Treat error nodes as atoms

Fixes #408

CHANGELOG.md

@@ -1,5 +1,12 @@
 ## 0.38 (unreleased)
 
+### Parsing
+
+Syntax error nodes are now always treated as atoms. This produces
+better results when difftastic doesn't fully support the syntax
+(e.g. PostCSS is a superset of CSS) or when there are actual syntax
+errors.
+
 ## 0.37 (released 14th October 2022)
 
 ### Manual

sample_files/compare.expected

@@ -47,7 +47,7 @@ sample_files/hack_before.php sample_files/hack_after.php
 50babcb945cf3b3ba12d5481a1bf8ccf  -
 
 sample_files/hare_before.ha sample_files/hare_after.ha
-8edc2f3ae2eaeba7bcf328ed00398767  -
+5fee7a686838b20327ed135ae07db0c2  -
 
 sample_files/haskell_before.hs sample_files/haskell_after.hs
 324d1c9b2a04133a75b1975272ffb6ad  -
@@ -179,11 +179,14 @@ sample_files/swift_before.swift sample_files/swift_after.swift
 eeab25a68552f051a6392b5e713fbd29  -
 
 sample_files/syntax_error_before.js sample_files/syntax_error_after.js
-fe636ad27b1aa75e0b153dfe248023bb  -
+5210fe27bf588a4a4a0ed826dfcac76b  -
 
 sample_files/tab_before.c sample_files/tab_after.c
 36ba3231eeba6f0b67a6be9db454de19  -
 
+sample_files/tailwind_before.css sample_files/tailwind_after.css
+a0ba45ac66d0c0b97377821a872d83fc  -
+
 sample_files/text_before.txt sample_files/text_after.txt
 dfc3495b8d5931029b479f0c878a3219  -
 

sample_files/tailwind_after.css

@@ -0,0 +1,6 @@
+select {
+  /* This isn't valid vanilla CSS, and tree-sitter-css produces an ERROR
+     node with children. Ensure the whole ERROR is treated as an
+     atom. */
+  @apply rounded-md bg-hss-dark-gray;
+}

sample_files/tailwind_before.css

@@ -0,0 +1,6 @@
+select {
+  /* This isn't valid vanilla CSS, and tree-sitter-css produces an ERROR
+     node with children. Ensure the whole ERROR is treated as an
+     atom. */
+  @apply rounded-md bg-gray-600;
+}

src/parse/tree_sitter_parser.rs

@@ -1077,15 +1077,15 @@ fn syntax_from_cursor<'a>(
     if node.is_error() {
         let position = nl_pos.from_offsets(node.start_byte(), node.end_byte());
         let content = &src[node.start_byte()..node.end_byte()];
-
         debug!(
             "Tree-sitter syntax error at {:?}: {}",
             position.get(0),
             content
         );
-    }
 
-    if config.atom_nodes.contains(node.kind()) {
+        // Treat error nodes as atoms, even if they have children.
+        atom_from_cursor(arena, src, nl_pos, cursor, highlights)
+    } else if config.atom_nodes.contains(node.kind()) {
         // Treat nodes like string literals as atoms, regardless
         // of whether they have children.
         atom_from_cursor(arena, src, nl_pos, cursor, highlights)