fdf
This commit is contained in:
@@ -210,6 +210,135 @@ def _resolve_archive_id(
|
||||
return ""
|
||||
|
||||
|
||||
def _fetch_openlibrary_edition_metadata(
|
||||
session: requests.Session,
|
||||
edition_id: str,
|
||||
) -> Dict[str, Any]:
|
||||
if not edition_id:
|
||||
return {}
|
||||
|
||||
try:
|
||||
resp = session.get(
|
||||
f"https://openlibrary.org/books/{edition_id}.json",
|
||||
timeout=6,
|
||||
)
|
||||
resp.raise_for_status()
|
||||
data = resp.json() or {}
|
||||
except Exception:
|
||||
return {}
|
||||
|
||||
if not isinstance(data, dict):
|
||||
return {}
|
||||
|
||||
identifiers = data.get("identifiers")
|
||||
if not isinstance(identifiers, dict):
|
||||
identifiers = {}
|
||||
|
||||
def _first_clean(value: Any) -> str:
|
||||
raw = _first_str(value)
|
||||
return str(raw or "").strip()
|
||||
|
||||
isbn_10 = _first_clean(identifiers.get("isbn_10"))
|
||||
isbn_13 = _first_clean(identifiers.get("isbn_13"))
|
||||
archive_id = str(data.get("ocaid") or "").strip()
|
||||
if not archive_id:
|
||||
archive_id = _first_clean(identifiers.get("internet_archive"))
|
||||
|
||||
out: Dict[str, Any] = {
|
||||
"openlibrary_id": str(edition_id).strip(),
|
||||
"openlibrary": str(edition_id).strip(),
|
||||
}
|
||||
if isbn_10:
|
||||
out["isbn_10"] = isbn_10
|
||||
if isbn_13:
|
||||
out["isbn_13"] = isbn_13
|
||||
if archive_id:
|
||||
out["archive_id"] = archive_id
|
||||
return out
|
||||
|
||||
|
||||
def _select_preferred_isbns(values: Any) -> Tuple[str, str]:
|
||||
items: List[Any]
|
||||
if isinstance(values, list):
|
||||
items = values
|
||||
elif values in (None, ""):
|
||||
items = []
|
||||
else:
|
||||
items = [values]
|
||||
|
||||
isbn_10 = ""
|
||||
isbn_13 = ""
|
||||
for raw in items:
|
||||
token = re.sub(r"[^0-9Xx]", "", str(raw or "")).upper().strip()
|
||||
if not token:
|
||||
continue
|
||||
if len(token) == 13 and not isbn_13:
|
||||
isbn_13 = token
|
||||
elif len(token) == 10 and not isbn_10:
|
||||
isbn_10 = token
|
||||
return isbn_10, isbn_13
|
||||
|
||||
|
||||
def _build_pipeline_progress_callback(
|
||||
progress: Any,
|
||||
title: str,
|
||||
) -> Callable[[str, int, Optional[int], str], None]:
|
||||
transfer_label = str(title or "book").strip() or "book"
|
||||
state = {"active": False, "finished": False}
|
||||
|
||||
def _ensure_started(total: Optional[int]) -> None:
|
||||
if state["active"]:
|
||||
return
|
||||
try:
|
||||
progress.begin_transfer(label=transfer_label, total=total)
|
||||
state["active"] = True
|
||||
state["finished"] = False
|
||||
except Exception:
|
||||
pass
|
||||
|
||||
def _finish() -> None:
|
||||
if not state["active"] or state["finished"]:
|
||||
return
|
||||
try:
|
||||
progress.finish_transfer(label=transfer_label)
|
||||
except Exception:
|
||||
pass
|
||||
state["finished"] = True
|
||||
state["active"] = False
|
||||
|
||||
def _callback(kind: str, completed: int, total: Optional[int], label: str) -> None:
|
||||
text = str(label or kind or "download").strip() or "download"
|
||||
try:
|
||||
progress.set_status(f"openlibrary: {text}")
|
||||
except Exception:
|
||||
pass
|
||||
|
||||
if kind == "step":
|
||||
if text != "download pages":
|
||||
_finish()
|
||||
return
|
||||
|
||||
if kind in {"pages", "bytes"}:
|
||||
_ensure_started(total)
|
||||
try:
|
||||
progress.update_transfer(
|
||||
label=transfer_label,
|
||||
completed=int(completed) if completed is not None else None,
|
||||
total=int(total) if total is not None else None,
|
||||
)
|
||||
except Exception:
|
||||
pass
|
||||
if total is not None:
|
||||
try:
|
||||
if int(completed) >= int(total):
|
||||
_finish()
|
||||
except Exception:
|
||||
pass
|
||||
|
||||
setattr(_callback, "_finish_transfer", _finish)
|
||||
return _callback
|
||||
|
||||
|
||||
def _archive_id_from_url(url: str) -> str:
|
||||
"""Best-effort extraction of an Archive.org item identifier from a URL."""
|
||||
|
||||
@@ -1082,6 +1211,12 @@ class OpenLibrary(Provider):
|
||||
|
||||
meta = result.full_metadata or {}
|
||||
edition_id = str(meta.get("openlibrary_id") or "").strip()
|
||||
edition_meta = _fetch_openlibrary_edition_metadata(self._session, edition_id)
|
||||
if edition_meta and isinstance(meta, dict):
|
||||
for key, value in edition_meta.items():
|
||||
if value and not meta.get(key):
|
||||
meta[key] = value
|
||||
result.full_metadata = meta
|
||||
|
||||
# Accept direct Archive.org URLs too (details/borrow/download) even when no OL edition id is known.
|
||||
archive_id = str(meta.get("archive_id") or "").strip()
|
||||
@@ -1097,7 +1232,9 @@ class OpenLibrary(Provider):
|
||||
archive_id = _first_str(ia_candidates) or ""
|
||||
|
||||
if not archive_id and edition_id:
|
||||
archive_id = _resolve_archive_id(self._session, edition_id, ia_candidates)
|
||||
archive_id = str(edition_meta.get("archive_id") or "").strip()
|
||||
if not archive_id:
|
||||
archive_id = _resolve_archive_id(self._session, edition_id, ia_candidates)
|
||||
|
||||
if not archive_id:
|
||||
# Try to extract identifier from the SearchResult path (URL).
|
||||
@@ -1114,17 +1251,49 @@ class OpenLibrary(Provider):
|
||||
try:
|
||||
archive_meta = fetch_archive_item_metadata(archive_id)
|
||||
tags = archive_item_metadata_to_tags(archive_id, archive_meta)
|
||||
if edition_id:
|
||||
tags.append(f"openlibrary:{edition_id}")
|
||||
if tags:
|
||||
try:
|
||||
result.tag.update(tags)
|
||||
except Exception:
|
||||
# Fallback for callers that pass plain dicts.
|
||||
pass
|
||||
|
||||
isbn_10 = str(meta.get("isbn_10") or edition_meta.get("isbn_10") or "").strip()
|
||||
isbn_13 = str(meta.get("isbn_13") or edition_meta.get("isbn_13") or "").strip()
|
||||
if not isbn_10 and not isbn_13:
|
||||
isbn_10, isbn_13 = _select_preferred_isbns(archive_meta.get("isbn"))
|
||||
|
||||
if isinstance(meta, dict):
|
||||
meta["archive_id"] = archive_id
|
||||
if archive_meta:
|
||||
meta["archive_metadata"] = archive_meta
|
||||
if edition_id:
|
||||
meta.setdefault("openlibrary_id", edition_id)
|
||||
meta.setdefault("openlibrary", edition_id)
|
||||
if isbn_10:
|
||||
meta.setdefault("isbn_10", isbn_10)
|
||||
if isbn_13:
|
||||
meta.setdefault("isbn_13", isbn_13)
|
||||
if not meta.get("isbn"):
|
||||
meta["isbn"] = isbn_13 or isbn_10
|
||||
result.full_metadata = meta
|
||||
|
||||
extra_identifier_tags: List[str] = []
|
||||
if edition_id:
|
||||
extra_identifier_tags.append(f"openlibrary:{edition_id}")
|
||||
if isbn_13:
|
||||
extra_identifier_tags.append(f"isbn_13:{isbn_13}")
|
||||
extra_identifier_tags.append(f"isbn:{isbn_13}")
|
||||
elif isbn_10:
|
||||
extra_identifier_tags.append(f"isbn_10:{isbn_10}")
|
||||
extra_identifier_tags.append(f"isbn:{isbn_10}")
|
||||
if extra_identifier_tags:
|
||||
try:
|
||||
result.tag.update(extra_identifier_tags)
|
||||
except Exception:
|
||||
pass
|
||||
except Exception:
|
||||
# Never block downloads on metadata fetch.
|
||||
pass
|
||||
@@ -1133,6 +1302,13 @@ class OpenLibrary(Provider):
|
||||
if not safe_title or "http" in safe_title.lower():
|
||||
safe_title = sanitize_filename(archive_id) or "archive"
|
||||
|
||||
internal_progress_finish = None
|
||||
if progress_callback is None and isinstance(self.config, dict):
|
||||
pipeline_progress = self.config.get("_pipeline_progress")
|
||||
if pipeline_progress is not None:
|
||||
progress_callback = _build_pipeline_progress_callback(pipeline_progress, safe_title)
|
||||
internal_progress_finish = getattr(progress_callback, "_finish_transfer", None)
|
||||
|
||||
# 1) Direct download if available.
|
||||
try:
|
||||
can_direct, pdf_url = self._archive_check_direct_download(archive_id)
|
||||
@@ -1318,6 +1494,12 @@ class OpenLibrary(Provider):
|
||||
except Exception as exc:
|
||||
log(f"[openlibrary] Borrow workflow error: {exc}", file=sys.stderr)
|
||||
return None
|
||||
finally:
|
||||
if callable(internal_progress_finish):
|
||||
try:
|
||||
internal_progress_finish()
|
||||
except Exception:
|
||||
pass
|
||||
|
||||
def validate(self) -> bool:
|
||||
return True
|
||||
|
||||
Reference in New Issue
Block a user