Fix line diff by using rune index without a separator - #136
Merged
Conversation
Current implementation produces wrong result because it calls `DiffMain` on the following 2 arguments: * `1,2,3,4,5,6,7,8,9,10` * `1,2,3,4,5,6,7,8,9,11` This numbers represent indices into the lines array. The algorithm finds that equal part of those strings is `1,2,3,4,5,6,7,8,9,1` and which is followed by `Delete 0` and `Insert `1`.
kdarkhan
commented
Jan 22, 2023
kdarkhan
force-pushed
the
master
branch
2 times, most recently
from
January 23, 2023 23:27
a9d3bea to
dfb3555
Compare
kdarkhan
commented
Jan 23, 2023
| ) | ||
|
|
||
| go 1.12 | ||
| go 1.13 |
Contributor
Author
There was a problem hiding this comment.
Upgrading to go 1.13 in order to use binary and hex integer literals.
[The suggested approach](https://github.com/google/diff-match-patch/wiki/Line-or-Word-Diffs#line-mode ) for doing line level diffing is the following set of steps: 1. `ti1, ti2, linesIdx = DiffLinesToChars(t1, t2)` 2. `diffs = DiffMain(ti1, ti2)` 3. `DiffCharsToLines(diff, linesIdx)` The original implementation in `google/diff-match-patch` uses unicode codepoints for storing indices in `ti1` and `ti2` joined by an empty string. Current implementation in this repo stores them as integers joined by a comma. While this implementation makes `ti1` and `ti2` more readable, it introduces bugs when trying to rely on it when doing line level diffing with `DiffMain`. The root cause of the issue is that an integer line index might span more than one character/rune, and `DiffMain` can assume that two different lines having the same index prefix match partially. For example, indices 123 and 129 will have partial match `12`. In that example, the diff will show lines 3 and 9 which is not correct. A simple failing test case demonstrating this issue is available at `TestDiffPartialLineIndex`. In this PR I am adjusting the algorithm to use the same approach as in [diff-match-patch](https://github.com/google/diff-match-patch/blob/62f2e689f498f9c92dbc588c58750addec9b1654/javascript/diff_match_patch_uncompressed.js#L508-L510 ) by storing each line index as a rune. While a rune in Golang is a type alias to uint32, not every uint32 can be a valid rune. During string to rune slice conversion invalid runes will be replaced with `utf.RuneError`. The integer to rune generation logic is based on the table in https://en.wikipedia.org/wiki/UTF-8#Encoding The first 127 lines will work the fastest as they are represented as a single bytes. Higher numbers are represented as 2-4 bytes. In addition to that, the range `U+D800 - U+DFFF` contains [invalid codepoints](https://en.wikipedia.org/wiki/UTF-8#Invalid_sequences_and_error_handling). and all codepoints higher or equal to `0xD800` are incremented by `0xDFFF - 0xD800`. The maximum representable integer using this approach is 1'112'060. This improves on Javascript implementation which currently [bails out](https://github.com/google/diff-match-patch/blob/62f2e689f498f9c92dbc588c58750addec9b1654/javascript/diff_match_patch_uncompressed.js#L503-L505 ) when files have more than 65535 lines.
This was referenced Jan 23, 2023
Contributor
Author
|
Here is the benchmark comparison with the original code and the new code from this PR: |
Contributor
Author
|
I did not realize that there is an existing PR #120 which implements somewhat similar solution. My PR implements a solution closer to the Javascript version |
Owner
|
LGTM |
|
@sergi could we publish a tag with this in it? |
|
Re-ping for @sergi |
atiratree
added a commit
to atiratree/api
that referenced
this pull request
Aug 2, 2024
- removed exclude for github.com/sergi/go-diff as it should have been fixed by sergi/go-diff#136
atiratree
added a commit
to atiratree/api
that referenced
this pull request
Aug 5, 2024
- removed exclude for github.com/sergi/go-diff as it should have been fixed by sergi/go-diff#136
atiratree
added a commit
to atiratree/api
that referenced
this pull request
Aug 6, 2024
- removed exclude for github.com/sergi/go-diff as it should have been fixed by sergi/go-diff#136
atiratree
added a commit
to atiratree/api
that referenced
this pull request
Aug 6, 2024
- removed exclude for github.com/sergi/go-diff as it should have been fixed by sergi/go-diff#136
atiratree
added a commit
to atiratree/api
that referenced
this pull request
Aug 6, 2024
- removed exclude for github.com/sergi/go-diff as it should have been fixed by sergi/go-diff#136
atiratree
added a commit
to atiratree/api
that referenced
this pull request
Aug 6, 2024
- removed exclude for github.com/sergi/go-diff as it should have been fixed by sergi/go-diff#136
atiratree
added a commit
to atiratree/api
that referenced
this pull request
Aug 15, 2024
- removed exclude for github.com/sergi/go-diff as it should have been fixed by sergi/go-diff#136
atiratree
added a commit
to atiratree/api
that referenced
this pull request
Aug 20, 2024
- removed exclude for github.com/sergi/go-diff as it should have been fixed by sergi/go-diff#136
bertinatto
pushed a commit
to bertinatto/api
that referenced
this pull request
Aug 22, 2024
- removed exclude for github.com/sergi/go-diff as it should have been fixed by sergi/go-diff#136
bertinatto
pushed a commit
to bertinatto/api
that referenced
this pull request
Aug 22, 2024
- removed exclude for github.com/sergi/go-diff as it should have been fixed by sergi/go-diff#136
bertinatto
pushed a commit
to bertinatto/api
that referenced
this pull request
Aug 23, 2024
- removed exclude for github.com/sergi/go-diff as it should have been fixed by sergi/go-diff#136
atiratree
added a commit
to atiratree/api
that referenced
this pull request
Aug 30, 2024
- removed exclude for github.com/sergi/go-diff as it should have been fixed by sergi/go-diff#136
atiratree
added a commit
to atiratree/api
that referenced
this pull request
Sep 11, 2024
- removed exclude for github.com/sergi/go-diff as it should have been fixed by sergi/go-diff#136
atiratree
added a commit
to atiratree/api
that referenced
this pull request
Sep 20, 2024
- removed exclude for github.com/sergi/go-diff as it should have been fixed by sergi/go-diff#136
atiratree
added a commit
to atiratree/api
that referenced
this pull request
Sep 20, 2024
- removed exclude for github.com/sergi/go-diff as it should have been fixed by sergi/go-diff#136
atiratree
added a commit
to atiratree/api
that referenced
this pull request
Sep 24, 2024
- removed exclude for github.com/sergi/go-diff as it should have been fixed by sergi/go-diff#136
atiratree
added a commit
to atiratree/api
that referenced
this pull request
Sep 25, 2024
- removed exclude for github.com/sergi/go-diff as it should have been fixed by sergi/go-diff#136
atiratree
added a commit
to atiratree/api
that referenced
this pull request
Sep 26, 2024
- removed exclude for github.com/sergi/go-diff as it should have been fixed by sergi/go-diff#136
atiratree
added a commit
to atiratree/api
that referenced
this pull request
Sep 27, 2024
- removed exclude for github.com/sergi/go-diff as it should have been fixed by sergi/go-diff#136
atiratree
added a commit
to atiratree/api
that referenced
this pull request
Oct 1, 2024
- removed exclude for github.com/sergi/go-diff as it should have been fixed by sergi/go-diff#136
Open
2 tasks
harbourmaster
pushed a commit
to homeport/dyff
that referenced
this pull request
Feb 22, 2026
Ref: sergi/go-diff#123 Ref: sergi/go-diff#136 Move import into direct import sections, since comment is not required anymore.
HeavyWombat
added a commit
to homeport/dyff
that referenced
this pull request
Feb 22, 2026
Ref: sergi/go-diff#123 Ref: sergi/go-diff#136 Move import into direct import sections, since comment is not required anymore.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
The suggested approach for doing line level diffing is the following set of steps:
ti1, ti2, linesIdx = DiffLinesToChars(t1, t2)diffs = DiffMain(ti1, ti2)DiffCharsToLines(diff, linesIdx)The original implementation in
google/diff-match-patchusesunicode codepoints for storing indices in
ti1andti2joined by an empty string.Current implementation in this repo stores them as integers joined by a
comma. While this implementation makes
ti1andti2more readable, itintroduces bugs when trying to rely on it when doing line level diffing
with
DiffMain. The root cause of the issue is that an integer lineindex might span more than one character/rune, and
DiffMaincan assumethat two different lines having the same index prefix match partially. For
example, indices 123 and 129 will have partial match
12. In thatexample, the diff will show lines 3 and 9 which is not correct. A simple
failing test case demonstrating this issue is available at
TestDiffPartialLineIndex.In this PR I am adjusting the algorithm to use the same approach as in
diff-match-patch by storing each line index as a rune.
While a rune in Golang is a type alias to uint32, not every uint32
can be a valid rune. During string to rune slice conversion invalid runes will
be replaced with
utf.RuneError.The integer to rune generation logic is based on the table in https://en.wikipedia.org/wiki/UTF-8#Encoding
The first 127 lines will work the fastest as they are represented as a
single bytes. Higher numbers are represented as 2-4 bytes.
In addition to that, the range
U+D800 - U+DFFFcontainsinvalid codepoints.
and all codepoints higher or equal to
0xD800are incremented by0xDFFF - 0xD800.The maximum representable integer using this approach is 1'112'060.
This improves on Javascript implementation which currently
bails out when files have more than 65535 lines.