def update_docket_info_iquery(self, d_pk):
cookies = get_or_cache_pacer_cookies(
"pacer_scraper",
settings.PACER_USERNAME,
password=settings.PACER_PASSWORD,
)
s = PacerSession(
cookies=cookies,
username=settings.PACER_USERNAME,
password=settings.PACER_PASSWORD,
)
d = Docket.objects.get(pk=d_pk)
report = CaseQuery(map_cl_to_pacer_id(d.court_id), s)
try:
report.query(d.pacer_case_id)
except (requests.Timeout, requests.RequestException) as exc:
logger.warning(
"Timeout or unknown RequestException on iquery crawl. "
"Trying again if retries not exceeded."
)
if self.request.retries == self.max_retries:
return
raise self.retry(exc=exc)
d = update_docket_metadata(d, report.data)
d.save()
add_bankruptcy_data_to_docket(d, report.data)
add_items_to_solr([d.pk], "search.Docket")
def validate(self, attrs):
# Is it a good court value?
district_court_ids = Court.objects.filter(
Q(jurisdiction__in=[
Court.FEDERAL_DISTRICT,
Court.FEDERAL_BANKRUPTCY,
])
| Q(pk__in=["cit", "jpml", "uscfc"]), ).values_list("pk",
flat=True)
if attrs.get("court") and attrs["court"].pk not in district_court_ids:
raise ValidationError("Invalid court id: %s" % attrs["court"].pk)
# Docket validations
if attrs.get("pacer_case_id") and not attrs.get("court"):
# If a pacer_case_id is included, is a court also?
raise ValidationError("Cannot use 'pacer_case_id' parameter "
"without 'court' parameter.")
if attrs.get("docket_number") and not attrs.get("court"):
# If a docket_number is included, is a court also?
raise ValidationError("Cannot use 'docket_number' parameter "
"without 'court' parameter.")
if attrs.get("show_terminated_parties"
) and not attrs.get("show_parties_and_counsel"):
raise ValidationError(
"You've requested to show_terminated_parties parties while "
"show_parties_and_counsel is False. To show terminated "
"parties, you must also request showing parties and counsel "
"generally.")
# Attachment page and PDF validation
if attrs["request_type"] in [
REQUEST_TYPE.PDF,
REQUEST_TYPE.ATTACHMENT_PAGE,
]:
if not attrs.get("recap_document"):
raise ValidationError(
"recap_document is a required field for attachment page "
"and PDF fetches.")
# PDF validations
if attrs["request_type"] == REQUEST_TYPE.PDF:
rd = attrs["recap_document"]
if rd.is_available:
raise ValidationError(
"Cannot fetch a PDF for recap_document %s. That document "
"is already marked as available in our database "
"(is_available = True)." % rd.pk)
# Do the PACER credentials work?
try:
_ = get_or_cache_pacer_cookies(
attrs["user"].pk,
username=attrs.pop("pacer_username"),
password=attrs.pop("pacer_password"),
)
except PacerLoginException as e:
raise ValidationError("PacerLoginException: %s" % e.message)
return attrs
def validate(self, attrs):
# Is it a good court value?
district_court_ids = Court.objects.filter(
Q(jurisdiction__in=[
Court.FEDERAL_DISTRICT,
Court.FEDERAL_BANKRUPTCY,
]) | Q(pk__in=['cit', 'jpml', 'uscfc']),
).values_list('pk', flat=True)
if attrs.get('court') and attrs['court'].pk not in district_court_ids:
raise ValidationError("Invalid court id: %s" % attrs['court'].pk)
# Docket validations
if attrs.get('pacer_case_id') and not attrs.get('court'):
# If a pacer_case_id is included, is a court also?
raise ValidationError("Cannot use 'pacer_case_id' parameter "
"without 'court' parameter.")
if attrs.get('docket_number') and not attrs.get('court'):
# If a docket_number is included, is a court also?
raise ValidationError("Cannot use 'docket_number' parameter "
"without 'court' parameter.")
if attrs.get('show_terminated_parties') and not \
attrs.get('show_parties_and_counsel'):
raise ValidationError(
"You've requested to show_terminated_parties parties while "
"show_parties_and_counsel is False. To show terminated "
"parties, you must also request showing parties and counsel "
"generally."
)
# PDF validations
if attrs.get('recap_document'):
rd = attrs['recap_document']
if rd.is_available:
raise ValidationError(
"Cannot fetch a PDF for recap_document %s. That document "
"is already marked as available in our database "
"(is_available = True)." % rd.pk
)
# Do the PACER credentials work?
try:
_ = get_or_cache_pacer_cookies(
attrs['user'].pk,
username=attrs.pop('pacer_username'),
password=attrs.pop('pacer_password'),
)
except PacerLoginException as e:
raise ValidationError("PacerLoginException: %s" % e.message)
return attrs
def update_docket_info_iquery(self, d_pk: int, court_id: str) -> None:
"""Update the docket info from iquery
:param self: The Celery task
:param d_pk: The ID of the docket
:param court_id: The court of the docket. Needed for throttling by court.
:return: None
"""
cookies = get_or_cache_pacer_cookies(
"pacer_scraper",
settings.PACER_USERNAME,
password=settings.PACER_PASSWORD,
)
s = PacerSession(
cookies=cookies,
username=settings.PACER_USERNAME,
password=settings.PACER_PASSWORD,
)
d = Docket.objects.get(pk=d_pk, court_id=court_id)
report = CaseQuery(map_cl_to_pacer_id(d.court_id), s)
try:
report.query(d.pacer_case_id)
except (requests.Timeout, requests.RequestException) as exc:
logger.warning(
"Timeout or unknown RequestException on iquery crawl. "
"Trying again if retries not exceeded."
)
if self.request.retries == self.max_retries:
return
raise self.retry(exc=exc)
if not report.data:
return
save_iquery_to_docket(
self,
report.data,
d,
tag_names=None,
add_to_solr=True,
)
