Repair elided empty array slots like [1,,2] (0.15.0)

[1,,2] silently mangled to [[1],2] and [1,,,,2] raised. The empty slot
made parse_array terminate early (emitting [1]), then the root
comma-sequence wrap turned [1] and 2 into NDJSON [[1],2] — data
corruption. Both are upstream-shared (jsonrepair v3.14.0 produces the
same mangle / raise).

Drop the elided slots instead: after a separator comma, parse_array now
skips any further commas via a new skip_elided_commas helper, so
[1,,2] -> [1,2], [1,,,,2] -> [1,2], [1, ,2] -> [1,2] (whitespace and
comments between the commas are skipped too). Dropping — rather than
inserting null — matches the existing leading-comma repair ([,1] ->
[1]) and dirty-json. Divergence from upstream, commented at the site.

Arrays only. An elided comma in an object ({"a":1,,"b":2}) still raises
rather than guessing a missing key: it is a clean error with no data to
mangle, so we hold upstream's behavior. Pinned in the spec.

Differential vs main over a 42-input comma grid: every change is an
array with elided commas going from mangle/raise to a correct drop;
objects and all non-elided shapes (leading/trailing commas, ellipsis,
missing commas, truncated nested containers) are unchanged. Bench flat
across all four scenarios. rspec (204 examples, 100% line+branch),
rubocop, rbs validate, steep all green.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

Author: sashazykov

CHANGELOG.md

@@ -1,5 +1,22 @@
 # Changes
 
+### 2026-06-17 (0.15.0)
+
+* Repair elided empty array slots: `[1,,2]` → `[1,2]`, `[1,,,,2]` →
+  `[1,2]`, `[1, ,2]` → `[1,2]`. After a separator comma, any further
+  commas mark empty slots with no value between them and are dropped —
+  matching how a leading comma is dropped (`[,1]` → `[1]`), and
+  consistent with `dirty-json`. Divergence from upstream
+  [jsonrepair](https://github.com/josdejong/jsonrepair), which as of
+  v3.14.0 mangles `[1,,2]` into `[[1],2]` (its root comma-sequence wrap
+  misfires) and raises on `[1,,,,2]`; commented at the site. Arrays
+  only: an elided comma in an object (`{"a":1,,"b":2}`) still raises
+  rather than guessing a missing key — there is no data to mangle, so
+  the clean error is kept (and pinned in the spec). Differential vs
+  main over a 42-input comma grid: every change is an array with elided
+  commas going from mangle/raise to a correct drop, with objects and
+  all non-elided shapes unchanged. Benchmarks flat.
+
 ### 2026-06-17 (0.14.0)
 
 * Repair a leading `+` on numbers, like in JSON5: `+1.23` → `1.23`,

Gemfile.lock

@@ -1,7 +1,7 @@
 PATH
   remote: .
   specs:
-    json-repair (0.14.0)
+    json-repair (0.15.0)
 
 GEM
   remote: https://rubygems.org/

lib/json/repair/version.rb

@@ -2,6 +2,6 @@
 
 module JSON
   module Repair
-    VERSION = '0.14.0'
+    VERSION = '0.15.0'
   end
 end

lib/json/repairer.rb

@@ -434,6 +434,22 @@ def skip_ellipsis
       end
     end
 
+    # Repair elided empty array slots like [1,,2] or [1,,,,2]. Called right
+    # after a separator comma: skip every further comma (each marks a slot
+    # with no value between it and the previous one), so [1,,2] -> [1,2] and
+    # [1,,,,2] -> [1,2]. Dropping the slot — rather than inserting null —
+    # matches how a leading comma is dropped ([,1] -> [1]). Divergence from
+    # upstream, which mangles [1,,2] into [[1],2] via its root comma-sequence
+    # wrap and raises on [1,,,,2] (as of v3.14.0).
+    def skip_elided_commas
+      loop do
+        parse_whitespace_and_skip_comments
+        break unless @json[@index] == COMMA
+
+        @index += 1
+      end
+    end
+
     # Parse a string enclosed by double quotes "...". Can contain escaped quotes
     # Repair strings enclosed in single quotes or special quotes
     # Repair an escaped string
@@ -857,6 +873,8 @@ def parse_array
             processed_comma = parse_character(COMMA)
             # repair missing comma
             @output = insert_before_last_whitespace(@output, ',') unless processed_comma
+            # repair: drop elided empty slots like [1,,2] -> [1,2]
+            skip_elided_commas if processed_comma
           end
 
           skip_ellipsis

sig/json/repairer.rbs

@@ -66,6 +66,9 @@ module JSON
     # or a similar construct in objects.
     def skip_ellipsis: () -> void
 
+    # Drop elided empty array slots like [1,,2] -> [1,2].
+    def skip_elided_commas: () -> void
+
     # Parse a string enclosed by double quotes "...". Can contain escaped quotes
     # Repair strings enclosed in single quotes or special quotes
     # Repair an escaped string

spec/json_spec.rb

@@ -504,6 +504,25 @@
         expect(JSON.repair('[ , 1,2,3]')).to eq('[1,2,3]')
       end
 
+      it 'drops elided empty slots in an array' do
+        expect(JSON.repair('[1,,2]')).to eq('[1,2]')
+        expect(JSON.repair('[1,,,,2]')).to eq('[1,2]')
+        expect(JSON.repair('[1, ,2]')).to eq('[1,2]')
+        expect(JSON.repair('[1,/* gap */,2]')).to eq('[1,2]')
+        expect(JSON.repair('["a",,"b"]')).to eq('["a","b"]')
+        # works inside a nested array too
+        expect(JSON.repair('[[1,,2],3]')).to eq('[[1,2],3]')
+        # trailing empty slots collapse with the trailing-comma repair
+        expect(JSON.repair('[1,,]')).to eq('[1]')
+      end
+
+      it 'still raises on elided commas in an object (no silent mangle)' do
+        # arrays drop the slot (above); objects keep raising rather than
+        # guessing a key — a comma with no key:value pair is ambiguous and
+        # there is no data to mangle, so we hold upstream's clean error
+        expect { JSON.repair('{"a":1,,"b":2}') }.to raise_error(JSON::JSONRepairError)
+      end
+
       it 'strips a leading comma from an object' do
         expect(JSON.repair('{,"message": "hi"}')).to eq('{"message":"hi"}')
         expect(JSON.repair('{/* a */,/* b */"message": "hi"}')).to eq('{"message":"hi"}')