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

[PATCH v4 7/9] add -p: calculate offset delta for edited patches

From
PWPhillip Wood <phillip.wood@talktalk.net>
Date
Mar 1, 2018, 10:51 UTC
Message-ID
<20180301105103.24268-8-phillip.wood@talktalk.net>
In-Reply-To
<20180301105103.24268-1-phillip.wood@talktalk.net>
From: Phillip Wood <phillip.wood@dunelm.org.uk>

Recount the number of preimage and postimage lines in a hunk after it has been edited so any change in the number of insertions or deletions can be used to adjust the offsets of subsequent hunks. If an edited hunk is subsequently split then the offset correction will be lost. It would be possible to fix this if it is a problem, however the code here is still an improvement on the status quo for the common case where an edited hunk is applied without being split.

This is also a necessary step to removing '--recount' and '--allow-overlap' from the invocation of 'git apply'. Before '--recount' can be removed the splitting and coalescing counting needs to be fixed to handle a missing newline at the end of a file. In order to remove '--allow-overlap' there needs to be i) some way of verifying the offset data in the edited hunk (probably by correlating the preimage (or postimage if the patch is going to be applied in reverse) lines of the edited and unedited versions to see if they are offset or if any leading/trailing context lines have been removed) and ii) a way of dealing with edited hunks that change context lines that are shared with neighbouring hunks.

Signed-off-by: Phillip Wood <phillip.wood@dunelm.org.uk>
---
Notes:
    changes since v1
     - edited hunks are now recounted before seeing if they apply in
       preparation for removing '--recount' when invoking 'git apply'
     - added sentence to commit message about the offset data being lost
       if an edited hunk is split
 git-add--interactive.perl  | 55 ++++++++++++++++++++++++++++++++++++++--------
 t/t3701-add-interactive.sh |  2 +-
 2 files changed, 47 insertions(+), 10 deletions(-)
diff --git a/git-add--interactive.perl b/git-add--interactive.perl
index 7a0a5896bb..0df0c2aa06 100755
--- a/git-add--interactive.perl
+++ b/git-add--interactive.perl
@@ -938,13 +938,19 @@ sub coalesce_overlapping_hunks {
 						parse_hunk_header($text->[0]);
 		unless ($_->{USE}) {
 			$ofs_delta += $o_cnt - $n_cnt;
+			# If this hunk has been edited then subtract
+			# the delta that is due to the edit.
+			$_->{OFS_DELTA} and $ofs_delta -= $_->{OFS_DELTA};
 			next;
 		}
 		if ($ofs_delta) {
 			$n_ofs += $ofs_delta;
 			$_->{TEXT}->[0] = format_hunk_header($o_ofs, $o_cnt,
 							     $n_ofs, $n_cnt);
 		}
+		# If this hunk was edited then adjust the offset delta
+		# to reflect the edit.
+		$_->{OFS_DELTA} and $ofs_delta += $_->{OFS_DELTA};
 		if (defined $last_o_ctx &&
 		    $o_ofs <= $last_o_ctx &&
 		    !$_->{DIRTY} &&
@@ -1016,6 +1022,30 @@ marked for discarding."),
 marked for applying."),
 );
 
+sub recount_edited_hunk {
+	local $_;
+	my ($oldtext, $newtext) = @_;
+	my ($o_cnt, $n_cnt) = (0, 0);
+	for (@{$newtext}[1..$#{$newtext}]) {
+		my $mode = substr($_, 0, 1);
+		if ($mode eq '-') {
+			$o_cnt++;
+		} elsif ($mode eq '+') {
+			$n_cnt++;
+		} elsif ($mode eq ' ') {
+			$o_cnt++;
+			$n_cnt++;
+		}
+	}
+	my ($o_ofs, undef, $n_ofs, undef) =
+					parse_hunk_header($newtext->[0]);
+	$newtext->[0] = format_hunk_header($o_ofs, $o_cnt, $n_ofs, $n_cnt);
+	my (undef, $orig_o_cnt, undef, $orig_n_cnt) =
+					parse_hunk_header($oldtext->[0]);
+	# Return the change in the number of lines inserted by this hunk
+	return $orig_o_cnt - $orig_n_cnt - $o_cnt + $n_cnt;
+}
+
 sub edit_hunk_manually {
 	my ($oldtext) = @_;
 
@@ -1114,25 +1144,32 @@ sub prompt_yesno {
 }
 
 sub edit_hunk_loop {
-	my ($head, $hunk, $ix) = @_;
-	my $text = $hunk->[$ix]->{TEXT};
+	my ($head, $hunks, $ix) = @_;
+	my $hunk = $hunks->[$ix];
+	my $text = $hunk->{TEXT};
 
 	while (1) {
-		$text = edit_hunk_manually($text);
-		if (!defined $text) {
+		my $newtext = edit_hunk_manually($text);
+		if (!defined $newtext) {
 			return undef;
 		}
 		my $newhunk = {
-			TEXT => $text,
-			TYPE => $hunk->[$ix]->{TYPE},
+			TEXT => $newtext,
+			TYPE => $hunk->{TYPE},
 			USE => 1,
 			DIRTY => 1,
 		};
+		$newhunk->{OFS_DELTA} = recount_edited_hunk($text, $newtext);
+		# If this hunk has already been edited then add the
+		# offset delta of the previous edit to get the real
+		# delta from the original unedited hunk.
+		$hunk->{OFS_DELTA} and
+				$newhunk->{OFS_DELTA} += $hunk->{OFS_DELTA};
 		if (diff_applies($head,
-				 @{$hunk}[0..$ix-1],
+				 @{$hunks}[0..$ix-1],
 				 $newhunk,
-				 @{$hunk}[$ix+1..$#{$hunk}])) {
-			$newhunk->{DISPLAY} = [color_diff(@{$text})];
+				 @{$hunks}[$ix+1..$#{$hunks}])) {
+			$newhunk->{DISPLAY} = [color_diff(@{$newtext})];
 			return $newhunk;
 		}
 		else {
diff --git a/t/t3701-add-interactive.sh b/t/t3701-add-interactive.sh
index e3150a4e07..4cc9517eda 100755
--- a/t/t3701-add-interactive.sh
+++ b/t/t3701-add-interactive.sh
@@ -552,7 +552,7 @@ test_expect_success 'add -p works with pathological context lines' '
 	test_cmp expected-1 actual
 '
 
-test_expect_failure 'add -p patch editing works with pathological context lines' '
+test_expect_success 'add -p patch editing works with pathological context lines' '
 	git reset &&
 	# n q q below is in case edit fails
 	printf "%s\n" e y    n q q |

-- 
2.16.1
Previous: Phillip WoodNext: Junio C Hamano
Message 53 of 79 in “Correct offsets of hunks when one is skipped”
  1. 0/4 Correct offsets of hunks when one is skippedPhillip Wood, Feb 13, 2018
  2. 1/4 add -i: add function to format hunk headerPhillip Wood, Feb 13, 2018
  3. 3/4 add -p: Adjust offsets of subsequent hunks when one is skippedPhillip Wood, Feb 13, 2018
  4. 4/4 add -p: calculate offset delta for edited patchesPhillip Wood, Feb 13, 2018
  5. 2/4 t3701: add failing test for pathological context linesPhillip Wood, Feb 13, 2018
  6. Junio C HamanoFeb 13, 2018
  7. brian m. carlsonFeb 13, 2018
  8. Phillip WoodFeb 19, 2018
  9. 0/9 Correct offsets of hunks when one is skippedPhillip Wood, Feb 19, 2018
  10. 1/9 add -i: add function to format hunk headerPhillip Wood, Feb 19, 2018
  11. Junio C HamanoFeb 20, 2018
  12. 2/9 t3701: indent here documentsPhillip Wood, Feb 19, 2018
  13. Eric SunshineFeb 19, 2018
  14. 7/9 add -p: calculate offset delta for edited patchesPhillip Wood, Feb 19, 2018
  15. 9/9 add -p: don't rely on apply's '--recount' optionPhillip Wood, Feb 19, 2018
  16. Junio C HamanoFeb 20, 2018
  17. 8/9 add -p: fix counting when splitting and coalescingPhillip Wood, Feb 19, 2018
  18. 4/9 t3701: don't hard code sha1 hash valuesPhillip Wood, Feb 19, 2018
  19. Eric SunshineFeb 19, 2018
  20. Junio C HamanoFeb 20, 2018
  21. Phillip WoodFeb 21, 2018
  22. Junio C HamanoFeb 21, 2018
  23. 3/9 t3701: use test_write_lines and write_scriptPhillip Wood, Feb 19, 2018
  24. Eric SunshineFeb 19, 2018
  25. Junio C HamanoFeb 20, 2018
  26. Phillip WoodFeb 21, 2018
  27. 6/9 add -p: Adjust offsets of subsequent hunks when one is skippedPhillip Wood, Feb 19, 2018
  28. 5/9 t3701: add failing test for pathological context linesPhillip Wood, Feb 19, 2018
  29. Phillip WoodFeb 21, 2018
  30. 0/9 Correct offsets of hunks when one is skippedPhillip Wood, Feb 27, 2018
  31. 1/9 add -i: add function to format hunk headerPhillip Wood, Feb 27, 2018
  32. 4/9 t3701: don't hard code sha1 hash valuesPhillip Wood, Feb 27, 2018
  33. Junio C HamanoFeb 27, 2018
  34. Phillip WoodFeb 28, 2018
  35. Phillip WoodFeb 28, 2018
  36. 3/9 t3701: use test_write_lines and write_scriptPhillip Wood, Feb 27, 2018
  37. 5/9 t3701: add failing test for pathological context linesPhillip Wood, Feb 27, 2018
  38. 9/9 add -p: don't rely on apply's '--recount' optionPhillip Wood, Feb 27, 2018
  39. 2/9 t3701: indent here documentsPhillip Wood, Feb 27, 2018
  40. Junio C HamanoFeb 27, 2018
  41. Phillip WoodFeb 28, 2018
  42. Junio C HamanoFeb 28, 2018
  43. Phillip WoodMar 1, 2018
  44. Junio C HamanoMar 1, 2018
  45. 6/9 add -p: Adjust offsets of subsequent hunks when one is skippedPhillip Wood, Feb 27, 2018
  46. 7/9 add -p: calculate offset delta for edited patchesPhillip Wood, Feb 27, 2018
  47. 8/9 add -p: fix counting when splitting and coalescingPhillip Wood, Feb 27, 2018
  48. 0/9 Correct offsets of hunks when one is skippedPhillip Wood, Mar 1, 2018
  49. 1/9 add -i: add function to format hunk headerPhillip Wood, Mar 1, 2018
  50. 9/9 add -p: don't rely on apply's '--recount' optionPhillip Wood, Mar 1, 2018
  51. Junio C HamanoMar 1, 2018
  52. Phillip WoodMar 2, 2018
  53. 7/9 add -p: calculate offset delta for edited patchesPhillip Wood, Mar 1, 2018
  54. Junio C HamanoMar 1, 2018
  55. Phillip WoodMar 2, 2018
  56. 3/9 t3701: use test_write_lines and write_scriptPhillip Wood, Mar 1, 2018
  57. 4/9 t3701: don't hard code sha1 hash valuesPhillip Wood, Mar 1, 2018
  58. SZEDER GáborMar 2, 2018
  59. Junio C HamanoMar 2, 2018
  60. Phillip WoodMar 5, 2018
  61. SZEDER GáborMar 5, 2018
  62. 2/9 t3701: indent here documentsPhillip Wood, Mar 1, 2018
  63. 5/9 t3701: add failing test for pathological context linesPhillip Wood, Mar 1, 2018
  64. 6/9 add -p: adjust offsets of subsequent hunks when one is skippedPhillip Wood, Mar 1, 2018
  65. 8/9 add -p: fix counting when splitting and coalescingPhillip Wood, Mar 1, 2018
  66. Junio C HamanoMar 1, 2018
  67. Phillip WoodMar 2, 2018
  68. 0/9 Correct offsets of hunks when one is skippedPhillip Wood, Mar 5, 2018
  69. 1/9 add -i: add function to format hunk headerPhillip Wood, Mar 5, 2018
  70. 2/9 t3701: indent here documentsPhillip Wood, Mar 5, 2018
  71. 3/9 t3701: use test_write_lines and write_scriptPhillip Wood, Mar 5, 2018
  72. 8/9 add -p: fix counting when splitting and coalescingPhillip Wood, Mar 5, 2018
  73. 6/9 add -p: adjust offsets of subsequent hunks when one is skippedPhillip Wood, Mar 5, 2018
  74. 5/9 t3701: add failing test for pathological context linesPhillip Wood, Mar 5, 2018
  75. 4/9 t3701: don't hard code sha1 hash valuesPhillip Wood, Mar 5, 2018
  76. 7/9 add -p: calculate offset delta for edited patchesPhillip Wood, Mar 5, 2018
  77. 9/9 add -p: don't rely on apply's '--recount' optionPhillip Wood, Mar 5, 2018
  78. Junio C HamanoMar 5, 2018
  79. Phillip WoodMar 6, 2018

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.