Big changes

This commit is contained in:
Justin Visser 2026-02-21 17:33:05 +01:00
parent 4240ad38e2
commit e3f0d63fec
99 changed files with 8804 additions and 1731 deletions

View file

@ -2,6 +2,7 @@ from __future__ import annotations
from app.services.domains.publications.counts import (
count_for_user,
count_favorite_for_user,
count_latest_for_user,
count_unread_for_user,
)
@ -12,7 +13,9 @@ from app.services.domains.publications.listing import (
list_unread_for_user,
)
from app.services.domains.publications.enrichment import (
hydrate_pdf_enrichment_state,
schedule_missing_pdf_enrichment_for_user,
schedule_retry_pdf_enrichment_for_row,
)
from app.services.domains.publications.modes import (
MODE_ALL,
@ -30,6 +33,14 @@ from app.services.domains.publications.queries import (
from app.services.domains.publications.read_state import (
mark_all_unread_as_read_for_user,
mark_selected_as_read_for_user,
set_publication_favorite_for_user,
)
from app.services.domains.publications.pdf_queue import (
count_pdf_queue_items,
enqueue_all_missing_pdf_jobs,
enqueue_retry_pdf_job_for_publication_id,
list_pdf_queue_page,
list_pdf_queue_items,
)
from app.services.domains.publications.types import PublicationListItem, UnreadPublicationItem
@ -49,10 +60,19 @@ __all__ = [
"list_unread_for_user",
"list_new_for_latest_run_for_user",
"retry_pdf_for_user",
"hydrate_pdf_enrichment_state",
"schedule_retry_pdf_enrichment_for_row",
"list_pdf_queue_items",
"list_pdf_queue_page",
"count_pdf_queue_items",
"enqueue_all_missing_pdf_jobs",
"enqueue_retry_pdf_job_for_publication_id",
"schedule_missing_pdf_enrichment_for_user",
"count_for_user",
"count_favorite_for_user",
"count_unread_for_user",
"count_latest_for_user",
"mark_all_unread_as_read_for_user",
"mark_selected_as_read_for_user",
"set_publication_favorite_for_user",
]

View file

@ -19,6 +19,7 @@ async def count_for_user(
user_id: int,
mode: str = MODE_ALL,
scholar_profile_id: int | None = None,
favorite_only: bool = False,
) -> int:
resolved_mode = resolve_publication_view_mode(mode)
latest_run_id = await get_latest_completed_run_id_for_user(db_session, user_id=user_id)
@ -30,6 +31,8 @@ async def count_for_user(
)
if scholar_profile_id is not None:
stmt = stmt.where(ScholarProfile.id == scholar_profile_id)
if favorite_only:
stmt = stmt.where(ScholarPublication.is_favorite.is_(True))
if resolved_mode == MODE_UNREAD:
stmt = stmt.where(ScholarPublication.is_read.is_(False))
if resolved_mode == MODE_LATEST:
@ -45,12 +48,14 @@ async def count_unread_for_user(
*,
user_id: int,
scholar_profile_id: int | None = None,
favorite_only: bool = False,
) -> int:
return await count_for_user(
db_session,
user_id=user_id,
mode=MODE_UNREAD,
scholar_profile_id=scholar_profile_id,
favorite_only=favorite_only,
)
@ -59,10 +64,27 @@ async def count_latest_for_user(
*,
user_id: int,
scholar_profile_id: int | None = None,
favorite_only: bool = False,
) -> int:
return await count_for_user(
db_session,
user_id=user_id,
mode=MODE_LATEST,
scholar_profile_id=scholar_profile_id,
favorite_only=favorite_only,
)
async def count_favorite_for_user(
db_session: AsyncSession,
*,
user_id: int,
scholar_profile_id: int | None = None,
) -> int:
return await count_for_user(
db_session,
user_id=user_id,
mode=MODE_ALL,
scholar_profile_id=scholar_profile_id,
favorite_only=True,
)

View file

@ -1,135 +1,73 @@
from __future__ import annotations
import asyncio
import logging
import time
from app.db.session import get_session_factory
from app.services.domains.publications.listing import (
missing_pdf_items,
resolve_and_persist_oa_metadata,
from sqlalchemy.ext.asyncio import AsyncSession
from app.services.domains.publications.pdf_queue import (
enqueue_missing_pdf_jobs,
enqueue_retry_pdf_job,
overlay_pdf_job_state,
)
from app.services.domains.publications.types import PublicationListItem
from app.settings import settings
logger = logging.getLogger(__name__)
_enrichment_lock = asyncio.Lock()
_inflight_publications: set[tuple[int, int]] = set()
_recent_attempt_seconds: dict[tuple[int, int], float] = {}
_scheduled_tasks: set[asyncio.Task[None]] = set()
def _cooldown_seconds() -> float:
return max(float(settings.unpaywall_retry_cooldown_seconds), 1.0)
def _prune_recent_attempts(now_seconds: float, *, cooldown_seconds: float) -> None:
expiry = cooldown_seconds * 3
stale_keys = [
key for key, attempted_seconds in _recent_attempt_seconds.items()
if now_seconds - attempted_seconds >= expiry
]
for key in stale_keys:
_recent_attempt_seconds.pop(key, None)
async def _claim_items(
*,
user_id: int,
items: list[PublicationListItem],
max_items: int,
) -> list[PublicationListItem]:
candidates = missing_pdf_items(items, limit=max_items)
if not candidates:
return []
now_seconds = time.monotonic()
cooldown_seconds = _cooldown_seconds()
claimed: list[PublicationListItem] = []
async with _enrichment_lock:
_prune_recent_attempts(now_seconds, cooldown_seconds=cooldown_seconds)
for item in candidates:
key = (user_id, item.publication_id)
attempted_seconds = _recent_attempt_seconds.get(key)
if key in _inflight_publications:
continue
if attempted_seconds is not None and now_seconds - attempted_seconds < cooldown_seconds:
continue
_inflight_publications.add(key)
_recent_attempt_seconds[key] = now_seconds
claimed.append(item)
return claimed
async def _release_claims(*, user_id: int, publication_ids: list[int]) -> None:
async with _enrichment_lock:
for publication_id in publication_ids:
_inflight_publications.discard((user_id, publication_id))
def _on_task_done(task: asyncio.Task[None]) -> None:
_scheduled_tasks.discard(task)
try:
task.result()
except Exception:
logger.exception(
"publications.enrichment.task_failed",
extra={"event": "publications.enrichment.task_failed"},
)
async def _run_enrichment(
*,
user_id: int,
request_email: str | None,
items: list[PublicationListItem],
) -> None:
publication_ids = [item.publication_id for item in items]
try:
session_factory = get_session_factory()
async with session_factory() as db_session:
await resolve_and_persist_oa_metadata(
db_session,
rows=items,
unpaywall_email=request_email,
)
finally:
await _release_claims(user_id=user_id, publication_ids=publication_ids)
logger.info(
"publications.enrichment.completed",
extra={
"event": "publications.enrichment.completed",
"user_id": user_id,
"publication_count": len(items),
},
)
async def schedule_missing_pdf_enrichment_for_user(
db_session: AsyncSession,
*,
user_id: int,
request_email: str | None,
items: list[PublicationListItem],
max_items: int,
) -> int:
claimed_items = await _claim_items(user_id=user_id, items=items, max_items=max_items)
if not claimed_items:
return 0
task = asyncio.create_task(
_run_enrichment(
user_id=user_id,
request_email=request_email,
items=claimed_items,
)
queued_ids = await enqueue_missing_pdf_jobs(
db_session,
user_id=user_id,
request_email=request_email,
rows=items,
max_items=max_items,
)
_scheduled_tasks.add(task)
task.add_done_callback(_on_task_done)
logger.info(
"publications.enrichment.scheduled",
extra={
"event": "publications.enrichment.scheduled",
"user_id": user_id,
"publication_count": len(claimed_items),
"publication_count": len(queued_ids),
},
)
return len(claimed_items)
return len(queued_ids)
async def schedule_retry_pdf_enrichment_for_row(
db_session: AsyncSession,
*,
user_id: int,
request_email: str | None,
item: PublicationListItem,
) -> bool:
queued = await enqueue_retry_pdf_job(
db_session,
user_id=user_id,
request_email=request_email,
row=item,
)
logger.info(
"publications.enrichment.retry_scheduled",
extra={
"event": "publications.enrichment.retry_scheduled",
"user_id": user_id,
"publication_id": item.publication_id,
"queued": queued,
},
)
return queued
async def hydrate_pdf_enrichment_state(
db_session: AsyncSession,
*,
items: list[PublicationListItem],
) -> list[PublicationListItem]:
return await overlay_pdf_job_state(db_session, rows=items)

View file

@ -16,94 +16,6 @@ from app.services.domains.publications.queries import (
unread_item_from_row,
)
from app.services.domains.publications.types import PublicationListItem, UnreadPublicationItem
from app.services.domains.unpaywall.application import resolve_publication_oa_metadata
from sqlalchemy import update
from app.db.models import Publication
def _with_oa_overrides(
rows: list[PublicationListItem],
oa_data: dict[int, tuple[str | None, str | None]],
) -> list[PublicationListItem]:
return [
PublicationListItem(
publication_id=row.publication_id,
scholar_profile_id=row.scholar_profile_id,
scholar_label=row.scholar_label,
title=row.title,
year=row.year,
citation_count=row.citation_count,
venue_text=row.venue_text,
pub_url=row.pub_url,
doi=(oa_data.get(row.publication_id) or (None, None))[0] or row.doi,
pdf_url=(oa_data.get(row.publication_id) or (None, None))[1] or row.pdf_url,
is_read=row.is_read,
first_seen_at=row.first_seen_at,
is_new_in_latest_run=row.is_new_in_latest_run,
)
for row in rows
]
def _resolved_fields(
*,
row: PublicationListItem,
resolved: tuple[str | None, str | None] | None,
) -> tuple[str | None, str | None]:
if resolved is None:
return row.doi, row.pdf_url
resolved_doi, resolved_pdf = resolved
return resolved_doi or row.doi, resolved_pdf or row.pdf_url
async def _persist_resolved_metadata(
db_session: AsyncSession,
*,
rows: list[PublicationListItem],
oa_data: dict[int, tuple[str | None, str | None]],
) -> None:
by_id = {row.publication_id: row for row in rows}
updates: list[tuple[int, str | None, str | None]] = []
for publication_id, resolved in oa_data.items():
existing = by_id.get(publication_id)
if existing is None:
continue
next_doi, next_pdf = _resolved_fields(row=existing, resolved=resolved)
if next_doi == existing.doi and next_pdf == existing.pdf_url:
continue
updates.append((publication_id, next_doi, next_pdf))
for publication_id, doi, pdf_url in updates:
await db_session.execute(
update(Publication)
.where(Publication.id == publication_id)
.values(doi=doi, pdf_url=pdf_url)
)
if updates:
await db_session.commit()
def missing_pdf_items(
rows: list[PublicationListItem],
*,
limit: int,
) -> list[PublicationListItem]:
bounded_limit = max(0, int(limit))
if bounded_limit == 0:
return []
return [row for row in rows if not row.pdf_url][:bounded_limit]
async def resolve_and_persist_oa_metadata(
db_session: AsyncSession,
*,
rows: list[PublicationListItem],
unpaywall_email: str | None = None,
) -> dict[int, tuple[str | None, str | None]]:
if not rows:
return {}
oa_data = await resolve_publication_oa_metadata(rows, request_email=unpaywall_email)
await _persist_resolved_metadata(db_session, rows=rows, oa_data=oa_data)
return oa_data
async def list_for_user(
@ -112,7 +24,9 @@ async def list_for_user(
user_id: int,
mode: str = MODE_ALL,
scholar_profile_id: int | None = None,
favorite_only: bool = False,
limit: int = 300,
offset: int = 0,
) -> list[PublicationListItem]:
resolved_mode = resolve_publication_view_mode(mode)
latest_run_id = await get_latest_completed_run_id_for_user(db_session, user_id=user_id)
@ -122,7 +36,9 @@ async def list_for_user(
mode=resolved_mode,
latest_run_id=latest_run_id,
scholar_profile_id=scholar_profile_id,
favorite_only=favorite_only,
limit=limit,
offset=offset,
)
)
rows = [
@ -138,22 +54,13 @@ async def retry_pdf_for_user(
user_id: int,
scholar_profile_id: int,
publication_id: int,
unpaywall_email: str | None = None,
) -> PublicationListItem | None:
item = await get_publication_item_for_user(
return await get_publication_item_for_user(
db_session,
user_id=user_id,
scholar_profile_id=scholar_profile_id,
publication_id=publication_id,
)
if item is None:
return None
oa_data = await resolve_and_persist_oa_metadata(
db_session,
rows=[item],
unpaywall_email=unpaywall_email,
)
return _with_oa_overrides([item], oa_data)[0]
async def list_unread_for_user(
@ -168,7 +75,9 @@ async def list_unread_for_user(
mode=MODE_UNREAD,
latest_run_id=None,
scholar_profile_id=None,
favorite_only=False,
limit=limit,
offset=0,
)
)
return [unread_item_from_row(row) for row in result.all()]

View file

@ -0,0 +1,877 @@
from __future__ import annotations
import asyncio
from dataclasses import dataclass
from datetime import datetime, timezone
import logging
from sqlalchemy import Select, func, literal, or_, select, union_all
from sqlalchemy.ext.asyncio import AsyncSession
from app.db.models import (
Publication,
PublicationPdfJob,
PublicationPdfJobEvent,
ScholarProfile,
ScholarPublication,
User,
)
from app.db.session import get_session_factory
from app.services.domains.publications.types import PublicationListItem
from app.services.domains.unpaywall.application import (
FAILURE_RESOLUTION_EXCEPTION,
OaResolutionOutcome,
resolve_publication_oa_outcomes,
)
from app.settings import settings
PDF_STATUS_UNTRACKED = "untracked"
PDF_STATUS_QUEUED = "queued"
PDF_STATUS_RUNNING = "running"
PDF_STATUS_RESOLVED = "resolved"
PDF_STATUS_FAILED = "failed"
PDF_EVENT_QUEUED = "queued"
PDF_EVENT_ATTEMPT_STARTED = "attempt_started"
PDF_EVENT_RESOLVED = "resolved"
PDF_EVENT_FAILED = "failed"
logger = logging.getLogger(__name__)
_scheduled_tasks: set[asyncio.Task[None]] = set()
@dataclass(frozen=True)
class PdfQueueListItem:
publication_id: int
title: str
doi: str | None
pdf_url: str | None
status: str
attempt_count: int
last_failure_reason: str | None
last_failure_detail: str | None
last_source: str | None
requested_by_user_id: int | None
requested_by_email: str | None
queued_at: datetime | None
last_attempt_at: datetime | None
resolved_at: datetime | None
updated_at: datetime
@dataclass(frozen=True)
class PdfRequeueResult:
publication_exists: bool
queued: bool
@dataclass(frozen=True)
class PdfBulkQueueResult:
requested_count: int
queued_count: int
@dataclass(frozen=True)
class PdfQueuePage:
items: list[PdfQueueListItem]
total_count: int
limit: int
offset: int
def _utcnow() -> datetime:
return datetime.now(timezone.utc)
def _publication_ids(rows: list[PublicationListItem]) -> list[int]:
return sorted({row.publication_id for row in rows})
def _status_from_job(row: PublicationListItem, job: PublicationPdfJob | None) -> str:
if row.pdf_url:
return PDF_STATUS_RESOLVED
if job is None:
return PDF_STATUS_UNTRACKED
return job.status
def _item_from_row_and_job(
row: PublicationListItem,
job: PublicationPdfJob | None,
) -> PublicationListItem:
return PublicationListItem(
publication_id=row.publication_id,
scholar_profile_id=row.scholar_profile_id,
scholar_label=row.scholar_label,
title=row.title,
year=row.year,
citation_count=row.citation_count,
venue_text=row.venue_text,
pub_url=row.pub_url,
doi=row.doi,
pdf_url=row.pdf_url,
is_read=row.is_read,
is_favorite=row.is_favorite,
first_seen_at=row.first_seen_at,
is_new_in_latest_run=row.is_new_in_latest_run,
pdf_status=_status_from_job(row, job),
pdf_attempt_count=int(job.attempt_count) if job is not None else 0,
pdf_failure_reason=job.last_failure_reason if job is not None else None,
pdf_failure_detail=job.last_failure_detail if job is not None else None,
)
def _queueable_rows(
rows: list[PublicationListItem],
*,
max_items: int,
) -> list[PublicationListItem]:
bounded = max(0, int(max_items))
if bounded == 0:
return []
candidates = [row for row in rows if not row.pdf_url]
return candidates[:bounded]
def _bounded_limit(limit: int, *, max_value: int = 500) -> int:
return max(1, min(int(limit), max_value))
def _bounded_offset(offset: int) -> int:
return max(int(offset), 0)
def _auto_retry_interval_seconds() -> int:
return max(int(settings.pdf_auto_retry_interval_seconds), 1)
def _auto_retry_first_interval_seconds() -> int:
return max(int(settings.pdf_auto_retry_first_interval_seconds), 1)
def _auto_retry_max_attempts() -> int:
return max(int(settings.pdf_auto_retry_max_attempts), 1)
def _retry_interval_seconds_for_attempt_count(attempt_count: int) -> int:
if int(attempt_count) <= 1:
return _auto_retry_first_interval_seconds()
return _auto_retry_interval_seconds()
def _cooldown_active(
*,
last_attempt_at: datetime | None,
attempt_count: int,
) -> bool:
if last_attempt_at is None:
return False
elapsed = (_utcnow() - last_attempt_at).total_seconds()
return elapsed < _retry_interval_seconds_for_attempt_count(int(attempt_count))
def _can_enqueue_job(
job: PublicationPdfJob | None,
*,
force_retry: bool,
) -> bool:
if job is None:
return True
if job.status in {PDF_STATUS_QUEUED, PDF_STATUS_RUNNING}:
return False
if force_retry:
return job.status in {PDF_STATUS_FAILED, PDF_STATUS_RESOLVED, PDF_STATUS_UNTRACKED}
if job.status == PDF_STATUS_RESOLVED:
return False
if int(job.attempt_count) >= _auto_retry_max_attempts():
return False
if _cooldown_active(
last_attempt_at=job.last_attempt_at,
attempt_count=int(job.attempt_count),
):
return False
return True
def _event_row(
*,
publication_id: int,
user_id: int | None,
event_type: str,
status: str | None,
source: str | None = None,
failure_reason: str | None = None,
message: str | None = None,
) -> PublicationPdfJobEvent:
return PublicationPdfJobEvent(
publication_id=publication_id,
user_id=user_id,
event_type=event_type,
status=status,
source=source,
failure_reason=failure_reason,
message=message,
)
def _queued_job(
*,
publication_id: int,
user_id: int,
) -> PublicationPdfJob:
now = _utcnow()
return PublicationPdfJob(
publication_id=publication_id,
status=PDF_STATUS_QUEUED,
queued_at=now,
last_requested_by_user_id=user_id,
)
def _mark_job_queued(job: PublicationPdfJob, *, user_id: int) -> None:
now = _utcnow()
job.status = PDF_STATUS_QUEUED
job.queued_at = now
job.last_requested_by_user_id = user_id
job.last_failure_reason = None
job.last_failure_detail = None
job.last_source = None
def _state_map(jobs: list[PublicationPdfJob]) -> dict[int, PublicationPdfJob]:
return {int(job.publication_id): job for job in jobs}
async def _jobs_for_publication_ids(
db_session: AsyncSession,
*,
publication_ids: list[int],
) -> dict[int, PublicationPdfJob]:
if not publication_ids:
return {}
result = await db_session.execute(
select(PublicationPdfJob).where(PublicationPdfJob.publication_id.in_(publication_ids))
)
return _state_map(list(result.scalars()))
async def overlay_pdf_job_state(
db_session: AsyncSession,
*,
rows: list[PublicationListItem],
) -> list[PublicationListItem]:
if not rows:
return []
jobs = await _jobs_for_publication_ids(
db_session,
publication_ids=_publication_ids(rows),
)
return [_item_from_row_and_job(row, jobs.get(row.publication_id)) for row in rows]
async def _enqueue_rows(
db_session: AsyncSession,
*,
user_id: int,
rows: list[PublicationListItem],
force_retry: bool,
) -> list[PublicationListItem]:
if not rows:
return []
queued: list[PublicationListItem] = []
jobs = await _jobs_for_publication_ids(
db_session,
publication_ids=_publication_ids(rows),
)
for row in rows:
job = jobs.get(row.publication_id)
if not _can_enqueue_job(job, force_retry=force_retry):
continue
if job is None:
job = _queued_job(publication_id=row.publication_id, user_id=user_id)
jobs[row.publication_id] = job
db_session.add(job)
else:
_mark_job_queued(job, user_id=user_id)
db_session.add(
_event_row(
publication_id=row.publication_id,
user_id=user_id,
event_type=PDF_EVENT_QUEUED,
status=PDF_STATUS_QUEUED,
)
)
queued.append(row)
if queued:
await db_session.commit()
return queued
def _register_task(task: asyncio.Task[None]) -> None:
_scheduled_tasks.add(task)
def _drop_finished_task(task: asyncio.Task[None]) -> None:
_scheduled_tasks.discard(task)
try:
task.result()
except Exception:
logger.exception(
"publications.pdf_queue.task_failed",
extra={"event": "publications.pdf_queue.task_failed"},
)
async def _mark_attempt_started(
*,
publication_id: int,
user_id: int,
) -> None:
session_factory = get_session_factory()
async with session_factory() as db_session:
job = await db_session.get(PublicationPdfJob, publication_id)
if job is None:
job = _queued_job(publication_id=publication_id, user_id=user_id)
db_session.add(job)
job.status = PDF_STATUS_RUNNING
job.last_attempt_at = _utcnow()
job.attempt_count = int(job.attempt_count) + 1
db_session.add(
_event_row(
publication_id=publication_id,
user_id=user_id,
event_type=PDF_EVENT_ATTEMPT_STARTED,
status=PDF_STATUS_RUNNING,
)
)
await db_session.commit()
def _failed_outcome(
*,
row: PublicationListItem,
) -> OaResolutionOutcome:
return OaResolutionOutcome(
publication_id=row.publication_id,
doi=row.doi,
pdf_url=None,
failure_reason=FAILURE_RESOLUTION_EXCEPTION,
source=None,
used_crossref=False,
)
async def _fetch_outcome_for_row(
*,
row: PublicationListItem,
request_email: str | None,
) -> OaResolutionOutcome:
outcomes = await resolve_publication_oa_outcomes([row], request_email=request_email)
outcome = outcomes.get(row.publication_id)
if outcome is not None:
return outcome
return _failed_outcome(row=row)
def _apply_publication_update(
publication: Publication,
*,
doi: str | None,
pdf_url: str | None,
) -> None:
if doi and publication.doi != doi:
publication.doi = doi
if pdf_url and publication.pdf_url != pdf_url:
publication.pdf_url = pdf_url
def _apply_job_outcome(job: PublicationPdfJob, *, outcome: OaResolutionOutcome) -> None:
job.last_source = outcome.source
if outcome.pdf_url:
job.status = PDF_STATUS_RESOLVED
job.resolved_at = _utcnow()
job.last_failure_reason = None
job.last_failure_detail = None
return
job.status = PDF_STATUS_FAILED
job.last_failure_reason = outcome.failure_reason
job.last_failure_detail = outcome.failure_reason
def _result_event(outcome: OaResolutionOutcome) -> tuple[str, str]:
if outcome.pdf_url:
return PDF_EVENT_RESOLVED, PDF_STATUS_RESOLVED
return PDF_EVENT_FAILED, PDF_STATUS_FAILED
async def _persist_outcome(
*,
publication_id: int,
user_id: int,
outcome: OaResolutionOutcome,
) -> None:
session_factory = get_session_factory()
async with session_factory() as db_session:
publication = await db_session.get(Publication, publication_id)
job = await db_session.get(PublicationPdfJob, publication_id)
if publication is None or job is None:
return
_apply_publication_update(publication, doi=outcome.doi, pdf_url=outcome.pdf_url)
_apply_job_outcome(job, outcome=outcome)
event_type, status = _result_event(outcome)
db_session.add(
_event_row(
publication_id=publication_id,
user_id=user_id,
event_type=event_type,
status=status,
source=outcome.source,
failure_reason=outcome.failure_reason,
message=outcome.failure_reason,
)
)
await db_session.commit()
async def _resolve_publication_row(
*,
user_id: int,
request_email: str | None,
row: PublicationListItem,
) -> None:
await _mark_attempt_started(publication_id=row.publication_id, user_id=user_id)
try:
outcome = await _fetch_outcome_for_row(row=row, request_email=request_email)
except Exception as exc: # pragma: no cover - defensive network boundary
logger.warning(
"publications.pdf_queue.resolve_failed",
extra={
"event": "publications.pdf_queue.resolve_failed",
"publication_id": row.publication_id,
"error": str(exc),
},
)
outcome = _failed_outcome(row=row)
await _persist_outcome(
publication_id=row.publication_id,
user_id=user_id,
outcome=outcome,
)
async def _run_resolution_task(
*,
user_id: int,
request_email: str | None,
rows: list[PublicationListItem],
) -> None:
for row in rows:
await _resolve_publication_row(
user_id=user_id,
request_email=request_email,
row=row,
)
def _schedule_rows(
*,
user_id: int,
request_email: str | None,
rows: list[PublicationListItem],
) -> None:
if not rows:
return
task = asyncio.create_task(
_run_resolution_task(
user_id=user_id,
request_email=request_email,
rows=rows,
)
)
_register_task(task)
task.add_done_callback(_drop_finished_task)
async def enqueue_missing_pdf_jobs(
db_session: AsyncSession,
*,
user_id: int,
request_email: str | None,
rows: list[PublicationListItem],
max_items: int,
) -> list[int]:
queueable = _queueable_rows(rows, max_items=max_items)
queued_rows = await _enqueue_rows(
db_session,
user_id=user_id,
rows=queueable,
force_retry=False,
)
_schedule_rows(user_id=user_id, request_email=request_email, rows=queued_rows)
return [row.publication_id for row in queued_rows]
async def enqueue_retry_pdf_job(
db_session: AsyncSession,
*,
user_id: int,
request_email: str | None,
row: PublicationListItem,
) -> bool:
queued_rows = await _enqueue_rows(
db_session,
user_id=user_id,
rows=[row],
force_retry=True,
)
_schedule_rows(user_id=user_id, request_email=request_email, rows=queued_rows)
return bool(queued_rows)
def _retry_item_label(display_name: str | None, scholar_id: str | None) -> str:
return str(display_name or scholar_id or "unknown")
def _retry_item_from_publication(
publication: Publication,
*,
link_row: tuple | None,
) -> PublicationListItem:
if link_row is None:
scholar_profile_id = 0
scholar_label = "unknown"
is_read = True
first_seen_at = publication.created_at or _utcnow()
else:
scholar_profile_id = int(link_row[0])
scholar_label = _retry_item_label(link_row[1], link_row[2])
is_read = bool(link_row[3])
first_seen_at = link_row[4] or publication.created_at or _utcnow()
return PublicationListItem(
publication_id=int(publication.id),
scholar_profile_id=scholar_profile_id,
scholar_label=scholar_label,
title=publication.title_raw,
year=publication.year,
citation_count=int(publication.citation_count or 0),
venue_text=publication.venue_text,
pub_url=publication.pub_url,
doi=publication.doi,
pdf_url=publication.pdf_url,
is_read=is_read,
first_seen_at=first_seen_at,
is_new_in_latest_run=False,
)
async def _retry_item_for_publication_id(
db_session: AsyncSession,
*,
publication_id: int,
) -> PublicationListItem | None:
publication = await db_session.get(Publication, publication_id)
if publication is None:
return None
result = await db_session.execute(
select(
ScholarProfile.id,
ScholarProfile.display_name,
ScholarProfile.scholar_id,
ScholarPublication.is_read,
ScholarPublication.created_at,
)
.join(ScholarProfile, ScholarProfile.id == ScholarPublication.scholar_profile_id)
.where(ScholarPublication.publication_id == publication_id)
.order_by(ScholarPublication.created_at.asc())
.limit(1)
)
return _retry_item_from_publication(publication, link_row=result.one_or_none())
async def enqueue_retry_pdf_job_for_publication_id(
db_session: AsyncSession,
*,
user_id: int,
request_email: str | None,
publication_id: int,
) -> PdfRequeueResult:
row = await _retry_item_for_publication_id(
db_session,
publication_id=publication_id,
)
if row is None:
return PdfRequeueResult(publication_exists=False, queued=False)
queued = await enqueue_retry_pdf_job(
db_session,
user_id=user_id,
request_email=request_email,
row=row,
)
return PdfRequeueResult(publication_exists=True, queued=queued)
def _queue_candidate_from_publication(publication: Publication) -> PublicationListItem:
return PublicationListItem(
publication_id=int(publication.id),
scholar_profile_id=0,
scholar_label="admin",
title=publication.title_raw,
year=publication.year,
citation_count=int(publication.citation_count or 0),
venue_text=publication.venue_text,
pub_url=publication.pub_url,
doi=publication.doi,
pdf_url=publication.pdf_url,
is_read=True,
first_seen_at=publication.created_at or _utcnow(),
is_new_in_latest_run=False,
)
async def _missing_pdf_candidates(
db_session: AsyncSession,
*,
limit: int,
) -> list[PublicationListItem]:
bounded_limit = max(1, min(int(limit), 5000))
result = await db_session.execute(
select(Publication)
.outerjoin(PublicationPdfJob, PublicationPdfJob.publication_id == Publication.id)
.where(Publication.pdf_url.is_(None))
.where(
or_(
PublicationPdfJob.publication_id.is_(None),
PublicationPdfJob.status.notin_([PDF_STATUS_QUEUED, PDF_STATUS_RUNNING]),
)
)
.order_by(Publication.updated_at.desc(), Publication.id.desc())
.limit(bounded_limit)
)
return [
_queue_candidate_from_publication(publication)
for publication in result.scalars()
]
async def enqueue_all_missing_pdf_jobs(
db_session: AsyncSession,
*,
user_id: int,
request_email: str | None,
limit: int = 1000,
) -> PdfBulkQueueResult:
candidates = await _missing_pdf_candidates(db_session, limit=limit)
queued_rows = await _enqueue_rows(
db_session,
user_id=user_id,
rows=candidates,
force_retry=True,
)
_schedule_rows(user_id=user_id, request_email=request_email, rows=queued_rows)
return PdfBulkQueueResult(
requested_count=len(candidates),
queued_count=len(queued_rows),
)
def _tracked_queue_select_base(*, status: str | None) -> Select[tuple]:
stmt = (
select(
PublicationPdfJob.publication_id,
Publication.title_raw,
Publication.doi,
Publication.pdf_url,
PublicationPdfJob.status,
PublicationPdfJob.attempt_count,
PublicationPdfJob.last_failure_reason,
PublicationPdfJob.last_failure_detail,
PublicationPdfJob.last_source,
PublicationPdfJob.last_requested_by_user_id,
User.email,
PublicationPdfJob.queued_at,
PublicationPdfJob.last_attempt_at,
PublicationPdfJob.resolved_at,
PublicationPdfJob.updated_at,
)
.join(Publication, Publication.id == PublicationPdfJob.publication_id)
.outerjoin(User, User.id == PublicationPdfJob.last_requested_by_user_id)
)
if status:
stmt = stmt.where(PublicationPdfJob.status == status)
return stmt
def _tracked_queue_select(*, limit: int, offset: int, status: str | None) -> Select[tuple]:
return (
_tracked_queue_select_base(status=status)
.order_by(PublicationPdfJob.updated_at.desc())
.limit(_bounded_limit(limit))
.offset(_bounded_offset(offset))
)
def _untracked_queue_select_base() -> Select[tuple]:
return (
select(
Publication.id,
Publication.title_raw,
Publication.doi,
Publication.pdf_url,
literal(PDF_STATUS_UNTRACKED),
literal(0),
literal(None),
literal(None),
literal(None),
literal(None),
literal(None),
literal(None),
literal(None),
literal(None),
Publication.updated_at,
)
.outerjoin(PublicationPdfJob, PublicationPdfJob.publication_id == Publication.id)
.where(Publication.pdf_url.is_(None))
.where(PublicationPdfJob.publication_id.is_(None))
)
def _untracked_queue_select(*, limit: int, offset: int) -> Select[tuple]:
return (
_untracked_queue_select_base()
.order_by(Publication.updated_at.desc(), Publication.id.desc())
.limit(_bounded_limit(limit))
.offset(_bounded_offset(offset))
)
def _all_queue_select(*, limit: int, offset: int) -> Select[tuple]:
union_stmt = union_all(
_tracked_queue_select_base(status=None),
_untracked_queue_select_base(),
).subquery()
return (
select(union_stmt)
.order_by(union_stmt.c.updated_at.desc())
.limit(_bounded_limit(limit))
.offset(_bounded_offset(offset))
)
def _tracked_queue_count_select(*, status: str | None) -> Select[tuple]:
stmt = select(func.count()).select_from(PublicationPdfJob)
if status:
stmt = stmt.where(PublicationPdfJob.status == status)
return stmt
def _untracked_queue_count_select() -> Select[tuple]:
return (
select(func.count())
.select_from(Publication)
.outerjoin(PublicationPdfJob, PublicationPdfJob.publication_id == Publication.id)
.where(Publication.pdf_url.is_(None))
.where(PublicationPdfJob.publication_id.is_(None))
)
def _queue_item_from_row(row: tuple) -> PdfQueueListItem:
return PdfQueueListItem(
publication_id=int(row[0]),
title=str(row[1] or ""),
doi=row[2],
pdf_url=row[3],
status=str(row[4] or PDF_STATUS_UNTRACKED),
attempt_count=int(row[5] or 0),
last_failure_reason=row[6],
last_failure_detail=row[7],
last_source=row[8],
requested_by_user_id=int(row[9]) if row[9] is not None else None,
requested_by_email=row[10],
queued_at=row[11],
last_attempt_at=row[12],
resolved_at=row[13],
updated_at=row[14],
)
async def list_pdf_queue_items(
db_session: AsyncSession,
*,
limit: int = 100,
offset: int = 0,
status: str | None = None,
) -> list[PdfQueueListItem]:
bounded_limit = _bounded_limit(limit)
bounded_offset = _bounded_offset(offset)
normalized_status = (status or "").strip().lower() or None
if normalized_status == PDF_STATUS_UNTRACKED:
result = await db_session.execute(
_untracked_queue_select(
limit=bounded_limit,
offset=bounded_offset,
)
)
return [_queue_item_from_row(row) for row in result.all()]
if normalized_status is None:
result = await db_session.execute(
_all_queue_select(
limit=bounded_limit,
offset=bounded_offset,
)
)
return [_queue_item_from_row(row) for row in result.all()]
result = await db_session.execute(
_tracked_queue_select(
limit=bounded_limit,
offset=bounded_offset,
status=normalized_status,
)
)
return [_queue_item_from_row(row) for row in result.all()]
async def count_pdf_queue_items(
db_session: AsyncSession,
*,
status: str | None = None,
) -> int:
normalized_status = (status or "").strip().lower() or None
if normalized_status == PDF_STATUS_UNTRACKED:
result = await db_session.execute(_untracked_queue_count_select())
return int(result.scalar_one() or 0)
tracked_result = await db_session.execute(
_tracked_queue_count_select(status=normalized_status)
)
tracked_count = int(tracked_result.scalar_one() or 0)
if normalized_status is not None:
return tracked_count
untracked_result = await db_session.execute(_untracked_queue_count_select())
untracked_count = int(untracked_result.scalar_one() or 0)
return tracked_count + untracked_count
async def list_pdf_queue_page(
db_session: AsyncSession,
*,
limit: int = 100,
offset: int = 0,
status: str | None = None,
) -> PdfQueuePage:
bounded_limit = _bounded_limit(limit)
bounded_offset = _bounded_offset(offset)
items = await list_pdf_queue_items(
db_session,
limit=bounded_limit,
offset=bounded_offset,
status=status,
)
total_count = await count_pdf_queue_items(
db_session,
status=status,
)
return PdfQueuePage(
items=items,
total_count=total_count,
limit=bounded_limit,
offset=bounded_offset,
)

View file

@ -36,7 +36,9 @@ def publications_query(
mode: str,
latest_run_id: int | None,
scholar_profile_id: int | None,
favorite_only: bool,
limit: int,
offset: int = 0,
) -> Select[tuple]:
scholar_label = ScholarProfile.display_name
stmt = (
@ -53,6 +55,7 @@ def publications_query(
Publication.doi,
Publication.pdf_url,
ScholarPublication.is_read,
ScholarPublication.is_favorite,
ScholarPublication.first_seen_run_id,
ScholarPublication.created_at,
)
@ -60,10 +63,13 @@ def publications_query(
.join(ScholarProfile, ScholarProfile.id == ScholarPublication.scholar_profile_id)
.where(ScholarProfile.user_id == user_id)
.order_by(ScholarPublication.created_at.desc(), Publication.id.desc())
.offset(max(int(offset), 0))
.limit(limit)
)
if scholar_profile_id is not None:
stmt = stmt.where(ScholarProfile.id == scholar_profile_id)
if favorite_only:
stmt = stmt.where(ScholarPublication.is_favorite.is_(True))
if mode == MODE_UNREAD:
stmt = stmt.where(ScholarPublication.is_read.is_(False))
if mode == MODE_LATEST:
@ -93,6 +99,7 @@ def publication_query_for_user(
Publication.doi,
Publication.pdf_url,
ScholarPublication.is_read,
ScholarPublication.is_favorite,
ScholarPublication.first_seen_run_id,
ScholarPublication.created_at,
)
@ -146,6 +153,7 @@ def publication_list_item_from_row(
doi,
pdf_url,
is_read,
is_favorite,
first_seen_run_id,
created_at,
) = row
@ -161,6 +169,7 @@ def publication_list_item_from_row(
doi=doi,
pdf_url=pdf_url,
is_read=bool(is_read),
is_favorite=bool(is_favorite),
first_seen_at=created_at,
is_new_in_latest_run=(
latest_run_id is not None and int(first_seen_run_id or 0) == latest_run_id
@ -182,6 +191,7 @@ def unread_item_from_row(row: tuple) -> UnreadPublicationItem:
doi,
pdf_url,
_is_read,
_is_favorite,
_first_seen_run_id,
_created_at,
) = row

View file

@ -16,16 +16,20 @@ def _normalized_selection_pairs(selections: list[tuple[int, int]]) -> set[tuple[
return pairs
def _scoped_scholar_ids_query(*, user_id: int):
return (
select(ScholarProfile.id)
.where(ScholarProfile.user_id == user_id)
.scalar_subquery()
)
async def mark_all_unread_as_read_for_user(
db_session: AsyncSession,
*,
user_id: int,
) -> int:
scholar_ids = (
select(ScholarProfile.id)
.where(ScholarProfile.user_id == user_id)
.scalar_subquery()
)
scholar_ids = _scoped_scholar_ids_query(user_id=user_id)
stmt = (
update(ScholarPublication)
.where(
@ -49,11 +53,7 @@ async def mark_selected_as_read_for_user(
if not normalized_pairs:
return 0
scholar_ids = (
select(ScholarProfile.id)
.where(ScholarProfile.user_id == user_id)
.scalar_subquery()
)
scholar_ids = _scoped_scholar_ids_query(user_id=user_id)
stmt = (
update(ScholarPublication)
.where(
@ -69,3 +69,26 @@ async def mark_selected_as_read_for_user(
result = await db_session.execute(stmt)
await db_session.commit()
return int(result.rowcount or 0)
async def set_publication_favorite_for_user(
db_session: AsyncSession,
*,
user_id: int,
scholar_profile_id: int,
publication_id: int,
is_favorite: bool,
) -> int:
scholar_ids = _scoped_scholar_ids_query(user_id=user_id)
stmt = (
update(ScholarPublication)
.where(
ScholarPublication.scholar_profile_id.in_(scholar_ids),
ScholarPublication.scholar_profile_id == int(scholar_profile_id),
ScholarPublication.publication_id == int(publication_id),
)
.values(is_favorite=bool(is_favorite))
)
result = await db_session.execute(stmt)
await db_session.commit()
return int(result.rowcount or 0)

View file

@ -19,6 +19,11 @@ class PublicationListItem:
is_read: bool
first_seen_at: datetime
is_new_in_latest_run: bool
is_favorite: bool = False
pdf_status: str = "untracked"
pdf_attempt_count: int = 0
pdf_failure_reason: str | None = None
pdf_failure_detail: str | None = None
@dataclass(frozen=True)