[pretty_format_json] Add compact array feature

- Find non-nested numeric arrays using JSON spec:
https://www.json.org/json-en.html

Author: adithyabsk

pre_commit_hooks/pretty_format_json.py

@@ -2,6 +2,7 @@
 
 import argparse
 import json
+import re
 import sys
 from collections.abc import Mapping
 from collections.abc import Sequence
@@ -14,6 +15,7 @@ def _get_pretty_format(
         ensure_ascii: bool = True,
         sort_keys: bool = True,
         top_keys: Sequence[str] = (),
+        compact_arrays: bool = False,
 ) -> str:
     def pairs_first(pairs: Sequence[tuple[str, str]]) -> Mapping[str, str]:
         before = [pair for pair in pairs if pair[0] in top_keys]
@@ -22,14 +24,58 @@ def pairs_first(pairs: Sequence[tuple[str, str]]) -> Mapping[str, str]:
         if sort_keys:
             after.sort()
         return dict(before + after)
+
     json_pretty = json.dumps(
         json.loads(contents, object_pairs_hook=pairs_first),
         indent=indent,
         ensure_ascii=ensure_ascii,
     )
+
+    if compact_arrays:
+        json_pretty = _compact_arrays(json_pretty)
+
     return f'{json_pretty}\n'
 
 
+def _compact_arrays(json_text: str) -> str:
+    """Convert arrays with simple values to a single line format."""
+    pattern = re.compile(
+        r'''
+        (                             # Capturing group for the entire array
+            \[                        # Opening bracket
+            \s*                       # Optional whitespace
+            (?:                       # Non-capturing group for array elements
+                (?:                   # Non-capturing group for each value type
+                    "[^"]*"           # String: anything in quotes
+                    |
+                    -?                # Optional negative sign
+                    (?:
+                        0|[1-9]\d*    # Integer part: 0 or non-zero digit
+                                      # followed by digits
+                    )
+                    (?:\.\d+)?        # Optional fractional part
+                    (?:[eE][+-]?\d+)? # Optional exponent part
+                    |
+                    true|false        # Boolean
+                    |
+                    null              # Null
+                )
+                (?:\s*,\s*)?          # Optional comma and whitespace
+            )++                       # One or more elements
+            \s*                       # Optional whitespace
+            \]                        # Closing bracket
+        )
+    ''', re.VERBOSE,
+    )
+
+    def compact_match(match: re.Match[str]) -> str:
+        array_content = match.group(0)
+        compact = re.sub(r'\s*\n\s*', ' ', array_content)
+        return compact
+
+    return re.sub(pattern, compact_match, json_text)
+
+
 def _autofix(filename: str, new_contents: str) -> None:
     print(f'Fixing file {filename}')
     with open(filename, 'w', encoding='UTF-8') as f:
@@ -96,6 +142,16 @@ def main(argv: Sequence[str] | None = None) -> int:
         default=[],
         help='Ordered list of keys to keep at the top of JSON hashes',
     )
+    parser.add_argument(
+        '--compact-arrays',
+        action='store_true',
+        dest='compact_arrays',
+        default=False,
+        help=(
+            'Format simple arrays on a single line for more '
+            'compact representation'
+        ),
+    )
     parser.add_argument('filenames', nargs='*', help='Filenames to fix')
     args = parser.parse_args(argv)
 
@@ -109,6 +165,7 @@ def main(argv: Sequence[str] | None = None) -> int:
             pretty_contents = _get_pretty_format(
                 contents, args.indent, ensure_ascii=not args.no_ensure_ascii,
                 sort_keys=not args.no_sort_keys, top_keys=args.top_keys,
+                compact_arrays=args.compact_arrays,
             )
         except ValueError:
             print(

tests/pretty_format_json_test.py

@@ -155,3 +155,96 @@ def test_diffing_output(capsys):
     assert actual_retval == expected_retval
     assert actual_out == expected_out
     assert actual_err == ''
+
+
+def test_compact_arrays_main(tmpdir):
+    # TODO: Intentionally don't address round trip bug caused by
+    # using `json.loads(json.dumps(data))`. This will need to be
+    # resolved separately.
+    srcfile = tmpdir.join('to_be_compacted.json')
+    srcfile.write(
+        '{\n'
+        '  "simple_array": [\n'
+        '    1,\n'
+        '    2,\n'
+        '    3\n'
+        '  ],\n'
+        '  "string_array": [\n'
+        '    "a",\n'
+        '    "b",\n'
+        '    "c"\n'
+        '  ],\n'
+        '  "mixed_array": [\n'
+        '    1,\n'
+        '    "string",\n'
+        '    true,\n'
+        '    null\n'
+        '  ],\n'
+        '  "nested_objects": [\n'
+        '    {\n'
+        '      "a": 1\n'
+        '    },\n'
+        '    {\n'
+        '      "b": 2\n'
+        '    }\n'
+        '  ]\n'
+        '}',
+    )
+
+    ret = main(['--compact-arrays', '--autofix', str(srcfile)])
+    assert ret == 1
+
+    with open(str(srcfile), encoding='UTF-8') as f:
+        contents = f.read()
+
+    # Simple arrays should be compacted
+    assert '"simple_array": [ 1, 2, 3 ]' in contents
+    assert '"string_array": [ "a", "b", "c" ]' in contents
+    assert '"mixed_array": [ 1, "string", true, null ]' in contents
+
+    # Nested array objects should remain expanded
+    assert '  "nested_objects": [\n' in contents
+    assert '      "a": 1\n' in contents
+
+
+def test_compact_arrays_diff_output(tmpdir, capsys):
+    srcfile = tmpdir.join('expanded_arrays.json')
+    srcfile.write(
+        '{\n'
+        '  "array": [\n'
+        '    1,\n'
+        '    2,\n'
+        '    3\n'
+        '  ]\n'
+        '}',
+    )
+
+    ret = main(['--compact-arrays', str(srcfile)])
+    assert ret == 1
+
+    out, _ = capsys.readouterr()
+    assert '+  "array": [ 1, 2, 3 ]' in out
+
+    # Validate diff output
+    assert '-    1,' in out
+    assert '-    2,' in out
+    assert '-    3' in out
+    assert '-  "array": [' in out
+    assert '-  ]' in out
+
+
+def test_compact_arrays_disabled(tmpdir):
+    """Test that compacting arrays does not impact default formatting."""
+    srcfile = tmpdir.join('already_compact.json')
+    srcfile.write('{\n  "array": [ 1, 2, 3 ]\n}')
+
+    ret = main(['--autofix', str(srcfile)])
+    assert ret == 1
+
+    with open(str(srcfile), encoding='UTF-8') as f:
+        contents = f.read()
+
+    assert '"array": [\n' in contents
+    assert '    1,' in contents
+    assert '    2,' in contents
+    assert '    3\n  ]' in contents