fix(hy3): tolerate malformed relay F3 legs and orphaned H1 DQ lines

hytek_parser/hy3/enums.py

@@ -23,6 +23,18 @@ class Stroke(Enum):
     BREASTSTROKE = "C", "3", 3
     BUTTERFLY = "D", "4", 4
     MEDLEY = "E", "5", 5
+    # Diving. Hy-Tek encodes the BOARD in the stroke column:
+    #   F = 1-metre springboard, G = 3-metre springboard, H = platform.
+    # Verified on a multi-conference corpus: the chars nest strictly
+    # (F, then F+G, then F+G+H -- never G without F, never H without G),
+    # carry the same dive count within a meet, and their median scores
+    # ascend with degree of difficulty.
+    # Without these members all three fall through select_from_enum() to
+    # UNKNOWN, which is the catch-all for any unrecognized byte -- making
+    # diving indistinguishable from file corruption.
+    DIVING_1M = "F"
+    DIVING_3M = "G"
+    DIVING_PLATFORM = "H"
 
     UNKNOWN = "U", "0", 0
 

hytek_parser/hy3/line_parsers/f_relay_parsers.py

@@ -172,7 +172,14 @@ def f3_parser(
             # Out of swimmers
             break
 
-        swimmer = file.meet.swimmers[swimmer_meet_id]
+        swimmer = file.meet.swimmers.get(swimmer_meet_id)
+        if swimmer is None:
+            # The F3 leg references a swimmer meet-id with no D1 roster record in
+            # this file — seen in some incomplete meet exports (e.g. relay legs for
+            # athletes the roster section omits). Skip the leg rather than raising a
+            # KeyError; the relay keeps its other legs. Mirrors the tolerance the
+            # empty-swimmers and absent-leg-1 cases already get below.
+            continue
         swimmer_leg = safe_cast(int, extract(line, 15 + offset, 1))
 
         # Hy-Tek encodes legs 1..8; preserve the leg number as-is.

hytek_parser/hy3/line_parsers/h_dq_parsers.py

@@ -17,9 +17,13 @@ def h1_parser(
     dq_code = select_from_enum(DisqualificationCode, extract(line, 3, 2))
     dq_info = extract(line, 5, 124)  # Whitespace is stripped
 
-    assert (
-        entry.prelim_dq_info or entry.swimoff_dq_info or entry.finals_dq_info
-    ), "There must be a DQ for there to be an H1 line"
+    # No-op if the entry carries no DQ slot to attach to — mirrors h2_parser.
+    # An H1 can appear whose last_entry is not the DQ'd swim it describes (e.g. a
+    # relay DQ, or a non-DQ entry emitted between the DQ result and its H1); skip
+    # the detail rather than raising. The DQ result itself is unaffected, only the
+    # human-readable reason string is dropped.
+    if not (entry.prelim_dq_info or entry.swimoff_dq_info or entry.finals_dq_info):
+        return file
 
     if entry.finals_dq_info:
         # DQ happened in prelims

tests/hy3/line_parsers/test_f_relay_parsers.py

@@ -134,6 +134,34 @@ def test_f3_with_8_swimmers_keeps_all_8_keyed(self):
         entry = result.meet.last_event[1].last_entry
         self.assertEqual(set(entry.swimmers.keys()), {1, 2, 3, 4, 5, 6, 7, 8})
 
+    @staticmethod
+    def _f3_line(pairs):
+        # Build an F3 line from (meet_id, leg) pairs. Each 13-char block is a
+        # 5-char right-justified meet_id (cols 4-8), 6 pad, 1-char leg (col 15), pad.
+        line = "F3 "
+        for mid, leg in pairs:
+            line += f"{mid:>5d}" + " " * 6 + f"{leg:d}" + " "
+        return line
+
+    def test_f3_unrostered_swimmer_id_is_skipped_not_raised(self):
+        # Regression: an F3 relay leg referencing a swimmer meet-id with no D1
+        # roster record in the file must be skipped, not raise KeyError. Roster
+        # has ids 1..8; id 42 is absent, so leg 1 is dropped and 2..4 are kept.
+        file, opts = self._file_with_relay_entry_and_4_swimmers()
+        f3_line = self._f3_line([(42, 1), (2, 2), (3, 3), (4, 4)])
+        result = f3_parser(f3_line, file, opts)  # must not raise
+        entry = result.meet.last_event[1].last_entry
+        self.assertEqual(set(entry.swimmers.keys()), {2, 3, 4})
+
+    def test_f3_all_unrostered_yields_empty_swimmers_no_raise(self):
+        # Every leg references an absent id → empty swimmers dict, still no raise
+        # (the empty-dict age guard already handles this downstream).
+        file, opts = self._file_with_relay_entry_and_4_swimmers()
+        f3_line = self._f3_line([(40, 1), (41, 2), (42, 3), (43, 4)])
+        result = f3_parser(f3_line, file, opts)  # must not raise
+        entry = result.meet.last_event[1].last_entry
+        self.assertEqual(entry.swimmers, {})
+
 
 class TestF2BackupTimingFields(unittest.TestCase):
     """F2 timing fields. Same offsets as E2 for the five timing

tests/hy3/line_parsers/test_h_dq_parsers.py

@@ -69,5 +69,40 @@ def test_h2_empty_detail_is_none(self) -> None:
         self.assertIsNone(entry.finals_dq_info.info_str_detail)
 
 
+class TestH1DqParser(unittest.TestCase):
+
+    def _file_with_entry(self):
+        opts = {"default_country": "USA"}
+        file = ParsedHytekFile()
+        file.meet = Meet()
+        file.meet.last_team = ("FOO", Team("Foo Bar", "FOO", "foo", "", "", "", "", "", "", "", "", "", "", "", {}))
+        d_line = "D1M   27Hansen              Mads                                                        10272010 13                             27"
+        e_line = "E1M   27HanseXX    50D 11109  0U  0.00 22X   37.41S   37.41S    0.00    0.00  0NN               N                               70"
+        file = d1_parser(d_line, file, opts)
+        file = e1_parser(e_line, file, opts)
+        return file, opts
+
+    def test_h1_sets_reason_on_finals_dq(self) -> None:
+        from hytek_parser.hy3.line_parsers.h_dq_parsers import h1_parser
+        file, opts = self._file_with_entry()
+        entry = file.meet.last_event[1].last_entry
+        entry.finals_dq_info = DisqualificationInfo(DisqualificationCode.FLY_KICK_ALTERNATING, "")
+        # "H1" + "1A" (FLY_KICK_ALTERNATING, matches the slot's code) + reason text
+        result = h1_parser("H11AAlternating Kick", file, opts)
+        self.assertEqual("Alternating Kick", result.meet.last_event[1].last_entry.finals_dq_info.info_str)
+
+    def test_h1_no_dq_is_noop(self) -> None:
+        # Regression: an H1 line whose entry carries no DQ slot must be a no-op,
+        # not raise — mirroring h2_parser. (A relay DQ, or a non-DQ entry emitted
+        # between the DQ result and its H1, resolves last_entry to a non-DQ entry.)
+        from hytek_parser.hy3.line_parsers.h_dq_parsers import h1_parser
+        file, opts = self._file_with_entry()
+        result = h1_parser("H11AAlternating Kick", file, opts)  # must not raise
+        entry = result.meet.last_event[1].last_entry
+        self.assertIsNone(entry.finals_dq_info)
+        self.assertIsNone(entry.swimoff_dq_info)
+        self.assertIsNone(entry.prelim_dq_info)
+
+
 if __name__ == "__main__":
     unittest.main()