Files
VacatureRadar/apps/sources/adapters/email_alert.py
T
Jens b8091e59bd
deploy / deploy (push) Canceled after 0s
Initial deploy setup
2026-07-21 14:00:00 +02:00

101 lines
4.0 KiB
Python

from __future__ import annotations
import html
import re
from email import policy
from email.message import Message
from email.parser import BytesParser
from urllib.parse import urljoin, urlsplit
from bs4 import BeautifulSoup
from apps.sources.services.canonicalize import canonicalize_url
from .base import ExtractedJob, ExtractionResult, FieldEvidence
URL_RE = re.compile(r"https?://[^\s<>\"']+", re.I)
SKIP_TEXT = re.compile(r"unsubscribe|afmelden|uitschrijven|privacy|view in browser", re.I)
class EmailAlertAdapter:
parser_key = "email-alert"
parser_version = "1.0.0"
@staticmethod
def _decode_parts(message: Message) -> tuple[str, str]:
plain_parts: list[str] = []
html_parts: list[str] = []
for part in message.walk() if message.is_multipart() else [message]:
disposition = str(part.get("Content-Disposition") or "")
if "attachment" in disposition.lower():
continue
content_type = part.get_content_type()
try:
payload = part.get_content()
except Exception:
raw = part.get_payload(decode=True) or b""
payload = raw.decode(part.get_content_charset() or "utf-8", errors="replace")
if content_type == "text/plain":
plain_parts.append(str(payload))
elif content_type == "text/html":
html_parts.append(str(payload))
return "\n".join(plain_parts), "\n".join(html_parts)
def extract_message(self, raw_message: bytes) -> ExtractionResult:
message = BytesParser(policy=policy.default).parsebytes(raw_message)
plain, html_body = self._decode_parts(message)
candidates: list[tuple[str, str]] = []
if html_body:
soup = BeautifulSoup(html_body, "lxml")
for anchor in soup.find_all("a", href=True):
label = " ".join(anchor.get_text(" ", strip=True).split())
href = html.unescape(str(anchor["href"]).strip())
if not href.lower().startswith(("http://", "https://")):
continue
if SKIP_TEXT.search(label) or SKIP_TEXT.search(href):
continue
candidates.append((label, href))
for href in URL_RE.findall(plain):
clean_href = href.rstrip(".,);]")
if SKIP_TEXT.search(clean_href):
continue
candidates.append(("", clean_href))
jobs: list[ExtractedJob] = []
seen: set[str] = set()
subject = str(message.get("subject") or "Vacature uit e-mail").strip()
for label, href in candidates:
canonical = canonicalize_url(urljoin("https://invalid.local/", href))
if not canonical or canonical in seen:
continue
seen.add(canonical)
hostname = urlsplit(canonical).hostname or ""
title = label if len(label) >= 4 else subject
if len(title) > 300:
title = title[:300]
jobs.append(
ExtractedJob(
url=canonical,
title=title,
employer_name="",
description_text=plain[:5000]
or BeautifulSoup(html_body, "lxml").get_text("\n", strip=True)[:5000],
raw={
"email_subject": subject,
"email_sender": str(message.get("from") or ""),
"target_domain": hostname,
},
evidence=[FieldEvidence("url", "email-anchor", 0.75, label[:240])],
)
)
return ExtractionResult(
jobs,
self.parser_key,
self.parser_version,
0.65 if jobs else 0.0,
[] if jobs else ["Geen vacaturelinks in e-mail gevonden"],
)
def extract(self, content: str, *, url: str = "") -> ExtractionResult:
return self.extract_message(content.encode("utf-8", errors="replace"))