hytek-parser 2.3.0__tar.gz → 3.0.0__tar.gz

{hytek_parser-2.3.0 → hytek_parser-3.0.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: hytek-parser
-Version: 2.3.0
+Version: 3.0.0
 Summary: Parsers for the files produced by Hytek's Meet Manager.
 Home-page: https://www.github.com/SwimComm/hytek-parser
 License: MIT

hytek_parser-3.0.0/hytek_parser/hy3/_utils.py

@@ -0,0 +1,69 @@
+import math
+from typing import Optional, Union
+
+from hytek_parser._utils import safe_cast, select_from_enum
+from hytek_parser.hy3.enums import ReplacedTimeTimeCode
+
+
+def parse_time(raw_time: str) -> Union[float, ReplacedTimeTimeCode]:
+    """Parse a time into either a number or time code.
+
+    Args:
+        raw_time (str): The time string extracted from the Hytek file.
+
+    Returns:
+        Union[float, ReplacedTimeTimeCode]: Either a numerical time or a time code.
+    """
+    if (casted := safe_cast(float, raw_time, default=-1)) != -1:
+        # Number
+        return casted
+    else:
+        return select_from_enum(ReplacedTimeTimeCode, raw_time)
+
+
+def parse_time_or_none(raw_time: str) -> Optional[float]:
+    """Parse a timing field where 0.00/blank means 'not recorded'.
+
+    Unlike parse_time (which returns 0.0 for "0.00" and a ReplacedTimeTimeCode
+    for blank/non-numeric input), this returns None unless the value is a
+    positive float. Used for pad and backup-button times, where an unused
+    slot is written as 0.00 and should surface as None rather than 0.0.
+    """
+    val = parse_time(raw_time)
+    return val if isinstance(val, float) and val > 0.0 else None
+
+
+def parse_reaction_time(raw: str) -> Optional[float]:
+    """Parse a reaction/takeoff-time column (E2 col 83-87, F2 col 83-102).
+
+    NEGATIVE VALUES ARE MEANINGFUL and load-bearing here: a relay takeover slot
+    records an early exchange as a negative number (7,868 values corpus-wide).
+    ``parse_time_or_none`` requires > 0.0 and would silently destroy every one
+    of them -- do not substitute it.
+
+    Sentinels, all meaning "not recorded": blank, 0.00 in any sign spelling,
+    and the literal NRT ("No Reaction Time") that Meet Manager writes into
+    takeover slots when the exchange was not measured.
+
+    Values above the plausible reaction range are returned unchanged. A
+    minority of files put something else in these columns; its meaning is
+    unresolved, and filtering it here would make it permanently invisible.
+    """
+    val = raw.strip()
+    if not val or val.upper() == "NRT":
+        return None
+    try:
+        num = float(val)
+    except ValueError:
+        # Observed malformed forms: a bare "+" sign, stray high bytes.
+        return None
+    if not math.isfinite(num):
+        # float() accepts "nan"/"inf"/"-inf", all five characters or fewer,
+        # so they fit this column like any other token. Neither is a
+        # reaction time -- and a bare int() downstream would raise on them
+        # (ValueError on nan, OverflowError on inf) instead of yielding None
+        # like every other malformed token, dropping the whole file.
+        return None
+    # float() maps "0.00", "+0.00" and "-0.00" all to zero; all three are the
+    # "not recorded" sentinel.
+    return None if num == 0.0 else num

{hytek_parser-2.3.0 → hytek_parser-3.0.0}/hytek_parser/hy3/enums.py

@@ -12,6 +12,25 @@ class Gender(Enum):
     FEMALE = "F"
     UNKNOWN = "U"
 
+    @classmethod
+    def _missing_(cls, value):
+        # Hy-Tek exporters occasionally emit a lowercase sex byte. Case is
+        # not meaningful in this column, and UNKNOWN specifically means
+        # "the file did not say" -- so a lowercase 'm' has to resolve to
+        # MALE rather than become indistinguishable from a blank column.
+        #
+        # Retry ONLY when upcasing changes the value. Without that guard an
+        # unrecognized uppercase byte re-enters this hook with an identical
+        # value and recurses forever.
+        #
+        # _missing_ rather than aenum's native _missing_value_: both work
+        # here, but _missing_ is also the stdlib hook, so it keeps working
+        # if the enum base ever changes. _missing_value_ would silently
+        # become a no-op and reintroduce the bug with no signal.
+        if isinstance(value, str) and value.upper() != value:
+            return cls(value.upper())
+        return None
+
 
 class Stroke(Enum):
     """Types of swimming strokes."""
@@ -23,6 +42,18 @@ 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-2.3.0 → hytek_parser-3.0.0}/hytek_parser/hy3/line_parsers/e_event_parsers.py

@@ -2,7 +2,7 @@ from datetime import datetime
 from typing import Any
 
 from hytek_parser._utils import extract, get_age_group, safe_cast, select_from_enum
-from hytek_parser.hy3._utils import parse_time, parse_time_or_none
+from hytek_parser.hy3._utils import parse_reaction_time, parse_time, parse_time_or_none
 from hytek_parser.hy3.enums import (
     Course,
     DisqualificationCode,
@@ -12,6 +12,7 @@ from hytek_parser.hy3.enums import (
     Stroke,
     WithTimeTimeCode,
 )
+from hytek_parser.hy3.line_parsers.h_dq_parsers import LAST_DQ_SLOT_KEY
 from hytek_parser.hy3.schemas import DisqualificationInfo, ParsedHytekFile
 
 
@@ -26,7 +27,7 @@ def e1_parser(
     # Event info
     event_gender = select_from_enum(Gender, extract(line, 14, 1))
     event_gender_age = select_from_enum(GenderAge, extract(line, 15, 1))
-    distance = safe_cast(int, extract(line, 16, 6))
+    distance = safe_cast(float, extract(line, 16, 6))
     stroke = select_from_enum(Stroke, extract(line, 22, 1))
     age_min, age_max = get_age_group(
         age_min=safe_cast(int, extract(line, 23, 3)),
@@ -117,6 +118,8 @@ def e2_parser(
     button_3_time = parse_time_or_none(extract(line, 55, 8))
     backup_4_time = parse_time_or_none(extract(line, 75, 8))
     alt_time_code = extract(line, 96, 1) or None  # observed: 'A' / 'K' / blank
+    # col 83-87. Signed; sentinels are blank / 0.00 in any sign spelling.
+    reaction_time = parse_reaction_time(extract(line, 83, 5))
 
     raw_date = extract(line, 88, 8).strip()
     date_ = datetime.strptime(raw_date, "%m%d%Y").date() if raw_date else None
@@ -150,8 +153,15 @@ def e2_parser(
     setattr(entry, f"{prefix}_button_2_time", button_2_time)
     setattr(entry, f"{prefix}_button_3_time", button_3_time)
     setattr(entry, f"{prefix}_backup_4_time", backup_4_time)
+    setattr(entry, f"{prefix}_reaction_time", reaction_time)
     setattr(entry, f"{prefix}_alt_time_code", alt_time_code)
 
+    # Anchor the H1 reason / H2 detail lines that follow to this DQ slot (see
+    # h_dq_parsers). File position is the only signal that survives a same-code
+    # prelim+finals double-DQ; clear it on a non-DQ result so a later orphaned
+    # H1/H2 does not attach to a stale slot.
+    opts[LAST_DQ_SLOT_KEY] = f"{prefix}_dq_info" if dq_info is not None else None
+
     event.last_entry = entry
     file.meet.last_event = (event_num, event)
 

{hytek_parser-2.3.0 → hytek_parser-3.0.0}/hytek_parser/hy3/line_parsers/f_relay_parsers.py

@@ -2,7 +2,7 @@ from datetime import datetime
 from typing import Any
 
 from hytek_parser._utils import extract, get_age_group, safe_cast, select_from_enum
-from hytek_parser.hy3._utils import parse_time, parse_time_or_none
+from hytek_parser.hy3._utils import parse_reaction_time, parse_time, parse_time_or_none
 from hytek_parser.hy3.enums import (
     Course,
     DisqualificationCode,
@@ -11,6 +11,7 @@ from hytek_parser.hy3.enums import (
     Stroke,
     WithTimeTimeCode,
 )
+from hytek_parser.hy3.line_parsers.h_dq_parsers import LAST_DQ_SLOT_KEY
 from hytek_parser.hy3.schemas import (
     DisqualificationInfo,
     Gender,
@@ -30,7 +31,7 @@ def f1_parser(
     # Get event info
     event_gender = select_from_enum(Gender, extract(line, 14, 1))
     event_gender_age = select_from_enum(GenderAge, extract(line, 15, 1))
-    distance = safe_cast(int, extract(line, 16, 6))
+    distance = safe_cast(float, extract(line, 16, 6))
     stroke = select_from_enum(Stroke, extract(line, 22, 1))
 
     # Have to set these now since swimmer ages are not available yet
@@ -108,9 +109,11 @@ def f2_parser(
     overall_place = safe_cast(int, extract(line, 30, 4))
 
     # previously-dropped F2 timing fields. The five timing-column
-    # offsets are IDENTICAL to e2_parser; only alt_time_code differs because F2
-    # has a 15-column gap before its date field (date at col 103, not 88).
-    # F2 alt_time_code lives at col 111, not col 96.
+    # offsets are IDENTICAL to e2_parser. F2 then carries FOUR reaction-time
+    # slots at cols 83-102 (5 chars each) where E2 carries one at 83-87, which
+    # is why F2's date sits at col 103 and its alt_time_code at col 111.
+    # Slot 1 is the leadoff block start; slots 2-4 are exchange takeovers and
+    # are legitimately negative when a swimmer leaves early.
     pad_time      = parse_time_or_none(extract(line, 63, 12))
     button_1_time = parse_time_or_none(extract(line, 39, 8))
     button_2_time = parse_time_or_none(extract(line, 47, 8))
@@ -118,6 +121,10 @@ def f2_parser(
     backup_4_time = parse_time_or_none(extract(line, 75, 8))
     alt_time_code = extract(line, 111, 1) or None  # F2 offset; observed: 'A'/'K'/blank
 
+    reaction_times = [
+        parse_reaction_time(extract(line, 83 + 5 * i, 5)) for i in range(4)
+    ]
+
     raw_date = extract(line, 103, 8).strip()
     date_ = datetime.strptime(raw_date, "%m%d%Y").date() if raw_date else None
 
@@ -150,8 +157,15 @@ def f2_parser(
     setattr(entry, f"{prefix}_button_2_time", button_2_time)
     setattr(entry, f"{prefix}_button_3_time", button_3_time)
     setattr(entry, f"{prefix}_backup_4_time", backup_4_time)
+    setattr(entry, f"{prefix}_reaction_times", reaction_times)
     setattr(entry, f"{prefix}_alt_time_code", alt_time_code)
 
+    # Anchor the H1 reason / H2 detail lines that follow to this DQ slot (see
+    # h_dq_parsers). File position is the only signal that survives a same-code
+    # prelim+finals double-DQ; clear it on a non-DQ result so a later orphaned
+    # H1/H2 does not attach to a stale slot.
+    opts[LAST_DQ_SLOT_KEY] = f"{prefix}_dq_info" if dq_info is not None else None
+
     event.last_entry = entry
     file.meet.last_event = (event_num, event)
 
@@ -172,7 +186,14 @@ 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-3.0.0/hytek_parser/hy3/line_parsers/h_dq_parsers.py

@@ -0,0 +1,75 @@
+from typing import Any
+
+from hytek_parser._utils import extract
+from hytek_parser.hy3.schemas import ParsedHytekFile
+
+# Key under which the E2/F2 result parser records the DQ slot it just populated,
+# so the H1 reason line and H2 detail line that immediately follow attach to that
+# same slot. A swim DQ'd in both prelims and finals resolves to ONE shared entry
+# with two populated slots; the result line preceding each round's H1/H2 in file
+# order is the authoritative anchor for which slot those detail lines belong to.
+# This holds even when both rounds carry the SAME DQ code (e.g. a false start in
+# both prelims and finals): matching on the code alone cannot tell the two slots
+# apart, but file position always can.
+LAST_DQ_SLOT_KEY = "_last_dq_slot"
+
+
+def h1_parser(
+    line: str, file: ParsedHytekFile, opts: dict[str, Any]
+) -> ParsedHytekFile:
+    """Parse an H1 line: the DQ reason string, routed to its own DQ slot.
+
+    Attaches the reason to the slot the immediately-preceding E2/F2 result line
+    populated (recorded in ``opts`` under ``LAST_DQ_SLOT_KEY``). That result line
+    is the authoritative anchor: a swim DQ'd in both prelims and finals carries
+    two populated slots on one shared entry, and file position — not the DQ code —
+    tells the two H1s apart, including when both rounds share the same code.
+
+    No-op when no slot is anchored or the anchored slot is unpopulated — the DQ
+    status and code are recorded by the E2/F2 result line regardless, so only the
+    human-readable reason string is ever at stake. Leaves the anchor in place so
+    the H2 detail line that follows attaches to the same slot.
+    """
+    event_num, event = file.meet.last_event
+    entry = event.last_entry
+
+    info_str = extract(line, 5, 124)  # Whitespace is stripped
+
+    slot = opts.get(LAST_DQ_SLOT_KEY)
+    if slot is not None:
+        info = getattr(entry, slot)
+        if info is not None:
+            info.info_str = info_str
+
+    event.last_entry = entry
+    file.meet.last_event = (event_num, event)
+
+    return file
+
+
+def h2_parser(
+    line: str, file: ParsedHytekFile, opts: dict[str, Any]
+) -> ParsedHytekFile:
+    """Parse an H2 line: the specific, human-readable DQ infraction detail.
+
+    Attaches the detail to the same slot the preceding E2/F2 result line anchored
+    (``LAST_DQ_SLOT_KEY``) — the slot h1_parser also wrote to. H2 cannot
+    self-identify its slot: its own 2-char code is the stroke/leg infraction (e.g.
+    "2L"), not the slot's ``DisqualificationCode`` (e.g. the relay-leg code "6A"),
+    so the anchor is the only reliable signal. No-op when no slot is anchored or
+    the anchored slot is unpopulated.
+    """
+    event_num, event = file.meet.last_event
+    entry = event.last_entry
+
+    detail = extract(line, 5, 124) or None
+
+    slot = opts.get(LAST_DQ_SLOT_KEY)
+    if slot is not None:
+        info = getattr(entry, slot)
+        if info is not None:
+            info.info_str_detail = detail
+
+    event.last_entry = entry
+    file.meet.last_event = (event_num, event)
+    return file

{hytek_parser-2.3.0 → hytek_parser-3.0.0}/hytek_parser/hy3/schemas.py

@@ -1,5 +1,5 @@
 from datetime import date, datetime
-from typing import Optional, Union
+from typing import List, Optional, Union
 
 from attrs import Factory, define, field
 
@@ -119,6 +119,8 @@ class EventEntry:
     prelim_button_2_time: Optional[float] = None
     prelim_button_3_time: Optional[float] = None
     prelim_backup_4_time: Optional[float] = None
+    prelim_reaction_time: Optional[float] = None
+    prelim_reaction_times: Optional[List[Optional[float]]] = None
     # col 96; semantics unverified — observed 'A'/'K'/blank
     prelim_alt_time_code: Optional[str] = None
 
@@ -139,6 +141,8 @@ class EventEntry:
     swimoff_button_2_time: Optional[float] = None
     swimoff_button_3_time: Optional[float] = None
     swimoff_backup_4_time: Optional[float] = None
+    swimoff_reaction_time: Optional[float] = None
+    swimoff_reaction_times: Optional[List[Optional[float]]] = None
     # col 96; semantics unverified — observed 'A'/'K'/blank
     swimoff_alt_time_code: Optional[str] = None
 
@@ -159,6 +163,8 @@ class EventEntry:
     finals_button_2_time: Optional[float] = None
     finals_button_3_time: Optional[float] = None
     finals_backup_4_time: Optional[float] = None
+    finals_reaction_time: Optional[float] = None
+    finals_reaction_times: Optional[List[Optional[float]]] = None
     # col 96; semantics unverified — observed 'A'/'K'/blank
     finals_alt_time_code: Optional[str] = None
 
@@ -216,6 +222,8 @@ class EventEntry:
             setattr(self, f"{course}_button_2_time", None)
             setattr(self, f"{course}_button_3_time", None)
             setattr(self, f"{course}_backup_4_time", None)
+            setattr(self, f"{course}_reaction_time", None)
+            setattr(self, f"{course}_reaction_times", None)
             setattr(self, f"{course}_alt_time_code", None)
 
         self.prelim_dq_info = None
@@ -256,7 +264,7 @@ class Event:
 
     # ID info
     number: str
-    distance: int
+    distance: float
     stroke: Stroke
     course: Course
     date_: Optional[date]
@@ -424,7 +432,7 @@ class Meet:
     def get_or_create_event(
         self,
         number: str,
-        distance: int,
+        distance: float,
         stroke: Stroke,
         course: Course,
         gender: Gender,

{hytek_parser-2.3.0 → hytek_parser-3.0.0}/pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "hytek-parser"
-version = "2.3.0"
+version = "3.0.0"
 description = "Parsers for the files produced by Hytek's Meet Manager."
 license = "MIT"
 authors = ["Nino Maruszewski <nino.maruszewski@hotmail.com>"]

hytek_parser-2.3.0/hytek_parser/hy3/_utils.py

@@ -1,32 +0,0 @@
-from typing import Optional, Union
-
-from hytek_parser._utils import safe_cast, select_from_enum
-from hytek_parser.hy3.enums import ReplacedTimeTimeCode
-
-
-def parse_time(raw_time: str) -> Union[float, ReplacedTimeTimeCode]:
-    """Parse a time into either a number or time code.
-
-    Args:
-        raw_time (str): The time string extracted from the Hytek file.
-
-    Returns:
-        Union[float, ReplacedTimeTimeCode]: Either a numerical time or a time code.
-    """
-    if (casted := safe_cast(float, raw_time, default=-1)) != -1:
-        # Number
-        return casted
-    else:
-        return select_from_enum(ReplacedTimeTimeCode, raw_time)
-
-
-def parse_time_or_none(raw_time: str) -> Optional[float]:
-    """Parse a timing field where 0.00/blank means 'not recorded'.
-
-    Unlike parse_time (which returns 0.0 for "0.00" and a ReplacedTimeTimeCode
-    for blank/non-numeric input), this returns None unless the value is a
-    positive float. Used for pad and backup-button times, where an unused
-    slot is written as 0.00 and should surface as None rather than 0.0.
-    """
-    val = parse_time(raw_time)
-    return val if isinstance(val, float) and val > 0.0 else None

hytek_parser-2.3.0/hytek_parser/hy3/line_parsers/h_dq_parsers.py

@@ -1,77 +0,0 @@
-from typing import Any
-
-from hytek_parser._utils import extract, select_from_enum
-from hytek_parser.hy3.enums import DisqualificationCode
-from hytek_parser.hy3.schemas import ParsedHytekFile
-
-
-def h1_parser(
-    line: str, file: ParsedHytekFile, opts: dict[str, Any]
-) -> ParsedHytekFile:
-    """Parse a G1 split time entry."""
-    # Get event
-    event_num, event = file.meet.last_event
-    entry = event.last_entry
-
-    # Get dq code and message
-    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"
-
-    if entry.finals_dq_info:
-        # DQ happened in prelims
-        assert (
-            entry.finals_dq_info.code == dq_code
-        ), "DQ Codes should match in the H1 line"
-
-        entry.finals_dq_info.info_str = dq_info
-    elif entry.swimoff_dq_info:
-        # DQ happened in prelims
-        assert (
-            entry.swimoff_dq_info.code == dq_code
-        ), "DQ Codes should match in the H1 line"
-
-        entry.swimoff_dq_info.info_str = dq_info
-    elif entry.prelim_dq_info:
-        # DQ happened in prelims
-        assert (
-            entry.prelim_dq_info.code == dq_code
-        ), "DQ Codes should match in the H1 line"
-
-        entry.prelim_dq_info.info_str = dq_info
-
-    event.last_entry = entry
-    file.meet.last_event = (event_num, event)
-
-    return file
-
-
-def h2_parser(
-    line: str, file: ParsedHytekFile, opts: dict[str, Any]
-) -> ParsedHytekFile:
-    """Parse an H2 line: the specific, human-readable DQ infraction detail.
-
-    Attaches the detail text to whichever DQ slot the preceding H1 populated
-    (finals -> swimoff -> prelim), mirroring h1_parser's resolution. Unlike H1,
-    the H2 2-char code is NOT asserted against the DQ code (it is a stroke/leg
-    sub-classification, not the same DisqualificationCode). No-op if no DQ slot
-    is set.
-    """
-    event_num, event = file.meet.last_event
-    entry = event.last_entry
-
-    detail = extract(line, 5, 124) or None
-
-    if entry.finals_dq_info:
-        entry.finals_dq_info.info_str_detail = detail
-    elif entry.swimoff_dq_info:
-        entry.swimoff_dq_info.info_str_detail = detail
-    elif entry.prelim_dq_info:
-        entry.prelim_dq_info.info_str_detail = detail
-
-    event.last_entry = entry
-    file.meet.last_event = (event_num, event)
-    return file