Backup automatico script del 2026-08-23 12:04

services/loogle-mcp/app/context/gitea_link.py

@@ -0,0 +1,154 @@
+# -*- coding: utf-8 -*-
+"""Collegamento progetti MCP ↔ repository Gitea."""
+
+from __future__ import annotations
+
+import logging
+from typing import Optional
+
+from ..knowledge import gitea
+
+LOGGER = logging.getLogger("loogle_mcp.context.gitea_link")
+
+README_CANDIDATES = ("README.md", "readme.md", "Readme.md", "README.MD")
+DOCS_DIR = "docs"
+MAX_README_CHARS = 12_000
+MAX_DOC_FILES = 25
+
+
+def normalize_gitea_repo(repo: Optional[str]) -> Optional[str]:
+    if repo is None:
+        return None
+    cleaned = repo.strip()
+    if not cleaned:
+        return None
+    owner, name = gitea.parse_repo(cleaned)
+    return f"{owner}/{name}"
+
+
+def verify_repo_access(username: str, repo: str) -> None:
+    owner, name = gitea.parse_repo(repo)
+    gitea._request("GET", f"/repos/{owner}/{name}", username=username)
+
+
+def fetch_readme(username: str, repo: str) -> Optional[dict]:
+    for path in README_CANDIDATES:
+        try:
+            data = gitea.get_file(repo, path, username=username)
+        except FileNotFoundError:
+            continue
+        except Exception as exc:
+            LOGGER.warning("README %s/%s non leggibile: %s", repo, path, exc)
+            continue
+        if data.get("type") != "file":
+            continue
+        content = (data.get("content") or "").strip()
+        if not content:
+            continue
+        truncated = len(content) > MAX_README_CHARS
+        return {
+            "path": path,
+            "sha": data.get("sha"),
+            "html_url": data.get("html_url"),
+            "content": content[:MAX_README_CHARS],
+            "truncated": truncated,
+        }
+    return None
+
+
+def fetch_docs_index(username: str, repo: str) -> list[dict]:
+    try:
+        data = gitea.get_file(repo, DOCS_DIR, username=username)
+    except FileNotFoundError:
+        return []
+    except Exception as exc:
+        LOGGER.warning("Directory docs/ non leggibile per %s: %s", repo, exc)
+        return []
+
+    if data.get("type") != "dir":
+        return []
+
+    entries = []
+    for item in data.get("entries") or []:
+        if item.get("type") != "file":
+            continue
+        path = item.get("path") or ""
+        if not path.lower().endswith((".md", ".txt", ".rst")):
+            continue
+        entries.append(
+            {
+                "path": path,
+                "size": item.get("size"),
+            }
+        )
+        if len(entries) >= MAX_DOC_FILES:
+            break
+    return entries
+
+
+def project_enrichment(username: str, gitea_repo: str) -> dict:
+    repo = normalize_gitea_repo(gitea_repo)
+    if not repo:
+        return {"linked": False}
+
+    base = {
+        "linked": True,
+        "repo": repo,
+        "html_url": f"{gitea.public_base_url()}/{repo}",
+    }
+
+    if not gitea.is_configured(username):
+        return {
+            **base,
+            "available": False,
+            "error": "Gitea non configurato per questo utente",
+        }
+
+    try:
+        verify_repo_access(username, repo)
+        readme = fetch_readme(username, repo)
+        docs = fetch_docs_index(username, repo)
+        open_issues = None
+        try:
+            issues = gitea.list_issues(repo, state="open", limit=1, username=username)
+            open_issues = issues.get("count", 0)
+        except Exception:
+            pass
+        return {
+            **base,
+            "available": True,
+            "readme": readme,
+            "docs_files": docs,
+            "open_issues": open_issues,
+        }
+    except Exception as exc:
+        return {
+            **base,
+            "available": False,
+            "error": str(exc),
+        }
+
+
+def readme_seed_block(repo: str, readme: dict) -> str:
+    path = readme.get("path") or "README.md"
+    content = readme.get("content") or ""
+    truncated_note = "\n\n*(README troncato — usa get_file per il testo completo)*" if readme.get("truncated") else ""
+    return (
+        f"<!-- seed:gitea {repo} {path} -->\n\n"
+        f"## Sorgente Gitea: `{repo}`\n\n"
+        f"Contenuto iniziale da `{path}`.\n\n"
+        f"{content.rstrip()}{truncated_note}\n"
+    )
+
+
+def seed_context_from_readme(username: str, repo: str, context_md: str) -> tuple[str, bool]:
+    """Importa README nel context se non già presente un seed Gitea."""
+    if "<!-- seed:gitea " in context_md:
+        return context_md, False
+    readme = fetch_readme(username, repo)
+    if not readme:
+        return context_md, False
+    block = readme_seed_block(repo, readme)
+    if context_md.strip():
+        return context_md.rstrip() + "\n\n---\n\n" + block, True
+    return block, True