Feature/us 09 repo web analyzer - 리포지토리 및 웹 이력서 분석 모듈 작성

ai/.env.example

@@ -18,3 +18,18 @@ LLM_API_KEY=
 LLM_BASE_URL=https://factchat-cloud.mindlogic.ai/v1/gateway
 LLM_PRO_MODEL=gemini-3.1-pro-preview
 LLM_PRO_TEMPERATURE=0.2
+
+# Github 관련
+CORE_INTERNAL_BASE_URL=http://localhost:38010
+CORE_INTERNAL_API_KEY=
+CORE_INTERNAL_TIMEOUT_SEC=10
+
+GITHUB_FALLBACK_TOKEN=
+GITHUB_API_BASE_URL=https://api.github.com
+REPO_MAX_SOURCE_FILES=8
+REPO_MAX_SOURCE_FILE_BYTES=50000
+REPO_FETCH_TIMEOUT_SEC=30
+
+# 웹 이력서 관련
+WEB_FETCH_TIMEOUT_SEC=20
+WEB_MAX_HTML_BYTES=2000000

ai/CLAUDE.md

@@ -88,11 +88,13 @@ ai/
 
 본 서버는 RabbitMQ **consumer**로 작동.
 
-| Queue | Bind |
-|-------|------|
-| `q.ai.resume` | `ai.request.resume.*` |
-| `q.ai.repo` | `ai.request.repo.*` |
-| `q.ai.session` | `ai.request.session.*` |
+| Queue | Routing key | 상태 |
+|-------|-------------|------|
+| `ai.analyze.resume` | `analyze.resume` | 본 구현 (PDF → MD) |
+| `ai.analyze.repository` | `analyze.repository` | 본 구현 (GitHub README + tree + 소스 sampling) |
+| `ai.analyze.web` | `analyze.web` | 본 구현 (URL → trafilatura) |
+| `ai.generate.questions` | `generate.questions` | 큐만, 코드 미구현 |
+| `ai.generate.followup` | `generate.followup` | 큐만, 코드 미구현 |
 
 콜백 발행: `ai.callback.{type}` 익스체인지.
 상세 envelope/스키마/재시도: [`/docs/messaging.md`](../docs/messaging.md).

ai/pyproject.toml

@@ -16,6 +16,7 @@ dependencies = [
     "boto3>=1.42.77",
     "aiofiles>=24.1.0",
     "pypdf>=5.1.0",
+    "trafilatura>=2.0.0",
     "langchain>=1.2.13",
     "langchain-core>=1.2.22",
     "langchain-community>=0.4.1",

ai/src/ai_server/analyzer/repository_analyzer.py

@@ -0,0 +1,128 @@
+from __future__ import annotations
+
+from dataclasses import dataclass
+
+import structlog
+
+from ai_server.analyzer.sources.github_repo import (
+    GitHubRepoSourceExtractor,
+    RepositoryFetchError,
+)
+from ai_server.chain.document_analysis_chain import DocumentAnalyzer
+from ai_server.core.client import CoreClient, CoreTokenError
+from ai_server.storage.base import ObjectStorage
+
+log = structlog.get_logger(__name__)
+
+
+class RepositoryAnalyzeError(Exception):
+    def __init__(self, *, code: str, message: str, retriable: bool) -> None:
+        super().__init__(message)
+        self.code = code
+        self.message = message
+        self.retriable = retriable
+
+
+@dataclass(frozen=True)
+class RepositoryAnalysisResult:
+    summary: str
+    tech_stack: list[str]
+    document_path: str
+
+
+# 코어서버에서 토큰 받아오고 레포 가져온다
+# 이후 LLM 분석하고 마크다운 저장 
+class RepositoryAnalyzer:
+    def __init__(
+        self,
+        *,
+        extractor: GitHubRepoSourceExtractor,
+        core_client: CoreClient,
+        chain: DocumentAnalyzer,
+        storage: ObjectStorage,
+        analyzed_key_template: str,
+    ) -> None:
+        self._extractor = extractor
+        self._core_client = core_client
+        self._chain = chain
+        self._storage = storage
+        self._analyzed_key_template = analyzed_key_template
+
+    async def analyze(
+        self,
+        *,
+        repository_id: int,
+        repo_full_name: str,
+        default_branch: str = "main",
+        user_id: int | None,
+    ) -> RepositoryAnalysisResult:
+        if user_id is None:
+            raise RepositoryAnalyzeError(
+                code="MISSING_USER_ID",
+                message="envelope.context.userId 없이는 GitHub token을 가져올 수 없음",
+                retriable=False,
+            )
+
+        log.info(
+            "repository.token.fetch",
+            user_id=user_id,
+            repository_id=repository_id,
+        )
+        try:
+            access_token = await self._core_client.fetch_github_token(user_id)
+        except CoreTokenError as err:
+            raise RepositoryAnalyzeError(
+                code=err.code,
+                message=err.message,
+                retriable=err.retriable,
+            ) from err
+
+        log.info(
+            "repository.extract.start",
+            repository_id=repository_id,
+            repo_full_name=repo_full_name,
+            default_branch=default_branch,
+        )
+        try:
+            source = await self._extractor.extract(
+                repo_full_name,
+                access_token=access_token,
+            )
+        except RepositoryFetchError as err:
+            raise RepositoryAnalyzeError(
+                code=err.code,
+                message=err.message,
+                retriable=err.retriable,
+            ) from err
+
+        if not source.text.strip():
+            raise RepositoryAnalyzeError(
+                code="EMPTY_REPO_CONTENT",
+                message="레포에서 추출된 텍스트가 비어 있음",
+                retriable=False,
+            )
+
+        log.info(
+            "repository.llm.start",
+            repository_id=repository_id,
+            text_chars=len(source.text),
+        )
+        analysis = await self._chain.analyze(
+            text=source.text,
+            source_type=source.source_type,
+        )
+
+        out_key = self._analyzed_key_template.format(repository_id=repository_id)
+        await self._storage.put_text(out_key, analysis.markdown)
+        log.info(
+            "repository.markdown.saved",
+            repository_id=repository_id,
+            key=out_key,
+            md_chars=len(analysis.markdown),
+        )
+
+        return RepositoryAnalysisResult(
+            summary=analysis.summary,
+            tech_stack=list(analysis.tech_stack),
+            document_path=out_key,
+        )

ai/src/ai_server/analyzer/sources/__init__.py

@@ -3,11 +3,20 @@
     SourceExtractor,
     SourceType,
 )
+from ai_server.analyzer.sources.github_repo import (
+    GitHubRepoSourceExtractor,
+    RepositoryFetchError,
+)
 from ai_server.analyzer.sources.pdf import PdfSourceExtractor
+from ai_server.analyzer.sources.web import WebFetchError, WebSourceExtractor
 
 __all__ = [
     "ExtractedSource",
     "SourceExtractor",
     "SourceType",
     "PdfSourceExtractor",
+    "GitHubRepoSourceExtractor",
+    "RepositoryFetchError",
+    "WebSourceExtractor",
+    "WebFetchError",
 ]