From 5d9e0977f01465166f25abf0826bd61e7c8cac25 Mon Sep 17 00:00:00 2001
From: Mads Lorentzen <50207393+MadsLorentzen@users.noreply.github.com>
Date: Sat, 17 Oct 2026 20:12:49 +0200
Subject: [PATCH] fix(salary): convert the IDA workbook again ("Firmanavn og
 by" header) and list `unreachable` in the scrape schema (#615)

"firma" joins COMPOUND_PATTERNS so the glued Danish header "Firmanavn og by"
matches again; whole-token matching since #75 skipped both sheets of the
repository's own demonstration workbook with "Could not find header row".
Every other workbook converts byte-identically. Also folds the `unreachable`
status from #564 into the seen_jobs.json schema enum in the scrape spec.

Co-authored-by: Claude Fable 5.1 <noreply@anthropic.com>
---
 .claude/skills/job-scraper/SKILL.md |  2 +-
 CHANGELOG.md                        | 11 +++++++++++
 tests/test_convert_salary_excel.py  | 17 +++++++++++++++++
 tools/convert_salary_excel.py       |  5 +++--
 4 files changed, 32 insertions(+), 3 deletions(-)

diff --git a/.claude/skills/job-scraper/SKILL.md b/.claude/skills/job-scraper/SKILL.md
index 1dbcf671..297e1092 100644
--- a/.claude/skills/job-scraper/SKILL.md
+++ b/.claude/skills/job-scraper/SKILL.md
@@ -161,7 +161,7 @@ It prints one line: the canonical key for that posting. The key must be a pure f
       "posted_date": "YYYY-MM-DD" | null,
       "deadline": "YYYY-MM-DD" | null,
       "fit": "high/medium/low",
-      "status": "new/skipped/ranked/expired",
+      "status": "new/skipped/ranked/expired/unreachable",
       "portal": "<source portal skill, e.g. jobindex-search>",
       "source": "cli/websearch"
     }
diff --git a/CHANGELOG.md b/CHANGELOG.md
index fe3e60ab..a8783729 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -68,6 +68,17 @@ per-file diff commands.
 
 ### Fixed
 
+- **`convert_salary_excel.py` converts the IDA workbook again** (#615, `tools/convert_salary_excel.py`,
+  `tests/test_convert_salary_excel.py`) - the workbook heads its company column "Firmanavn og by",
+  and since #75 made header matching whole-token only, "firma" no longer matched inside "Firmanavn",
+  so both sheets were skipped with "Could not find header row" and the converter exited 1 on the
+  repository's own demonstration data. "firma" is now a compound token, like the other glued Danish
+  headers; every other workbook converts byte-identically.
+
+- **`/scrape`'s entry schema lists the `unreachable` status** (#615, `.claude/skills/job-scraper/SKILL.md`) -
+  follow-up to #564: the `status` enum in the `seen_jobs.json` schema block still read
+  `new/skipped/ranked/expired` while the dedup rule two sections down already named `unreachable`.
+
 - **`/scrape` names `job_scraper/seen_jobs.json` by its full path in every step** (#608,
   `.claude/skills/job-scraper/SKILL.md`, `tests/test_scrape_seen_jobs_path.py`) - seven references
   after Step 0, including the Step 4 write instruction, used the bare filename, and an assistant
diff --git a/tests/test_convert_salary_excel.py b/tests/test_convert_salary_excel.py
index 08122577..5518e541 100644
--- a/tests/test_convert_salary_excel.py
+++ b/tests/test_convert_salary_excel.py
@@ -7,6 +7,7 @@ from salary_lookup import format_entry
 from tools.convert_salary_excel import (
     INDEX_PATTERNS,
     detect_column_type,
+    COMPANY_PATTERNS,
     header_matches,
     parse_numeric_cell,
     parse_sheet,
@@ -108,6 +109,22 @@ class DetectColumnTypeTests(unittest.TestCase):
                     companies[0]["categories"]["salary"], {"index": 105.5}
                 )
 
+    def test_parse_sheet_detects_danish_compound_company_header(self):
+        # The IDA workbook heads its company column "Firmanavn og by": "firma"
+        # glued to "navn". Whole-token matching alone skipped the entire sheet
+        # ("Could not find header row") from #75 until "firma" joined
+        # COMPOUND_PATTERNS, so the shipped demonstration data no longer
+        # converted at all.
+        self.assertTrue(header_matches("Firmanavn og by", COMPANY_PATTERNS))
+        ws = FakeWorksheet([
+            ("Firmanavn og by", "Antal", "Indeks"),
+            ("Example A/S, Aarhus", 12, 105.0),
+        ])
+        companies = parse_sheet(ws)
+        self.assertEqual(len(companies), 1)
+        self.assertEqual(companies[0]["company"], "Example A/S, Aarhus")
+        self.assertEqual(companies[0]["categories"]["all_employees"], {"count": 12, "index": 105.0})
+
     def test_parse_sheet_detects_city_column_with_token_header(self):
         # City headers are matched with the same token-based header_matches()
         # used for the company column, not exact string equality. Real-world
diff --git a/tools/convert_salary_excel.py b/tools/convert_salary_excel.py
index 4cf58b30..f6e521b3 100644
--- a/tools/convert_salary_excel.py
+++ b/tools/convert_salary_excel.py
@@ -46,8 +46,9 @@ INDEX_PATTERNS = {"indeks", "index", "idx", "salary", "løn", "median", "average
 # header token, for languages that glue words together (e.g. Danish "lønindeks"
 # -> løn + indeks). Languages that write headers as separate words need none.
 # Ships populated for this repo's Danish demonstration data; a fork targeting
-# another locale edits this constant.
-COMPOUND_PATTERNS = {"antal", "indeks", "løn", "gennemsnit", "medarbejdere"}
+# another locale edits this constant. "firma" is here because the IDA workbook
+# heads its company column "Firmanavn og by" (firma + navn glued together).
+COMPOUND_PATTERNS = {"firma", "antal", "indeks", "løn", "gennemsnit", "medarbejdere"}
 # Identifier columns (employee id, Danish "personnummer", etc.) are never salary
 # data. They are dropped at classification so they are not mistaken for a salary
 # category. Matched as whole tokens only, like other pattern sets.