feat: release regional radar and mailbox integrations
deploy / deploy (push) Canceled after 0s

This commit is contained in:
Jens
2026-07-22 05:12:07 +02:00
parent 598d3ec18a
commit 551d0f46c2
131 changed files with 6209 additions and 336 deletions
+41 -9
View File
@@ -13,7 +13,8 @@ from django.utils import timezone
from apps.jobs.services.normalization import normalize_extracted_job
from apps.jobs.services.pipeline import persist_draft
from apps.sources.adapters.email_alert import EmailAlertAdapter
from apps.sources.models import EmailMessageRecord, RawDocument, Source
from apps.sources.models import EmailMessageRecord, MailboxConnection, RawDocument, Source
from apps.sources.platforms import platform_label
def message_identity(raw_message: bytes) -> str:
@@ -24,22 +25,32 @@ def message_identity(raw_message: bytes) -> str:
@transaction.atomic
def ingest_email(raw_message: bytes, *, mailbox: str = "INBOX") -> EmailMessageRecord:
def ingest_email(
raw_message: bytes,
*,
mailbox: str = "INBOX",
mailbox_connection: MailboxConnection | None = None,
platform: str | None = None,
) -> EmailMessageRecord:
parsed = BytesParser(policy=policy.default).parsebytes(raw_message)
message_id = message_identity(raw_message)
existing = EmailMessageRecord.objects.filter(message_id=message_id).first()
existing = EmailMessageRecord.objects.filter(
message_id=message_id, mailbox_connection=mailbox_connection
).first()
if existing:
return existing
source_domain = f"{platform}.mailbox.local" if platform else "mailbox.local"
source_name = f"{platform_label(platform)} vacaturemailbox" if platform else "Vacaturemailbox"
source, _ = Source.objects.get_or_create(
domain="mailbox.local",
domain=source_domain,
source_type=Source.Type.EMAIL,
defaults={
"name": "Vacaturemailbox",
"name": source_name,
"status": Source.Status.ACTIVE,
"policy": Source.Policy.ALLOW,
"parser_key": "email-alert",
"crawl_interval_minutes": 10,
"crawl_interval_minutes": 1440,
},
)
content_hash = hashlib.sha256(raw_message).hexdigest()
@@ -52,7 +63,11 @@ def ingest_email(raw_message: bytes, *, mailbox: str = "INBOX") -> EmailMessageR
body_text=raw_message.decode("utf-8", errors="replace"),
byte_length=len(raw_message),
retain_until=retain_until,
metadata={"mailbox": mailbox},
metadata={
"mailbox": mailbox,
"mailbox_connection_id": mailbox_connection.pk if mailbox_connection else None,
"alert_provider": platform or "unknown",
},
)
received_at = None
if parsed.get("date"):
@@ -64,12 +79,28 @@ def ingest_email(raw_message: bytes, *, mailbox: str = "INBOX") -> EmailMessageR
received_at = None
adapter = EmailAlertAdapter()
result = adapter.extract_message(raw_message)
result = adapter.extract_message(raw_message, expected_provider=platform)
detected_provider = next(
(
str(extracted.raw.get("alert_provider"))
for extracted in result.jobs
if extracted.raw.get("alert_provider")
),
"unknown",
)
alert_provider = platform or detected_provider
document.parser_key = result.parser_key
document.parser_version = result.parser_version
document.extraction_confidence = result.confidence
document.metadata = {**document.metadata, "alert_provider": alert_provider}
document.save(
update_fields=["parser_key", "parser_version", "extraction_confidence", "updated_at"]
update_fields=[
"parser_key",
"parser_version",
"extraction_confidence",
"metadata",
"updated_at",
]
)
links: list[str] = []
errors: list[str] = []
@@ -87,6 +118,7 @@ def ingest_email(raw_message: bytes, *, mailbox: str = "INBOX") -> EmailMessageR
except Exception as exc:
errors.append(f"{exc.__class__.__name__}: {exc}")
return EmailMessageRecord.objects.create(
mailbox_connection=mailbox_connection,
message_id=message_id,
mailbox=mailbox,
sender=str(parsed.get("from") or "")[:500],