feat: restrict search to 100% remote jobs with hybrid/on-site filter

LinkedIn's f_WT=2 remote filter is leaky, so besides requesting only
remote work types, drop jobs whose card location says Hybrid/On-site
before detail fetches, and drop jobs whose description reveals hybrid
or in-office requirements after. Bare 'hybrid' is not matched in
descriptions since networking JDs legitimately say 'hybrid cloud'.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

config.yaml

@@ -47,7 +47,8 @@ scraper:
     - "Cloud Network Engineer"
     - "Systems Engineer"
   location: "Canada"
-  work_types: ["remote", "hybrid"]
+  work_types: ["remote"]
+  remote_only: true   # also drop jobs whose location/description reveals hybrid or on-site
   pages_per_keyword: 3
   delay_between_pages: [3, 6]
   delay_between_keywords: [5, 10]

daemon.py

@@ -226,6 +226,41 @@ class LinkedInJobScraper:
         return jobs
 
 
+# LinkedIn's f_WT=2 (remote) URL filter is leaky — hybrid/on-site posts slip
+# through. These patterns catch them from the card location and description.
+# "hybrid" alone is NOT matched in descriptions: networking JDs legitimately
+# say "hybrid cloud" / "hybrid infrastructure".
+NON_REMOTE_LOCATION_RE = re.compile(r"\b(hybrid|on[- ]?site)\b", re.I)
+NON_REMOTE_DESC_RE = re.compile(
+    r"hybrid\s+(work(ing)?|model|role|schedule|position|arrangement|setup)"
+    r"|\b(work|be)\s+(from|in)\s+(the\s+)?office"
+    r"|\bin[- ]office\b"
+    r"|\b\d+\s*(?:-\s*\d+\s*)?days?\s+(?:per\s+week\s+)?(?:in|at)\s+(?:the\s+)?office"
+    r"|this\s+(?:is\s+an?|role\s+is)\s+(?:hybrid|on[- ]?site)"
+    r"|must\s+(?:be\s+able\s+to\s+)?(?:commute|relocate)",
+    re.I,
+)
+
+
+def filter_remote_only(jobs, check_descriptions=False):
+    """Drop jobs that are visibly hybrid/on-site despite LinkedIn's remote filter."""
+    kept = []
+    for job in jobs:
+        if NON_REMOTE_LOCATION_RE.search(job.get("location") or ""):
+            logger.info(f"  🚫 Not remote (location): {job['title']} @ {job['company']} [{job['location']}]")
+            continue
+        if check_descriptions and job.get("description"):
+            hit = NON_REMOTE_DESC_RE.search(job["description"])
+            if hit:
+                logger.info(f"  🚫 Not remote (description: '{hit.group(0)}'): {job['title']} @ {job['company']}")
+                continue
+        kept.append(job)
+    dropped = len(jobs) - len(kept)
+    if dropped:
+        logger.info(f"Remote-only filter: {dropped} hybrid/on-site jobs dropped, {len(kept)} remaining.")
+    return kept
+
+
 def pre_filter_jobs(jobs, config):
     title_blacklist = [t.lower() for t in config.get("title_blacklist", [])]
     company_blacklist = [c.lower() for c in config.get("company_blacklist", [])]
@@ -478,9 +513,16 @@ def run_once(config, profile):
 
     jobs = pre_filter_jobs(jobs, config)
 
+    remote_only = scraper_cfg.get("remote_only", False)
+    if remote_only:
+        jobs = filter_remote_only(jobs)
+
     logger.info("📄 Fetching job descriptions...")
     jobs = scraper.fetch_descriptions_batch(jobs, max_fetches=max_details)
 
+    if remote_only:
+        jobs = filter_remote_only(jobs, check_descriptions=True)
+
     desc_count = sum(1 for j in jobs if j.get("description"))
     logger.info(f"📄 Jobs with descriptions: {desc_count}/{len(jobs)}")