Strip pre-2000 leading points to prevent epoch-zero start time and absurd duration

bincio/extract/models.py

@@ -5,9 +5,27 @@ It gets fed into metrics computation and the BAS JSON writer.
 """
 
 from dataclasses import dataclass, field
-from datetime import datetime
+from datetime import datetime, timezone
 from typing import Optional
 
+# Any timestamp before this is almost certainly an uninitialised sensor value
+# (epoch 0, FIT "no-data" sentinel, RTC not yet synced, etc.).
+_MIN_TIMESTAMP = datetime(2000, 1, 1, tzinfo=timezone.utc)
+
+
+def strip_bogus_leading_points(points: list["DataPoint"]) -> list["DataPoint"]:
+    """Drop leading points whose timestamp predates the year 2000.
+
+    FIT files occasionally emit a record with timestamp=0 (or another
+    pre-2000 value) as an uninitialised sentinel before the real data
+    begins.  Keeping such a point as points[0] produces a 1970 start
+    time and an absurdly large duration_s.
+    """
+    i = 0
+    while i < len(points) and points[i].timestamp < _MIN_TIMESTAMP:
+        i += 1
+    return points[i:]
+
 
 @dataclass
 class DataPoint:

bincio/extract/parsers/fit.py

@@ -6,7 +6,7 @@ from typing import Any
 
 import fitdecode
 
-from bincio.extract.models import DataPoint, LapData, ParsedActivity
+from bincio.extract.models import DataPoint, LapData, ParsedActivity, strip_bogus_leading_points
 from bincio.extract.sport import normalise_sport
 
 
@@ -101,6 +101,7 @@ class FitParser:
                             )
                         )
 
+        points = strip_bogus_leading_points(points)
         if not points:
             raise ValueError(f"No record messages found in {path.name}")
 

bincio/extract/parsers/gpx.py

@@ -6,7 +6,7 @@ from pathlib import Path
 import gpxpy
 import gpxpy.gpx
 
-from bincio.extract.models import DataPoint, ParsedActivity
+from bincio.extract.models import DataPoint, ParsedActivity, strip_bogus_leading_points
 from bincio.extract.parsers.base import BaseParser
 from bincio.extract.sport import normalise_sport, normalise_sub_sport
 
@@ -38,6 +38,7 @@ class GpxParser(BaseParser):
                     _apply_extensions(pt, dp)
                     points.append(dp)
 
+        points = strip_bogus_leading_points(points)
         if not points:
             raise ValueError(f"No trackpoints found in {path.name}")
 

bincio/extract/parsers/tcx.py

@@ -5,7 +5,7 @@ from pathlib import Path
 
 from lxml import etree
 
-from bincio.extract.models import DataPoint, ParsedActivity
+from bincio.extract.models import DataPoint, ParsedActivity, strip_bogus_leading_points
 from bincio.extract.sport import normalise_sport, normalise_sub_sport
 
 _NS_HTTP  = {
@@ -73,6 +73,7 @@ class TcxParser:
             )
             points.append(dp)
 
+        points = strip_bogus_leading_points(points)
         if not points:
             raise ValueError(f"No trackpoints found in {path.name}")