try to avoid parsing the GPX more than once

5 files changed, 31 insertions, 12 deletions

diff --git a/fietsboek/actions.py b/fietsboek/actions.py
index 4974a2c..2058f4d 100644
--- a/fietsboek/actions.py
+++ b/fietsboek/actions.py
@@ -8,7 +8,7 @@ the test functions.
 import datetime
 import logging
 import re
-from typing import List
+from typing import List, Optional
 
 import brotli
 import gpxpy
@@ -97,7 +97,7 @@ def add_track(
 
     # Best time to build the cache is right after the upload, but *after* the
     # transformers have been applied!
-    track.ensure_cache(manager.decompress_gpx())
+    track.ensure_cache(gpx)
     dbsession.add(track.cache)
 
     manager.engrave_metadata(
@@ -105,6 +105,7 @@ def add_track(
         description=track.description,
         author_name=track.owner.name,
         time=track.date,
+        gpx=gpx,
     )
 
     return track
@@ -170,14 +171,18 @@ def edit_images(request: Request, track: models.Track):
         request.dbsession.add(image_meta)
 
 
-def execute_transformers(request: Request, track: models.Track):
+def execute_transformers(request: Request, track: models.Track) -> Optional[gpxpy.gpx.GPX]:
     """Execute the transformers for the given track.
 
     Note that this function "short circuits" if the saved transformer settings
     already match the settings given in the request.
 
+    This function saves the modified data, but does also return it in case you
+    need to do further processing (unless no transformations have taken place).
+
     :param request: The request.
     :param track: The track.
+    :return: The transformed track.
     """
     # pylint: disable=too-many-locals
     LOGGER.debug("Executing transformers for %d", track.id)
@@ -187,7 +192,7 @@ def execute_transformers(request: Request, track: models.Track):
     serialized = [[tfm.identifier(), tfm.parameters.dict()] for tfm in settings]
     if serialized == track.transformers:
         LOGGER.debug("Applied transformations mach on %d, skipping", track.id)
-        return
+        return None
 
     # We always start with the backup, that way we don't get "deepfried GPX"
     # files by having the same filters run multiple times on the same input.
@@ -210,5 +215,6 @@ def execute_transformers(request: Request, track: models.Track):
     LOGGER.debug("Rebuilding cache for %d", track.id)
     request.dbsession.delete(track.cache)
     track.cache = None
-    track.ensure_cache(manager.decompress_gpx())
+    track.ensure_cache(gpx)
     request.dbsession.add(track.cache)
+    return gpx
diff --git a/fietsboek/data.py b/fietsboek/data.py
index 1a1b66b..6906c84 100644
--- a/fietsboek/data.py
+++ b/fietsboek/data.py
@@ -158,7 +158,13 @@ class TrackDataDir:
         return brotli.decompress(self.gpx_path().read_bytes())
 
     def engrave_metadata(
-        self, title: str, description: str, author_name: str, time: datetime.datetime
+        self,
+        title: str,
+        description: str,
+        author_name: str,
+        time: datetime.datetime,
+        *,
+        gpx: Optional[gpxpy.gpx.GPX] = None,
     ):
         """Engrave the given metadata into the GPX file.
 
@@ -168,8 +174,10 @@ class TrackDataDir:
         :param description: The description of the track.
         :param creator: Name of the track's creator.
         :param time: Time of the track.
+        :param gpx: The pre-parsed GPX track, to save time if it is already parsed.
         """
-        gpx = gpxpy.parse(self.decompress_gpx())
+        if gpx is None:
+            gpx = gpxpy.parse(self.decompress_gpx())
         # First we delete the existing metadata
         for track in gpx.tracks:
             track.name = None
diff --git a/fietsboek/models/track.py b/fietsboek/models/track.py
index 9342cff..e0d2820 100644
--- a/fietsboek/models/track.py
+++ b/fietsboek/models/track.py
@@ -18,6 +18,7 @@ import logging
 from itertools import chain
 from typing import TYPE_CHECKING, List, Optional, Set, Union
 
+import gpxpy
 from babel.numbers import format_decimal
 from markupsafe import Markup
 from pyramid.authorization import (
@@ -330,7 +331,7 @@ class Track(Base):
         result = ACLHelper().permits(self, principals, "track.view")
         return isinstance(result, ACLAllowed)
 
-    def ensure_cache(self, gpx_data: Union[str, bytes]):
+    def ensure_cache(self, gpx_data: Union[str, bytes, gpxpy.gpx.GPX]):
         """Ensure that a cached version of this track's metadata exists.
 
         :param gpx_data: GPX data (uncompressed) from which to build the cache.
diff --git a/fietsboek/util.py b/fietsboek/util.py
index c741550..68ba769 100644
--- a/fietsboek/util.py
+++ b/fietsboek/util.py
@@ -152,19 +152,22 @@ def guess_gpx_timezone(gpx: gpxpy.gpx.GPX) -> datetime.tzinfo:
     return datetime.timezone.utc
 
 
-def tour_metadata(gpx_data: Union[str, bytes]) -> dict:
+def tour_metadata(gpx_data: Union[str, bytes, gpxpy.gpx.GPX]) -> dict:
     """Calculate the metadata of the tour.
 
     Returns a dict with ``length``, ``uphill``, ``downhill``, ``moving_time``,
     ``stopped_time``, ``max_speed``, ``avg_speed``, ``start_time`` and
     ``end_time``.
 
-    :param gpx_data: The GPX data of the tour.
+    :param gpx_data: The GPX data of the tour. Can be pre-parsed to save time.
     :return: A dictionary with the computed values.
     """
     if isinstance(gpx_data, bytes):
         gpx_data = gpx_data.decode("utf-8")
-    gpx = gpxpy.parse(gpx_data)
+    if isinstance(gpx_data, gpxpy.gpx.GPX):
+        gpx = gpx_data
+    else:
+        gpx = gpxpy.parse(gpx_data)
     timezone = guess_gpx_timezone(gpx)
     uphill, downhill = gpx.get_uphill_downhill()
     moving_data = gpx.get_moving_data()
diff --git a/fietsboek/views/edit.py b/fietsboek/views/edit.py
index d60cced..9cc666f 100644
--- a/fietsboek/views/edit.py
+++ b/fietsboek/views/edit.py
@@ -95,12 +95,13 @@ def do_edit(request):
         track.sync_tags(tags)
 
         actions.edit_images(request, request.context)
-        actions.execute_transformers(request, request.context)
+        gpx = actions.execute_transformers(request, request.context)
         data.engrave_metadata(
             title=track.title,
             description=track.description,
             author_name=track.owner.name,
             time=track.date,
+            gpx=gpx,
         )
 
     return HTTPFound(request.route_url("details", track_id=track.id))