sig);
 +      next:
 +              free(buf);
 +      }
 +      if (tagbuf.len) {
 +              strbuf_addch(out, '\n');
 +              strbuf_addbuf(out, &tagbuf);
 +      }
 +      strbuf_release(&tagbuf);
 +}
 +
 +static void find_merge_parents(struct merge_parents *result,
 +                             struct strbuf *in, struct object_id *head)
 +{
 +      struct commit_list *parents;
 +      struct commit *head_commit;
 +      int pos = 0, i, j;
 +
 +      parents = NULL;
 +      while (pos < in->len) {
 +              int len;
 +              char *p = in->buf + pos;
 +              char *newline = strchr(p, '\n');
 +              const char *q;
 +              struct object_id oid;
 +              struct commit *parent;
 +              struct object *obj;
 +
 +              len = newline ? newline - p : strlen(p);
 +              pos += len + !!newline;
 +
 +              if (parse_oid_hex(p, &oid, &q) ||
 +                  q[0] != '\t' ||
 +                  q[1] != '\t')
 +                      continue; /* skip not-for-merge */
 +              /*
 +               * Do not use get_merge_parent() here; we do not have
 +               * "name" here and we do not want to contaminate its
 +               * util field yet.
 +               */
 +              obj = parse_object(the_repository, &oid);
 +              parent = (struct commit *)peel_to_type(NULL, 0, obj, OBJ_COMMIT);
 +              if (!parent)
 +                      continue;
 +              commit_list_insert(parent, &parents);
 +              add_merge_parent(result, &obj->oid, &parent->object.oid);
 +      }
 +      head_commit = lookup_commit(the_repository, head);
 +      if (head_commit)
 +              commit_list_insert(head_commit, &parents);
 +      reduce_heads_replace(&parents);
 +
 +      while (parents) {
 +              struct commit *cmit = pop_commit(&parents);
 +              for (i = 0; i < result->nr; i++)
 +                      if (oideq(&result->item[i].commit, &cmit->object.oid))
 +                              result->item[i].used = 1;
 +      }
 +
 +      for (i = j = 0; i < result->nr; i++) {
 +              if (result->item[i].used) {
 +                      if (i != j)
 +                              result->item[j] = result->item[i];
 +                      j++;
 +              }
 +      }
 +      result->nr = j;
 +}
 +
 +
 +int fmt_merge_msg(struct strbuf *in, struct strbuf *out,
 +                struct fmt_merge_msg_opts *opts)
 +{
 +      int i = 0, pos = 0;
 +      struct object_id head_oid;
 +      const char *current_branch;
 +      void *current_branch_to_free;
 +      struct merge_parents merge_parents;
 +
 +      memset(&merge_parents, 0, sizeof(merge_parents));
 +
 +      /* get current branch */
 +      current_branch = current_branch_to_free =
 +              resolve_refdup("HEAD", RESOLVE_REF_READING, &head_oid, NULL);
 +      if (!current_branch)
 +              die("No current branch");
 +      if (starts_with(current_branch, "refs/heads/"))
 +              current_branch += 11;
 +
 +      find_merge_parents(&merge_parents, in, &head_oid);
 +
 +      /* get a line */
 +      while (pos < in->len) {
 +              int len;
 +              char *newline, *p = in->buf + pos;
 +
 +              newline = strchr(p, '\n');
 +              len = newline ? newline - p : strlen(p);
 +              pos += len + !!newline;
 +              i++;
 +              p[len] = 0;
 +              if (handle_line(p, &merge_parents))
 +                      die("error in line %d: %.*s", i, len, p);
 +      }
 +
 +      if (opts->add_title && srcs.nr)
 +              fmt_merge_msg_title(out, current_branch);
 +
 +      if (origins.nr)
 +              fmt_merge_msg_sigs(out);
 +
 +      if (opts->shortlog_len) {
 +              struct commit *head;
 +              struct rev_info rev;
 +
 +              head = lookup_commit_or_die(&head_oid, "HEAD");
 +              repo_init_revisions(the_repository, &rev, NULL);
 +              rev.commit_format = CMIT_FMT_ONELINE;
 +              rev.ignore_merges = 1;
 +              rev.limited = 1;
 +
 +              strbuf_complete_line(out);
 +
 +              for (i = 0; i < origins.nr; i++)
 +                      shortlog(origins.items[i].string,
 +                               origins.items[i].util,
 +                               head, &rev, opts, out);
 +      }
 +
 +      strbuf_complete_line(out);
 +      free(current_branch_to_free);
 +      free(merge_parents.item);
 +      return 0;
 +}