Discovery & scoring fixes (location filter discarding all jobs, real company name, retryable scoring)

CHANGELOG.md

@@ -5,6 +5,17 @@ All notable changes to ApplyPilot will be documented in this file.
 The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.1.0/),
 and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
 
+## [Unreleased]
+
+### Fixed
+- Location filter no longer discards nearly every job — it reads the documented
+  `location.accept_patterns` schema and treats an empty accept list as "keep".
+- The real company name is stored and used in scoring, tailoring, cover letters,
+  the apply prompt, and the dashboard (was showing the job board, e.g. "linkedin").
+- Scoring failures (rate limits, parse errors) leave jobs pending for retry
+  instead of writing a permanent `fit_score=0`; scores commit incrementally so an
+  interrupt doesn't discard the run; markdown-decorated scores parse correctly.
+
 ## [0.2.0] - 2026-02-17
 
 ### Added

src/applypilot/apply/launcher.py

@@ -106,7 +106,7 @@ def acquire_job(target_url: str | None = None, min_score: int = 7,
         if target_url:
             like = f"%{target_url.split('?')[0].rstrip('/')}%"
             row = conn.execute("""
-                SELECT url, title, site, application_url, tailored_resume_path,
+                SELECT url, title, company, site, application_url, tailored_resume_path,
                        fit_score, location, full_description, cover_letter_path
                 FROM jobs
                 WHERE (url = ? OR application_url = ? OR application_url LIKE ? OR url LIKE ?)
@@ -128,7 +128,7 @@ def acquire_job(target_url: str | None = None, min_score: int = 7,
                 url_clauses = " ".join(f"AND url NOT LIKE ?" for _ in blocked_patterns)
                 params.extend(blocked_patterns)
             row = conn.execute(f"""
-                SELECT url, title, site, application_url, tailored_resume_path,
+                SELECT url, title, company, site, application_url, tailored_resume_path,
                        fit_score, location, full_description, cover_letter_path
                 FROM jobs
                 WHERE tailored_resume_path IS NOT NULL
@@ -350,7 +350,7 @@ def run_job(job: dict, port: int, worker_id: int = 0,
     worker_dir = reset_worker_dir(worker_id)
 
     update_state(worker_id, status="applying", job_title=job["title"],
-                 company=job.get("site", ""), score=job.get("fit_score", 0),
+                 company=job.get("company") or job.get("site", ""), score=job.get("fit_score", 0),
                  start_time=time.time(), actions=0, last_action="starting")
     add_event(f"[W{worker_id}] Starting: {job['title'][:40]} @ {job.get('site', '')}")
 

src/applypilot/apply/prompt.py

@@ -518,7 +518,7 @@ def build_prompt(job: dict, tailored_resume: str,
 == JOB ==
 URL: {job.get('application_url') or job['url']}
 Title: {job['title']}
-Company: {job.get('site', 'Unknown')}
+Company: {job.get('company') or job.get('site', 'Unknown')}
 Fit Score: {job.get('fit_score', 'N/A')}/10
 
 == FILES ==

src/applypilot/database.py

@@ -92,6 +92,7 @@ def init_db(db_path: Path | str | None = None) -> sqlite3.Connection:
             -- Discovery stage (smart_extract / job_search)
             url                   TEXT PRIMARY KEY,
             title                 TEXT,
+            company               TEXT,
             salary                TEXT,
             description           TEXT,
             location              TEXT,
@@ -147,6 +148,7 @@ def init_db(db_path: Path | str | None = None) -> sqlite3.Connection:
     # Discovery
     "url": "TEXT PRIMARY KEY",
     "title": "TEXT",
+    "company": "TEXT",
     "salary": "TEXT",
     "description": "TEXT",
     "location": "TEXT",
@@ -349,9 +351,9 @@ def store_jobs(conn: sqlite3.Connection, jobs: list[dict],
             continue
         try:
             conn.execute(
-                "INSERT INTO jobs (url, title, salary, description, location, site, strategy, discovered_at) "
-                "VALUES (?, ?, ?, ?, ?, ?, ?, ?)",
-                (url, job.get("title"), job.get("salary"), job.get("description"),
+                "INSERT INTO jobs (url, title, company, salary, description, location, site, strategy, discovered_at) "
+                "VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?)",
+                (url, job.get("title"), job.get("company"), job.get("salary"), job.get("description"),
                  job.get("location"), site, strategy, now),
             )
             new += 1

src/applypilot/discovery/jobspy.py

@@ -16,6 +16,7 @@
 
 from applypilot import config
 from applypilot.database import get_connection, init_db, store_jobs
+from applypilot.locfilter import load_location_filter, location_ok
 
 log = logging.getLogger(__name__)
 
@@ -75,44 +76,10 @@ def _scrape_with_retry(kwargs: dict, max_retries: int = 2, backoff: float = 5.0)
 
 
 # -- Location filtering ------------------------------------------------------
-
-def _load_location_config(search_cfg: dict) -> tuple[list[str], list[str]]:
-    """Extract accept/reject location lists from search config.
-
-    Falls back to sensible defaults if not defined in the YAML.
-    """
-    accept = search_cfg.get("location_accept", [])
-    reject = search_cfg.get("location_reject_non_remote", [])
-    return accept, reject
-
-
-def _location_ok(location: str | None, accept: list[str], reject: list[str]) -> bool:
-    """Check if a job location passes the user's location filter.
-
-    Remote jobs are always accepted. Non-remote jobs must match an accept
-    pattern and not match a reject pattern.
-    """
-    if not location:
-        return True  # unknown location -- keep it, let scorer decide
-
-    loc = location.lower()
-
-    # Remote jobs always OK
-    if any(r in loc for r in ("remote", "anywhere", "work from home", "wfh", "distributed")):
-        return True
-
-    # Reject non-remote matches
-    for r in reject:
-        if r.lower() in loc:
-            return False
-
-    # Accept matches
-    for a in accept:
-        if a.lower() in loc:
-            return True
-
-    # No match -- reject unknown
-    return False
+# Canonical implementation lives in applypilot.locfilter; these aliases keep the
+# existing call sites unchanged.
+_load_location_config = load_location_filter
+_location_ok = location_ok
 
 
 # -- DB storage (JobSpy DataFrame -> SQLite) ---------------------------------
@@ -168,10 +135,10 @@ def store_jobspy_results(conn: sqlite3.Connection, df, source_label: str) -> tup
 
         try:
             conn.execute(
-                "INSERT INTO jobs (url, title, salary, description, location, site, strategy, discovered_at, "
+                "INSERT INTO jobs (url, title, company, salary, description, location, site, strategy, discovered_at, "
                 "full_description, application_url, detail_scraped_at) "
-                "VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?)",
-                (url, title, salary, description, location_str, site_label, strategy, now,
+                "VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?)",
+                (url, title, company, salary, description, location_str, site_label, strategy, now,
                  full_description, apply_url, detail_scraped_at),
             )
             new += 1

src/applypilot/discovery/smartextract.py

@@ -32,6 +32,7 @@
 from applypilot.config import CONFIG_DIR
 from applypilot.database import get_connection, init_db, store_jobs, get_stats
 from applypilot.llm import get_client
+from applypilot.locfilter import load_location_filter, location_ok as _location_ok
 
 log = logging.getLogger(__name__)
 
@@ -49,28 +50,13 @@
 # -- Location filtering -------------------------------------------------------
 
 def _load_location_filter(search_cfg: dict | None = None):
-    """Load location accept/reject lists from search config."""
+    """Load location accept/reject lists, defaulting to the user's search config.
+
+    Delegates the schema parsing to applypilot.locfilter.
+    """
     if search_cfg is None:
         search_cfg = config.load_search_config()
-    accept = search_cfg.get("location_accept", [])
-    reject = search_cfg.get("location_reject_non_remote", [])
-    return accept, reject
-
-
-def _location_ok(location: str | None, accept: list[str], reject: list[str]) -> bool:
-    """Check if a job location passes the user's location filter."""
-    if not location:
-        return True
-    loc = location.lower()
-    if any(r in loc for r in ("remote", "anywhere", "work from home", "wfh", "distributed")):
-        return True
-    for r in reject:
-        if r.lower() in loc:
-            return False
-    for a in accept:
-        if a.lower() in loc:
-            return True
-    return False
+    return load_location_filter(search_cfg)
 
 
 # -- Site configuration from YAML --------------------------------------------
@@ -107,10 +93,13 @@ def _store_jobs_filtered(
             filtered += 1
             continue
         try:
+            # Prefer a parsed company; fall back to the site name (direct career
+            # sites are themselves the employer).
+            company = job.get("company") or site
             conn.execute(
-                "INSERT INTO jobs (url, title, salary, description, location, site, strategy, discovered_at) "
-                "VALUES (?, ?, ?, ?, ?, ?, ?, ?)",
-                (url, job.get("title"), job.get("salary"), job.get("description"),
+                "INSERT INTO jobs (url, title, company, salary, description, location, site, strategy, discovered_at) "
+                "VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?)",
+                (url, job.get("title"), company, job.get("salary"), job.get("description"),
                  job.get("location"), site, strategy, now),
             )
             new += 1

src/applypilot/discovery/workday.py

@@ -22,6 +22,7 @@
 from applypilot import config
 from applypilot.config import CONFIG_DIR
 from applypilot.database import get_connection, init_db
+from applypilot.locfilter import load_location_filter, location_ok as _location_ok
 
 log = logging.getLogger(__name__)
 
@@ -41,34 +42,13 @@ def load_employers() -> dict:
 # -- Location filtering from search config -----------------------------------
 
 def _load_location_filter(search_cfg: dict | None = None):
-    """Load location accept/reject lists from search config."""
+    """Load location accept/reject lists, defaulting to the user's search config.
+
+    Delegates the schema parsing to applypilot.locfilter.
+    """
     if search_cfg is None:
         search_cfg = config.load_search_config()
-
-    accept = search_cfg.get("location_accept", [])
-    reject = search_cfg.get("location_reject_non_remote", [])
-    return accept, reject
-
-
-def _location_ok(location: str | None, accept: list[str], reject: list[str]) -> bool:
-    """Check if a job location passes the user's location filter."""
-    if not location:
-        return True
-
-    loc = location.lower()
-
-    if any(r in loc for r in ("remote", "anywhere", "work from home", "wfh", "distributed")):
-        return True
-
-    for r in reject:
-        if r.lower() in loc:
-            return False
-
-    for a in accept:
-        if a.lower() in loc:
-            return True
-
-    return False
+    return load_location_filter(search_cfg)
 
 
 # -- HTML stripper -----------------------------------------------------------
@@ -322,14 +302,15 @@ def store_results(conn: sqlite3.Connection, jobs: list[dict], employers: dict) -
         detail_error = job.get("detail_error")
 
         site = job.get("employer_name", "Corporate")
+        company = job.get("employer_name")  # Workday portals are the employer
         strategy = "workday_api"
 
         try:
             conn.execute(
-                "INSERT INTO jobs (url, title, salary, description, location, site, strategy, "
+                "INSERT INTO jobs (url, title, company, salary, description, location, site, strategy, "
                 "discovered_at, full_description, application_url, detail_scraped_at, detail_error) "
-                "VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?)",
-                (url, job.get("title"), None, short_desc, job.get("location"),
+                "VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?)",
+                (url, job.get("title"), company, None, short_desc, job.get("location"),
                  site, strategy, now, full_description, url, detail_scraped_at, detail_error),
             )
             new += 1

src/applypilot/locfilter.py

@@ -0,0 +1,54 @@
+"""Canonical job-location filtering shared by all discovery sources.
+
+Historically each discovery module carried its own copy of this logic and read
+config keys (``location_accept`` / ``location_reject_non_remote``) that nothing
+ever wrote -- the shipped config uses ``location.accept_patterns`` /
+``location.reject_patterns``. With empty lists the old code rejected every
+non-remote job, silently discarding almost all results. This module reads both
+schemas and treats an empty accept list as "accept everything not rejected".
+"""
+from __future__ import annotations
+
+_REMOTE_MARKERS = ("remote", "anywhere", "work from home", "wfh", "distributed")
+
+
+def load_location_filter(search_cfg: dict | None) -> tuple[list[str], list[str]]:
+    """Read accept/reject location patterns from a search config dict.
+
+    Supports the current ``location: {accept_patterns, reject_patterns}`` schema
+    and the legacy flat ``location_accept`` / ``location_reject_non_remote`` keys.
+    """
+    cfg = search_cfg or {}
+    loc = cfg.get("location", {}) or {}
+    accept = loc.get("accept_patterns") or cfg.get("location_accept", []) or []
+    reject = loc.get("reject_patterns") or cfg.get("location_reject_non_remote", []) or []
+    return accept, reject
+
+
+def location_ok(location: str | None, accept: list[str], reject: list[str]) -> bool:
+    """Decide whether a job location passes the filter.
+
+    Remote locations are always accepted. A reject-pattern match always fails.
+    Empty accept list = accept everything not explicitly rejected. A non-empty
+    accept list is exclusive: a non-remote location must match it to pass.
+    """
+    if not location:
+        return True  # unknown location -- keep it, let the scorer decide
+
+    loc = location.lower()
+
+    if any(marker in loc for marker in _REMOTE_MARKERS):
+        return True
+
+    for r in reject:
+        if r.lower() in loc:
+            return False
+
+    if not accept:
+        return True  # nothing explicitly rejected and no accept list to enforce
+
+    for a in accept:
+        if a.lower() in loc:
+            return True
+
+    return False

src/applypilot/scoring/cover_letter.py

@@ -138,7 +138,7 @@ def generate_cover_letter(
     """
     job_text = (
         f"TITLE: {job['title']}\n"
-        f"COMPANY: {job['site']}\n"
+        f"COMPANY: {job.get('company') or job['site']}\n"
         f"LOCATION: {job.get('location', 'N/A')}\n\n"
         f"DESCRIPTION:\n{(job.get('full_description') or '')[:6000]}"
     )