Update the industry classification

data/by_ticker/part-00000.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:831cbc7123530308eb4de065ebdf0c8e2ce18b4ac67be041f51368f71db334aa
+oid sha256:549f85cfe3ef4025adb81dbe9f2c5b75a6d7befdd9d0b46aa8353115ec67baaf
 size 727024

data/classification/part-00000.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:724491615796030c045ab30a070eb5dd9519564c17b060d25724fe905ecebe09
+oid sha256:43772ae97083bd4586d02bc44b18eb16033d1cdfeb30fe3e20b4ef39e4d8daec
 size 830399

data/naics_map/part-00000.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:73683e3d179237e97f3dd0553dac85c261106ab8d5e0e81d020da12ea2851ef0
+oid sha256:eb8a74e0bdb486e698d99925da16818ba7772ee662935db6dcfbb8e12c727217
 size 67201

data/observations/part-00000.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:4d08effad90c3f6de818a6c1dda5b9fc96e97eaf2fc0fd57da0eb28425f03706
+oid sha256:75da0ea036cb36c572ad2ebba6b8997b273465d49bb052ad53155a1e1e6ca7c2
 size 6255799

data/sic_codes/part-00000.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:950199b74a4376da1f722fb692f9f8254b0562cb9585948df970b9d44cfb8af4
+oid sha256:c9005bb8f4f2f496d1e8cdb6d1be4272220f4575d432f7497ce406e7fb67a957
 size 29383

data/spans/part-00000.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:b8276701bffbdc4a806a6a719788f2cfcbb28de837d912ffad8fb0cbbe26e0a6
+oid sha256:5acee064ee42b65f13d97166246b25e44ac39e34b27ce592d1ffb1a1c0520735
 size 270249

data/taxonomy/part-00000.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:ace4f48cf120b06ec15cca0987f163f2258e78af624f3c53878b27eff46f730a
+oid sha256:b02de6ba1db8afb9bac035f6200ed6c0ae28dcdfe1b9583101a4828b68a1fafa
 size 6420

jobs/run.py

@@ -48,7 +48,13 @@ def restore_repository(repo_id: str, token: str | None, workspace: Path) -> Path
             # Only the recipe and tests are needed: every table is rebuilt from
             # the sources each run, so restoring the data would download tens of
             # megabytes to overwrite them.
-            allow_patterns=["recipe/**", "tests/**", "jobs/**", "*.md", "*.json", "*.toml"],
+            # LICENSE and NOTICE have no extension, so they need naming: the
+            # tests read NOTICE, and a restore that skipped it failed the job
+            # before anything was fetched.
+            allow_patterns=[
+                "recipe/**", "tests/**", "jobs/**",
+                "*.md", "*.json", "*.toml", "LICENSE", "NOTICE",
+            ],
         )
         log(f"restored {repo_id} into {workspace}")
     except RepositoryNotFoundError:

manifest.json

@@ -163,7 +163,7 @@
     "max_statement": "2026-10-13",
     "min_statement": "2009-04-15"
   },
-  "generated_at": "2026-10-13T19:43:19.146667+00:00",
+  "generated_at": "2026-10-13T20:02:11.096197+00:00",
   "industries": {
     "CONC_AUTO": {
       "sector": "CONC",
@@ -548,7 +548,7 @@
   "source": {
     "build_version": "1.0.0",
     "package_version": "1.0.0",
-    "recipe_hash": "sha256:675e969a373cd6ec1fad32888bf6c51ecc4c536cd8164ec4106c5cec24cd989a",
+    "recipe_hash": "sha256:ed2bf8d7a99d9f87ff829fcb67f9908fa595d4b0586f937a72c1d36478b2eaa6",
     "taxonomy_version": "1.0.0"
   }
 }

recipe/sources.py

@@ -459,8 +459,38 @@ async def fetch_concordance(cache_dir: Path) -> bytes:
     return payload
 
 
+CONCORDANCE_COLUMNS = ("sic", "part", "title", "naics", "naics_title")
+
+
+def read_concordance_sheet(payload: bytes) -> pl.DataFrame:
+    """The workbook's single sheet, every cell a string.
+
+    A 2008-vintage .xls, which only a BIFF-capable reader opens; polars
+    delegates to one. Kept apart from the parsing below so that the parsing
+    can be tested without a spreadsheet writer in the loop -- the first
+    version built a workbook in its fixtures and passed locally only because
+    `xlsxwriter` happened to be installed, then failed in the job.
+    """
+    import io
+
+    frame = pl.read_excel(io.BytesIO(payload), has_header=True, infer_schema_length=0)
+    if frame.width < len(CONCORDANCE_COLUMNS):
+        raise RuntimeError(
+            f"concordance has {frame.width} columns, expected {len(CONCORDANCE_COLUMNS)}"
+        )
+    frame.columns = list(CONCORDANCE_COLUMNS)[: frame.width]
+    return frame
+
+
 def parse_concordance(payload: bytes) -> tuple[pl.DataFrame, dict[str, str], list[str]]:
-    """The crosswalk, the major group titles, and the division titles in order.
+    """The crosswalk, the major group titles, and the division titles in order."""
+    return parse_concordance_frame(read_concordance_sheet(payload))
+
+
+def parse_concordance_frame(
+    frame: pl.DataFrame,
+) -> tuple[pl.DataFrame, dict[str, str], list[str]]:
+    """Separate the hierarchy from the crosswalk.
 
     One file answers three questions because of how it is laid out: a row with
     no SIC is a division heading, a two-digit SIC is a major group heading, and
@@ -469,12 +499,6 @@ def parse_concordance(payload: bytes) -> tuple[pl.DataFrame, dict[str, str], lis
     the agency that maintains the classification rather than being typed in
     here from a manual.
     """
-    import io
-
-    frame = pl.read_excel(io.BytesIO(payload), has_header=True, infer_schema_length=0)
-    if frame.width < 5:
-        raise RuntimeError(f"concordance has {frame.width} columns, expected 5")
-    frame.columns = ["sic", "part", "title", "naics", "naics_title"][: frame.width]
     frame = frame.with_columns(pl.col("sic").str.strip_chars())
 
     divisions = [
@@ -531,8 +555,10 @@ __all__ = [
     "from_fundamental_filings",
     "issuers",
     "parse_concordance",
+    "parse_concordance_frame",
     "parse_sic_list",
     "read_bulk_sic",
+    "read_concordance_sheet",
     "sibling_frame",
     "sibling_path",
     "submissions_snapshot",

tests/test_sources.py

@@ -5,7 +5,7 @@ from __future__ import annotations
 import polars as pl
 
 from recipe.sic import DIVISIONS
-from recipe.sources import parse_concordance
+from recipe.sources import parse_concordance_frame
 
 
 class _FakePage:
@@ -48,17 +48,21 @@ async def test_sic_list_refuses_a_page_that_lost_its_table():
         raise AssertionError("a three-row page should not have been accepted")
 
 
+# The sheet as the reader hands it over: every cell a string, columns already
+# named. Built directly rather than round-tripped through a spreadsheet, so
+# the tests need no Excel writer -- the first version did, and passed here
+# while failing in the job, where `xlsxwriter` is not installed.
 CONCORDANCE = pl.DataFrame(
     {
-        "SIC": [None, "28", "2834", "2836", "2836", None, "99", "9999"],
-        "Part Indicator": [None, None, None, "pt", "pt", None, None, None],
-        "SIC Titles and Part Descriptions": [
+        "sic": [None, "28", "2834", "2836", "2836", None, "99", "9999"],
+        "part": [None, None, None, "pt", "pt", None, None, None],
+        "title": [
             "Manufacturing", "Chemicals and allied products", "Pharmaceutical Preparations",
             ".   Vaccines", ".   Other", "Nonclassifiable establishments",
             "Nonclassifiable", "Nonclassifiable Establishments",
         ],
-        "1997 NAICS": [None, None, "325412", "325414", "325411", None, None, "99999"],
-        "1997 NAICS Titles and Part Indicators": [
+        "naics": [None, None, "325412", "325414", "325411", None, None, "99999"],
+        "naics_title": [
             None, None, "Pharmaceutical Preparation Manufacturing",
             "Biological Product Manufacturing", "Medicinal and Botanical Manufacturing",
             None, None, "Unclassified Establishments",
@@ -67,16 +71,8 @@ CONCORDANCE = pl.DataFrame(
 )
 
 
-def _workbook() -> bytes:
-    import io
-
-    buffer = io.BytesIO()
-    CONCORDANCE.write_excel(buffer)
-    return buffer.getvalue()
-
-
 def test_concordance_separates_hierarchy_from_crosswalk():
-    crosswalk, major_groups, divisions = parse_concordance(_workbook())
+    crosswalk, major_groups, divisions = parse_concordance_frame(CONCORDANCE)
     assert divisions == ["Manufacturing", "Nonclassifiable establishments"]
     assert major_groups["28"] == "Chemicals and allied products"
     # Supplied here because the Census file never prints a header row for them.
@@ -85,7 +81,7 @@ def test_concordance_separates_hierarchy_from_crosswalk():
 
 
 def test_a_split_industry_keeps_every_candidate():
-    crosswalk, _, _ = parse_concordance(_workbook())
+    crosswalk, _, _ = parse_concordance_frame(CONCORDANCE)
     split = crosswalk.filter(pl.col("sic") == "2836")
     assert split.height == 2
     assert split["is_part"].to_list() == [True, True]
@@ -104,8 +100,8 @@ def test_division_ranges_are_ordered_and_do_not_overlap():
         previous_high = high
 
 
-def test_concordance_divisions_match_the_declared_count():
-    _, _, divisions = parse_concordance(_workbook())
+def test_concordance_divisions_are_distinct():
+    _, _, divisions = parse_concordance_frame(CONCORDANCE)
     assert len(set(divisions)) == len(divisions)
     assert divisions[0] == "Manufacturing"