attocode-code-intel 0.1.27__tar.gz → 0.1.28__tar.gz

{attocode_code_intel-0.1.27 → attocode_code_intel-0.1.28}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.5
 Name: attocode-code-intel
-Version: 0.1.27
+Version: 0.1.28
 Summary: Local code search and navigation for coding agents and people
 Project-URL: Homepage, https://github.com/eren23/attocode
 Project-URL: Repository, https://github.com/eren23/attocode
@@ -156,6 +156,8 @@ To use the latest source, install from a clone of this repository: `uv tool inst
 
 ## What your agent gets
 
+Version 0.1.28 summarizes duplicate compact-search metadata diagnostics after evidence fitting. Full lexical-index details remain at `data.ranking.index`, with readiness and coverage hints in `metadata.ranking.index`. Results and source excerpts are unchanged. See the [release notes](https://github.com/eren23/attocode/blob/main/docs/intelligence-0.1.28.md).
+
 Version 0.1.27 fixes a dictionary-size error during symbol inspection when AST hydration changes the cache concurrently. Namespace resolution snapshots cached file membership; later changes appear in the next build. See the [release notes](https://github.com/eren23/attocode/blob/main/docs/intelligence-0.1.27.md).
 
 Version 0.1.26 gives Rust precision lookups a shared five-second budget and fixes cancelled startup cleanup when a child retains inherited pipes. Other languages keep two seconds; longer reloads still return partial evidence.

{attocode_code_intel-0.1.27 → attocode_code_intel-0.1.28}/README.md

@@ -43,6 +43,8 @@ To use the latest source, install from a clone of this repository: `uv tool inst
 
 ## What your agent gets
 
+Version 0.1.28 summarizes duplicate compact-search metadata diagnostics after evidence fitting. Full lexical-index details remain at `data.ranking.index`, with readiness and coverage hints in `metadata.ranking.index`. Results and source excerpts are unchanged. See the [release notes](https://github.com/eren23/attocode/blob/main/docs/intelligence-0.1.28.md).
+
 Version 0.1.27 fixes a dictionary-size error during symbol inspection when AST hydration changes the cache concurrently. Namespace resolution snapshots cached file membership; later changes appear in the next build. See the [release notes](https://github.com/eren23/attocode/blob/main/docs/intelligence-0.1.27.md).
 
 Version 0.1.26 gives Rust precision lookups a shared five-second budget and fixes cancelled startup cleanup when a child retains inherited pipes. Other languages keep two seconds; longer reloads still return partial evidence.

{attocode_code_intel-0.1.27 → attocode_code_intel-0.1.28}/pyproject.toml

@@ -6,7 +6,7 @@ build-backend = "hatchling.build"
 
 [project]
 name = "attocode-code-intel"
-version = "0.1.27"
+version = "0.1.28"
 description = "Local code search and navigation for coding agents and people"
 readme = "README.md"
 requires-python = ">=3.12"

{attocode_code_intel-0.1.27 → attocode_code_intel-0.1.28}/src/attocode_intel/__init__.py

@@ -49,4 +49,4 @@ def __getattr__(name: str) -> Any:
     raise AttributeError(f"module {__name__!r} has no attribute {name!r}")
 
 # The other copy is packages/code-intel/pyproject.toml. A test checks that both match.
-__version__ = "0.1.27"
+__version__ = "0.1.28"

{attocode_code_intel-0.1.27 → attocode_code_intel-0.1.28}/src/attocode_intel/output.py

@@ -207,6 +207,30 @@ def _restore_search_snippets(result, originals, max_tokens, *, upstream_truncate
     return result
 
 
+def _compact_search_ranking(result, max_tokens):
+    """Summarize a duplicate metadata index after all evidence fitting is done.
+
+    The full record remains in data.ranking.index. Preserve readiness/coverage
+    hints for metadata consumers, and retain the original page if no saving fits.
+    """
+    page = json.loads(result.content[0].text)
+    ranking = page['metadata'].get('ranking')
+    if not isinstance(ranking, dict) or ranking != page['data'].get('ranking'):
+        return result
+    index = ranking.get('index')
+    if not isinstance(index, dict):
+        return result
+    summary = {key: index[key] for key in (
+        'status', 'body_index_available', 'body_index_ready', 'body_index_coverage',
+        'discovery_truncated', 'vector_status', 'vector_coverage') if key in index}
+    summary['details_in'] = 'data.ranking.index'
+    ranking['index'] = summary
+    candidate = compact_result(page['metadata'], page['data'])
+    if response_tokens(candidate) <= min(max_tokens, response_tokens(result)):
+        return candidate
+    return result
+
+
 def bounded_search(metadata, data, max_tokens, *, all_results, follow_up=None):
     """Fit search delivery, reserving a progressing retry after final omissions."""
     payload = deepcopy(data)
@@ -246,8 +270,9 @@ def bounded_search(metadata, data, max_tokens, *, all_results, follow_up=None):
             and len(delivered["data"]["results"]) >= delivered["data"]["ranking"]["candidate_pool_count"]):
         delivered["data"].pop("follow_up", None)
         result = compact_result(delivered["metadata"], delivered["data"])
-    return _restore_search_snippets(result, all_results, max_tokens,
-                                    upstream_truncated=metadata.get("truncated", False))
+    restored = _restore_search_snippets(result, all_results, max_tokens,
+                                         upstream_truncated=metadata.get("truncated", False))
+    return _compact_search_ranking(restored, max_tokens)
 
 
 def bounded_compact(metadata: dict, data, max_tokens: int, *, search_results=None) -> types.CallToolResult:

attocode_code_intel-0.1.28/tests/test_search_ranking_summary.py

@@ -0,0 +1,109 @@
+"""Compact metadata need not serialize a complete ranking index twice."""
+import json
+from copy import deepcopy
+
+import pytest
+from attocode_intel import output
+
+
+def page():
+    index = {'status': 'ready', 'body_index_available': True, 'body_index_ready': True,
+             'body_index_coverage': .99, 'discovery_truncated': False, 'vector_status': 'idle',
+             'vector_coverage': 0., 'keyword_indexed_files': 9000, 'body_indexed_files': 8910,
+             'discovery_selected_files': 9000, 'discovery_discovered_files': 9000,
+             'discovery_max_files': 50000, 'warmup_reused_asts': 8000, 'warmup_stage': 'ready',
+             'warmup_seconds': 12.123, 'keyword_seconds': 7.123, 'body_seconds': 5.0,
+             'importance_seconds': .001, 'keyword_index_ready': True}
+    ranking = {'method': 'source_aware_deterministic', 'provider': 'off', 'index': index,
+               'candidate_pool_count': 1, 'returned_count': 1, 'delivered_count': 1}
+    metadata = {'workspace': '/repo', 'source': 'local', 'revision': 'working-tree',
+                'freshness': 'working_tree', 'truncated': False, 'ranking': deepcopy(ranking)}
+    data = {'query': 'quota', 'ranking': ranking, 'results': [
+        {'file_path': 'a.py', 'name': 'quota', 'chunk_type': 'function', 'score': .5,
+         'line': 1, 'end_line': 2, 'snippet': 'def quota():\n    return 3\n',
+         'snippet_start_line': 1, 'snippet_end_line': 2, 'snippet_truncated': False,
+         'matched_terms': ['quota'], 'match_fields': ['source_body']}]}
+    return metadata, data
+
+
+def legacy(monkeypatch, metadata, data, budget):
+    with monkeypatch.context() as patch:
+        if hasattr(output, '_compact_search_ranking'):
+            patch.setattr(output, '_compact_search_ranking', lambda result, _budget: result)
+        return output.bounded_search(metadata, data, budget, all_results=data['results'])
+
+
+def test_duplicate_index_summary_keeps_entire_data(monkeypatch):
+    metadata, data = page()
+    before = legacy(monkeypatch, metadata, data, 8000)
+    result = output.bounded_search(metadata, data, 8000, all_results=data['results'])
+    old, new = (json.loads(r.content[0].text) for r in (before, result))
+    assert new['data'] == old['data'] == data
+    summary = new['metadata']['ranking']['index']
+    assert summary['details_in'] == 'data.ranking.index'
+    assert all(summary[k] == data['ranking']['index'][k] for k in (
+        'status', 'body_index_available', 'body_index_ready', 'body_index_coverage',
+        'discovery_truncated', 'vector_status', 'vector_coverage'))
+    assert {k:v for k,v in new['metadata']['ranking'].items() if k != 'index'} == {
+        k:v for k,v in old['metadata']['ranking'].items() if k != 'index'}
+    assert output.response_tokens(result) < output.response_tokens(before)
+
+
+@pytest.mark.parametrize('budget', [800, 1200, 2000, 8000])
+def test_budget_and_full_data_floor_are_identical(monkeypatch, budget):
+    metadata, data = page()
+    before = legacy(monkeypatch, metadata, data, budget)
+    result = output.bounded_search(metadata, data, budget, all_results=data['results'])
+    old, new = (json.loads(r.content[0].text) for r in (before, result))
+    assert new['data'] == old['data']
+    assert output.response_tokens(result) <= output.response_tokens(before) <= budget
+    assert {k:v for k,v in new['metadata'].items() if k != 'ranking'} == {
+        k:v for k,v in old['metadata'].items() if k != 'ranking'}
+
+
+@pytest.mark.parametrize('kind', ['missing', 'divergent', 'tiny', 'non_mapping'])
+def test_unrelated_or_nonbeneficial_alias_is_byte_identical(monkeypatch, kind):
+    metadata, data = page()
+    if kind == 'missing':
+        metadata.pop('ranking')
+    elif kind == 'divergent':
+        metadata['ranking']['method'] = 'different stage'
+    elif kind == 'tiny':
+        metadata['ranking']['index'] = data['ranking']['index'] = {'status': 'ready'}
+    else:
+        metadata['ranking']['index'] = data['ranking']['index'] = 'ready'
+    expected = legacy(monkeypatch, metadata, data, 8000)
+    actual = output.bounded_search(metadata, data, 8000, all_results=data['results'])
+    assert actual.model_dump() == expected.model_dump()
+
+
+def test_input_objects_are_not_mutated():
+    metadata, data = page()
+    saved = deepcopy((metadata,data))
+    output.bounded_search(metadata,data,8000,all_results=data['results'])
+    assert (metadata,data) == saved
+
+
+@pytest.mark.asyncio
+async def test_real_gateway_preserves_payload_and_readiness(monkeypatch, tmp_path):
+    from attocode_intel.gateway import OperationGateway
+
+    from eval.quality.corpus import ready_search
+
+    (tmp_path/'quota.py').write_text('def quota():\n    return "quota limit"\n')
+    gateway=OperationGateway(str(tmp_path),profile='daily',watch=False)
+    args={'query':'quota limit','top_k':10,'max_tokens':2000}
+    try:
+        await ready_search(gateway, {'query':'neutral readiness','top_k':1,'max_tokens':2000})
+        with monkeypatch.context() as patch:
+            if hasattr(output,'_compact_search_ranking'):
+                patch.setattr(output,'_compact_search_ranking',lambda result,_budget:result)
+            old=await gateway.execute_mcp('semantic_search',args)
+        new=await gateway.execute_mcp('semantic_search',args)
+        a,b=(json.loads(r.content[0].text) for r in (old,new))
+        assert a['data']==b['data']
+        assert b['metadata']['ranking']['index']['details_in']=='data.ranking.index'
+        assert b['metadata']['ranking']['index']['status']==b['data']['ranking']['index']['status']
+        assert output.response_tokens(new)<output.response_tokens(old)<=2000
+    finally:
+        await gateway.close()