def async_job(user, limit=None, after=None, before=None):
user.indexing(True)
activities_list = []
count = 1
try:
for a in self.client().get_activities():
d = strava2dict(a)
if d.get("summary_polyline"):
activities_list.append(d)
if (limit or
(after and (d["beginTimestamp"] >= after)) or
(before and (d["beginTimestamp"] <= before))):
d2 = dict(d)
d2["beginTimestamp"] = str(d2["beginTimestamp"])
Q.put(d2)
app.logger.info("put {} on queue".format(d2["id"]))
if limit:
limit -= 1
if not limit:
Q.put({"stop_rendering": "1"})
else:
Q.put({"msg": "indexing...{} activities".format(count)})
count += 1
gevent.sleep(0)
except Exception as e:
Q.put({"error": str(e)})
else:
Q.put({"msg": "done indexing {} activities.".format(count)})
activity_index = (pd.DataFrame(activities_list)
.set_index("beginTimestamp")
.sort_index(ascending=False)
.astype(dtypes))
app.logger.debug("done with indexing for {}".format(self))
dt_last_indexed = datetime.utcnow()
packed = activity_index.to_msgpack(compress='blosc')
cache.set(self.index_key(),
(dt_last_indexed, packed),
CACHE_INDEX_TIMEOUT)
app.logger.info("cached {}, size={}".format(self.index_key(),
len(packed)))
finally:
user.indexing(False)
Q.put(StopIteration)
def indexing(self, status=None):
# Indicate to other processes that we are currently indexing
# This should not take any longer than 30 seconds
key = "indexing {}".format(self.strava_id)
if status is None:
return cache.get(key)
else:
return cache.set(key, status, 30)
def index(self, activity_ids=None, limit=None, after=None, before=None):
def strava2dict(a):
return {
"id": a.id,
"name": a.name,
"type": a.type,
"summary_polyline": a.map.summary_polyline,
"beginTimestamp": a.start_date_local,
"total_distance": float(a.distance),
"elapsed_time": int(a.elapsed_time.total_seconds()),
"average_speed": float(a.average_speed)
}
dtypes = {
"id": "uint32",
"type": "category",
"total_distance": "float32",
"elapsed_time": "uint32",
"average_speed": "float16"
}
if self.indexing():
return [{
"error": "Indexing activities for user {}...<br>Please try again in a few seconds.<br>"
.format(self.strava_id)
}]
ind = cache.get(self.index_key())
if ind:
dt_last_indexed, packed = ind
activity_index = pd.read_msgpack(packed).astype({"type": str})
elapsed = (datetime.utcnow() -
dt_last_indexed).total_seconds()
# update the index if we need to
if (elapsed > CACHE_INDEX_UPDATE_TIMEOUT) and (not OFFLINE):
latest = activity_index.index[0]
app.logger.info("updating activity index for {}"
.format(self.strava_id))
already_got = set(activity_index.id)
try:
activities_list = [strava2dict(
a) for a in self.client().get_activities(after=latest)
if a.id not in already_got]
except Exception as e:
return [{"error": str(e)}]
if activities_list:
df = pd.DataFrame(activities_list).set_index(
"beginTimestamp")
activity_index = (
df.append(activity_index)
.drop_duplicates()
.sort_index(ascending=False)
.astype(dtypes)
)
dt_last_indexed = datetime.utcnow()
cache.set(self.index_key(),
(dt_last_indexed,
activity_index.to_msgpack(compress='blosc')),
CACHE_INDEX_TIMEOUT)
if activity_ids:
df = activity_index[activity_index["id"].isin(activity_ids)]
else:
if limit:
df = activity_index.head(limit)
else:
df = activity_index
if after:
df = df[:after]
if before:
df = df[before:]
df = df.reset_index()
df.beginTimestamp = df.beginTimestamp.astype(str)
return df.to_dict("records")
# If we got here then the index hasn't been created yet
Q = Queue()
P = Pool()
def async_job(user, limit=None, after=None, before=None):
user.indexing(True)
activities_list = []
count = 1
try:
for a in self.client().get_activities():
d = strava2dict(a)
if d.get("summary_polyline"):
activities_list.append(d)
if (limit or
(after and (d["beginTimestamp"] >= after)) or
(before and (d["beginTimestamp"] <= before))):
d2 = dict(d)
d2["beginTimestamp"] = str(d2["beginTimestamp"])
Q.put(d2)
app.logger.info("put {} on queue".format(d2["id"]))
if limit:
limit -= 1
if not limit:
Q.put({"stop_rendering": "1"})
else:
Q.put({"msg": "indexing...{} activities".format(count)})
count += 1
gevent.sleep(0)
except Exception as e:
Q.put({"error": str(e)})
else:
Q.put({"msg": "done indexing {} activities.".format(count)})
activity_index = (pd.DataFrame(activities_list)
.set_index("beginTimestamp")
.sort_index(ascending=False)
.astype(dtypes))
app.logger.debug("done with indexing for {}".format(self))
dt_last_indexed = datetime.utcnow()
packed = activity_index.to_msgpack(compress='blosc')
cache.set(self.index_key(),
(dt_last_indexed, packed),
CACHE_INDEX_TIMEOUT)
app.logger.info("cached {}, size={}".format(self.index_key(),
len(packed)))
finally:
user.indexing(False)
Q.put(StopIteration)
P.apply_async(async_job, [self, limit, after, before])
return Q
def cache(self, identifier=None, timeout=CACHE_USERS_TIMEOUT):
key = User.key(identifier or self.strava_id)
cache.set(key, self, timeout)
app.logger.debug(
"cached {} with key '{}' for {} sec".format(self, key, timeout))
return self
