barrett/diffs.nvim
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

feat(highlight): wire highlights.context into treesitter pipeline (#151)
Some checks failed
luarocks / quality (push) Has been cancelled
Details
luarocks / publish (push) Has been cancelled
Details

Browse source 
## Problem

`highlights.context.enabled` and `highlights.context.lines` were
defined, validated, and range-checked but never read during
highlighting. Hunks inside incomplete constructs (e.g., a table literal
or function body whose opening is beyond the hunk's own context lines)
parsed incorrectly because treesitter had no surrounding code.

## Solution

`compute_hunk_context` in `init.lua` reads the working tree file using
the hunk's `@@ +start,count @@` line numbers to collect up to `lines`
(default 25) surrounding code lines in each direction. Files are read
once via `io.open` and cached across hunks in the same file.
`highlight_treesitter` in `highlight.lua` accepts an optional context
parameter that prepends/appends context lines to the parse string and
offsets capture rows by the prefix count, so extmarks only land on
actual hunk lines. Wired through `highlight_hunk` for the two
code-language treesitter calls (not headers, not `highlight_text`, not
vim syntax).

Closes #148.
This commit is contained in:
Barrett Ruth 2026-03-05 11:14:31 -05:00 • committed by GitHub
parent 29e624d9f0
commit e7d56e3bbe
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
5 changed files with 534 additions and 17 deletions
Download patch file Download diff file Expand all files Collapse all files

32
doc/diffs.nvim.txt
View file

@ -225,16 +225,20 @@ Configuration is done via `vim.g.diffs`. Set this before the plugin loads:
*diffs.ContextConfig*
Context config fields: ~
{enabled} (boolean, default: true)
Read lines from disk before and after each hunk
to provide surrounding syntax context. Improves
accuracy at hunk boundaries where incomplete
constructs (e.g., a function definition with no
body) would otherwise confuse the parser.
Read surrounding code from the working tree
file and feed it into the treesitter string
parser. Uses the hunk's `@@ +start,count @@`
line numbers to read lines before and after
the hunk from disk. Improves syntax accuracy
when the hunk is inside an incomplete construct
(e.g., a table literal or function body whose
opening is not visible in the hunk's own
context lines).
{lines} (integer, default: 25)
Number of context lines to read in each
direction. Lines are read with early exit —
cost scales with this value, not file size.
Max context lines to read in each direction.
Files are read once per parse and cached across
hunks in the same file.
*diffs.PrioritiesConfig*
Priorities config fields: ~
@ -695,10 +699,14 @@ KNOWN LIMITATIONS *diffs-limitations*
Incomplete Syntax Context ~
*diffs-syntax-context*
Treesitter parses each diff hunk in isolation. Context lines within the hunk
(lines with a ` ` prefix) provide syntactic context for the parser. In rare
cases, hunks that start or end mid-expression may produce imperfect highlights
due to treesitter error recovery.
Treesitter parses each diff hunk in isolation. When `highlights.context` is
enabled (the default), surrounding code is read from the working tree file
and fed into the parser to improve accuracy at hunk boundaries. This helps
when a hunk is inside a table, function body, or loop whose opening is
beyond the hunk's own context lines. Requires `repo_root` and
`file_new_start` to be available on the hunk (true for standard unified
diffs). In rare cases, hunks that start or end mid-expression may still
produce imperfect highlights due to treesitter error recovery.
Syntax Highlighting Flash ~
*diffs-flash*