remove caching

This commit is contained in:
georg.sinn-schirwitz
2025-09-08 14:44:46 +02:00
parent f8e23d0fba
commit 042a196718
13 changed files with 144 additions and 525 deletions

View File

@@ -1,5 +1,5 @@
import os
from flask import Flask, request, jsonify, render_template, redirect, url_for, session, flash, send_file
from flask import Flask, request, jsonify, render_template, redirect, url_for, session, flash, Response
from flask_wtf import CSRFProtect
from typing import Dict, List
@@ -34,7 +34,6 @@ from web.utils import (
initialize_users_from_settings,
filter_jobs,
get_job_by_id,
get_cache_dir,
)
from web.db import get_all_regions, get_all_keywords
@@ -232,39 +231,6 @@ def job_by_id(job_id):
return jsonify({"error": "Job not found"}), 404
@app.route('/cached/<job_id>', methods=['GET'])
def serve_cached(job_id):
"""Serve the cached HTML file for a job if available.
Uses the job record's `file_path_abs` when present, or resolves the DB `file_path` via helper.
Ensures the returned file is located under the configured cache directory to avoid path-traversal.
"""
try:
from web.db import db_get_cached_abs_path
j = get_job_by_id(job_id)
if not j:
return "Job not found", 404
# Prefer file_path_abs, fall back to resolving the DB-stored file_path
abs_fp = j.get('file_path_abs') or None
if not abs_fp:
db_fp = j.get('file_path')
abs_fp = db_get_cached_abs_path(db_fp) if db_fp else None
if not abs_fp or not os.path.isfile(abs_fp):
return "Cached file not available", 404
cache_dir = os.path.abspath(get_cache_dir())
abs_fp = os.path.abspath(abs_fp)
# Ensure the file is inside the cache directory
if os.path.commonpath([cache_dir, abs_fp]) != cache_dir:
return "Forbidden", 403
return send_file(abs_fp)
except Exception:
return "Error serving cached file", 500
@app.route('/jobs/<job_id>/favorite', methods=['POST'])
def set_favorite(job_id):
"""Mark or unmark a job as favorite for a given user.
@@ -290,9 +256,21 @@ csrf.exempt(set_favorite)
@app.route('/scrape', methods=['GET'])
def scrape():
"""Trigger the web scraping process."""
scraper()
return jsonify({"status": "Scraping completed"})
"""Trigger the web scraping process with streaming output."""
def generate():
try:
for message in scraper():
yield message
except Exception as e:
yield f"Error during scraping: {str(e)}\n"
return Response(generate(), mimetype='text/plain')
@app.route('/scrape-page', methods=['GET'])
def scrape_page():
"""Serve the scrape page with streaming output display."""
return render_template('scrape.html', title='Scrape Jobs')
# ---------------- Auth & Admin UI ------------------------------------------