Files
loogle-scripts/services/loogle-mcp/app/context/store.py
T

407 lines
15 KiB
Python

# -*- coding: utf-8 -*-
"""Filesystem-backed project context store."""
import json
import os
import re
import secrets
import threading
from datetime import datetime
from typing import Optional
from . import gitea_link
from ..knowledge import gitea
SLUG_RE = re.compile(r"^[a-z0-9][a-z0-9-]{0,62}$")
# Contesti condivisi: stessa struttura dei progetti personali, sotto un namespace comune.
# Tutta la famiglia li legge e li aggiorna; archiviazione e ritorno a personale solo
# per proprietario o admin.
SHARED_NS = "_shared"
SHARED_WITH_FAMILY = "family"
_SHARED_META_KEYS = ("shared", "shared_with", "owner", "shared_at", "shared_by", "updated_by")
class ConflictError(ValueError):
"""Il contesto è stato modificato da qualcun altro dopo l'ultima lettura."""
# I tool MCP girano in parallelo nel threadpool: serializza le scritture per progetto
_project_locks: dict[tuple[str, str], threading.Lock] = {}
_project_locks_guard = threading.Lock()
def _project_lock(username: str, project_id: str) -> threading.Lock:
with _project_locks_guard:
return _project_locks.setdefault((username, project_id), threading.Lock())
def _context_root() -> str:
return os.environ.get("MCP_CONTEXT_ROOT", "/data/context")
def _user_root(username: str) -> str:
path = os.path.join(_context_root(), username, "projects")
os.makedirs(path, exist_ok=True)
return path
def _shared_root() -> str:
path = os.path.join(_context_root(), SHARED_NS, "projects")
os.makedirs(path, exist_ok=True)
return path
def _project_dir(username: str, project_id: str) -> str:
if not SLUG_RE.match(project_id):
raise ValueError("ID progetto non valido")
return os.path.join(_user_root(username), project_id)
def _id_taken(project_id: str, username: str, shared: bool) -> bool:
"""Un ID condiviso non deve coincidere con progetti personali di nessun utente
(altrimenti per quell'utente il personale nasconderebbe il condiviso)."""
root = _context_root()
if os.path.exists(os.path.join(_shared_root(), project_id)):
return True
users = [u for u in os.listdir(root) if u != SHARED_NS] if shared else [username]
return any(os.path.exists(os.path.join(root, u, "projects", project_id)) for u in users)
def _read_meta(meta_path: str) -> dict:
with open(meta_path, encoding="utf-8") as f:
return json.load(f)
def _is_member(meta: dict) -> bool:
return meta.get("shared_with") == SHARED_WITH_FAMILY
def _resolve_project(username: str, project_id: str) -> tuple[str, str]:
"""(namespace, cartella) del progetto: prima quello personale, poi quello condiviso."""
personal = _project_dir(username, project_id)
if os.path.isfile(os.path.join(personal, "meta.json")):
return username, personal
shared = os.path.join(_shared_root(), project_id)
meta_path = os.path.join(shared, "meta.json")
if os.path.isfile(meta_path) and _is_member(_read_meta(meta_path)):
return SHARED_NS, shared
raise FileNotFoundError("Progetto non trovato")
def _load_meta(username: str, project_id: str) -> tuple[str, dict]:
_, proj_dir = _resolve_project(username, project_id)
meta_path = os.path.join(proj_dir, "meta.json")
return meta_path, _read_meta(meta_path)
def _require_owner_or_admin(username: str, meta: dict, is_admin: bool, action: str) -> None:
if meta.get("shared") and not is_admin and meta.get("owner") != username:
raise PermissionError(
f"Solo il proprietario ({meta.get('owner')}) o un admin può {action} un contesto condiviso"
)
def _save_meta(meta_path: str, meta: dict) -> None:
meta["updated_at"] = datetime.now().astimezone().isoformat(timespec="seconds")
with open(meta_path, "w", encoding="utf-8") as f:
json.dump(meta, f, ensure_ascii=False, indent=2)
def _apply_gitea_link(
username: str,
meta: dict,
gitea_repo: Optional[str],
*,
verify: bool = True,
seed_from_gitea: bool = False,
) -> dict:
repo = gitea_link.normalize_gitea_repo(gitea_repo)
if verify and repo:
if not gitea.is_configured(username):
raise RuntimeError(
"Gitea non configurato per questo utente — impossibile collegare il repository"
)
gitea_link.verify_repo_access(username, repo)
if repo:
meta["gitea_repo"] = repo
else:
meta.pop("gitea_repo", None)
if seed_from_gitea and repo:
_, proj_dir = _resolve_project(username, meta["id"])
ctx_path = os.path.join(proj_dir, "context.md")
context_md = open(ctx_path, encoding="utf-8").read() if os.path.isfile(ctx_path) else ""
new_md, changed = gitea_link.seed_context_from_readme(username, repo, context_md)
if changed:
with open(ctx_path, "w", encoding="utf-8") as f:
f.write(new_md)
meta["gitea_seeded_at"] = datetime.now().astimezone().isoformat(timespec="seconds")
return meta
def list_projects(username: str, include_archived: bool = False) -> list:
"""Progetti personali dell'utente seguiti dai contesti condivisi (campo shared=true)."""
projects = []
for root, shared in ((_user_root(username), False), (_shared_root(), True)):
for name in sorted(os.listdir(root)):
meta_path = os.path.join(root, name, "meta.json")
if not os.path.isfile(meta_path):
continue
meta = _read_meta(meta_path)
if shared and not _is_member(meta):
continue
if meta.get("archived") and not include_archived:
continue
projects.append(meta)
return projects
def create_project(
username: str,
title: str,
tags: Optional[list[str]] = None,
gitea_repo: Optional[str] = None,
seed_from_gitea: bool = False,
shared: bool = False,
) -> dict:
slug_base = re.sub(r"[^a-z0-9]+", "-", title.lower()).strip("-") or "progetto"
project_id = slug_base[:40].strip("-")
# ID univoco sia tra i progetti personali sia tra i condivisi (la risoluzione li unisce)
while _id_taken(project_id, username, shared):
project_id = f"{slug_base[:32].strip('-')}-{secrets.token_hex(2)}"
proj_dir = os.path.join(_shared_root() if shared else _user_root(username), project_id)
os.makedirs(os.path.join(proj_dir, "sessions"), exist_ok=True)
os.makedirs(os.path.join(proj_dir, "artifacts"), exist_ok=True)
now = datetime.now().astimezone().isoformat(timespec="seconds")
meta = {
"id": project_id,
"title": title.strip(),
"tags": tags or [],
"status": "active",
"archived": False,
"created_at": now,
"updated_at": now,
}
if shared:
meta.update(_shared_fields(username, now))
repo = gitea_link.normalize_gitea_repo(gitea_repo)
if repo:
meta["gitea_repo"] = repo
with open(os.path.join(proj_dir, "meta.json"), "w", encoding="utf-8") as f:
json.dump(meta, f, ensure_ascii=False, indent=2)
with open(os.path.join(proj_dir, "context.md"), "w", encoding="utf-8") as f:
f.write(f"# {title.strip()}\n\n")
if repo:
meta = _apply_gitea_link(
username,
meta,
repo,
verify=True,
seed_from_gitea=seed_from_gitea,
)
_save_meta(os.path.join(proj_dir, "meta.json"), meta)
return meta
def _shared_fields(username: str, now: str) -> dict:
return {
"shared": True,
"shared_with": SHARED_WITH_FAMILY,
"owner": username,
"shared_by": username,
"shared_at": now,
}
def share_project(username: str, project_id: str, shared: bool = True, is_admin: bool = False) -> dict:
"""Sposta un progetto tra lo spazio personale del proprietario e quello condiviso.
La storia (context.md, sessions/, artifacts/) si sposta con il progetto.
"""
ns, proj_dir = _resolve_project(username, project_id)
meta = _read_meta(os.path.join(proj_dir, "meta.json"))
if shared:
if ns == SHARED_NS:
return meta
target = os.path.join(_shared_root(), project_id)
others = [
u for u in os.listdir(_context_root())
if u not in (SHARED_NS, username) and os.path.exists(os.path.join(_context_root(), u, "projects", project_id))
]
if others:
raise ValueError(
f"L'ID '{project_id}' è usato anche da un progetto personale di {', '.join(others)}: "
"crea un contesto condiviso nuovo (create_project shared=true) e copiaci il contenuto"
)
else:
if ns != SHARED_NS:
return meta
_require_owner_or_admin(username, meta, is_admin, "rendere personale")
target = os.path.join(_user_root(meta["owner"]), project_id)
if os.path.exists(target):
raise ValueError(f"Esiste già un progetto '{project_id}' nella destinazione: rinomina prima")
with _project_lock(ns, project_id):
os.rename(proj_dir, target)
meta_path = os.path.join(target, "meta.json")
meta = _read_meta(meta_path)
if shared:
meta.update(_shared_fields(username, datetime.now().astimezone().isoformat(timespec="seconds")))
else:
for key in _SHARED_META_KEYS:
meta.pop(key, None)
_save_meta(meta_path, meta)
return meta
def link_project_repo(
username: str,
project_id: str,
gitea_repo: Optional[str] = None,
seed_from_gitea: bool = False,
) -> dict:
ns, _ = _resolve_project(username, project_id)
with _project_lock(ns, project_id):
meta_path, meta = _load_meta(username, project_id)
meta = _apply_gitea_link(
username,
meta,
gitea_repo,
verify=bool(gitea_link.normalize_gitea_repo(gitea_repo)),
seed_from_gitea=seed_from_gitea,
)
_save_meta(meta_path, meta)
return meta
def get_project_context(
username: str,
project_id: str,
session_limit: int = 5,
include_gitea: bool = True,
) -> dict:
_, proj_dir = _resolve_project(username, project_id)
meta = _read_meta(os.path.join(proj_dir, "meta.json"))
ctx_path = os.path.join(proj_dir, "context.md")
context_md = open(ctx_path, encoding="utf-8").read() if os.path.isfile(ctx_path) else ""
sessions_dir = os.path.join(proj_dir, "sessions")
sessions = []
if os.path.isdir(sessions_dir):
files = sorted(os.listdir(sessions_dir), reverse=True)[:session_limit]
for fname in files:
path = os.path.join(sessions_dir, fname)
if os.path.isfile(path):
sessions.append({"name": fname, "content": open(path, encoding="utf-8").read()})
result = {"meta": meta, "context_md": context_md, "recent_sessions": sessions}
gitea_repo = meta.get("gitea_repo")
if include_gitea and gitea_repo:
result["gitea"] = gitea_link.project_enrichment(username, gitea_repo)
return result
def save_context(
username: str,
project_id: str,
content: str,
mode: str = "append",
if_revision: Optional[int] = None,
) -> dict:
"""if_revision: meta.revision letto prima di modificare; se nel frattempo qualcuno
ha salvato, ConflictError invece di sovrascrivere (utile nei contesti condivisi)."""
ns, proj_dir = _resolve_project(username, project_id)
with _project_lock(ns, project_id):
return _save_context_locked(username, proj_dir, content, mode, if_revision)
def _save_context_locked(
username: str, proj_dir: str, content: str, mode: str, if_revision: Optional[int]
) -> dict:
meta_path = os.path.join(proj_dir, "meta.json")
if not os.path.isfile(meta_path):
raise FileNotFoundError("Progetto non trovato")
meta = _read_meta(meta_path)
if if_revision is not None and int(meta.get("revision", 0)) != int(if_revision):
raise ConflictError(
f"Contesto modificato (revisione {meta.get('revision', 0)}) il {meta.get('updated_at')}"
+ (f" da {meta['updated_by']}" if meta.get("updated_by") else "")
+ ": rileggilo con get_project_context e riapplica la modifica"
)
ctx_path = os.path.join(proj_dir, "context.md")
if mode == "replace":
text = content
else:
existing = open(ctx_path, encoding="utf-8").read() if os.path.isfile(ctx_path) else ""
text = existing.rstrip() + "\n\n" + content.strip() + "\n"
with open(ctx_path, "w", encoding="utf-8") as f:
f.write(text)
meta["revision"] = int(meta.get("revision", 0)) + 1
if meta.get("shared"):
meta["updated_by"] = username
_save_meta(meta_path, meta)
# Nei condivisi il nome della sessione dice chi ha scritto
stamp = datetime.now().strftime("%Y%m%d-%H%M%S") + (f"-{username}" if meta.get("shared") else "")
snapshot = stamp + ".md"
n = 1
while os.path.exists(os.path.join(proj_dir, "sessions", snapshot)):
snapshot = f"{stamp}-{n}.md"
n += 1
with open(os.path.join(proj_dir, "sessions", snapshot), "w", encoding="utf-8") as f:
f.write(content)
return meta
def archive_project(
username: str, project_id: str, archived: bool = True, is_admin: bool = False
) -> dict:
ns, _ = _resolve_project(username, project_id)
with _project_lock(ns, project_id):
meta_path, meta = _load_meta(username, project_id)
_require_owner_or_admin(username, meta, is_admin, "archiviare")
meta["archived"] = archived
meta["status"] = "archived" if archived else "active"
_save_meta(meta_path, meta)
return meta
def list_resources(username: str) -> list:
resources = []
for meta in list_projects(username, include_archived=False):
ns = SHARED_NS if meta.get("shared") else username
desc = f"Contesto {'condiviso con la famiglia' if meta.get('shared') else 'progetto'} {meta['id']}"
if meta.get("gitea_repo"):
desc += f" (Gitea: {meta['gitea_repo']})"
resources.append(
{
"uri": f"loogle://context/{ns}/{meta['id']}",
"name": meta["title"],
"description": desc,
"mimeType": "text/markdown",
}
)
return resources
def read_resource(username: str, uri: str) -> dict:
for ns in (username, SHARED_NS):
prefix = f"loogle://context/{ns}/"
if uri.startswith(prefix):
project_id = uri[len(prefix):]
break
else:
raise FileNotFoundError("Risorsa non trovata")
data = get_project_context(username, project_id, session_limit=0, include_gitea=False)
text = data["context_md"]
gitea_repo = data["meta"].get("gitea_repo")
if gitea_repo:
text = f"<!-- gitea_repo: {gitea_repo} -->\n\n" + text
return {
"uri": uri,
"mimeType": "text/markdown",
"text": text,
}