add the code
Browse files- code_retriever.py +70 -0
code_retriever.py
CHANGED
|
@@ -0,0 +1,70 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
from typing import List
|
| 2 |
+
|
| 3 |
+
from github_search import search_github
|
| 4 |
+
from stack_search import search_stackoverflow
|
| 5 |
+
from schemas import CodeTaskType, CodeXRequest, RetrievedEvidence
|
| 6 |
+
from config import settings
|
| 7 |
+
|
| 8 |
+
|
| 9 |
+
def normalize_text(text: str) -> str:
|
| 10 |
+
return (text or "").strip().lower()
|
| 11 |
+
|
| 12 |
+
|
| 13 |
+
def deduplicate_evidence(evidence_list: List[RetrievedEvidence]) -> List[RetrievedEvidence]:
|
| 14 |
+
seen = set()
|
| 15 |
+
unique_items: List[RetrievedEvidence] = []
|
| 16 |
+
|
| 17 |
+
for item in evidence_list:
|
| 18 |
+
key = (
|
| 19 |
+
normalize_text(item.title),
|
| 20 |
+
normalize_text(item.url or ""),
|
| 21 |
+
item.source_type.value,
|
| 22 |
+
)
|
| 23 |
+
|
| 24 |
+
if key in seen:
|
| 25 |
+
continue
|
| 26 |
+
|
| 27 |
+
seen.add(key)
|
| 28 |
+
unique_items.append(item)
|
| 29 |
+
|
| 30 |
+
return unique_items
|
| 31 |
+
|
| 32 |
+
|
| 33 |
+
def sort_evidence(evidence_list: List[RetrievedEvidence]) -> List[RetrievedEvidence]:
|
| 34 |
+
return sorted(
|
| 35 |
+
evidence_list,
|
| 36 |
+
key=lambda item: (item.score is not None, item.score if item.score is not None else -1),
|
| 37 |
+
reverse=True,
|
| 38 |
+
)
|
| 39 |
+
|
| 40 |
+
|
| 41 |
+
def retrieve_code_evidence(task_type: CodeTaskType, request: CodeXRequest) -> List[RetrievedEvidence]:
|
| 42 |
+
if task_type != CodeTaskType.FIX:
|
| 43 |
+
return []
|
| 44 |
+
|
| 45 |
+
collected: List[RetrievedEvidence] = []
|
| 46 |
+
|
| 47 |
+
if settings.ENABLE_STACK_SEARCH:
|
| 48 |
+
stack_results = search_stackoverflow(
|
| 49 |
+
message=request.message,
|
| 50 |
+
error_message=request.error_message,
|
| 51 |
+
language=request.language,
|
| 52 |
+
framework=request.framework,
|
| 53 |
+
max_results=settings.MAX_STACK_RESULTS,
|
| 54 |
+
)
|
| 55 |
+
collected.extend(stack_results)
|
| 56 |
+
|
| 57 |
+
if settings.ENABLE_GITHUB_SEARCH:
|
| 58 |
+
github_results = search_github(
|
| 59 |
+
message=request.message,
|
| 60 |
+
error_message=request.error_message,
|
| 61 |
+
language=request.language,
|
| 62 |
+
framework=request.framework,
|
| 63 |
+
max_results=settings.MAX_GITHUB_RESULTS,
|
| 64 |
+
)
|
| 65 |
+
collected.extend(github_results)
|
| 66 |
+
|
| 67 |
+
unique_items = deduplicate_evidence(collected)
|
| 68 |
+
ranked_items = sort_evidence(unique_items)
|
| 69 |
+
|
| 70 |
+
return ranked_items[: settings.MAX_RETRIEVED_ITEMS]
|