speed up track xml serialization

The comment explains it.

2 files changed, 78 insertions, 25 deletions

diff --git a/fietsboek/models/track.py b/fietsboek/models/track.py
index 3bf7eee..5f1e081 100644
--- a/fietsboek/models/track.py
+++ b/fietsboek/models/track.py
@@ -17,6 +17,7 @@ meta information.
 import datetime
 import enum
 import gzip
+import io
 import json
 import logging
 from itertools import chain
@@ -435,34 +436,64 @@ class Track(Base):
 
         :return: The XML representation (a GPX file).
         """
-        gpx = gpxpy.gpx.GPX()
-        gpx.description = self.description
-        gpx.name = self.title
-        segment = gpxpy.gpx.GPXTrackSegment()
+        # This is a cumbersome way to do it, as we're re-implementing XML
+        # serialization logic. However, recreating the track in gpxpy and
+        # letting it serialize it is much slower:
+        # For a track with around 50,000 points, the gpxpy method takes
+        # ~5.9 seconds here, while the "manual" buffer takes only ~2.4 seconds.
+        # This is a speed-up we're happy to take!
+        buf = io.BytesIO()
+        buf.write(b'<?xml version="1.0" encoding="UTF-8" standalone="yes" ?>')
+        buf.write(b'<gpx version="1.1" xmlns="http://www.topografix.com/GPX/1/1">')
+
+        buf.write(b"<metadata>")
+        if self.title:
+            buf.write(b"<name>%s</name>" % util.xml_escape(self.title))
+        if self.description:
+            buf.write(b"<desc>%s</desc>" % util.xml_escape(self.description))
+        buf.write(b"</metadata>")
+
+        # Cache for easy access, especially the date is important since it's a
+        # dynamic property
+        date = self.date
+        write = buf.write
+
+        write(b"<trk>")
+        write(b"<trkseg>")
         for point in self.path().points:
-            segment.points.append(
-                gpxpy.gpx.GPXTrackPoint(
-                    latitude=point.latitude,
-                    longitude=point.longitude,
-                    elevation=point.elevation,
-                    time=self.date + datetime.timedelta(seconds=point.time_offset),
-                )
-            )
-        track = gpxpy.gpx.GPXTrack()
-        track.segments.append(segment)
-        gpx.tracks.append(track)
+            write(b'<trkpt lat="')
+            write(str(point.latitude).encode("ascii"))
+            write(b'" lon="')
+            write(str(point.longitude).encode("ascii"))
+            write(b'">')
+            write(b"<ele>")
+            write(str(point.elevation).encode("ascii"))
+            write(b"</ele>")
+            write(b"<time>")
+            write(str(date + datetime.timedelta(seconds=point.time_offset)).encode("ascii"))
+            write(b"</time>")
+            write(b"</trkpt>\n")
+        write(b"</trkseg>")
+        write(b"</trk>")
+
+        # This loop is not as hot:
         for wpt in self.waypoints:
-            gpx.waypoints.append(
-                gpxpy.gpx.GPXWaypoint(
-                    longitude=wpt.longitude,
-                    latitude=wpt.latitude,
-                    elevation=wpt.elevation,
-                    name=wpt.name,
-                    comment=wpt.description,
-                    description=wpt.description,
-                )
+            write(
+                b'<wpt lat="%s" lon="%s">'
+                % (util.xml_escape(str(wpt.latitude)), util.xml_escape(str(wpt.longitude)))
             )
-        return gpx.to_xml(prettyprint=False).encode("utf-8")
+            if wpt.elevation is not None:
+                write(b"<ele>%s</ele>" % util.xml_escape(str(wpt.elevation)))
+            if wpt.name is not None:
+                write(b"<name>%s</name>" % util.xml_escape(wpt.name))
+            if wpt.description is not None:
+                write(b"<cmt>%s</cmt>" % util.xml_escape(wpt.description))
+                write(b"<desc>%s</desc>" % util.xml_escape(wpt.description))
+            write(b"</wpt>")
+
+        write(b"</gpx>")
+
+        return buf.getvalue()
 
     @property
     def date(self):
diff --git a/fietsboek/util.py b/fietsboek/util.py
index 156b7d4..37e0943 100644
--- a/fietsboek/util.py
+++ b/fietsboek/util.py
@@ -64,6 +64,9 @@ _windows_device_files = (
 )
 
 
+_valid_xml_chars = set("ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789.,-: ")
+
+
 def safe_markdown(md_source: str) -> Markup:
     """Transform a markdown document into a safe HTML document.
 
@@ -480,6 +483,24 @@ def recursive_size(path: Path) -> int:
     return size
 
 
+def xml_escape(value: str) -> bytes:
+    """Escapes and encodes a string to be embedded in a XML document.
+
+    This replaces characters like < and > with their entities.
+
+    :param value: The value.
+    :return: The escaped and encoded string.
+    """
+    return b"".join(
+        (
+            char.encode("ascii")
+            if char in _valid_xml_chars
+            else b"&#x%s;" % hex(ord(char))[2:].encode("ascii")
+        )
+        for char in value
+    )
+
+
 __all__ = [
     "ALLOWED_TAGS",
     "ALLOWED_ATTRIBUTES",
@@ -505,4 +526,5 @@ __all__ = [
     "encode_gpx",
     "secure_filename",
     "recursive_size",
+    "xml_escape",
 ]