hyperboria/nexus/actions/document_operations_pb/update_document_scimag_pb.py

325 lines
12 KiB
Python

import asyncio
import logging
from typing import (
Optional,
Set,
)
import orjson as json
from aiocrossref import CrossrefClient
from aiocrossref.exceptions import (
NotFoundError,
WrongContentTypeError,
)
from aiokafka import AIOKafkaProducer
from aiosumma import SummaClient
from izihawa_utils.common import filter_none
from izihawa_utils.pb_to_json import MessageToDict
from library.aiopostgres.pool_holder import AioPostgresPoolHolder
from nexus.actions import scimag_pb
from nexus.actions.base import BaseAction
from nexus.actions.common import canonize_doi
from nexus.actions.crossref_api import ToScimagPbAction
from nexus.actions.exceptions import InterruptProcessing
from nexus.models.proto.operation_pb2 import \
CrossReferenceOperation as CrossReferenceOperationPb
from nexus.models.proto.operation_pb2 import \
DocumentOperation as DocumentOperationPb
from nexus.models.proto.scimag_pb2 import Scimag as ScimagPb
from pypika import (
PostgreSQLQuery,
Table,
)
from pypika.terms import Array
from summa.proto import index_service_pb2 as index_service_pb
class ToPostgresAction(BaseAction):
scimag_table = Table('scimag')
db_multi_fields = {
'authors',
'ipfs_multihashes',
'isbns',
'issns',
'tags',
}
db_single_fields = {
'id',
'abstract',
'container_title',
'content',
'doi',
'embedding',
'filesize',
'first_page',
'is_deleted',
'issued_at',
'issue',
'journal_id',
'language',
'last_page',
'meta_language',
'md5',
'page_rank',
'referenced_by_count',
'scimag_bulk_id',
'title',
'type',
'updated_at',
'volume',
}
db_fields = db_single_fields | db_multi_fields
def __init__(self, database):
super().__init__()
self.pool_holder = AioPostgresPoolHolder(
conninfo=f'dbname={database["database"]} '
f'user={database["username"]} '
f'password={database["password"]} '
f'host={database["host"]}',
)
self.starts.append(self.pool_holder)
def cast_field_value(self, field_name: str, field_value):
if field_name in self.db_multi_fields:
field_value = Array(*field_value)
return field_name, field_value
def is_field_set(self, scimag_pb: ScimagPb, field_name: str):
field_value = getattr(scimag_pb, field_name)
if field_name in {'scimag_bulk_id', 'issued_at'}:
return scimag_pb.HasField(field_name)
return field_value
def generate_insert_sql(self, scimag_pb: ScimagPb, fields: Optional[Set[str]] = None):
columns = []
params = []
fields = fields or self.db_fields
for field_name in fields:
if self.is_field_set(scimag_pb, field_name):
field_value = getattr(scimag_pb, field_name)
field_name, field_value = self.cast_field_value(field_name, field_value)
columns.append(field_name)
params.append(field_value)
query = PostgreSQLQuery.into(self.scimag_table).columns(*columns).insert(*params)
if columns:
query = query.on_conflict('doi')
for field, val in zip(columns, params):
query = query.do_update(field, val)
return query.returning(self.scimag_table.id).get_sql()
def generate_update_sql(
self,
scimag_pb: ScimagPb,
fields: Optional[Set[str]] = None,
) -> str:
query = (
PostgreSQLQuery
.update(self.scimag_table)
)
fields = fields or self.db_fields
for field_name in fields:
if self.is_field_set(scimag_pb, field_name):
field_value = getattr(scimag_pb, field_name)
field_name, field_value = self.cast_field_value(field_name, field_value)
query = query.set(field_name, field_value)
return query.where(self.scimag_table.id == scimag_pb.id).get_sql()
async def do(self, document_operation_pb: DocumentOperationPb) -> DocumentOperationPb:
update_document_pb = document_operation_pb.update_document
scimag_pb = update_document_pb.typed_document.scimag
fields = update_document_pb.fields or self.db_fields
if scimag_pb.id:
sql = self.generate_update_sql(
scimag_pb,
fields=fields,
)
await self.pool_holder.execute(sql)
else:
sql = self.generate_insert_sql(
scimag_pb=scimag_pb,
fields=fields,
)
result = [row async for row in self.pool_holder.iterate(sql)]
scimag_pb.id = result[0][0]
return document_operation_pb
class ToSummaAction(BaseAction):
forbidden_types = {
'book-series',
'book-set',
'book-track',
'component',
'dataset',
'journal',
'journal-issue',
'journal-volume',
'peer-review',
'proceedings',
'report-series',
}
def __init__(self, kafka, summa):
super().__init__()
self.kafka = kafka
self.producer = None
self.summa_config = summa
self.summa_client = SummaClient(endpoint=summa['endpoint'])
async def start(self):
self.producer = self.get_producer()
await self.producer.start()
await self.summa_client.start()
async def stop(self):
await self.summa_client.stop()
if self.producer:
await self.producer.stop()
self.producer = None
def get_producer(self):
return AIOKafkaProducer(
loop=asyncio.get_running_loop(),
bootstrap_servers=self.kafka['bootstrap_servers'],
max_request_size=self.kafka['max_request_size'],
)
async def async_index(self, scimag_pb: ScimagPb):
for topic_name in self.kafka['topic_names']:
await self.producer.send_and_wait(
topic_name,
index_service_pb.IndexOperation(
index_document=index_service_pb.IndexDocumentOperation(
document=json.dumps(filter_none(MessageToDict(scimag_pb, preserving_proto_field_name=True))),
),
).SerializeToString(),
)
async def sync_index(self, scimag_pb: ScimagPb):
document = filter_none(MessageToDict(scimag_pb, preserving_proto_field_name=True))
logging.getLogger('statbox').info({'action': 'sync_index', 'document': document})
await self.summa_client.index_document(index_alias=self.summa_config['index_alias'], document=document)
await self.summa_client.commit_index(index_alias=self.summa_config['index_alias'])
async def do(self, document_operation_pb: DocumentOperationPb) -> DocumentOperationPb:
update_document_pb = document_operation_pb.update_document
scimag_pb = update_document_pb.typed_document.scimag
if scimag_pb.type in self.forbidden_types:
return document_operation_pb
if not scimag_pb.HasField('issued_at'):
scimag_pb.issued_at = -62135596800
if update_document_pb.full_text_index:
if update_document_pb.full_text_index_commit:
await self.sync_index(scimag_pb=scimag_pb)
else:
await self.async_index(scimag_pb=scimag_pb)
return document_operation_pb
class ReferencesToKafkaAction(BaseAction):
def __init__(self, kafka):
super().__init__()
self.kafka = kafka
self.producer = None
async def start(self):
self.producer = self.get_producer()
await self.producer.start()
async def stop(self):
if self.producer:
await self.producer.stop()
self.producer = None
def get_producer(self):
return AIOKafkaProducer(
loop=asyncio.get_running_loop(),
bootstrap_servers=self.kafka['bootstrap_servers'],
)
async def do(self, document_operation_pb: DocumentOperationPb) -> DocumentOperationPb:
update_document_pb = document_operation_pb.update_document
scimag_pb = update_document_pb.typed_document.scimag
for reference in scimag_pb.references:
reference_operation = CrossReferenceOperationPb(
source=scimag_pb.doi,
target=reference,
)
for topic_name in self.kafka['topic_names']:
await self.producer.send_and_wait(
topic_name,
reference_operation.SerializeToString(),
)
return document_operation_pb
class FillFromExternalSourceAction(BaseAction):
def __init__(self, crossref):
super().__init__()
self.crossref_client = CrossrefClient(
max_retries=crossref.get('max_retries', 15),
proxy_url=crossref.get('proxy_url'),
retry_delay=crossref.get('retry_delay', 0.5),
timeout=crossref.get('timeout'),
user_agent=crossref.get('user_agent'),
ttl_dns_cache=crossref.get('ttl_dns_cache'),
)
self.doi_client = self.crossref_client
self.crossref_api_to_scimag_pb_action = ToScimagPbAction()
self.starts.append(self.crossref_client)
async def try_resolve(self, doi, look_at_doi_org=False):
try:
return await self.crossref_client.works(doi=doi)
except (WrongContentTypeError, NotFoundError) as e:
if look_at_doi_org:
doi_org_response = await self.doi_client.get(doi=doi)
if doi_org_response:
resolved_doi = canonize_doi(doi_org_response.get('published-print', {}).get('DOI'))
if resolved_doi:
try:
return await self.crossref_client.works(doi=resolved_doi)
except (WrongContentTypeError, NotFoundError) as e:
raise InterruptProcessing(document_id=doi, reason=str(e))
raise InterruptProcessing(document_id=doi, reason=str(e))
async def do(self, document_operation_pb: DocumentOperationPb) -> DocumentOperationPb:
update_document_pb = document_operation_pb.update_document
scimag_pb = update_document_pb.typed_document.scimag
if not update_document_pb.should_fill_from_external_source:
return document_operation_pb
crossref_api_response = await self.try_resolve(doi=scimag_pb.doi)
new_scimag_pb = await self.crossref_api_to_scimag_pb_action.do(crossref_api_response)
scimag_pb.MergeFrom(new_scimag_pb)
return document_operation_pb
class CleanAction(BaseAction):
def __init__(self):
super().__init__()
self.cleaner = scimag_pb.CleanAction()
self.language_detect = scimag_pb.DetectLanguageAction()
self.starts.append(self.cleaner)
async def do(self, document_operation_pb: DocumentOperationPb) -> DocumentOperationPb:
update_document_pb = document_operation_pb.update_document
scimag_pb = update_document_pb.typed_document.scimag
scimag_pb = await self.cleaner.do(scimag_pb)
scimag_pb = await self.language_detect.do(scimag_pb)
if update_document_pb.fields and (scimag_pb.language or scimag_pb.meta_language):
fields = set(update_document_pb.fields)
if scimag_pb.language:
fields.add('language')
if scimag_pb.meta_language:
fields.add('meta_language')
del update_document_pb.fields[:]
update_document_pb.fields.extend(fields)
update_document_pb.typed_document.scimag.CopyFrom(scimag_pb)
return document_operation_pb