def _send_email(self):
try:
if not self.accumulation:
return
with Emailer(self.settings) as emailer:
# WHO ARE WE SENDING TO
emails = Data()
for template, params in self.accumulation:
content = expand_template(template, params)
emails[literal_field(
self.settings.to_address)] += [content]
for c in self.cc:
if any(d in params.params.error for d in c.contains):
emails[literal_field(c.to_address)] += [content]
# SEND TO EACH
for to_address, content in emails.items():
emailer.send_email(source=self.settings.from_address,
to_addresses=listwrap(to_address),
subject=self.settings.subject,
body="\n\n".join(content),
format="text")
self.accumulation = []
except Exception as e:
Log.warning("Could not send", e)
finally:
self.next_send = Date.now() + self.settings.average_interval * (
2 * Random.float())
def _send_email(self):
try:
if not self.accumulation:
return
with Emailer(self.settings) as emailer:
# WHO ARE WE SENDING TO
emails = Data()
for template, params in self.accumulation:
content = expand_template(template, params)
emails[literal_field(self.settings.to_address)] += [content]
for c in self.cc:
if any(d in params.params.error for d in c.contains):
emails[literal_field(c.to_address)] += [content]
# SEND TO EACH
for to_address, content in emails.items():
emailer.send_email(
from_address=self.settings.from_address,
to_address=listwrap(to_address),
subject=self.settings.subject,
text_data="\n\n".join(content)
)
self.accumulation = []
except Exception as e:
Log.warning("Could not send", e)
finally:
self.next_send = Date.now() + self.settings.average_interval * (2 * Random.float())
def test_wrap_2():
Log.alert("Random types")
switch = [
lambda: Random.int(20),
lambda: Random.string(20),
lambda: {"i": Random.int(2000)},
lambda: Data(i=Random.int(2000)),
lambda: FlatList([{"i": Random.int(2000)}]),
lambda: [{"i": Random.int(2000)}]
]
inputs = [switch[min(len(switch) - 1, int(floor(-log(Random.float(), 2))))]() for i in range(NUM_INPUT)]
for i in range(NUM_REPEAT):
results = []
gc.collect()
with Timer("more string: to_data"):
for v in inputs:
results.append(to_data(v))
results = []
gc.collect()
with Timer("more string: baseline"):
for v in inputs:
results.append(baseline(v))
Log.note("Done {{i}} of {{num}}", i=i, num=NUM_REPEAT)
def _daemon(self, please_stop):
while not please_stop:
with Explanation("looking for work"):
try:
branch, revisions, after = self.todo.pop(till=please_stop)
except Exception as e:
if please_stop:
break
else:
raise e
if branch.name in DAEMON_DO_NO_SCAN:
continue
revisions = set(revisions)
# FIND THE REVSIONS ON THIS BRANCH
for r in list(revisions):
try:
rev = self.get_revision(
Revision(branch=branch, changeset={"id": r}),
None, # local
False, # get_diff
True, # get_moves
)
if after and after > rev.etl.timestamp:
rev = self._get_from_hg(revision=rev)
if DAEMON_DEBUG:
Log.note(
"found revision with push date {{date|datetime}}",
date=rev.push.date,
)
revisions.discard(r)
if rev.etl.timestamp > Date.now() - (
DAEMON_RECENT_HG_PULL * SECOND):
# SOME PUSHES ARE BIG, RUNNING THE RISK OTHER MACHINES ARE
# ALSO INTERESTED AND PERFORMING THE SAME SCAN. THIS DELAY
# WILL HAVE SMALL EFFECT ON THE MAJORITY OF SMALL PUSHES
# https://bugzilla.mozilla.org/show_bug.cgi?id=1417720
Till(seconds=Random.float(DAEMON_HG_INTERVAL *
2)).wait()
except Exception as e:
Log.warning(
"Scanning {{branch}} {{revision|left(12)}}",
branch=branch.name,
revision=r,
cause=e,
)
if "Read timed out" in e:
Till(seconds=DAEMON_WAIT_AFTER_TIMEOUT).wait()
# FIND ANY BRANCH THAT MAY HAVE THIS REVISION
for r in list(revisions):
self._find_revision(r)
def _daemon(self, please_stop):
while not please_stop:
with Explanation("looking for work"):
try:
branch, revisions = self.todo.pop(till=please_stop)
except Exception as e:
if please_stop:
break
else:
raise e
if branch.name in DAEMON_DO_NO_SCAN:
continue
revisions = set(revisions)
# FIND THE REVSIONS ON THIS BRANCH
for r in list(revisions):
try:
rev = self.get_revision(Revision(branch=branch, changeset={"id": r}))
if DAEMON_DEBUG:
Log.note("found revision with push date {{date|datetime}}", date=rev.push.date)
revisions.discard(r)
if rev.etl.timestamp > Date.now() - (DAEMON_RECENT_HG_PULL * SECOND):
# SOME PUSHES ARE BIG, RUNNING THE RISK OTHER MACHINES ARE
# ALSO INTERESTED AND PERFORMING THE SAME SCAN. THIS DELAY
# WILL HAVE SMALL EFFECT ON THE MAJORITY OF SMALL PUSHES
# https://bugzilla.mozilla.org/show_bug.cgi?id=1417720
Till(seconds=Random.float(DAEMON_HG_INTERVAL*2)).wait()
except Exception as e:
Log.warning(
"Scanning {{branch}} {{revision|left(12)}}",
branch=branch.name,
revision=r,
cause=e
)
if "Read timed out" in e:
Till(seconds=DAEMON_WAIT_AFTER_TIMEOUT).wait()
# FIND ANY BRANCH THAT MAY HAVE THIS REVISION
for r in list(revisions):
self._find_revision(r)
def test_wrap_3():
switch = [
lambda: Random.string(20), lambda: {
"i": Random.int(2000)
}, lambda: Data(i=Random.int(2000)),
lambda: FlatList([{
"i": Random.int(2000)
}]), lambda: [{
"i": Random.int(2000)
}]
]
inputs = [
switch[min(len(switch) - 1, int(floor(-log(Random.float(), 2))))]()
for i in range(NUM_INPUT)
]
for i in range(NUM_REPEAT):
results = []
gc.collect()
with Profiler("more string: slow_wrap"):
for v in inputs:
results.append(slow_wrap(v))
results = []
gc.collect()
with Profiler("more string: wrap"):
for v in inputs:
results.append(wrap(v))
results = []
gc.collect()
with Profiler("more string: baseline"):
for v in inputs:
results.append(baseline(v))
Log.note("Done {{i}} of {{num}}", {"i": i, "num": NUM_REPEAT})
def get_revision(self,
revision,
locale=None,
get_diff=False,
get_moves=True,
after=None):
"""
EXPECTING INCOMPLETE revision OBJECT
RETURNS revision
"""
rev = revision.changeset.id
if not rev:
return Null
elif rev == "None":
return Null
elif revision.branch.name == None:
return Null
locale = coalesce(locale, revision.branch.locale, DEFAULT_LOCALE)
output = self._get_from_elasticsearch(revision,
locale=locale,
get_diff=get_diff,
get_moves=get_moves,
after=after)
if output:
if not get_diff: # DIFF IS BIG, DO NOT KEEP IT IF NOT NEEDED
output.changeset.diff = None
if not get_moves:
output.changeset.moves = None
DEBUG and Log.note(
"Got hg ({{branch}}, {{locale}}, {{revision}}) from ES",
branch=output.branch.name,
locale=locale,
revision=output.changeset.id,
)
if output.push.date >= Date.now() - MAX_TODO_AGE:
self.todo.add((output.branch, listwrap(output.parents), None))
self.todo.add((output.branch, listwrap(output.children), None))
if output.push.date:
return output
# RATE LIMIT CALLS TO HG (CACHE MISSES)
next_cache_miss = self.last_cache_miss + (
Random.float(WAIT_AFTER_CACHE_MISS * 2) * SECOND)
self.last_cache_miss = Date.now()
if next_cache_miss > self.last_cache_miss:
Log.note(
"delaying next hg call for {{seconds|round(decimal=1)}} seconds",
seconds=next_cache_miss - self.last_cache_miss,
)
Till(till=next_cache_miss.unix).wait()
found_revision = copy(revision)
if isinstance(found_revision.branch, (text, binary_type)):
lower_name = found_revision.branch.lower()
else:
lower_name = found_revision.branch.name.lower()
if not lower_name:
Log.error("Defective revision? {{rev|json}}",
rev=found_revision.branch)
b = found_revision.branch = self.branches[(lower_name, locale)]
if not b:
b = found_revision.branch = self.branches[(lower_name,
DEFAULT_LOCALE)]
if not b:
Log.warning(
"can not find branch ({{branch}}, {{locale}})",
branch=lower_name,
locale=locale,
)
return Null
if Date.now() - Date(b.etl.timestamp) > _hg_branches.OLD_BRANCH:
self.branches = _hg_branches.get_branches(kwargs=self.settings)
push = self._get_push(found_revision.branch,
found_revision.changeset.id)
id12 = found_revision.changeset.id[0:12]
url1 = found_revision.branch.url.rstrip(
"/") + "/json-info?node=" + id12
url2 = found_revision.branch.url.rstrip("/") + "/json-rev/" + id12
url3 = (found_revision.branch.url.rstrip("/") +
"/json-automationrelevance/" + id12)
with Explanation("get revision from {{url}}", url=url1, debug=DEBUG):
raw_rev2 = Null
automation_details = Null
try:
raw_rev1 = self._get_raw_json_info(url1, found_revision.branch)
raw_rev2 = self._get_raw_json_rev(url2, found_revision.branch)
automation_details = self._get_raw_json_rev(
url3, found_revision.branch)
except Exception as e:
if "Hg denies it exists" in e:
raw_rev1 = Data(node=revision.changeset.id)
else:
raise e
raw_rev3_changeset = first(r for r in automation_details.changesets
if r.node[:12] == id12)
if last(automation_details.changesets) != raw_rev3_changeset:
Log.note("interesting")
output = self._normalize_revision(
set_default(raw_rev1, raw_rev2, raw_rev3_changeset),
found_revision,
push,
get_diff,
get_moves,
)
if output.push.date >= Date.now() - MAX_TODO_AGE:
self.todo.add((output.branch, listwrap(output.parents), None))
self.todo.add((output.branch, listwrap(output.children), None))
self.todo.add((output.branch, listwrap(output.backsoutnodes),
output.push.date))
if not get_diff: # DIFF IS BIG, DO NOT KEEP IT IF NOT NEEDED
output.changeset.diff = None
if not get_moves:
output.changeset.moves = None
return output
def _get_from_hg(self,
revision,
locale=None,
get_diff=False,
get_moves=True):
# RATE LIMIT CALLS TO HG (CACHE MISSES)
next_cache_miss = self.last_cache_miss + (
Random.float(WAIT_AFTER_CACHE_MISS * 2) * SECOND)
self.last_cache_miss = Date.now()
if next_cache_miss > self.last_cache_miss:
Log.note(
"delaying next hg call for {{seconds|round(decimal=1)}} seconds",
seconds=next_cache_miss - self.last_cache_miss,
)
Till(till=next_cache_miss.unix).wait()
# CLEAN UP BRANCH NAME
found_revision = copy(revision)
if isinstance(found_revision.branch, (text, binary_type)):
lower_name = found_revision.branch.lower()
else:
lower_name = found_revision.branch.name.lower()
if not lower_name:
Log.error("Defective revision? {{rev|json}}",
rev=found_revision.branch)
b = found_revision.branch = self.branches[(lower_name, locale)]
if not b:
b = found_revision.branch = self.branches[(lower_name,
DEFAULT_LOCALE)]
if not b:
Log.warning(
"can not find branch ({{branch}}, {{locale}})",
branch=lower_name,
locale=locale,
)
return Null
# REFRESH BRANCHES, IF TOO OLD
if Date.now() - Date(b.etl.timestamp) > _hg_branches.OLD_BRANCH:
self.branches = _hg_branches.get_branches(kwargs=self.settings)
# FIND THE PUSH
push = self._get_push(found_revision.branch,
found_revision.changeset.id)
id12 = found_revision.changeset.id[0:12]
base_url = URL(found_revision.branch.url)
with Explanation("get revision from {{url}}",
url=base_url,
debug=DEBUG):
raw_rev2 = Null
automation_details = Null
try:
raw_rev1 = self._get_raw_json_info((base_url / "json-info") +
{"node": id12})
raw_rev2 = self._get_raw_json_rev(base_url / "json-rev" / id12)
automation_details = self._get_raw_json_rev(
base_url / "json-automationrelevance" / id12)
except Exception as e:
if "Hg denies it exists" in e:
raw_rev1 = Data(node=revision.changeset.id)
else:
raise e
raw_rev3_changeset = first(r for r in automation_details.changesets
if r.node[:12] == id12)
if last(automation_details.changesets) != raw_rev3_changeset:
Log.note("interesting")
output = self._normalize_revision(
set_default(raw_rev1, raw_rev2, raw_rev3_changeset),
found_revision,
push,
get_diff,
get_moves,
)
if output.push.date >= Date.now() - MAX_TODO_AGE:
self.todo.extend([
(output.branch, listwrap(output.parents), None),
(output.branch, listwrap(output.children), None),
(
output.branch,
listwrap(output.backsoutnodes),
output.push.date,
),
])
if not get_diff: # DIFF IS BIG, DO NOT KEEP IT IF NOT NEEDED
output.changeset.diff = None
if not get_moves:
output.changeset.moves = None
return output
def get_revision(self, revision, locale=None, get_diff=False, get_moves=True):
"""
EXPECTING INCOMPLETE revision OBJECT
RETURNS revision
"""
rev = revision.changeset.id
if not rev:
return Null
elif rev == "None":
return Null
elif revision.branch.name == None:
return Null
locale = coalesce(locale, revision.branch.locale, DEFAULT_LOCALE)
output = self._get_from_elasticsearch(revision, locale=locale, get_diff=get_diff)
if output:
if not get_diff: # DIFF IS BIG, DO NOT KEEP IT IF NOT NEEDED
output.changeset.diff = None
if not get_moves:
output.changeset.moves = None
DEBUG and Log.note("Got hg ({{branch}}, {{locale}}, {{revision}}) from ES", branch=output.branch.name, locale=locale, revision=output.changeset.id)
if output.push.date >= Date.now()-MAX_TODO_AGE:
self.todo.add((output.branch, listwrap(output.parents)))
self.todo.add((output.branch, listwrap(output.children)))
if output.push.date:
return output
# RATE LIMIT CALLS TO HG (CACHE MISSES)
next_cache_miss = self.last_cache_miss + (Random.float(WAIT_AFTER_CACHE_MISS * 2) * SECOND)
self.last_cache_miss = Date.now()
if next_cache_miss > self.last_cache_miss:
Log.note("delaying next hg call for {{seconds|round(decimal=1)}}", seconds=next_cache_miss - self.last_cache_miss)
Till(till=next_cache_miss.unix).wait()
found_revision = copy(revision)
if isinstance(found_revision.branch, (text_type, binary_type)):
lower_name = found_revision.branch.lower()
else:
lower_name = found_revision.branch.name.lower()
if not lower_name:
Log.error("Defective revision? {{rev|json}}", rev=found_revision.branch)
b = found_revision.branch = self.branches[(lower_name, locale)]
if not b:
b = found_revision.branch = self.branches[(lower_name, DEFAULT_LOCALE)]
if not b:
Log.warning("can not find branch ({{branch}}, {{locale}})", branch=lower_name, locale=locale)
return Null
if Date.now() - Date(b.etl.timestamp) > _OLD_BRANCH:
self.branches = _hg_branches.get_branches(kwargs=self.settings)
push = self._get_push(found_revision.branch, found_revision.changeset.id)
url1 = found_revision.branch.url.rstrip("/") + "/json-info?node=" + found_revision.changeset.id[0:12]
url2 = found_revision.branch.url.rstrip("/") + "/json-rev/" + found_revision.changeset.id[0:12]
with Explanation("get revision from {{url}}", url=url1, debug=DEBUG):
raw_rev2 = Null
try:
raw_rev1 = self._get_raw_json_info(url1, found_revision.branch)
raw_rev2 = self._get_raw_json_rev(url2, found_revision.branch)
except Exception as e:
if "Hg denies it exists" in e:
raw_rev1 = Data(node=revision.changeset.id)
else:
raise e
output = self._normalize_revision(set_default(raw_rev1, raw_rev2), found_revision, push, get_diff, get_moves)
if output.push.date >= Date.now()-MAX_TODO_AGE:
self.todo.add((output.branch, listwrap(output.parents)))
self.todo.add((output.branch, listwrap(output.children)))
if not get_diff: # DIFF IS BIG, DO NOT KEEP IT IF NOT NEEDED
output.changeset.diff = None
if not get_moves:
output.changeset.moves = None
return output