git/list[1] front-page[2] threads[3] people[4] search[5] about
 

Re: [PATCH v14 07/13] ref-filter: add support for %(contents:lines=X)

From
Eric Sunshine <sunshine@sunshineco.com>
Date
Aug 30, 2015, 07:53 UTC
Message-ID
<CAPig+cQmAkFAjUcJNJNF2gwunX6TKK+oQsd2QnmJwxLT=EEvjQ@mail.gmail.com>
In-Reply-To
<1440857537-13968-8-git-send-email-Karthik.188@gmail.com>
On Sat, Aug 29, 2015 at 10:12 AM, Karthik Nayak <karthik.188@gmail.com> wrote:
Show 33 quoted lines
> In 'tag.c' we can print N lines from the annotation of the tag using
> the '-n<num>' option. Copy code from 'tag.c' to 'ref-filter' and
> modify it to support appending of N lines from the annotation of tags
> to the given strbuf.
>
> Implement %(contents:lines=X) where X lines of the given object are
> obtained.
>
> Add documentation and test for the same.
>
> Signed-off-by: Karthik Nayak <karthik.188@gmail.com>
> ---
> diff --git a/builtin/tag.c b/builtin/tag.c
> index 471d6b1..0fc7557 100644
> --- a/builtin/tag.c
> +++ b/builtin/tag.c
> @@ -63,6 +64,11 @@ struct align {
>         unsigned int width;
>  };
>
> +struct contents {
> +       unsigned int lines;
> +       struct object_id oid;
> +};
> +
>  #define REF_FORMATTING_STATE_INIT  { 0, NULL }
>
>  struct ref_formatting_stack {
> @@ -80,6 +86,7 @@ struct ref_formatting_state {
>  struct atom_value {
>         const char *s;
>         struct align *align;
> +       struct contents *contents;

Same question as for 'align': Does 'contents' need to be heap-allocated because it must exist beyond the lifetime of 'atom_value'? If not, making it just a plain member of 'atom_value' would simplify memory management (no need to free it).

Also, will 'align' and 'contents' ever be used at the same time? If not, it might make sense to place them in a 'union' (not for the memory saving, but to make it clear to the reader that their use is mutually exclusive).

More below.
Show 49 quoted lines
>         void (*handler)(struct atom_value *atomv, struct ref_formatting_state *state);
>         unsigned long ul; /* used for sorting when not FIELD_STR */
>  };
> @@ -569,6 +576,61 @@ static void find_subpos(const char *buf, unsigned long sz,
>         *nonsiglen = *sig - buf;
>  }
>
> +/*
> + * If 'lines' is greater than 0, append that many lines from the given
> + * object_id 'oid' to the given strbuf.
> + */
> +static void append_tag_lines(struct strbuf *out, const struct object_id *oid, int lines)
> +{
> +       int i;
> +       unsigned long size;
> +       enum object_type type;
> +       char *buf, *sp, *eol;
> +       size_t len;
> +
> +       buf = read_sha1_file(oid->hash, &type, &size);
> +       if (!buf)
> +               die_errno("unable to read object %s", oid_to_hex(oid));
> +       if (type != OBJ_COMMIT && type != OBJ_TAG)
> +               goto free_return;
> +       if (!size)
> +               die("an empty %s object %s?",
> +                   typename(type), oid_to_hex(oid));
> +
> +       /* skip header */
> +       sp = strstr(buf, "\n\n");
> +       if (!sp)
> +               goto free_return;
> +
> +       /* only take up to "lines" lines, and strip the signature from a tag */
> +       if (type == OBJ_TAG)
> +               size = parse_signature(buf, size);
> +       for (i = 0, sp += 2; i < lines && sp < buf + size; i++) {
> +               if (i)
> +                       strbuf_addstr(out, "\n    ");
> +               eol = memchr(sp, '\n', size - (sp - buf));
> +               len = eol ? eol - sp : size - (sp - buf);
> +               strbuf_add(out, sp, len);
> +               if (!eol)
> +                       break;
> +               sp = eol + 1;
> +       }
> +free_return:
> +       free(buf);
> +}

I understand that you want to re-use this code from tag.c:show_tag_lines(), but (from a very cursory read) isn't this duplicating logic and processing already done elsewhere in ref-filter.c? More about this below.

> +
> +static void contents_lines_handler(struct atom_value *atomv, struct ref_formatting_state *state)
> +{
> +       struct contents *contents = (struct contents *)atomv->contents;
Why is this cast needed?
Show 28 quoted lines
> +       struct strbuf s = STRBUF_INIT;
> +
> +       append_tag_lines(&s, &contents->oid, contents->lines);
> +       quote_formatting(&state->stack->output, s.buf, state->quote_style);
> +       strbuf_release(&s);
> +
> +       free(contents);
> +}
> @@ -588,7 +651,8 @@ static void grab_sub_body_contents(struct atom_value *val, int deref, struct obj
>                     strcmp(name, "contents") &&
>                     strcmp(name, "contents:subject") &&
>                     strcmp(name, "contents:body") &&
> -                   strcmp(name, "contents:signature"))
> +                   strcmp(name, "contents:signature") &&
> +                   !starts_with(name, "contents:lines="))
>                         continue;
>                 if (!subpos)
>                         find_subpos(buf, sz,
> @@ -608,6 +672,15 @@ static void grab_sub_body_contents(struct atom_value *val, int deref, struct obj
>                         v->s = xmemdupz(sigpos, siglen);
>                 else if (!strcmp(name, "contents"))
>                         v->s = xstrdup(subpos);
> +               else if (skip_prefix(name, "contents:lines=", &valp)) {
> +                       struct contents *contents = xmalloc(sizeof(struct contents));
> +
> +                       if (strtoul_ui(valp, 10, &contents->lines))
> +                               die(_("positive width expected align:%s"), valp);
> +                       hashcpy(contents->oid.hash, obj->sha1);

The logic in append_tag_lines() which was copied from tag.c:show_tag_lines() goes through the effort of loading the object and parsing it, but hasn't the object already been loaded and parsed by the time you get to this point? Assuming I'm reading this correctly, wouldn't it make more sense to take advantage of the work already done loading and parsing the object rather than repeating it all inside append_tag_lines()?

Show 26 quoted lines
> +                       v->handler = contents_lines_handler;
> +                       v->contents = contents;
> +               }
>         }
>  }
> diff --git a/t/t6302-for-each-ref-filter.sh b/t/t6302-for-each-ref-filter.sh
> index cef7a41..0277498 100755
> --- a/t/t6302-for-each-ref-filter.sh
> +++ b/t/t6302-for-each-ref-filter.sh
> @@ -166,4 +166,20 @@ test_expect_success 'nested alignment' '
>         test_cmp expect actual
>  '
>
> +test_expect_success 'check `%(contents:lines=X)`' '
> +       cat >expect <<-\EOF &&
> +       master three
> +       side four
> +       odd/spot three
> +       double-tag Annonated doubly
> +       four four
> +       one one
> +       signed-tag A signed tag message
> +       three three
> +       two two
> +       EOF
> +       git for-each-ref --format="%(refname:short) %(contents:lines=1)" >actual &&

Maybe also test some edge cases, such as line=0, lines=-1 (an invalid value), lines=2, lines=9999999 (a value larger than the number of lines in any object).

Show 6 quoted lines
> +       test_cmp expect actual
> +'
> +
>  test_done
> --
> 2.5.0
Previous: Karthik NayakNext: Karthik Nayak
Message 36 of 52 in “Port tag.c to use ref-filter.c”
  1. 00/13 Port tag.c to use ref-filter.cKarthik Nayak, Aug 29, 2015
  2. 01/13 ref-filter: move `struct atom_value` to ref-filter.cKarthik Nayak, Aug 29, 2015
  3. 02/13 ref-filter: introduce ref_formatting_state and ref_formatting_stackKarthik Nayak, Aug 29, 2015
  4. 03/13 utf8: add function to align a string into given strbufKarthik Nayak, Aug 29, 2015
  5. Torsten BögershausenAug 29, 2015
  6. Karthik NayakAug 29, 2015
  7. 04/13 ref-filter: implement an `align` atomKarthik Nayak, Aug 29, 2015
  8. Eric SunshineAug 30, 2015
  9. Karthik NayakAug 30, 2015
  10. Eric SunshineAug 30, 2015
  11. Karthik NayakAug 31, 2015
  12. Eric SunshineAug 31, 2015
  13. Matthieu MoyAug 31, 2015
  14. Eric SunshineAug 31, 2015
  15. Karthik NayakSep 1, 2015
  16. Matthieu MoySep 1, 2015
  17. Karthik NayakSep 1, 2015
  18. Karthik NayakAug 30, 2015
  19. Eric SunshineAug 30, 2015
  20. Karthik NayakAug 31, 2015
  21. Junio C HamanoAug 30, 2015
  22. Eric SunshineAug 30, 2015
  23. Karthik NayakAug 31, 2015
  24. Karthik NayakAug 31, 2015
  25. Matthieu MoyAug 31, 2015
  26. Karthik NayakAug 31, 2015
  27. 05/13 ref-filter: add option to filter out tags, branches and remotesKarthik Nayak, Aug 29, 2015
  28. Eric SunshineAug 30, 2015
  29. Karthik NayakAug 30, 2015
  30. Eric SunshineAug 30, 2015
  31. 06/13 ref-filter: introduce format_ref_array_item()Karthik Nayak, Aug 29, 2015
  32. Eric SunshineAug 30, 2015
  33. Karthik NayakAug 30, 2015
  34. Karthik NayakAug 30, 2015
  35. 07/13 ref-filter: add support for %(contents:lines=X)Karthik Nayak, Aug 29, 2015
  36. Eric SunshineAug 30, 2015
  37. Karthik NayakAug 30, 2015
  38. Eric SunshineAug 30, 2015
  39. Karthik NayakAug 30, 2015
  40. Eric SunshineAug 30, 2015
  41. Karthik NayakAug 31, 2015
  42. 08/13 ref-filter: add support to sort by versionKarthik Nayak, Aug 29, 2015
  43. 09/13 ref-filter: add option to match literal patternKarthik Nayak, Aug 29, 2015
  44. 10/13 tag.c: use 'ref-filter' data structuresKarthik Nayak, Aug 29, 2015
  45. 11/13 tag.c: use 'ref-filter' APIsKarthik Nayak, Aug 29, 2015
  46. 12/13 tag.c: implement '--format' optionKarthik Nayak, Aug 29, 2015
  47. 13/13 tag.c: implement '--merged' and '--no-merged' optionsKarthik Nayak, Aug 29, 2015
  48. Matthieu MoyAug 31, 2015
  49. Karthik NayakAug 31, 2015
  50. Matthieu MoyAug 31, 2015
  51. Karthik NayakAug 31, 2015
  52. Karthik NayakSep 1, 2015

Read the whole thread, see it on lore, or plain text.

$ cat FOOTERMessages come from the public archive at lore.kernel.org/git, fetched every hour. The front page is chosen and written each morning by an AI editor and can be wrong; the threads themselves are the record. About and API. For agents: an MCP server at https://gitlist.dev/mcp, and any thread, story or person page as Markdown by adding .md to its URL (or sending Accept: text/markdown). Details in /llms.txt.