Initial commit

tests/test_bill_id.py

@@ -0,0 +1,151 @@
+"""Unit tests for scraper.bill_id -- the canonical bill_id codec (spec section 4).
+
+These tests are the source of truth for the codec the parser depends on, so they
+run first and are the most brittle-to-change. Any change to the codec that
+breaks any of these must be a conscious decision, not an accidental refactor.
+"""
+from __future__ import annotations
+
+import pytest
+
+from scraper.bill_id import (
+     normalise_session,
+     origin_house,
+     parse_bill_id,
+     parse_display,
+     parse_supplement,
+)
+
+
+# ---------------------------------------------------------------- parse_bill_id
+
+class TestParseBillId:
+    def test_assembly_bill_ab28(self):
+        dec = parse_bill_id("202520260AB28")
+        assert dec is not None
+        assert dec["session"] == "2025-2026"
+        assert dec["session_raw"] == "20252026"
+        assert dec["chamber"] == "AB"
+        assert dec["number"] == 28
+        assert dec["display"] == "AB 28"
+        assert dec["house"] == "assembly"
+        assert dec["raw"] == "202520260AB28"
+
+    def test_senate_bill_sb795(self):
+        dec = parse_bill_id("202520260SB795")
+        assert dec is not None
+        assert dec["chamber"] == "SB"
+        assert dec["session"] == "2025-2026"
+        assert dec["number"] == 795
+        assert dec["display"] == "SB 795"
+        assert dec["house"] == "senate"
+
+    def test_assembly_prefixes(self):
+        for raw, chamber, house in [
+            ("202520260ACR5", "ACR", "assembly"),
+            ("202520260HR7", "HR", "assembly"),
+            ("202520260ABR1", "ABR", "assembly"),
+        ]:
+            dec = parse_bill_id(raw)
+            assert dec is not None
+            assert dec["chamber"] == chamber, raw
+            assert dec["house"] == house, raw
+
+    def test_senate_prefixes(self):
+        for raw, chamber in [
+            ("202520260SCR12", "SCR"),
+            ("202520260SJR3", "SJR"),
+        ]:
+            dec = parse_bill_id(raw)
+            assert dec is not None
+            assert dec["chamber"] == chamber, raw
+            assert dec["house"] == "senate", raw
+
+    def test_invalid_inputs_return_none(self):
+        for bad in [
+             "",
+             "garbage",
+             "202520260",        # no chamber/number tail
+             "202520260AB",      # chamber with no number
+             "202520260AB28JUNK",# trailing junk
+             "20250AB28",        # wrong session length
+             "202520260XY12",    # unknown chamber prefix
+        ]:
+            assert parse_bill_id(bad) is None, f"expected None for {bad!r}"
+
+    def test_whitespace_is_stripped(self):
+        dec = parse_bill_id("    202520260AB28    ")
+        assert dec is not None
+        assert dec["display"] == "AB 28"
+        assert dec["raw"] == "    202520260AB28    "
+
+    def test_session_decimal_boundary(self):
+        dec = parse_bill_id("202520260AB40")
+        assert dec is not None
+        assert dec["number"] == 40
+        assert dec["display"] == "AB 40"
+
+
+# ---------------------------------------------------------------- normalise_session
+
+class TestNormaliseSession:
+    def test_standard_session(self):
+        assert normalise_session("20252026") == "2025-2026"
+
+    def test_empty(self):
+        assert normalise_session("") == ""
+
+    def test_non_session_passthrough(self):
+        assert normalise_session("garbage") == "garbage"
+
+    def test_four_digit_passthrough(self):
+        assert normalise_session("2025") == "2025"
+
+
+# ---------------------------------------------------------------- origin_house
+
+class TestOriginHouse:
+    def test_assembly_prefixes(self):
+        for chamber in ("AB", "ACR", "HR", "ABR"):
+            assert origin_house(chamber) == "assembly", chamber
+
+    def test_senate_prefixes(self):
+        for chamber in ("SB", "SCR", "SJR"):
+            assert origin_house(chamber) == "senate", chamber
+
+    def test_unknown(self):
+        assert origin_house("ZZ") == "unknown"
+        assert origin_house("") == "unknown"
+
+
+# ---------------------------------------------------------------- parse_display
+
+class TestParseDisplay:
+    def test_padded(self):
+        assert parse_display("AB 00028") == ("AB", 28)
+
+    def test_unpadded(self):
+        assert parse_display("AB 28") == ("AB", 28)
+
+    def test_various_prefixes(self):
+        assert parse_display("SB 795") == ("SB", 795)
+        assert parse_display("ACR 5") == ("ACR", 5)
+        assert parse_display("SCR 12") == ("SCR", 12)
+
+    def test_invalid(self):
+        for garbage in ("", "SB", "AB 7 X", "XYZ 5", "   "):
+            assert parse_display(garbage) is None, garbage
+
+
+# ---------------------------------------------------------------- parse_supplement
+
+class TestParseSupplement:
+    def test_standard(self):
+        assert parse_supplement("Supplemental 2") == 2
+
+    def test_case_insensitive(self):
+        assert parse_supplement("supplemental 1") == 1
+
+    def test_no_match(self):
+        for text in ("", "plain text", "Supplemental", "Supp 3"):
+            assert parse_supplement(text) is None, text