DeepTutor: Lifelong Personalized Tutoring. https://deeptutor.info/.
Python ai-agentsai-tutorclawbotcli-tooldeepresearchinteractive-learninglearningmulti-agent-systemsrag
"""Turn generated workspace paths into durable, authorized presentation URLs."""
import copy
import re
def normalize_note_resources(markdown, blocks=None, *, workspace_id="", published=None):
cache = dict(published or {})
def resolve(url):
if not isinstance(url, str) or not url.startswith("outputs/"):
return url
if url not in cache:
from deeptutor.services.workspace import get_content_workspace_service
service = get_content_workspace_service()
binding = (
service.binding_by_id(workspace_id) if workspace_id else service.current_binding()
)
item = service.publish(binding, [{"path": url}])[0]
cache[url] = item.url
return cache[url]
def text(value):
# Leave fenced code examples verbatim; only Markdown destinations are links.
pieces = re.split(r"(^```[^\n]*\n[\s\S]*?^```[^\n]*$)", value, flags=re.M)
for i in range(0, len(pieces), 2):
pieces[i] = re.sub(
r"(\]\()(outputs/[^\s()]+)(\))", lambda m: m[1] + resolve(m[2]) + m[3], pieces[i]
)
return "".join(pieces)
def visit(items):
if not isinstance(items, list):
return
for block in items:
if not isinstance(block, dict):
continue
props = block.get("props", {})
if not isinstance(props, dict):
continue
for key in ("url", "sourceUrl"):
if key in props:
props[key] = resolve(props[key])
if isinstance(props.get("markdown"), str):
props["markdown"] = text(props["markdown"])
visit(block.get("children", []))
normalized = copy.deepcopy(blocks)
if normalized:
visit(normalized)
return text(markdown), normalized
async def saved_note_resources(markdown, blocks=None):
"""Recover legacy links from this account/scope's recorded attachments only."""
import json
from deeptutor.services.session import get_session_store
serialized = markdown + "\n" + json.dumps(blocks or [])
paths = set(re.findall(r"outputs/[^\s\\\"()]+", serialized))
sessions = {
match[1]
for path in paths
if (match := re.match(r"outputs/[^/]+/(unified_[a-zA-Z0-9_]+)/", path))
}
published = {}
for session_id in sessions:
session = await get_session_store().get_session_with_messages(session_id)
for message in (session or {}).get("messages", []):
for item in message.get("attachments") or []:
path = item.get("relative_path")
url = item.get("url", "")
if path in paths and url.startswith("/files/workspace-items/"):
published[path] = url
return published