parallelizing extraction, fix tcx files

bincio/extract/parsers/tcx.py

@@ -8,18 +8,24 @@ from lxml import etree
 from bincio.extract.models import DataPoint, ParsedActivity
 from bincio.extract.sport import normalise_sport
 
-_NS = {
+_NS_HTTP  = {
     "tcx": "http://www.garmin.com/xmlschemas/TrainingCenterDatabase/v2",
     "ext": "http://www.garmin.com/xmlschemas/ActivityExtension/v2",
 }
+_NS_HTTPS = {
+    "tcx": "https://www.garmin.com/xmlschemas/TrainingCenterDatabase/v2",
+    "ext": "https://www.garmin.com/xmlschemas/ActivityExtension/v2",
+}
 
 
 class TcxParser:
     def parse(self, path: Path, raw_bytes: bytes) -> ParsedActivity:
-        # Some exporters (e.g. Garmin) prepend whitespace before the XML
-        # declaration, which is technically invalid. Strip it.
+        # Some exporters prepend whitespace before the XML declaration. Strip it.
         root = etree.fromstring(raw_bytes.lstrip())
 
+        # Garmin sometimes uses https:// instead of http:// in the namespace URI.
+        _NS = _NS_HTTPS if b"https://www.garmin.com" in raw_bytes else _NS_HTTP
+
         activities = root.findall(".//tcx:Activity", _NS)
         if not activities:
             raise ValueError(f"No Activity elements found in {path.name}")