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

Re: [PATCH] add -p: introduce 'w' command to view hunk with --word-diff

From
Pablo <pabloosabaterr@gmail.com>
Date
May 7, 2026, 07:55 UTC
Message-ID
<CAN5EUNRT7V3BrtyU0UYwGVnJ51LWSsNi1OnzMB5WL=w8vhKmrw@mail.gmail.com>
In-Reply-To
<20260506235459.529862-1-javierbassi@gmail.com>
El jue, 7 may 2026 a las 1:58, Javier Bassi (<javierbassi@gmail.com>) escribió:
Show 5 quoted lines
>
> When using `git add --patch`, reviewing changes in long lines can be
> difficult with the default line-based diff. This is particularly
> noticeable in formats such as JSONP, CSV, LaTeX, Markdown, or other
> plain text where small inline edits can be hard to spot.
Very good explanation. I think it's a good idea.
>
> Added `w - print the current hunk with word-diff` during hunk selection
  ^^^^^^^
Nit: following Documentation/SubmittingPatches, try to use an
imperative mood to write what has been done.
Show 52 quoted lines
> to re-display the current hunk using `--word-diff`. This provides a
> clearer inline view of changes without modifying the hunk or affecting
> how patches are applied or staged.
>
> Signed-off-by: Javier Bassi <javierbassi@gmail.com>
> ---
>  Documentation/git-add.adoc |   1 +
>  add-patch.c                | 105 ++++++++++++++++++++++++++++++++++++-
>  t/t3701-add-interactive.sh |  59 +++++++++++++--------
>  3 files changed, 142 insertions(+), 23 deletions(-)
>
> diff --git a/Documentation/git-add.adoc b/Documentation/git-add.adoc
> index 941135dc63..d2ee1cf9a9 100644
> --- a/Documentation/git-add.adoc
> +++ b/Documentation/git-add.adoc
> @@ -351,6 +351,7 @@ patch::
>         K - go to the previous hunk, roll over at the top
>         s - split the current hunk into smaller hunks
>         e - manually edit the current hunk
> +       w - print the current hunk with word-diff
>         p - print the current hunk
>         P - print the current hunk using the pager
>         ? - print help
> diff --git a/add-patch.c b/add-patch.c
> index f27edcbe8d..0364f4bc97 100644
> --- a/add-patch.c
> +++ b/add-patch.c
> @@ -7,6 +7,7 @@
>  #include "commit.h"
>  #include "config.h"
>  #include "diff.h"
> +#include "diffcore.h"
>  #include "editor.h"
>  #include "environment.h"
>  #include "gettext.h"
> @@ -1508,6 +1509,105 @@ static void summarize_hunk(struct add_p_state *s, struct hunk *hunk,
>         strbuf_complete_line(out);
>  }
>
> +static void trim_trailing_lf(struct strbuf *buf)
> +{
> +       if (buf->len && buf->buf[buf->len - 1] == '\n')
> +               strbuf_setlen(buf, buf->len - 1);
> +}
> +
> +static void add_word_diff_line(struct strbuf *old, struct strbuf *new,
> +                              const char *line, size_t len, char marker)
> +{
> +       if (marker == '-' || marker == '+' || *line == ' ') {
> +               line++;
> +               len--;
> +       }

Maybe a tiny comment here would help, to know why '*line' is being checked here instead of 'marker'. They seem the same and one has to go to marker declaration and see the comment at 'normalize_marker()'

  /* Empty context lines may omit the leading ' ' */
Show 24 quoted lines
> +
> +       if (marker != '+')
> +               strbuf_add(old, line, len);
> +       if (marker != '-')
> +               strbuf_add(new, line, len);
> +}
> +
> +static void build_word_diff_files(struct add_p_state *s, struct hunk *hunk,
> +                                 struct strbuf *old, struct strbuf *new)
> +{
> +       size_t i;
> +       char last_marker = '\0';
> +
> +       for (i = hunk->start; i < hunk->end; i = find_next_line(&s->plain, i)) {
> +               size_t next = find_next_line(&s->plain, i);
> +               char marker = normalize_marker(s->plain.buf + i);
> +
> +               if (marker == '\\') {
> +                       if (last_marker != '+')
> +                               trim_trailing_lf(old);
> +                       if (last_marker != '-')
> +                               trim_trailing_lf(new);
> +                       continue;
> +               }

Here we check about "\No newline at end of file", after this point I believe that 'buf->buf[buf->len - 1] == '\n'' will always be true. Same should be for 'buf->len' because "\No newline at end of file" shouldn't come first and a '+' '-' line should have been added on a previous iteration, but the check it's fine, just in case I'm wrong.

What I want to point out is, is the 'trim_trailing_lf' function necessary? It's only called in the same place and it carries a check that could be on the caller instead, leaving the function only with 'strbuf_setlen(buf, buf->len - 1);" making sense to inline it at this point. You could keep the buf->len check:

  if (marker == '\\') {
          if (last_marker != '+' && old->len)
                   strbuf_setlen(old, old->len - 1);
          if (last_marker != '-' && new->len)
                   strbuf_setlen(new, new->len - 1);
          continue;
  }
Show 35 quoted lines
> +
> +               if (marker != ' ' && marker != '-' && marker != '+')
> +                       BUG("unhandled diff marker: '%c'", marker);
> +
> +               add_word_diff_line(old, new, s->plain.buf + i, next - i,
> +                                  marker);
> +               last_marker = marker;
> +       }
> +}
> +
> +static struct diff_filespec *word_diff_filespec(struct repository *r,
> +                                               const char *name,
> +                                               struct strbuf *buf)
> +{
> +       struct diff_filespec *spec = alloc_filespec(name);
> +       size_t size;
> +
> +       fill_filespec(spec, null_oid(r->hash_algo), 0, 0100644);
> +       spec->data = strbuf_detach(buf, &size);
> +       spec->size = size;
> +       spec->should_free = 1;
> +       spec->is_stdin = 1;
> +
> +       return spec;
> +}
> +
> +static void show_hunk_word_diff(struct add_p_state *s, struct hunk *hunk,
> +                               int colored)
> +{
> +       struct hunk_header *header = &hunk->header;
> +       struct strbuf old = STRBUF_INIT, new = STRBUF_INIT;
> +       struct diff_options opts;
> +       struct diff_queue_struct queue;
> +
> +       if (!header->old_offset && !header->new_offset) {
Nit: a comment here would help to understand this early return
Show 222 quoted lines
> +               strbuf_reset(&s->buf);
> +               render_hunk(s, hunk, 0, colored, &s->buf);
> +               fputs(s->buf.buf, stdout);
> +               return;
> +       }
> +
> +       build_word_diff_files(s, hunk, &old, &new);
> +
> +       repo_diff_setup(s->r, &opts);
> +       opts.output_format = DIFF_FORMAT_PATCH;
> +       opts.use_color = colored ? s->cfg.use_color_diff : GIT_COLOR_NEVER;
> +       opts.word_diff = DIFF_WORDS_PLAIN;
> +       opts.context = header->old_count > header->new_count ?
> +               header->old_count : header->new_count;
> +       opts.flags.suppress_diff_headers = 1;
> +       diff_setup_done(&opts);
> +
> +       memcpy(&queue, &diff_queued_diff, sizeof(diff_queued_diff));
> +       diff_queue_init(&diff_queued_diff);
> +       diff_queue(&diff_queued_diff,
> +                  word_diff_filespec(s->r, "a", &old),
> +                  word_diff_filespec(s->r, "b", &new));
> +       diffcore_std(&opts);
> +       diff_flush(&opts);
> +       memcpy(&diff_queued_diff, &queue, sizeof(diff_queued_diff));
> +}
> +
>  #define DISPLAY_HUNKS_LINES 20
>  static size_t display_hunks(struct add_p_state *s,
>                             struct file_diff *file_diff, size_t start_index)
> @@ -1540,6 +1640,7 @@ N_("j - go to the next undecided hunk, roll over at the bottom\n"
>     "/ - search for a hunk matching the given regex\n"
>     "s - split the current hunk into smaller hunks\n"
>     "e - manually edit the current hunk\n"
> +   "w - print the current hunk with word-diff\n"
>     "p - print the current hunk\n"
>     "P - print the current hunk using the pager\n"
>     "> - go to the next file, roll over at the bottom\n"
> @@ -1731,7 +1832,7 @@ static size_t patch_update_file(struct add_p_state *s,
>                                 permitted |= ALLOW_GOTO_PREVIOUS_FILE;
>                                 strbuf_addstr(&s->buf, ",<");
>                         }
> -                       strbuf_addstr(&s->buf, ",p,P");
> +                       strbuf_addstr(&s->buf, ",w,p,P");
>                 }
>                 if (file_diff->deleted)
>                         prompt_mode_type = PROMPT_DELETION;
> @@ -1953,6 +2054,8 @@ static size_t patch_update_file(struct add_p_state *s,
>                                 hunk->use = USE_HUNK;
>                                 goto soft_increment;
>                         }
> +               } else if (s->answer.buf[0] == 'w') {
> +                       show_hunk_word_diff(s, hunk, colored);
>                 } else if (ch == 'p') {
>                         rendered_hunk_index = -1;
>                         use_pager = (s->answer.buf[0] == 'P') ? 1 : 0;
> diff --git a/t/t3701-add-interactive.sh b/t/t3701-add-interactive.sh
> index 6e120a4001..e1ce98d62b 100755
> --- a/t/t3701-add-interactive.sh
> +++ b/t/t3701-add-interactive.sh
> @@ -48,8 +48,8 @@ test_expect_success 'unknown command' '
>         git add -N command &&
>         git diff command >expect &&
>         cat >>expect <<-EOF &&
> -       (1/1) Stage addition [y,n,q,a,d,e,p,P,?]? Unknown command ${SQ}W${SQ} (use ${SQ}?${SQ} for help)
> -       (1/1) Stage addition [y,n,q,a,d,e,p,P,?]?$SP
> +       (1/1) Stage addition [y,n,q,a,d,e,w,p,P,?]? Unknown command ${SQ}W${SQ} (use ${SQ}?${SQ} for help)
> +       (1/1) Stage addition [y,n,q,a,d,e,w,p,P,?]?$SP
>         EOF
>         git add -p -- command <command >actual 2>&1 &&
>         test_cmp expect actual
> @@ -332,9 +332,9 @@ test_expect_success 'different prompts for mode change/deleted' '
>         git -c core.filemode=true add -p >actual &&
>         sed -n "s/^\(([0-9/]*) Stage .*?\).*/\1/p" actual >actual.filtered &&
>         cat >expect <<-\EOF &&
> -       (1/1) Stage deletion [y,n,q,a,d,p,P,?]?
> -       (1/2) Stage mode change [y,n,q,a,d,k,K,j,J,g,/,p,P,?]?
> -       (2/2) Stage this hunk [y,n,q,a,d,K,J,g,/,e,p,P,?]?
> +       (1/1) Stage deletion [y,n,q,a,d,w,p,P,?]?
> +       (1/2) Stage mode change [y,n,q,a,d,k,K,j,J,g,/,w,p,P,?]?
> +       (2/2) Stage this hunk [y,n,q,a,d,K,J,g,/,e,w,p,P,?]?
>         EOF
>         test_cmp expect actual.filtered
>  '
> @@ -521,13 +521,13 @@ test_expect_success 'split hunk setup' '
>  test_expect_success 'goto hunk 1 with "g 1"' '
>         test_when_finished "git reset" &&
>         tr _ " " >expect <<-EOF &&
> -       (2/2) Stage this hunk [y,n,q,a,d,K,J,g,/,e,p,P,?]? + 1:  -1,2 +1,3          +15
> +       (2/2) Stage this hunk [y,n,q,a,d,K,J,g,/,e,w,p,P,?]? + 1:  -1,2 +1,3          +15
>         _ 2:  -2,4 +3,8          +21
>         go to which hunk? @@ -1,2 +1,3 @@
>         _10
>         +15
>         _20
> -       (1/2) Stage this hunk (was: y) [y,n,q,a,d,k,K,j,J,g,/,e,p,P,?]?_
> +       (1/2) Stage this hunk (was: y) [y,n,q,a,d,k,K,j,J,g,/,e,w,p,P,?]?_
>         EOF
>         test_write_lines s y g 1 | git add -p >actual &&
>         tail -n 7 <actual >actual.trimmed &&
> @@ -540,7 +540,7 @@ test_expect_success 'goto hunk 1 with "g1"' '
>         _10
>         +15
>         _20
> -       (1/2) Stage this hunk (was: y) [y,n,q,a,d,k,K,j,J,g,/,e,p,P,?]?_
> +       (1/2) Stage this hunk (was: y) [y,n,q,a,d,k,K,j,J,g,/,e,w,p,P,?]?_
>         EOF
>         test_write_lines s y g1 | git add -p >actual &&
>         tail -n 4 <actual >actual.trimmed &&
> @@ -550,11 +550,11 @@ test_expect_success 'goto hunk 1 with "g1"' '
>  test_expect_success 'navigate to hunk via regex /pattern' '
>         test_when_finished "git reset" &&
>         tr _ " " >expect <<-EOF &&
> -       (2/2) Stage this hunk [y,n,q,a,d,K,J,g,/,e,p,P,?]? @@ -1,2 +1,3 @@
> +       (2/2) Stage this hunk [y,n,q,a,d,K,J,g,/,e,w,p,P,?]? @@ -1,2 +1,3 @@
>         _10
>         +15
>         _20
> -       (1/2) Stage this hunk (was: y) [y,n,q,a,d,k,K,j,J,g,/,e,p,P,?]?_
> +       (1/2) Stage this hunk (was: y) [y,n,q,a,d,k,K,j,J,g,/,e,w,p,P,?]?_
>         EOF
>         test_write_lines s y /1,2 | git add -p >actual &&
>         tail -n 5 <actual >actual.trimmed &&
> @@ -567,7 +567,7 @@ test_expect_success 'navigate to hunk via regex / pattern' '
>         _10
>         +15
>         _20
> -       (1/2) Stage this hunk (was: y) [y,n,q,a,d,k,K,j,J,g,/,e,p,P,?]?_
> +       (1/2) Stage this hunk (was: y) [y,n,q,a,d,k,K,j,J,g,/,e,w,p,P,?]?_
>         EOF
>         test_write_lines s y / 1,2 | git add -p >actual &&
>         tail -n 4 <actual >actual.trimmed &&
> @@ -579,27 +579,42 @@ test_expect_success 'print again the hunk' '
>         tr _ " " >expect <<-EOF &&
>         +15
>          20
> -       (1/2) Stage this hunk (was: y) [y,n,q,a,d,k,K,j,J,g,/,e,p,P,?]? @@ -1,2 +1,3 @@
> +       (1/2) Stage this hunk (was: y) [y,n,q,a,d,k,K,j,J,g,/,e,w,p,P,?]? @@ -1,2 +1,3 @@
>          10
>         +15
>          20
> -       (1/2) Stage this hunk (was: y) [y,n,q,a,d,k,K,j,J,g,/,e,p,P,?]?_
> +       (1/2) Stage this hunk (was: y) [y,n,q,a,d,k,K,j,J,g,/,e,w,p,P,?]?_
>         EOF
>         test_write_lines s y g 1 p | git add -p >actual &&
>         tail -n 7 <actual >actual.trimmed &&
>         test_cmp expect actual.trimmed
>  '
>
> +test_expect_success 'print hunk with word-diff' '
> +       test_when_finished "rm -rf word-diff-repo" &&
> +       git init word-diff-repo &&
> +       (
> +               cd word-diff-repo &&
> +               test_write_lines "alpha old beta" context >word-diff &&
> +               git add word-diff &&
> +               git commit -m word-diff &&
> +               test_write_lines "alpha new beta" context >word-diff &&
> +               test_write_lines w n | git add -p word-diff >actual &&
> +               test_grep "alpha \\[-old-\\]{+new+} beta" actual &&
> +               git diff --cached --exit-code
> +       )
> +'
> +
>  test_expect_success TTY 'print again the hunk (PAGER)' '
>         test_when_finished "git reset" &&
>         cat >expect <<-EOF &&
>         <GREEN>+<RESET><GREEN>15<RESET>
>          20<RESET>
> -       <BOLD;BLUE>(1/2) Stage this hunk (was: y) [y,n,q,a,d,k,K,j,J,g,/,e,p,P,?]? <RESET>PAGER <CYAN>@@ -1,2 +1,3 @@<RESET>
> +       <BOLD;BLUE>(1/2) Stage this hunk (was: y) [y,n,q,a,d,k,K,j,J,g,/,e,w,p,P,?]? <RESET>PAGER <CYAN>@@ -1,2 +1,3 @@<RESET>
>         PAGER  10<RESET>
>         PAGER <GREEN>+<RESET><GREEN>15<RESET>
>         PAGER  20<RESET>
> -       <BOLD;BLUE>(1/2) Stage this hunk (was: y) [y,n,q,a,d,k,K,j,J,g,/,e,p,P,?]? <RESET>
> +       <BOLD;BLUE>(1/2) Stage this hunk (was: y) [y,n,q,a,d,k,K,j,J,g,/,e,w,p,P,?]? <RESET>
>         EOF
>         test_write_lines s y g 1 P |
>         (
> @@ -796,21 +811,21 @@ test_expect_success 'colors can be overridden' '
>         <BLUE>+<RESET><BLUE>new<RESET>
>         <CYAN> more-context<RESET>
>         <BLUE>+<RESET><BLUE>another-one<RESET>
> -       <YELLOW>(1/1) Stage this hunk [y,n,q,a,d,s,e,p,P,?]? <RESET><BOLD>Split into 2 hunks.<RESET>
> +       <YELLOW>(1/1) Stage this hunk [y,n,q,a,d,s,e,w,p,P,?]? <RESET><BOLD>Split into 2 hunks.<RESET>
>         <MAGENTA>@@ -1,3 +1,3 @@<RESET>
>         <CYAN> context<RESET>
>         <BOLD>-old<RESET>
>         <BLUE>+<RESET><BLUE>new<RESET>
>         <CYAN> more-context<RESET>
> -       <YELLOW>(1/2) Stage this hunk [y,n,q,a,d,k,K,j,J,g,/,e,p,P,?]? <RESET><MAGENTA>@@ -3 +3,2 @@<RESET>
> +       <YELLOW>(1/2) Stage this hunk [y,n,q,a,d,k,K,j,J,g,/,e,w,p,P,?]? <RESET><MAGENTA>@@ -3 +3,2 @@<RESET>
>         <CYAN> more-context<RESET>
>         <BLUE>+<RESET><BLUE>another-one<RESET>
> -       <YELLOW>(2/2) Stage this hunk [y,n,q,a,d,K,J,g,/,e,p,P,?]? <RESET><MAGENTA>@@ -1,3 +1,3 @@<RESET>
> +       <YELLOW>(2/2) Stage this hunk [y,n,q,a,d,K,J,g,/,e,w,p,P,?]? <RESET><MAGENTA>@@ -1,3 +1,3 @@<RESET>
>         <CYAN> context<RESET>
>         <BOLD>-old<RESET>
>         <BLUE>+new<RESET>
>         <CYAN> more-context<RESET>
> -       <YELLOW>(1/2) Stage this hunk (was: y) [y,n,q,a,d,k,K,j,J,g,/,e,p,P,?]? <RESET>
> +       <YELLOW>(1/2) Stage this hunk (was: y) [y,n,q,a,d,k,K,j,J,g,/,e,w,p,P,?]? <RESET>
>         EOF
>         test_cmp expect actual
>  '
> @@ -1424,9 +1439,9 @@ test_expect_success 'invalid option s is rejected' '
>         test_write_lines j s q | git add -p >out &&
>         sed -ne "s/ @@.*//" -e "s/ \$//" -e "/^(/p" <out >actual &&
>         cat >expect <<-EOF &&
> -       (1/2) Stage this hunk [y,n,q,a,d,k,K,j,J,g,/,s,e,p,P,?]?
> -       (2/2) Stage this hunk [y,n,q,a,d,k,K,j,J,g,/,e,p,P,?]? Sorry, cannot split this hunk
> -       (2/2) Stage this hunk [y,n,q,a,d,k,K,j,J,g,/,e,p,P,?]?
> +       (1/2) Stage this hunk [y,n,q,a,d,k,K,j,J,g,/,s,e,w,p,P,?]?
> +       (2/2) Stage this hunk [y,n,q,a,d,k,K,j,J,g,/,e,w,p,P,?]? Sorry, cannot split this hunk
> +       (2/2) Stage this hunk [y,n,q,a,d,k,K,j,J,g,/,e,w,p,P,?]?
>         EOF
>         test_cmp expect actual
>  '
> --
> 2.54.0
>
>

I tried and found out that if a character is contiguous to a word diff it will be carried as part of the diff. e.g.:

  diff --git a/file b/file
  index f33f47f..93dd39c 100644
  --- a/file
  +++ b/file
  @@ -1 +1 @@
  -foo, bar
  +baz, bar
  (1/1) Stage this hunk [y,n,q,a,d,e,w,p,P,?]? w
  @@ -1 +1 @@
  [-foo,-]{+baz,+} bar
the ',' remains at the same place but became part of the diff which adds noise.

Regards, -- Pablo

Previous: Javier BassiNext: Phillip Wood
Message 2 of 9 in “add -p: introduce 'w' command to view hunk with --word-diff”
  1. add -p: introduce 'w' command to view hunk with --word-diffJavier Bassi, May 6, 2026
  2. PabloMay 7, 2026
  3. Phillip WoodMay 7, 2026
  4. PabloMay 7, 2026
  5. Phillip WoodMay 7, 2026
  6. D. Ben KnobleMay 7, 2026
  7. Junio C HamanoMay 11, 2026
  8. D. Ben KnobleMay 11, 2026
  9. Junio C HamanoMay 12, 2026

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.