fix bugs

Author: yingyuankai

Makefile

@@ -33,8 +33,8 @@ run-prod: ## Run the production server
 
 # ─── Quality ─────────────────────────────────────────────
 
-test: ## Run all tests
-	poetry run pytest tests/ -v
+test: ## Run all tests (excludes adapter integration tests that require Docker)
+	poetry run pytest tests/ -v -m "not (elasticsearch or opensearch or solr or meilisearch)"
 
 test-unit: ## Run unit tests only
 	poetry run pytest tests/unit/ -v -m "not integration"

src/opensift/adapters/meilisearch/adapter.py

@@ -21,7 +21,7 @@
 import logging
 import time
 from datetime import UTC, datetime
-from typing import Any
+from typing import Any, cast
 
 import httpx
 
@@ -171,7 +171,7 @@ async def fetch_document(self, doc_id: str) -> dict[str, Any]:
             if resp.status_code == 404:
                 raise DocumentNotFoundError(f"Document '{doc_id}' not found.")
             resp.raise_for_status()
-            return resp.json()
+            return cast(dict[str, Any], resp.json())
         except DocumentNotFoundError:
             raise
         except httpx.HTTPError as e:

src/opensift/adapters/wikipedia/adapter.py

@@ -78,7 +78,7 @@ async def initialize(self) -> None:
             ConfigurationError: If the ``wikipedia-api`` package is not installed.
         """
         try:
-            import wikipediaapi
+            import wikipediaapi  # type: ignore[import-untyped]
         except ImportError as e:
             raise ConfigurationError(
                 "The 'wikipedia-api' package is required for the Wikipedia adapter. "

src/opensift/api/v1/endpoints/search.py

@@ -12,6 +12,7 @@
 
 import json
 import logging
+from collections.abc import AsyncIterator
 
 from fastapi import APIRouter, Depends, HTTPException
 from fastapi.responses import StreamingResponse
@@ -103,7 +104,7 @@ async def search(
         ) from e
 
 
-async def _sse_generator(engine: OpenSiftEngine, request: SearchRequest):
+async def _sse_generator(engine: OpenSiftEngine, request: SearchRequest) -> AsyncIterator[str]:
     """Async generator that yields SSE-formatted lines.
 
     Each event follows the SSE protocol::

src/opensift/client/client.py

@@ -16,11 +16,13 @@
 import asyncio
 import json
 import logging
-from collections.abc import AsyncIterator, Iterator
-from typing import Any
+from collections.abc import AsyncIterator, Coroutine, Iterator
+from typing import Any, TypeVar, cast
 
 import httpx
 
+_T = TypeVar("_T")
+
 logger = logging.getLogger(__name__)
 
 # ═══════════════════════════════════════════════════════════════════════════════
@@ -95,7 +97,7 @@ async def health(self) -> dict[str, Any]:
         """
         resp = await self._client.get("/v1/health")
         resp.raise_for_status()
-        return resp.json()
+        return cast(dict[str, Any], resp.json())
 
     async def adapter_health(self) -> dict[str, Any]:
         """Check adapter health.
@@ -105,7 +107,7 @@ async def adapter_health(self) -> dict[str, Any]:
         """
         resp = await self._client.get("/v1/health/adapters")
         resp.raise_for_status()
-        return resp.json()
+        return cast(dict[str, Any], resp.json())
 
     # ── Plan (standalone) ──
 
@@ -136,7 +138,7 @@ async def plan(
         }
         resp = await self._client.post("/v1/plan", json=payload)
         resp.raise_for_status()
-        return resp.json()
+        return cast(dict[str, Any], resp.json())
 
     # ── Search (complete mode) ──
 
@@ -180,7 +182,7 @@ async def search(
         }
         resp = await self._client.post("/v1/search", json=payload)
         resp.raise_for_status()
-        return resp.json()
+        return cast(dict[str, Any], resp.json())
 
     # ── Search (streaming mode) ──
 
@@ -270,7 +272,7 @@ async def batch_search(
 
         resp = await self._client.post("/v1/search/batch", json=payload)
         resp.raise_for_status()
-        return resp.json()
+        return cast(dict[str, Any], resp.json())
 
 
 # ═══════════════════════════════════════════════════════════════════════════════
@@ -306,7 +308,7 @@ def __init__(
         self._timeout = timeout
         self._httpx_kwargs = httpx_kwargs
 
-    def _run(self, coro):
+    def _run(self, coro: Coroutine[Any, Any, _T]) -> _T:
         """Run an async coroutine synchronously."""
         try:
             loop = asyncio.get_running_loop()

src/opensift/config/settings.py

@@ -74,12 +74,12 @@ def _parse_hosts(cls, v: Any) -> list[str]:
             try:
                 parsed = json.loads(v)
                 if isinstance(parsed, list):
-                    return parsed
+                    return [str(h) for h in parsed]
             except (json.JSONDecodeError, TypeError):
                 pass
             # Single host as plain string
             return [v] if v else []
-        return v
+        return list(v)
 
 
 class SearchSettings(BaseModel):
@@ -141,7 +141,7 @@ def from_yaml(cls, path: str | Path) -> Settings:
         Returns:
             Populated Settings instance.
         """
-        import yaml
+        import yaml  # type: ignore[import-untyped]
 
         config_path = Path(path)
         if not config_path.exists():

src/opensift/core/classifier.py

@@ -18,6 +18,7 @@
 
 from opensift.models.assessment import (
     AssessmentType,
+    CriterionAssessment,
     ResultClassification,
     ScoredResult,
     ValidationResult,
@@ -105,7 +106,7 @@ def classify_batch(
         return results
 
     @staticmethod
-    def _classify_single(assessments: list) -> ResultClassification:
+    def _classify_single(assessments: list[CriterionAssessment]) -> ResultClassification:
         """Classification for a single-criterion scenario.
 
         support → perfect
@@ -125,7 +126,9 @@ def _classify_single(assessments: list) -> ResultClassification:
             return ResultClassification.REJECT
 
     @staticmethod
-    def _classify_multiple(assessments: list, criteria_map: dict) -> ResultClassification:
+    def _classify_multiple(
+        assessments: list[CriterionAssessment], criteria_map: dict[str, Criterion]
+    ) -> ResultClassification:
         """Classification for multi-criteria scenario.
 
         ALL support → perfect
@@ -151,7 +154,7 @@ def _classify_multiple(assessments: list, criteria_map: dict) -> ResultClassific
         return ResultClassification.REJECT
 
     @staticmethod
-    def _calculate_weighted_score(assessments: list, criteria_map: dict) -> float:
+    def _calculate_weighted_score(assessments: list[CriterionAssessment], criteria_map: dict[str, Criterion]) -> float:
         """Calculate weighted score based on assessments and criteria weights.
 
         Score mapping:

src/opensift/core/engine.py

@@ -20,7 +20,7 @@
 import time
 import uuid
 from collections.abc import AsyncIterator
-from typing import TYPE_CHECKING
+from typing import TYPE_CHECKING, Any
 
 from opensift.adapters.base.registry import AdapterRegistry
 from opensift.core.classifier import ResultClassifier
@@ -444,13 +444,12 @@ async def batch_search(self, request: BatchSearchRequest) -> BatchSearchResponse
             for query in request.queries
         ]
 
-        responses = await asyncio.gather(*tasks, return_exceptions=True)
+        raw_responses: list[SearchResponse | BaseException] = await asyncio.gather(*tasks, return_exceptions=True)
 
         results: list[SearchResponse] = []
-        for i, resp in enumerate(responses):
-            if isinstance(resp, Exception):
+        for i, resp in enumerate(raw_responses):
+            if isinstance(resp, BaseException):
                 logger.warning("Batch query %d failed: %s", i, resp)
-                # Create an error response stub
                 results.append(
                     SearchResponse(
                         request_id=f"req_batch_{i}_error",
@@ -584,25 +583,29 @@ async def _execute_searches(
             logger.warning("No search adapter available, returning empty results")
             return []
 
-        tasks: list[asyncio.Task] = []
+        tasks: list[asyncio.Task[Any]] = []
         task_meta: list[tuple[str, bool]] = []  # (adapter_name, use_paper_path)
 
         for adapter in adapters:
             use_paper_path = hasattr(adapter, "search_papers") and callable(adapter.search_papers)
             for query in search_queries:
                 if use_paper_path:
-                    tasks.append(adapter.search_papers(query, request.options))  # type: ignore[union-attr]
+                    tasks.append(
+                        asyncio.ensure_future(
+                            adapter.search_papers(query, request.options)  # type: ignore[attr-defined]
+                        )
+                    )
                 else:
-                    tasks.append(adapter.search_and_normalize(query, request.options))
+                    tasks.append(asyncio.ensure_future(adapter.search_and_normalize(query, request.options)))
                 task_meta.append((adapter.name, use_paper_path))
 
-        raw_results = await asyncio.gather(*tasks, return_exceptions=True)
+        raw_results: list[Any] = await asyncio.gather(*tasks, return_exceptions=True)
 
         seen_titles: set[str] = set()
         items: list[ResultItem] = []
 
         for i, result in enumerate(raw_results):
-            if isinstance(result, Exception):
+            if isinstance(result, BaseException):
                 adapter_name, _ = task_meta[i]
                 logger.warning("Search query failed on adapter '%s': %s", adapter_name, result)
                 continue

src/opensift/core/llm/client.py

@@ -9,7 +9,7 @@
 import json
 import logging
 import re
-from typing import TYPE_CHECKING, Any
+from typing import TYPE_CHECKING, Any, cast
 
 from openai import AsyncOpenAI
 
@@ -199,7 +199,7 @@ async def chat_json(
                 content = self._strip_code_fences(content)
 
                 try:
-                    return json.loads(content)
+                    return cast(dict[str, Any], json.loads(content))
                 except json.JSONDecodeError:
                     repaired = self._repair_json(content)
                     if repaired is not None:
@@ -369,7 +369,7 @@ def _repair_json(text: str) -> dict[str, Any] | None:
 
         # Try parsing after basic fixes
         try:
-            return json.loads(text)
+            return cast(dict[str, Any], json.loads(text))
         except json.JSONDecodeError:
             pass
 
@@ -381,7 +381,7 @@ def _repair_json(text: str) -> dict[str, Any] | None:
         text = re.sub(r"(})\s*(\")", r"\1,\2", text)
 
         try:
-            return json.loads(text)
+            return cast(dict[str, Any], json.loads(text))
         except json.JSONDecodeError:
             pass
 
@@ -410,7 +410,7 @@ def _escape_newlines_in_strings(s: str) -> str:
         text = _escape_newlines_in_strings(text)
 
         try:
-            return json.loads(text)
+            return cast(dict[str, Any], json.loads(text))
         except json.JSONDecodeError:
             return None
 

src/opensift/core/planner/planner.py

@@ -12,7 +12,7 @@
 
 import logging
 import time
-from typing import TYPE_CHECKING
+from typing import TYPE_CHECKING, Any
 
 from opensift.core.llm.client import LLMClient, LLMError
 from opensift.core.llm.prompts import CRITERIA_SYSTEM_PROMPT, CRITERIA_USER_PROMPT
@@ -122,7 +122,7 @@ async def _generate_with_llm(self, query: str) -> CriteriaResult:
 
         return self._parse_criteria_response(raw)
 
-    def _parse_criteria_response(self, raw: dict) -> CriteriaResult:
+    def _parse_criteria_response(self, raw: dict[str, Any]) -> CriteriaResult:
         """Parse and validate the raw LLM response into a CriteriaResult.
 
         Args: