def single_covers(title, authors, identifiers, caches, tdir):
patch_plugins()
load_caches(caches)
log = GUILog()
results = Queue()
worker = Thread(target=run_download, args=(log, results, Event()),
kwargs=dict(title=title, authors=authors, identifiers=identifiers))
worker.daemon = True
worker.start()
c = Counter()
while worker.is_alive():
try:
plugin, width, height, fmt, data = results.get(True, 1)
except Empty:
continue
else:
name = plugin.name
if plugin.can_get_multiple_covers:
name += '{%d}'%c[plugin.name]
c[plugin.name] += 1
name = '%s,,%s,,%s,,%s.cover'%(name, width, height, fmt)
with open(os.path.join(tdir, name), 'wb') as f:
f.write(data)
os.mkdir(os.path.join(tdir, name+'.done'))
return log.dump()
class DBThread(Thread):
CLOSE = '-------close---------'
def __init__(self, path, row_factory):
Thread.__init__(self)
self.setDaemon(True)
self.path = path
self.unhandled_error = (None, '')
self.row_factory = row_factory
self.requests = Queue(1)
self.results = Queue(1)
self.conn = None
def connect(self):
self.conn = do_connect(self.path, self.row_factory)
def run(self):
try:
self.connect()
while True:
func, args, kwargs = self.requests.get()
if func == self.CLOSE:
self.conn.close()
break
if func == 'dump':
try:
ok, res = True, tuple(self.conn.iterdump())
except Exception as err:
ok, res = False, (err, traceback.format_exc())
elif func == 'create_dynamic_filter':
try:
f = DynamicFilter(args[0])
self.conn.create_function(args[0], 1, f)
ok, res = True, f
except Exception as err:
ok, res = False, (err, traceback.format_exc())
else:
bfunc = getattr(self.conn, func)
try:
for i in range(3):
try:
ok, res = True, bfunc(*args, **kwargs)
break
except OperationalError as err:
# Retry if unable to open db file
e = str(err)
if 'unable to open' not in e or i == 2:
if 'unable to open' in e:
prints(
'Unable to open database for func',
func, reprlib.repr(args),
reprlib.repr(kwargs))
raise
time.sleep(0.5)
except Exception as err:
ok, res = False, (err, traceback.format_exc())
self.results.put((ok, res))
except Exception as err:
self.unhandled_error = (err, traceback.format_exc())
class Watcher(WatcherBase):
def __init__(self, root_dirs, worker, log):
WatcherBase.__init__(self, worker, log)
self.stream = Stream(self.notify, *(x.encode('utf-8') for x in root_dirs), file_events=True)
self.wait_queue = Queue()
def wakeup(self):
self.wait_queue.put(True)
def loop(self):
observer = Observer()
observer.schedule(self.stream)
observer.daemon = True
observer.start()
try:
while True:
try:
# Cannot use blocking get() as it is not interrupted by
# Ctrl-C
if self.wait_queue.get(10000) is True:
self.force_restart()
except Empty:
pass
finally:
observer.unschedule(self.stream)
observer.stop()
def notify(self, ev):
name = ev.name
if isinstance(name, bytes):
name = name.decode('utf-8')
if self.file_is_watched(name):
self.handle_modified({name})
class Watcher(WatcherBase):
def __init__(self, root_dirs, worker, log):
WatcherBase.__init__(self, worker, log)
self.stream = Stream(self.notify,
*(x.encode('utf-8') for x in root_dirs),
file_events=True)
self.wait_queue = Queue()
def wakeup(self):
self.wait_queue.put(True)
def loop(self):
observer = Observer()
observer.schedule(self.stream)
observer.daemon = True
observer.start()
try:
while True:
try:
# Cannot use blocking get() as it is not interrupted by
# Ctrl-C
if self.wait_queue.get(10000) is True:
self.force_restart()
except Empty:
pass
finally:
observer.unschedule(self.stream)
observer.stop()
def notify(self, ev):
name = ev.name
if isinstance(name, bytes):
name = name.decode('utf-8')
if self.file_is_watched(name):
self.handle_modified({name})
def single_covers(title, authors, identifiers, caches, tdir):
patch_plugins()
load_caches(caches)
log = GUILog()
results = Queue()
worker = Thread(target=run_download, args=(log, results, Event()),
kwargs=dict(title=title, authors=authors, identifiers=identifiers))
worker.daemon = True
worker.start()
c = Counter()
while worker.is_alive():
try:
plugin, width, height, fmt, data = results.get(True, 1)
except Empty:
continue
else:
name = plugin.name
if plugin.can_get_multiple_covers:
name += '{%d}'%c[plugin.name]
c[plugin.name] += 1
name = '%s,,%s,,%s,,%s.cover'%(name, width, height, fmt)
with open(os.path.join(tdir, name), 'wb') as f:
f.write(data)
os.mkdir(os.path.join(tdir, name+'.done'))
return log.dump()
class DBThread(Thread):
CLOSE = '-------close---------'
def __init__(self, path, row_factory):
Thread.__init__(self)
self.setDaemon(True)
self.path = path
self.unhandled_error = (None, '')
self.row_factory = row_factory
self.requests = Queue(1)
self.results = Queue(1)
self.conn = None
def connect(self):
self.conn = do_connect(self.path, self.row_factory)
def run(self):
try:
self.connect()
while True:
func, args, kwargs = self.requests.get()
if func == self.CLOSE:
self.conn.close()
break
if func == 'dump':
try:
ok, res = True, tuple(self.conn.iterdump())
except Exception as err:
ok, res = False, (err, traceback.format_exc())
elif func == 'create_dynamic_filter':
try:
f = DynamicFilter(args[0])
self.conn.create_function(args[0], 1, f)
ok, res = True, f
except Exception as err:
ok, res = False, (err, traceback.format_exc())
else:
bfunc = getattr(self.conn, func)
try:
for i in range(3):
try:
ok, res = True, bfunc(*args, **kwargs)
break
except OperationalError as err:
# Retry if unable to open db file
e = str(err)
if 'unable to open' not in e or i == 2:
if 'unable to open' in e:
prints('Unable to open database for func',
func, reprlib.repr(args),
reprlib.repr(kwargs))
raise
time.sleep(0.5)
except Exception as err:
ok, res = False, (err, traceback.format_exc())
self.results.put((ok, res))
except Exception as err:
self.unhandled_error = (err, traceback.format_exc())
class AnnotationsSaveWorker(Thread):
def __init__(self):
Thread.__init__(self, name='AnnotSaveWorker')
self.daemon = True
self.queue = Queue()
def shutdown(self):
if self.is_alive():
self.queue.put(None)
self.join()
def run(self):
while True:
x = self.queue.get()
if x is None:
return
annotations_list = x['annotations_list']
annotations_path_key = x['annotations_path_key']
bld = x['book_library_details']
pathtoebook = x['pathtoebook']
in_book_file = x['in_book_file']
sync_annots_user = x['sync_annots_user']
try:
save_annotations(annotations_list, annotations_path_key, bld,
pathtoebook, in_book_file, sync_annots_user)
except Exception:
import traceback
traceback.print_exc()
def save_annotations(self,
current_book_data,
in_book_file=True,
sync_annots_user=''):
alist = tuple(
annotations_as_copied_list(current_book_data['annotations_map']))
ebp = current_book_data['pathtoebook']
can_save_in_book_file = ebp.lower().endswith('.epub')
self.queue.put({
'annotations_list':
alist,
'annotations_path_key':
current_book_data['annotations_path_key'],
'book_library_details':
current_book_data['book_library_details'],
'pathtoebook':
current_book_data['pathtoebook'],
'in_book_file':
in_book_file and can_save_in_book_file,
'sync_annots_user':
sync_annots_user,
})
class Progress(Thread):
def __init__(self, conn):
Thread.__init__(self)
self.daemon = True
self.conn = conn
self.queue = Queue()
def __call__(self, percent, msg=''):
self.queue.put((percent, msg))
def run(self):
while True:
x = self.queue.get()
if x is None:
break
try:
eintr_retry_call(self.conn.send, x)
except:
break
class Progress(Thread):
def __init__(self, conn):
Thread.__init__(self)
self.daemon = True
self.conn = conn
self.queue = Queue()
def __call__(self, percent, msg=''):
self.queue.put((percent, msg))
def run(self):
while True:
x = self.queue.get()
if x is None:
break
try:
eintr_retry_call(self.conn.send, x)
except:
break
class Watcher(WatcherBase):
def __init__(self, root_dirs, worker, log):
WatcherBase.__init__(self, worker, log)
self.watchers = []
self.modified_queue = Queue()
for d in frozenset(root_dirs):
self.watchers.append(TreeWatcher(d, self.modified_queue))
def wakeup(self):
self.modified_queue.put(True)
def loop(self):
for w in self.watchers:
w.start()
with HandleInterrupt(lambda: self.modified_queue.put(None)):
while True:
path = self.modified_queue.get()
if path is None:
break
if path is True:
self.force_restart()
else:
self.handle_modified({path})
class Watcher(WatcherBase):
def __init__(self, root_dirs, worker, log):
WatcherBase.__init__(self, worker, log)
self.watchers = []
self.modified_queue = Queue()
for d in frozenset(root_dirs):
self.watchers.append(TreeWatcher(d, self.modified_queue))
def wakeup(self):
self.modified_queue.put(True)
def loop(self):
for w in self.watchers:
w.start()
with HandleInterrupt(lambda : self.modified_queue.put(None)):
while True:
path = self.modified_queue.get()
if path is None:
break
if path is True:
self.force_restart()
else:
self.handle_modified({path})
class Server(Thread):
def __init__(self,
notify_on_job_done=lambda x: x,
pool_size=None,
limit=sys.maxsize,
enforce_cpu_limit=True):
Thread.__init__(self)
self.daemon = True
self.id = next(server_counter) + 1
if enforce_cpu_limit:
limit = min(limit, cpu_count())
self.pool_size = limit if pool_size is None else pool_size
self.notify_on_job_done = notify_on_job_done
self.add_jobs_queue, self.changed_jobs_queue = Queue(), Queue()
self.kill_queue = Queue()
self.waiting_jobs = []
self.workers = deque()
self.launched_worker_counter = count()
next(self.launched_worker_counter)
self.start()
def launch_worker(self, gui=False, redirect_output=None, job_name=None):
start = time.monotonic()
id = next(self.launched_worker_counter)
fd, rfile = tempfile.mkstemp(prefix='ipc_result_%d_%d_' %
(self.id, id),
dir=base_dir(),
suffix='.pickle')
os.close(fd)
if redirect_output is None:
redirect_output = not gui
cw = self.do_launch(gui, redirect_output, rfile, job_name=job_name)
if isinstance(cw, string_or_bytes):
raise CriticalError('Failed to launch worker process:\n' +
force_unicode(cw))
if DEBUG:
print(
f'Worker Launch took: {time.monotonic() - start:.2f} seconds')
return cw
def do_launch(self, gui, redirect_output, rfile, job_name=None):
a, b = Pipe()
with a:
env = {
'CALIBRE_WORKER_FD': str(a.fileno()),
'CALIBRE_WORKER_RESULT': environ_item(as_hex_unicode(rfile))
}
w = Worker(env, gui=gui, job_name=job_name)
try:
w(pass_fds=(a.fileno(), ), redirect_output=redirect_output)
except BaseException:
try:
w.kill()
except:
pass
b.close()
import traceback
return traceback.format_exc()
return ConnectedWorker(w, b, rfile)
def add_job(self, job):
job.done2 = self.notify_on_job_done
self.add_jobs_queue.put(job)
def run_job(self, job, gui=True, redirect_output=False):
w = self.launch_worker(gui=gui,
redirect_output=redirect_output,
job_name=getattr(job, 'name', None))
w.start_job(job)
def run(self):
while True:
try:
job = self.add_jobs_queue.get(True, 0.2)
if job is None:
break
self.waiting_jobs.insert(0, job)
except Empty:
pass
# Get notifications from worker process
for worker in self.workers:
while True:
try:
n = worker.notifications.get_nowait()
worker.job.notifications.put(n)
self.changed_jobs_queue.put(worker.job)
except Empty:
break
# Remove finished jobs
for worker in [w for w in self.workers if not w.is_alive]:
try:
worker.close_log_file()
except:
pass
self.workers.remove(worker)
job = worker.job
if worker.returncode != 0:
job.failed = True
job.returncode = worker.returncode
elif os.path.exists(worker.rfile):
try:
with lopen(worker.rfile, 'rb') as f:
job.result = pickle_loads(f.read())
os.remove(worker.rfile)
except:
pass
job.duration = time.time() - job.start_time
self.changed_jobs_queue.put(job)
# Start waiting jobs
sj = self.suitable_waiting_job()
if sj is not None:
job = self.waiting_jobs.pop(sj)
job.start_time = time.time()
if job.kill_on_start:
job.duration = 0.0
job.returncode = 1
job.killed = job.failed = True
job.result = None
else:
worker = self.launch_worker()
worker.start_job(job)
self.workers.append(worker)
job.log_path = worker.log_path
self.changed_jobs_queue.put(job)
while True:
try:
j = self.kill_queue.get_nowait()
self._kill_job(j)
except Empty:
break
def suitable_waiting_job(self):
available_workers = self.pool_size - len(self.workers)
for worker in self.workers:
job = worker.job
if job.core_usage == -1:
available_workers = 0
elif job.core_usage > 1:
available_workers -= job.core_usage - 1
if available_workers < 1:
return None
for i, job in enumerate(self.waiting_jobs):
if job.core_usage == -1:
if available_workers >= self.pool_size:
return i
elif job.core_usage <= available_workers:
return i
def kill_job(self, job):
self.kill_queue.put(job)
def killall(self):
for worker in self.workers:
self.kill_queue.put(worker.job)
def _kill_job(self, job):
if job.start_time is None:
job.kill_on_start = True
return
for worker in self.workers:
if job is worker.job:
worker.kill()
job.killed = True
break
def split(self, tasks):
'''
Split a list into a list of sub lists, with the number of sub lists being
no more than the number of workers this server supports. Each sublist contains
2-tuples of the form (i, x) where x is an element from the original list
and i is the index of the element x in the original list.
'''
ans, count, pos = [], 0, 0
delta = int(ceil(len(tasks) / float(self.pool_size)))
while count < len(tasks):
section = []
for t in tasks[pos:pos + delta]:
section.append((count, t))
count += 1
ans.append(section)
pos += delta
return ans
def close(self):
try:
self.add_jobs_queue.put(None)
except:
pass
try:
self.listener.close()
except:
pass
time.sleep(0.2)
for worker in list(self.workers):
try:
worker.kill()
except:
pass
def __enter__(self):
return self
def __exit__(self, *args):
self.close()
class JobsManager(object):
def __init__(self, opts, log):
mj = opts.max_jobs
if mj < 1:
mj = detect_ncpus()
self.log = log
self.max_jobs = max(1, mj)
self.max_job_time = max(0, opts.max_job_time * 60)
self.lock = RLock()
self.jobs = {}
self.finished_jobs = {}
self.events = Queue()
self.job_id = count()
self.waiting_job_ids = set()
self.waiting_jobs = deque()
self.max_block = None
self.shutting_down = False
self.event_loop = None
def start_job(self,
name,
module,
func,
args=(),
kwargs=None,
job_done_callback=None,
job_data=None):
with self.lock:
if self.shutting_down:
return None
if self.event_loop is None:
self.event_loop = t = Thread(name='JobsEventLoop',
target=self.run)
t.daemon = True
t.start()
job_id = next(self.job_id)
self.events.put(
StartEvent(job_id, name, module, func, args, kwargs or {},
job_done_callback, job_data))
self.waiting_job_ids.add(job_id)
return job_id
def job_status(self, job_id):
with self.lock:
if not self.shutting_down:
if job_id in self.finished_jobs:
job = self.finished_jobs[job_id]
return 'finished', job.result, job.traceback, job.was_aborted
if job_id in self.jobs:
return 'running', None, None, None
if job_id in self.waiting_job_ids:
return 'waiting', None, None, None
return None, None, None, None
def abort_job(self, job_id):
job = self.jobs.get(job_id)
if job is not None:
job.abort_event.set()
def wait_for_running_job(self, job_id, timeout=None):
job = self.jobs.get(job_id)
if job is not None:
job.wait_for_end.wait(timeout)
if not job.done:
return False
while job_id not in self.finished_jobs:
time.sleep(0.001)
return True
def shutdown(self, timeout=5.0):
with self.lock:
self.shutting_down = True
for job in itervalues(self.jobs):
job.abort_event.set()
self.events.put(False)
def wait_for_shutdown(self, wait_till):
for job in itervalues(self.jobs):
delta = wait_till - monotonic()
if delta > 0:
job.join(delta)
if self.event_loop is not None:
delta = wait_till - monotonic()
if delta > 0:
self.event_loop.join(delta)
# Internal API {{{
def run(self):
while not self.shutting_down:
if self.max_block is None:
ev = self.events.get()
else:
try:
ev = self.events.get(block=True, timeout=self.max_block)
except Empty:
ev = None
if self.shutting_down:
break
if ev is None:
self.abort_hanging_jobs()
elif isinstance(ev, StartEvent):
self.waiting_jobs.append(ev)
self.start_waiting_jobs()
elif isinstance(ev, DoneEvent):
self.job_finished(ev.job_id)
elif ev is False:
break
def start_waiting_jobs(self):
with self.lock:
while self.waiting_jobs and len(self.jobs) < self.max_jobs:
ev = self.waiting_jobs.popleft()
self.jobs[ev.job_id] = Job(ev, self.events)
self.waiting_job_ids.discard(ev.job_id)
self.update_max_block()
def update_max_block(self):
with self.lock:
mb = None
now = monotonic()
for job in itervalues(self.jobs):
if not job.done and not job.abort_event.is_set():
delta = self.max_job_time - (now - job.start_time)
if delta <= 0:
self.max_block = 0
return
if mb is None:
mb = delta
else:
mb = min(mb, delta)
self.max_block = mb
def abort_hanging_jobs(self):
now = monotonic()
found = False
for job in itervalues(self.jobs):
if not job.done and not job.abort_event.is_set():
delta = self.max_job_time - (now - job.start_time)
if delta <= 0:
job.abort_event.set()
found = True
if found:
self.update_max_block()
def job_finished(self, job_id):
with self.lock:
self.finished_jobs[job_id] = job = self.jobs.pop(job_id)
if job.callback is not None:
try:
job.callback(job)
except Exception:
import traceback
self.log.error('Error running callback for job: %s:\n%s' %
(job.name, traceback.format_exc()))
self.prune_finished_jobs()
if job.traceback and not job.was_aborted:
logdata = job.read_log()
self.log.error('The job: %s failed:\n%s\n%s' %
(job.job_name, logdata, job.traceback))
job.remove_log()
self.start_waiting_jobs()
def prune_finished_jobs(self):
with self.lock:
remove = []
now = monotonic()
for job_id, job in iteritems(self.finished_jobs):
if now - job.end_time > 3600:
remove.append(job_id)
for job_id in remove:
del self.finished_jobs[job_id]
class JobsManager(object):
def __init__(self, opts, log):
mj = opts.max_jobs
if mj < 1:
mj = detect_ncpus()
self.log = log
self.max_jobs = max(1, mj)
self.max_job_time = max(0, opts.max_job_time * 60)
self.lock = RLock()
self.jobs = {}
self.finished_jobs = {}
self.events = Queue()
self.job_id = count()
self.waiting_job_ids = set()
self.waiting_jobs = deque()
self.max_block = None
self.shutting_down = False
self.event_loop = None
def start_job(self, name, module, func, args=(), kwargs=None, job_done_callback=None, job_data=None):
with self.lock:
if self.shutting_down:
return None
if self.event_loop is None:
self.event_loop = t = Thread(name='JobsEventLoop', target=self.run)
t.daemon = True
t.start()
job_id = next(self.job_id)
self.events.put(StartEvent(job_id, name, module, func, args, kwargs or {}, job_done_callback, job_data))
self.waiting_job_ids.add(job_id)
return job_id
def job_status(self, job_id):
with self.lock:
if not self.shutting_down:
if job_id in self.finished_jobs:
job = self.finished_jobs[job_id]
return 'finished', job.result, job.traceback, job.was_aborted
if job_id in self.jobs:
return 'running', None, None, None
if job_id in self.waiting_job_ids:
return 'waiting', None, None, None
return None, None, None, None
def abort_job(self, job_id):
job = self.jobs.get(job_id)
if job is not None:
job.abort_event.set()
def wait_for_running_job(self, job_id, timeout=None):
job = self.jobs.get(job_id)
if job is not None:
job.wait_for_end.wait(timeout)
if not job.done:
return False
while job_id not in self.finished_jobs:
time.sleep(0.001)
return True
def shutdown(self, timeout=5.0):
with self.lock:
self.shutting_down = True
for job in itervalues(self.jobs):
job.abort_event.set()
self.events.put(False)
def wait_for_shutdown(self, wait_till):
for job in itervalues(self.jobs):
delta = wait_till - monotonic()
if delta > 0:
job.join(delta)
if self.event_loop is not None:
delta = wait_till - monotonic()
if delta > 0:
self.event_loop.join(delta)
# Internal API {{{
def run(self):
while not self.shutting_down:
if self.max_block is None:
ev = self.events.get()
else:
try:
ev = self.events.get(block=True, timeout=self.max_block)
except Empty:
ev = None
if self.shutting_down:
break
if ev is None:
self.abort_hanging_jobs()
elif isinstance(ev, StartEvent):
self.waiting_jobs.append(ev)
self.start_waiting_jobs()
elif isinstance(ev, DoneEvent):
self.job_finished(ev.job_id)
elif ev is False:
break
def start_waiting_jobs(self):
with self.lock:
while self.waiting_jobs and len(self.jobs) < self.max_jobs:
ev = self.waiting_jobs.popleft()
self.jobs[ev.job_id] = Job(ev, self.events)
self.waiting_job_ids.discard(ev.job_id)
self.update_max_block()
def update_max_block(self):
with self.lock:
mb = None
now = monotonic()
for job in itervalues(self.jobs):
if not job.done and not job.abort_event.is_set():
delta = self.max_job_time - (now - job.start_time)
if delta <= 0:
self.max_block = 0
return
if mb is None:
mb = delta
else:
mb = min(mb, delta)
self.max_block = mb
def abort_hanging_jobs(self):
now = monotonic()
found = False
for job in itervalues(self.jobs):
if not job.done and not job.abort_event.is_set():
delta = self.max_job_time - (now - job.start_time)
if delta <= 0:
job.abort_event.set()
found = True
if found:
self.update_max_block()
def job_finished(self, job_id):
with self.lock:
self.finished_jobs[job_id] = job = self.jobs.pop(job_id)
if job.callback is not None:
try:
job.callback(job)
except Exception:
import traceback
self.log.error('Error running callback for job: %s:\n%s' % (job.name, traceback.format_exc()))
self.prune_finished_jobs()
if job.traceback and not job.was_aborted:
logdata = job.read_log()
self.log.error('The job: %s failed:\n%s\n%s' % (job.job_name, logdata, job.traceback))
job.remove_log()
self.start_waiting_jobs()
def prune_finished_jobs(self):
with self.lock:
remove = []
now = monotonic()
for job_id, job in iteritems(self.finished_jobs):
if now - job.end_time > 3600:
remove.append(job_id)
for job_id in remove:
del self.finished_jobs[job_id]
class ParseWorker(Thread):
daemon = True
SLEEP_TIME = 1
def __init__(self):
Thread.__init__(self)
self.requests = Queue()
self.request_count = 0
self.parse_items = {}
self.launch_error = None
def run(self):
mod, func = 'calibre.gui2.tweak_book.preview', 'parse_html'
try:
# Connect to the worker and send a dummy job to initialize it
self.worker = offload_worker(priority='low')
self.worker(mod, func, '<p></p>')
except:
import traceback
traceback.print_exc()
self.launch_error = traceback.format_exc()
return
while True:
time.sleep(self.SLEEP_TIME)
x = self.requests.get()
requests = [x]
while True:
try:
requests.append(self.requests.get_nowait())
except Empty:
break
if shutdown in requests:
self.worker.shutdown()
break
request = sorted(requests, reverse=True)[0]
del requests
pi, data = request[1:]
try:
res = self.worker(mod, func, data)
except:
import traceback
traceback.print_exc()
else:
pi.parsing_done = True
parsed_data = res['result']
if res['tb']:
prints("Parser error:")
prints(res['tb'])
else:
pi.parsed_data = parsed_data
def add_request(self, name):
data = get_data(name)
ldata, hdata = len(data), hash(data)
pi = self.parse_items.get(name, None)
if pi is None:
self.parse_items[name] = pi = ParseItem(name)
else:
if pi.parsing_done and pi.length == ldata and pi.fingerprint == hdata:
return
pi.parsed_data = None
pi.parsing_done = False
pi.length, pi.fingerprint = ldata, hdata
self.requests.put((self.request_count, pi, data))
self.request_count += 1
def shutdown(self):
self.requests.put(shutdown)
def get_data(self, name):
return getattr(self.parse_items.get(name, None), 'parsed_data', None)
def clear(self):
self.parse_items.clear()
def is_alive(self):
return Thread.is_alive(self) or (hasattr(self, 'worker')
and self.worker.is_alive())
class SearchPanel(QWidget): # {{{
search_requested = pyqtSignal(object)
results_found = pyqtSignal(object)
show_search_result = pyqtSignal(object)
def __init__(self, parent=None):
QWidget.__init__(self, parent)
self.last_hidden_text_warning = None
self.current_search = None
self.l = l = QVBoxLayout(self)
l.setContentsMargins(0, 0, 0, 0)
self.search_input = si = SearchInput(self)
self.searcher = None
self.search_tasks = Queue()
self.results_found.connect(self.on_result_found, type=Qt.QueuedConnection)
si.do_search.connect(self.search_requested)
l.addWidget(si)
self.results = r = Results(self)
r.show_search_result.connect(self.do_show_search_result, type=Qt.QueuedConnection)
r.currentRowChanged.connect(self.update_hidden_message)
l.addWidget(r, 100)
self.spinner = s = BusySpinner(self)
s.setVisible(False)
l.addWidget(s)
self.hidden_message = la = QLabel(_('This text is hidden in the book and cannot be displayed'))
la.setStyleSheet('QLabel { margin-left: 1ex }')
la.setWordWrap(True)
la.setVisible(False)
l.addWidget(la)
def update_hidden_message(self):
self.hidden_message.setVisible(self.results.current_result_is_hidden)
def focus_input(self, text=None):
self.search_input.focus_input(text)
def start_search(self, search_query, current_name):
if self.current_search is not None and search_query == self.current_search:
self.find_next_requested(search_query.backwards)
return
if self.searcher is None:
self.searcher = Thread(name='Searcher', target=self.run_searches)
self.searcher.daemon = True
self.searcher.start()
self.results.clear()
self.hidden_message.setVisible(False)
self.spinner.start()
self.current_search = search_query
self.last_hidden_text_warning = None
self.search_tasks.put((search_query, current_name))
def run_searches(self):
while True:
x = self.search_tasks.get()
if x is None:
break
search_query, current_name = x
try:
manifest = get_manifest() or {}
spine = manifest.get('spine', ())
idx_map = {name: i for i, name in enumerate(spine)}
spine_idx = idx_map.get(current_name, -1)
except Exception:
import traceback
traceback.print_exc()
spine_idx = -1
if spine_idx < 0:
self.results_found.emit(SearchFinished(search_query))
continue
for name in spine:
counter = Counter()
spine_idx = idx_map[name]
try:
for i, result in enumerate(search_in_name(name, search_query)):
before, text, after = result
q = (before or '')[-5:] + text + (after or '')[:5]
self.results_found.emit(SearchResult(search_query, before, text, after, q, name, spine_idx, counter[q]))
counter[q] += 1
except Exception:
import traceback
traceback.print_exc()
self.results_found.emit(SearchFinished(search_query))
def on_result_found(self, result):
if self.current_search is None or result.search_query != self.current_search:
return
if isinstance(result, SearchFinished):
self.spinner.stop()
if not self.results.count():
self.show_no_results_found()
return
if self.results.add_result(result) == 1:
# first result
self.results.setCurrentRow(0)
self.results.item_activated()
self.update_hidden_message()
def visibility_changed(self, visible):
if visible:
self.focus_input()
def clear_searches(self):
self.current_search = None
self.last_hidden_text_warning = None
searchable_text_for_name.cache_clear()
self.spinner.stop()
self.results.clear()
def shutdown(self):
self.search_tasks.put(None)
self.spinner.stop()
self.current_search = None
self.last_hidden_text_warning = None
self.searcher = None
def find_next_requested(self, previous):
self.results.find_next(previous)
def do_show_search_result(self, sr):
self.show_search_result.emit(sr.for_js)
def search_result_not_found(self, sr):
self.results.search_result_not_found(sr)
self.update_hidden_message()
def show_no_results_found(self):
msg = _('No matches were found for:')
warning_dialog(self, _('No matches found'), msg + ' <b>{}</b>'.format(self.current_search.text), show=True)
class GenericDownloadThreadPool(object):
'''
add_task must be implemented in a subclass and must
GenericDownloadThreadPool.add_task must be called
at the end of the function.
'''
def __init__(self, thread_type, thread_count=1):
self.thread_type = thread_type
self.thread_count = thread_count
self.tasks = Queue()
self.results = Queue()
self.threads = []
def set_thread_count(self, thread_count):
self.thread_count = thread_count
def add_task(self):
'''
This must be implemented in a sub class and this function
must be called at the end of the add_task function in
the sub class.
The implementation of this function (in this base class)
starts any threads necessary to fill the pool if it is
not already full.
'''
for i in range(self.thread_count - self.running_threads_count()):
t = self.thread_type(self.tasks, self.results)
self.threads.append(t)
t.start()
def abort(self):
self.tasks = Queue()
self.results = Queue()
for t in self.threads:
t.abort()
self.threads = []
def has_tasks(self):
return not self.tasks.empty()
def get_result(self):
return self.results.get()
def get_result_no_wait(self):
return self.results.get_nowait()
def result_count(self):
return len(self.results)
def has_results(self):
return not self.results.empty()
def threads_running(self):
return self.running_threads_count() > 0
def running_threads_count(self):
count = 0
for t in self.threads:
if t.is_alive():
count += 1
return count
class ParseWorker(Thread):
daemon = True
SLEEP_TIME = 1
def __init__(self):
Thread.__init__(self)
self.requests = Queue()
self.request_count = 0
self.parse_items = {}
self.launch_error = None
def run(self):
mod, func = 'calibre.gui2.tweak_book.preview', 'parse_html'
try:
# Connect to the worker and send a dummy job to initialize it
self.worker = offload_worker(priority='low')
self.worker(mod, func, '<p></p>')
except:
import traceback
traceback.print_exc()
self.launch_error = traceback.format_exc()
return
while True:
time.sleep(self.SLEEP_TIME)
x = self.requests.get()
requests = [x]
while True:
try:
requests.append(self.requests.get_nowait())
except Empty:
break
if shutdown in requests:
self.worker.shutdown()
break
request = sorted(requests, reverse=True)[0]
del requests
pi, data = request[1:]
try:
res = self.worker(mod, func, data)
except:
import traceback
traceback.print_exc()
else:
pi.parsing_done = True
parsed_data = res['result']
if res['tb']:
prints("Parser error:")
prints(res['tb'])
else:
pi.parsed_data = parsed_data
def add_request(self, name):
data = get_data(name)
ldata, hdata = len(data), hash(data)
pi = self.parse_items.get(name, None)
if pi is None:
self.parse_items[name] = pi = ParseItem(name)
else:
if pi.parsing_done and pi.length == ldata and pi.fingerprint == hdata:
return
pi.parsed_data = None
pi.parsing_done = False
pi.length, pi.fingerprint = ldata, hdata
self.requests.put((self.request_count, pi, data))
self.request_count += 1
def shutdown(self):
self.requests.put(shutdown)
def get_data(self, name):
return getattr(self.parse_items.get(name, None), 'parsed_data', None)
def clear(self):
self.parse_items.clear()
def is_alive(self):
return Thread.is_alive(self) or (hasattr(self, 'worker') and self.worker.is_alive())
class Server(Thread):
def __init__(self, notify_on_job_done=lambda x: x, pool_size=None,
limit=sys.maxsize, enforce_cpu_limit=True):
Thread.__init__(self)
self.daemon = True
global _counter
self.id = _counter+1
_counter += 1
if enforce_cpu_limit:
limit = min(limit, cpu_count())
self.pool_size = limit if pool_size is None else pool_size
self.notify_on_job_done = notify_on_job_done
self.auth_key = os.urandom(32)
self.address, self.listener = create_listener(self.auth_key, backlog=4)
self.add_jobs_queue, self.changed_jobs_queue = Queue(), Queue()
self.kill_queue = Queue()
self.waiting_jobs = []
self.workers = deque()
self.launched_worker_count = 0
self._worker_launch_lock = RLock()
self.start()
def launch_worker(self, gui=False, redirect_output=None, job_name=None):
start = time.time()
with self._worker_launch_lock:
self.launched_worker_count += 1
id = self.launched_worker_count
fd, rfile = tempfile.mkstemp(prefix=u'ipc_result_%d_%d_'%(self.id, id),
dir=base_dir(), suffix=u'.pickle')
os.close(fd)
if redirect_output is None:
redirect_output = not gui
env = {
'CALIBRE_WORKER_ADDRESS' : environ_item(as_hex_unicode(msgpack_dumps(self.address))),
'CALIBRE_WORKER_KEY' : environ_item(as_hex_unicode(self.auth_key)),
'CALIBRE_WORKER_RESULT' : environ_item(as_hex_unicode(rfile)),
}
cw = self.do_launch(env, gui, redirect_output, rfile, job_name=job_name)
if isinstance(cw, string_or_bytes):
raise CriticalError('Failed to launch worker process:\n'+cw)
if DEBUG:
print('Worker Launch took:', time.time() - start)
return cw
def do_launch(self, env, gui, redirect_output, rfile, job_name=None):
w = Worker(env, gui=gui, job_name=job_name)
try:
w(redirect_output=redirect_output)
conn = eintr_retry_call(self.listener.accept)
if conn is None:
raise Exception('Failed to launch worker process')
except BaseException:
try:
w.kill()
except:
pass
import traceback
return traceback.format_exc()
return ConnectedWorker(w, conn, rfile)
def add_job(self, job):
job.done2 = self.notify_on_job_done
self.add_jobs_queue.put(job)
def run_job(self, job, gui=True, redirect_output=False):
w = self.launch_worker(gui=gui, redirect_output=redirect_output, job_name=getattr(job, 'name', None))
w.start_job(job)
def run(self):
while True:
try:
job = self.add_jobs_queue.get(True, 0.2)
if job is None:
break
self.waiting_jobs.insert(0, job)
except Empty:
pass
# Get notifications from worker process
for worker in self.workers:
while True:
try:
n = worker.notifications.get_nowait()
worker.job.notifications.put(n)
self.changed_jobs_queue.put(worker.job)
except Empty:
break
# Remove finished jobs
for worker in [w for w in self.workers if not w.is_alive]:
try:
worker.close_log_file()
except:
pass
self.workers.remove(worker)
job = worker.job
if worker.returncode != 0:
job.failed = True
job.returncode = worker.returncode
elif os.path.exists(worker.rfile):
try:
with lopen(worker.rfile, 'rb') as f:
job.result = pickle_loads(f.read())
os.remove(worker.rfile)
except:
pass
job.duration = time.time() - job.start_time
self.changed_jobs_queue.put(job)
# Start waiting jobs
sj = self.suitable_waiting_job()
if sj is not None:
job = self.waiting_jobs.pop(sj)
job.start_time = time.time()
if job.kill_on_start:
job.duration = 0.0
job.returncode = 1
job.killed = job.failed = True
job.result = None
else:
worker = self.launch_worker()
worker.start_job(job)
self.workers.append(worker)
job.log_path = worker.log_path
self.changed_jobs_queue.put(job)
while True:
try:
j = self.kill_queue.get_nowait()
self._kill_job(j)
except Empty:
break
def suitable_waiting_job(self):
available_workers = self.pool_size - len(self.workers)
for worker in self.workers:
job = worker.job
if job.core_usage == -1:
available_workers = 0
elif job.core_usage > 1:
available_workers -= job.core_usage - 1
if available_workers < 1:
return None
for i, job in enumerate(self.waiting_jobs):
if job.core_usage == -1:
if available_workers >= self.pool_size:
return i
elif job.core_usage <= available_workers:
return i
def kill_job(self, job):
self.kill_queue.put(job)
def killall(self):
for worker in self.workers:
self.kill_queue.put(worker.job)
def _kill_job(self, job):
if job.start_time is None:
job.kill_on_start = True
return
for worker in self.workers:
if job is worker.job:
worker.kill()
job.killed = True
break
def split(self, tasks):
'''
Split a list into a list of sub lists, with the number of sub lists being
no more than the number of workers this server supports. Each sublist contains
2-tuples of the form (i, x) where x is an element from the original list
and i is the index of the element x in the original list.
'''
ans, count, pos = [], 0, 0
delta = int(ceil(len(tasks)/float(self.pool_size)))
while count < len(tasks):
section = []
for t in tasks[pos:pos+delta]:
section.append((count, t))
count += 1
ans.append(section)
pos += delta
return ans
def close(self):
try:
self.add_jobs_queue.put(None)
except:
pass
try:
self.listener.close()
except:
pass
time.sleep(0.2)
for worker in list(self.workers):
try:
worker.kill()
except:
pass
def __enter__(self):
return self
def __exit__(self, *args):
self.close()
class GenericDownloadThreadPool(object):
'''
add_task must be implemented in a subclass and must
GenericDownloadThreadPool.add_task must be called
at the end of the function.
'''
def __init__(self, thread_type, thread_count=1):
self.thread_type = thread_type
self.thread_count = thread_count
self.tasks = Queue()
self.results = Queue()
self.threads = []
def set_thread_count(self, thread_count):
self.thread_count = thread_count
def add_task(self):
'''
This must be implemented in a sub class and this function
must be called at the end of the add_task function in
the sub class.
The implementation of this function (in this base class)
starts any threads necessary to fill the pool if it is
not already full.
'''
for i in range(self.thread_count - self.running_threads_count()):
t = self.thread_type(self.tasks, self.results)
self.threads.append(t)
t.start()
def abort(self):
self.tasks = Queue()
self.results = Queue()
for t in self.threads:
t.abort()
self.threads = []
def has_tasks(self):
return not self.tasks.empty()
def get_result(self):
return self.results.get()
def get_result_no_wait(self):
return self.results.get_nowait()
def result_count(self):
return len(self.results)
def has_results(self):
return not self.results.empty()
def threads_running(self):
return self.running_threads_count() > 0
def running_threads_count(self):
count = 0
for t in self.threads:
if t.is_alive():
count += 1
return count
class Pool(Thread):
daemon = True
def __init__(self, max_workers=None, name=None):
Thread.__init__(self, name=name)
self.max_workers = max_workers or detect_ncpus()
self.available_workers = []
self.busy_workers = {}
self.pending_jobs = []
self.events = Queue()
self.results = Queue()
self.tracker = Queue()
self.terminal_failure = None
self.common_data = pickle_dumps(None)
self.worker_data = None
self.shutting_down = False
self.start()
def set_common_data(self, data=None):
''' Set some data that will be passed to all subsequent jobs without
needing to be transmitted every time. You must call this method before
queueing any jobs, otherwise the behavior is undefined. You can call it
after all jobs are done, then it will be used for the new round of
jobs. Can raise the :class:`Failure` exception is data could not be
sent to workers.'''
if self.failed:
raise Failure(self.terminal_failure)
self.events.put(data)
def __call__(self, job_id, module, func, *args, **kwargs):
'''
Schedule a job. The job will be run in a worker process, with the
result placed in self.results. If a terminal failure has occurred
previously, this method will raise the :class:`Failure` exception.
:param job_id: A unique id for the job. The result will have this id.
:param module: Either a fully qualified python module name or python
source code which will be executed as a module.
Source code is detected by the presence of newlines in module.
:param func: Name of the function from ``module`` that will be
executed. ``args`` and ``kwargs`` will be passed to the function.
'''
if self.failed:
raise Failure(self.terminal_failure)
job = Job(job_id, module, func, args, kwargs)
self.tracker.put(None)
self.events.put(job)
def wait_for_tasks(self, timeout=None):
''' Wait for all queued jobs to be completed, if timeout is not None,
will raise a RuntimeError if jobs are not completed in the specified
time. Will raise a :class:`Failure` exception if a terminal failure has
occurred previously. '''
if self.failed:
raise Failure(self.terminal_failure)
if timeout is None:
self.tracker.join()
else:
join_with_timeout(self.tracker, timeout)
def shutdown(self, wait_time=0.1):
''' Shutdown this pool, terminating all worker process. The pool cannot
be used after a shutdown. '''
self.shutting_down = True
self.events.put(None)
self.shutdown_workers(wait_time=wait_time)
def create_worker(self):
p = start_worker('from {0} import run_main, {1}; run_main({1})'.format(self.__class__.__module__, 'worker_main'))
sys.stdout.flush()
eintr_retry_call(p.stdin.write, self.worker_data)
p.stdin.flush(), p.stdin.close()
conn = eintr_retry_call(self.listener.accept)
w = Worker(p, conn, self.events, self.name)
if self.common_data != pickle_dumps(None):
w.set_common_data(self.common_data)
return w
def start_worker(self):
try:
w = self.create_worker()
if not self.shutting_down:
self.available_workers.append(w)
except Exception:
import traceback
self.terminal_failure = TerminalFailure('Failed to start worker process', traceback.format_exc(), None)
self.terminal_error()
return False
def run(self):
from calibre.utils.ipc.server import create_listener
self.auth_key = os.urandom(32)
self.address, self.listener = create_listener(self.auth_key)
self.worker_data = msgpack_dumps((self.address, self.auth_key))
if self.start_worker() is False:
return
while True:
event = self.events.get()
if event is None or self.shutting_down:
break
if self.handle_event(event) is False:
break
def handle_event(self, event):
if isinstance(event, Job):
job = event
if not self.available_workers:
if len(self.busy_workers) >= self.max_workers:
self.pending_jobs.append(job)
return
if self.start_worker() is False:
return False
return self.run_job(job)
elif isinstance(event, WorkerResult):
worker_result = event
self.busy_workers.pop(worker_result.worker, None)
self.available_workers.append(worker_result.worker)
self.tracker.task_done()
if worker_result.is_terminal_failure:
self.terminal_failure = TerminalFailure('Worker process crashed while executing job', worker_result.result.traceback, worker_result.id)
self.terminal_error()
return False
self.results.put(worker_result)
else:
self.common_data = pickle_dumps(event)
if len(self.common_data) > MAX_SIZE:
self.cd_file = PersistentTemporaryFile('pool_common_data')
with self.cd_file as f:
f.write(self.common_data)
self.common_data = pickle_dumps(File(f.name))
for worker in self.available_workers:
try:
worker.set_common_data(self.common_data)
except Exception:
import traceback
self.terminal_failure = TerminalFailure('Worker process crashed while sending common data', traceback.format_exc(), None)
self.terminal_error()
return False
while self.pending_jobs and self.available_workers:
if self.run_job(self.pending_jobs.pop()) is False:
return False
def run_job(self, job):
worker = self.available_workers.pop()
try:
worker(job)
except Exception:
import traceback
self.terminal_failure = TerminalFailure('Worker process crashed while sending job', traceback.format_exc(), job.id)
self.terminal_error()
return False
self.busy_workers[worker] = job
@property
def failed(self):
return self.terminal_failure is not None
def terminal_error(self):
if self.shutting_down:
return
for worker, job in iteritems(self.busy_workers):
self.results.put(WorkerResult(job.id, Result(None, None, None), True, worker))
self.tracker.task_done()
while self.pending_jobs:
job = self.pending_jobs.pop()
self.results.put(WorkerResult(job.id, Result(None, None, None), True, None))
self.tracker.task_done()
self.shutdown()
def shutdown_workers(self, wait_time=0.1):
self.worker_data = self.common_data = None
for worker in self.busy_workers:
if worker.process.poll() is None:
try:
worker.process.terminate()
except EnvironmentError:
pass # If the process has already been killed
workers = [w.process for w in self.available_workers + list(self.busy_workers)]
aw = list(self.available_workers)
def join():
for w in aw:
try:
w(None)
except Exception:
pass
for w in workers:
try:
w.wait()
except Exception:
pass
reaper = Thread(target=join, name='ReapPoolWorkers')
reaper.daemon = True
reaper.start()
reaper.join(wait_time)
for w in self.available_workers + list(self.busy_workers):
try:
w.conn.close()
except Exception:
pass
for w in workers:
if w.poll() is None:
try:
w.kill()
except EnvironmentError:
pass
del self.available_workers[:]
self.busy_workers.clear()
if hasattr(self, 'cd_file'):
try:
os.remove(self.cd_file.name)
except EnvironmentError:
pass
class CompletionWorker(Thread):
daemon = True
def __init__(self, result_callback=lambda x: x, worker_entry_point='main'):
Thread.__init__(self)
self.worker_entry_point = worker_entry_point
self.start()
self.main_queue = Queue()
self.result_callback = result_callback
self.reap_thread = None
self.shutting_down = False
self.connected = Event()
self.current_completion_request = None
self.latest_completion_request_id = None
self.request_count = 0
self.lock = RLock()
def launch_worker_process(self):
from calibre.utils.ipc.server import create_listener
from calibre.utils.ipc.pool import start_worker
self.worker_process = p = start_worker(
'from {0} import run_main, {1}; run_main({1})'.format(
self.__class__.__module__, self.worker_entry_point))
auth_key = os.urandom(32)
address, self.listener = create_listener(auth_key)
eintr_retry_call(p.stdin.write, msgpack_dumps((address, auth_key)))
p.stdin.flush(), p.stdin.close()
self.control_conn = eintr_retry_call(self.listener.accept)
self.data_conn = eintr_retry_call(self.listener.accept)
self.data_thread = t = Thread(name='CWData',
target=self.handle_data_requests)
t.daemon = True
t.start()
self.connected.set()
def send(self, data, conn=None):
conn = conn or self.control_conn
try:
eintr_retry_call(conn.send, data)
except:
if not self.shutting_down:
raise
def recv(self, conn=None):
conn = conn or self.control_conn
try:
return eintr_retry_call(conn.recv)
except:
if not self.shutting_down:
raise
def wait_for_connection(self, timeout=None):
self.connected.wait(timeout)
def handle_data_requests(self):
from calibre.gui2.tweak_book.completion.basic import handle_data_request
while True:
try:
req = self.recv(self.data_conn)
except EOFError:
break
except Exception:
import traceback
traceback.print_exc()
break
if req is None or self.shutting_down:
break
result, tb = handle_data_request(req)
try:
self.send((result, tb), self.data_conn)
except EOFError:
break
except Exception:
import traceback
traceback.print_exc()
break
def run(self):
self.launch_worker_process()
while True:
obj = self.main_queue.get()
if obj is None:
break
req_type, req_data = obj
try:
if req_type is COMPLETION_REQUEST:
with self.lock:
if self.current_completion_request is not None:
ccr, self.current_completion_request = self.current_completion_request, None
self.send_completion_request(ccr)
elif req_type is CLEAR_REQUEST:
self.send(req_data)
except EOFError:
break
except Exception:
import traceback
traceback.print_exc()
def send_completion_request(self, request):
self.send(request)
result = self.recv()
if result.request_id == self.latest_completion_request_id:
try:
self.result_callback(result)
except Exception:
import traceback
traceback.print_exc()
def clear_caches(self, cache_type=None):
self.main_queue.put(
(CLEAR_REQUEST, Request(None, 'clear_caches', cache_type, None)))
def queue_completion(self,
request_id,
completion_type,
completion_data,
query=None):
with self.lock:
self.current_completion_request = Request(request_id,
completion_type,
completion_data, query)
self.latest_completion_request_id = self.current_completion_request.id
self.main_queue.put((COMPLETION_REQUEST, None))
def shutdown(self):
self.shutting_down = True
self.main_queue.put(None)
for conn in (getattr(self, 'control_conn',
None), getattr(self, 'data_conn', None)):
try:
conn.close()
except Exception:
pass
p = self.worker_process
if p.poll() is None:
self.worker_process.terminate()
t = self.reap_thread = Thread(target=p.wait)
t.daemon = True
t.start()
def join(self, timeout=0.2):
if self.reap_thread is not None:
self.reap_thread.join(timeout)
if not iswindows and self.worker_process.returncode is None:
self.worker_process.kill()
return self.worker_process.returncode
class SearchPanel(QWidget): # {{{
search_requested = pyqtSignal(object)
results_found = pyqtSignal(object)
show_search_result = pyqtSignal(object)
def __init__(self, parent=None):
QWidget.__init__(self, parent)
self.last_hidden_text_warning = None
self.current_search = None
self.l = l = QVBoxLayout(self)
l.setContentsMargins(0, 0, 0, 0)
self.search_input = si = SearchInput(self)
self.searcher = None
self.search_tasks = Queue()
self.results_found.connect(self.on_result_found,
type=Qt.QueuedConnection)
si.do_search.connect(self.search_requested)
l.addWidget(si)
self.results = r = Results(self)
r.show_search_result.connect(self.do_show_search_result,
type=Qt.QueuedConnection)
l.addWidget(r, 100)
self.spinner = s = BusySpinner(self)
s.setVisible(False)
l.addWidget(s)
def focus_input(self):
self.search_input.focus_input()
def start_search(self, search_query, current_name):
if self.current_search is not None and search_query == self.current_search:
self.find_next_requested(search_query.backwards)
return
if self.searcher is None:
self.searcher = Thread(name='Searcher', target=self.run_searches)
self.searcher.daemon = True
self.searcher.start()
self.results.clear()
self.spinner.start()
self.current_search = search_query
self.last_hidden_text_warning = None
self.search_tasks.put((search_query, current_name))
def run_searches(self):
while True:
x = self.search_tasks.get()
if x is None:
break
search_query, current_name = x
try:
manifest = get_manifest() or {}
spine = manifest.get('spine', ())
idx_map = {name: i for i, name in enumerate(spine)}
spine_idx = idx_map.get(current_name, -1)
except Exception:
import traceback
traceback.print_exc()
spine_idx = -1
if spine_idx < 0:
self.results_found.emit(SearchFinished(search_query))
continue
for name in spine:
counter = Counter()
spine_idx = idx_map[name]
try:
for i, result in enumerate(
search_in_name(name, search_query)):
before, text, after = result
self.results_found.emit(
SearchResult(search_query, before, text, after,
name, spine_idx, counter[text]))
counter[text] += 1
except Exception:
import traceback
traceback.print_exc()
self.results_found.emit(SearchFinished(search_query))
def on_result_found(self, result):
if self.current_search is None or result.search_query != self.current_search:
return
if isinstance(result, SearchFinished):
self.spinner.stop()
if not self.results.count():
self.show_no_results_found()
return
if self.results.add_result(result) == 1:
# first result
self.results.setCurrentRow(0)
self.results.item_activated()
def visibility_changed(self, visible):
if visible:
self.focus_input()
def clear_searches(self):
self.current_search = None
self.last_hidden_text_warning = None
searchable_text_for_name.cache_clear()
self.spinner.stop()
self.results.clear()
def shutdown(self):
self.search_tasks.put(None)
self.spinner.stop()
self.current_search = None
self.last_hidden_text_warning = None
self.searcher = None
def find_next_requested(self, previous):
self.results.find_next(previous)
def do_show_search_result(self, sr):
self.show_search_result.emit(sr.for_js)
def search_result_not_found(self, sr):
self.results.search_result_not_found(sr)
if self.results.count():
now = monotonic()
if self.last_hidden_text_warning is None or self.current_search != self.last_hidden_text_warning[
1] or now - self.last_hidden_text_warning[0] > 5:
self.last_hidden_text_warning = now, self.current_search
warning_dialog(
self,
_('Hidden text'),
_('Some search results were for hidden or non-reflowable text, they will be removed.'
),
show=True)
elif self.last_hidden_text_warning is not None:
self.last_hidden_text_warning = now, self.last_hidden_text_warning[
1]
if not self.results.count() and not self.spinner.is_running:
self.show_no_results_found()
def show_no_results_found(self):
has_hidden_text = self.last_hidden_text_warning is not None and self.last_hidden_text_warning[
1] == self.current_search
if self.current_search:
if has_hidden_text:
msg = _('No displayable matches were found for:')
else:
msg = _('No matches were found for:')
warning_dialog(self,
_('No matches found'),
msg +
' <b>{}</b>'.format(self.current_search.text),
show=True)
class CompletionWorker(Thread):
daemon = True
def __init__(self, result_callback=lambda x:x, worker_entry_point='main'):
Thread.__init__(self)
self.worker_entry_point = worker_entry_point
self.start()
self.main_queue = Queue()
self.result_callback = result_callback
self.reap_thread = None
self.shutting_down = False
self.connected = Event()
self.current_completion_request = None
self.latest_completion_request_id = None
self.request_count = 0
self.lock = RLock()
def launch_worker_process(self):
from calibre.utils.ipc.server import create_listener
from calibre.utils.ipc.pool import start_worker
self.worker_process = p = start_worker(
'from {0} import run_main, {1}; run_main({1})'.format(self.__class__.__module__, self.worker_entry_point))
auth_key = os.urandom(32)
address, self.listener = create_listener(auth_key)
eintr_retry_call(p.stdin.write, msgpack_dumps((address, auth_key)))
p.stdin.flush(), p.stdin.close()
self.control_conn = eintr_retry_call(self.listener.accept)
self.data_conn = eintr_retry_call(self.listener.accept)
self.data_thread = t = Thread(name='CWData', target=self.handle_data_requests)
t.daemon = True
t.start()
self.connected.set()
def send(self, data, conn=None):
conn = conn or self.control_conn
try:
eintr_retry_call(conn.send, data)
except:
if not self.shutting_down:
raise
def recv(self, conn=None):
conn = conn or self.control_conn
try:
return eintr_retry_call(conn.recv)
except:
if not self.shutting_down:
raise
def wait_for_connection(self, timeout=None):
self.connected.wait(timeout)
def handle_data_requests(self):
from calibre.gui2.tweak_book.completion.basic import handle_data_request
while True:
try:
req = self.recv(self.data_conn)
except EOFError:
break
except Exception:
import traceback
traceback.print_exc()
break
if req is None or self.shutting_down:
break
result, tb = handle_data_request(req)
try:
self.send((result, tb), self.data_conn)
except EOFError:
break
except Exception:
import traceback
traceback.print_exc()
break
def run(self):
self.launch_worker_process()
while True:
obj = self.main_queue.get()
if obj is None:
break
req_type, req_data = obj
try:
if req_type is COMPLETION_REQUEST:
with self.lock:
if self.current_completion_request is not None:
ccr, self.current_completion_request = self.current_completion_request, None
self.send_completion_request(ccr)
elif req_type is CLEAR_REQUEST:
self.send(req_data)
except EOFError:
break
except Exception:
import traceback
traceback.print_exc()
def send_completion_request(self, request):
self.send(request)
result = self.recv()
if result.request_id == self.latest_completion_request_id:
try:
self.result_callback(result)
except Exception:
import traceback
traceback.print_exc()
def clear_caches(self, cache_type=None):
self.main_queue.put((CLEAR_REQUEST, Request(None, 'clear_caches', cache_type, None)))
def queue_completion(self, request_id, completion_type, completion_data, query=None):
with self.lock:
self.current_completion_request = Request(request_id, completion_type, completion_data, query)
self.latest_completion_request_id = self.current_completion_request.id
self.main_queue.put((COMPLETION_REQUEST, None))
def shutdown(self):
self.shutting_down = True
self.main_queue.put(None)
for conn in (getattr(self, 'control_conn', None), getattr(self, 'data_conn', None)):
try:
conn.close()
except Exception:
pass
p = self.worker_process
if p.poll() is None:
self.worker_process.terminate()
t = self.reap_thread = Thread(target=p.wait)
t.daemon = True
t.start()
def join(self, timeout=0.2):
if self.reap_thread is not None:
self.reap_thread.join(timeout)
if not iswindows and self.worker_process.returncode is None:
self.worker_process.kill()
return self.worker_process.returncode
class SearchPanel(QWidget): # {{{
search_requested = pyqtSignal(object)
results_found = pyqtSignal(object)
show_search_result = pyqtSignal(object)
count_changed = pyqtSignal(object)
hide_search_panel = pyqtSignal()
goto_cfi = pyqtSignal(object)
def __init__(self, parent=None):
QWidget.__init__(self, parent)
self.discovery_counter = 0
self.last_hidden_text_warning = None
self.current_search = None
self.anchor_cfi = None
self.l = l = QVBoxLayout(self)
l.setContentsMargins(0, 0, 0, 0)
self.search_input = si = SearchInput(self)
self.searcher = None
self.search_tasks = Queue()
self.results_found.connect(self.on_result_found,
type=Qt.ConnectionType.QueuedConnection)
si.do_search.connect(self.search_requested)
si.cleared.connect(self.search_cleared)
si.go_back.connect(self.go_back)
l.addWidget(si)
self.results = r = Results(self)
r.count_changed.connect(self.count_changed)
r.show_search_result.connect(self.do_show_search_result,
type=Qt.ConnectionType.QueuedConnection)
r.current_result_changed.connect(self.update_hidden_message)
l.addWidget(r, 100)
self.spinner = s = BusySpinner(self)
s.setVisible(False)
l.addWidget(s)
self.hidden_message = la = QLabel(
_('This text is hidden in the book and cannot be displayed'))
la.setStyleSheet('QLabel { margin-left: 1ex }')
la.setWordWrap(True)
la.setVisible(False)
l.addWidget(la)
def go_back(self):
if self.anchor_cfi:
self.goto_cfi.emit(self.anchor_cfi)
def update_hidden_message(self):
self.hidden_message.setVisible(self.results.current_result_is_hidden)
def focus_input(self, text=None):
self.search_input.focus_input(text)
def search_cleared(self):
self.results.clear_all_results()
self.current_search = None
def start_search(self, search_query, current_name):
if self.current_search is not None and search_query == self.current_search:
self.find_next_requested(search_query.backwards)
return
if self.searcher is None:
self.searcher = Thread(name='Searcher', target=self.run_searches)
self.searcher.daemon = True
self.searcher.start()
self.results.clear_all_results()
self.hidden_message.setVisible(False)
self.spinner.start()
self.current_search = search_query
self.last_hidden_text_warning = None
self.search_tasks.put((search_query, current_name))
self.discovery_counter += 1
def set_anchor_cfi(self, pos_data):
self.anchor_cfi = pos_data['cfi']
def run_searches(self):
while True:
x = self.search_tasks.get()
if x is None:
break
search_query, current_name = x
try:
manifest = get_manifest() or {}
spine = manifest.get('spine', ())
idx_map = {name: i for i, name in enumerate(spine)}
spine_idx = idx_map.get(current_name, -1)
except Exception:
import traceback
traceback.print_exc()
spine_idx = -1
if spine_idx < 0:
self.results_found.emit(SearchFinished(search_query))
continue
num_in_spine = len(spine)
result_num = 0
for n in range(num_in_spine):
idx = (spine_idx + n) % num_in_spine
name = spine[idx]
counter = Counter()
try:
for i, result in enumerate(
search_in_name(name, search_query)):
before, text, after, offset = result
q = (before or '')[-15:] + text + (after or '')[:15]
result_num += 1
self.results_found.emit(
SearchResult(search_query, before, text, after, q,
name, idx, counter[q], offset,
result_num))
counter[q] += 1
except Exception:
import traceback
traceback.print_exc()
self.results_found.emit(SearchFinished(search_query))
def on_result_found(self, result):
if self.current_search is None or result.search_query != self.current_search:
return
if isinstance(result, SearchFinished):
self.spinner.stop()
if self.results.number_of_results:
self.results.ensure_current_result_visible()
else:
self.show_no_results_found()
return
self.results.add_result(result)
obj = result.for_js
obj['on_discovery'] = self.discovery_counter
self.show_search_result.emit(obj)
self.update_hidden_message()
def visibility_changed(self, visible):
if visible:
self.focus_input()
def clear_searches(self):
self.current_search = None
self.last_hidden_text_warning = None
searchable_text_for_name.cache_clear()
toc_offset_map_for_name.cache_clear()
get_toc_data.cache_clear()
self.spinner.stop()
self.results.clear_all_results()
def shutdown(self):
self.search_tasks.put(None)
self.spinner.stop()
self.current_search = None
self.last_hidden_text_warning = None
self.searcher = None
def find_next_requested(self, previous):
self.results.find_next(previous)
def trigger(self):
self.search_input.find_next()
def do_show_search_result(self, sr):
self.show_search_result.emit(sr.for_js)
def search_result_not_found(self, sr):
self.results.search_result_not_found(sr)
self.update_hidden_message()
def search_result_discovered(self, sr):
self.results.search_result_discovered(sr)
def show_no_results_found(self):
msg = _('No matches were found for:')
warning_dialog(self,
_('No matches found'),
msg + f' <b>{self.current_search.text}</b>',
show=True)
def keyPressEvent(self, ev):
if ev.key() == Qt.Key.Key_Escape:
self.hide_search_panel.emit()
ev.accept()
return
return QWidget.keyPressEvent(self, ev)
class DeleteService(Thread):
''' Provide a blocking file delete implementation with support for the
recycle bin. On windows, deleting files to the recycle bin spins the event
loop, which can cause locking errors in the main thread. We get around this
by only moving the files/folders to be deleted out of the library in the
main thread, they are deleted to recycle bin in a separate worker thread.
This has the added advantage that doing a restore from the recycle bin wont
cause metadata.db and the file system to get out of sync. Also, deleting
becomes much faster, since in the common case, the move is done by a simple
os.rename(). The downside is that if the user quits calibre while a long
move to recycle bin is happening, the files may not all be deleted.'''
daemon = True
def __init__(self):
Thread.__init__(self)
self.requests = Queue()
if isosx:
plugins['cocoa'][0].enable_cocoa_multithreading()
def shutdown(self, timeout=20):
self.requests.put(None)
self.join(timeout)
def create_staging(self, library_path):
base_path = os.path.dirname(library_path)
base = os.path.basename(library_path)
try:
ans = tempfile.mkdtemp(prefix=base + ' deleted ', dir=base_path)
except OSError:
ans = tempfile.mkdtemp(prefix=base + ' deleted ')
atexit.register(remove_dir, ans)
return ans
def remove_dir_if_empty(self, path):
try:
os.rmdir(path)
except OSError as e:
if e.errno == errno.ENOTEMPTY or len(os.listdir(path)) > 0:
# Some linux systems appear to raise an EPERM instead of an
# ENOTEMPTY, see https://bugs.launchpad.net/bugs/1240797
return
raise
def delete_books(self, paths, library_path):
tdir = self.create_staging(library_path)
self.queue_paths(tdir, paths, delete_empty_parent=True)
def queue_paths(self, tdir, paths, delete_empty_parent=True):
try:
self._queue_paths(tdir,
paths,
delete_empty_parent=delete_empty_parent)
except:
if os.path.exists(tdir):
shutil.rmtree(tdir, ignore_errors=True)
raise
def _queue_paths(self, tdir, paths, delete_empty_parent=True):
requests = []
for path in paths:
if os.path.exists(path):
basename = os.path.basename(path)
c = 0
while True:
dest = os.path.join(tdir, basename)
if not os.path.exists(dest):
break
c += 1
basename = '%d - %s' % (c, os.path.basename(path))
try:
shutil.move(path, dest)
except EnvironmentError:
if os.path.isdir(path):
# shutil.move may have partially copied the directory,
# so the subsequent call to move() will fail as the
# destination directory already exists
raise
# Wait a little in case something has locked a file
time.sleep(1)
shutil.move(path, dest)
if delete_empty_parent:
remove_dir_if_empty(os.path.dirname(path),
ignore_metadata_caches=True)
requests.append(dest)
if not requests:
remove_dir_if_empty(tdir)
else:
self.requests.put(tdir)
def delete_files(self, paths, library_path):
tdir = self.create_staging(library_path)
self.queue_paths(tdir, paths, delete_empty_parent=False)
def run(self):
while True:
x = self.requests.get()
try:
if x is None:
break
try:
self.do_delete(x)
except:
import traceback
traceback.print_exc()
finally:
self.requests.task_done()
def wait(self):
'Blocks until all pending deletes have completed'
self.requests.join()
def do_delete(self, tdir):
if os.path.exists(tdir):
try:
for x in os.listdir(tdir):
x = os.path.join(tdir, x)
if os.path.isdir(x):
delete_tree(x)
else:
delete_file(x)
finally:
shutil.rmtree(tdir)
class Repl(Thread):
LINE_CONTINUATION_CHARS = r'\:'
daemon = True
def __init__(self, ps1='>>> ', ps2='... ', show_js=False, libdir=None):
Thread.__init__(self, name='RapydScriptREPL')
self.to_python = to_python
self.JSError = JSError
self.enc = getattr(sys.stdin, 'encoding', None) or 'utf-8'
try:
import readline
self.readline = readline
except ImportError:
pass
self.output = ANSIStream(sys.stdout)
self.to_repl = Queue()
self.from_repl = Queue()
self.ps1, self.ps2 = ps1, ps2
self.show_js, self.libdir = show_js, libdir
self.prompt = ''
self.completions = None
self.start()
def init_ctx(self):
self.prompt = self.ps1
self.ctx = compiler()
self.ctx.g.Duktape.write = self.output.write
self.ctx.eval(
r'''console = { log: function() { Duktape.write(Array.prototype.slice.call(arguments).join(' ') + '\n');}};
console['error'] = console['log'];''')
self.ctx.g.repl_options = {
'show_js': self.show_js,
'histfile': False,
'input': True,
'output': True,
'ps1': self.ps1,
'ps2': self.ps2,
'terminal': self.output.isatty,
'enum_global': 'Object.keys(this)',
'lib_path': self.libdir or os.path.dirname(
P(COMPILER_PATH
)) # TODO: Change this to load pyj files from the src code
}
def get_from_repl(self):
while True:
try:
return self.from_repl.get(True, 1)
except Empty:
if not self.is_alive():
raise SystemExit(1)
def run(self):
self.init_ctx()
rl = None
def set_prompt(p):
self.prompt = p
def prompt(lw):
self.from_repl.put(to_python(lw))
self.ctx.g.set_prompt = set_prompt
self.ctx.g.prompt = prompt
self.ctx.eval('''
listeners = {};
rl = {
setPrompt:set_prompt,
write:Duktape.write,
clearLine: function() {},
on: function(ev, cb) { listeners[ev] = cb; return rl; },
prompt: prompt,
sync_prompt: true,
send_line: function(line) { listeners['line'](line); },
send_interrupt: function() { listeners['SIGINT'](); },
close: function() {listeners['close'](); },
};
repl_options.readline = { createInterface: function(options) { rl.completer = options.completer; return rl; }};
exports.init_repl(repl_options)
''',
fname='<init repl>')
rl = self.ctx.g.rl
completer = to_python(rl.completer)
send_interrupt = to_python(rl.send_interrupt)
send_line = to_python(rl.send_line)
while True:
ev, line = self.to_repl.get()
try:
if ev == 'SIGINT':
self.output.write('\n')
send_interrupt()
elif ev == 'line':
send_line(line)
else:
val = completer(line)
val = to_python(val)
self.from_repl.put(val[0])
except Exception as e:
if isinstance(e, JSError):
print(e.stack or error_message(e), file=sys.stderr)
else:
import traceback
traceback.print_exc()
for i in range(100):
# Do this many times to ensure we dont deadlock
self.from_repl.put(None)
def __call__(self):
if hasattr(self, 'readline'):
history = os.path.join(cache_dir(), 'pyj-repl-history.txt')
self.readline.parse_and_bind("tab: complete")
try:
self.readline.read_history_file(history)
except EnvironmentError as e:
if e.errno != errno.ENOENT:
raise
atexit.register(partial(self.readline.write_history_file, history))
def completer(text, num):
if self.completions is None:
self.to_repl.put(('complete', text))
self.completions = list(filter(None, self.get_from_repl()))
if not self.completions:
return None
try:
return self.completions[num]
except (IndexError, TypeError, AttributeError, KeyError):
self.completions = None
if hasattr(self, 'readline'):
self.readline.set_completer(completer)
while True:
lw = self.get_from_repl()
if lw is None:
raise SystemExit(1)
q = self.prompt
if hasattr(self, 'readline'):
self.readline.set_pre_input_hook(lambda: (
self.readline.insert_text(lw), self.readline.redisplay()))
else:
q += lw
try:
line = raw_input(q)
self.to_repl.put(('line', line))
except EOFError:
return
except KeyboardInterrupt:
self.to_repl.put(('SIGINT', None))
class Repl(Thread):
LINE_CONTINUATION_CHARS = r'\:'
daemon = True
def __init__(self, ps1='>>> ', ps2='... ', show_js=False, libdir=None):
Thread.__init__(self, name='RapydScriptREPL')
self.to_python = to_python
self.JSError = JSError
self.enc = getattr(sys.stdin, 'encoding', None) or 'utf-8'
try:
import readline
self.readline = readline
except ImportError:
pass
self.output = ANSIStream(sys.stdout)
self.to_repl = Queue()
self.from_repl = Queue()
self.ps1, self.ps2 = ps1, ps2
self.show_js, self.libdir = show_js, libdir
self.prompt = ''
self.completions = None
self.start()
def init_ctx(self):
self.prompt = self.ps1
self.ctx = compiler()
self.ctx.g.Duktape.write = self.output.write
self.ctx.eval(r'''console = { log: function() { Duktape.write(Array.prototype.slice.call(arguments).join(' ') + '\n');}};
console['error'] = console['log'];''')
self.ctx.g.repl_options = {
'show_js': self.show_js,
'histfile':False,
'input':True, 'output':True, 'ps1':self.ps1, 'ps2':self.ps2,
'terminal':self.output.isatty,
'enum_global': 'Object.keys(this)',
'lib_path': self.libdir or os.path.dirname(P(COMPILER_PATH)) # TODO: Change this to load pyj files from the src code
}
def get_from_repl(self):
while True:
try:
return self.from_repl.get(True, 1)
except Empty:
if not self.is_alive():
raise SystemExit(1)
def run(self):
self.init_ctx()
rl = None
def set_prompt(p):
self.prompt = p
def prompt(lw):
self.from_repl.put(to_python(lw))
self.ctx.g.set_prompt = set_prompt
self.ctx.g.prompt = prompt
self.ctx.eval('''
listeners = {};
rl = {
setPrompt:set_prompt,
write:Duktape.write,
clearLine: function() {},
on: function(ev, cb) { listeners[ev] = cb; return rl; },
prompt: prompt,
sync_prompt: true,
send_line: function(line) { listeners['line'](line); },
send_interrupt: function() { listeners['SIGINT'](); },
close: function() {listeners['close'](); },
};
repl_options.readline = { createInterface: function(options) { rl.completer = options.completer; return rl; }};
exports.init_repl(repl_options)
''', fname='<init repl>')
rl = self.ctx.g.rl
completer = to_python(rl.completer)
send_interrupt = to_python(rl.send_interrupt)
send_line = to_python(rl.send_line)
while True:
ev, line = self.to_repl.get()
try:
if ev == 'SIGINT':
self.output.write('\n')
send_interrupt()
elif ev == 'line':
send_line(line)
else:
val = completer(line)
val = to_python(val)
self.from_repl.put(val[0])
except Exception as e:
if isinstance(e, JSError):
print(e.stack or error_message(e), file=sys.stderr)
else:
import traceback
traceback.print_exc()
for i in range(100):
# Do this many times to ensure we dont deadlock
self.from_repl.put(None)
def __call__(self):
if hasattr(self, 'readline'):
history = os.path.join(cache_dir(), 'pyj-repl-history.txt')
self.readline.parse_and_bind("tab: complete")
try:
self.readline.read_history_file(history)
except EnvironmentError as e:
if e.errno != errno.ENOENT:
raise
atexit.register(partial(self.readline.write_history_file, history))
def completer(text, num):
if self.completions is None:
self.to_repl.put(('complete', text))
self.completions = list(filter(None, self.get_from_repl()))
if not self.completions:
return None
try:
return self.completions[num]
except (IndexError, TypeError, AttributeError, KeyError):
self.completions = None
if hasattr(self, 'readline'):
self.readline.set_completer(completer)
while True:
lw = self.get_from_repl()
if lw is None:
raise SystemExit(1)
q = self.prompt
if hasattr(self, 'readline'):
self.readline.set_pre_input_hook(lambda:(self.readline.insert_text(lw), self.readline.redisplay()))
else:
q += lw
try:
line = raw_input(q)
self.to_repl.put(('line', line))
except EOFError:
return
except KeyboardInterrupt:
self.to_repl.put(('SIGINT', None))
class DeleteService(Thread):
''' Provide a blocking file delete implementation with support for the
recycle bin. On windows, deleting files to the recycle bin spins the event
loop, which can cause locking errors in the main thread. We get around this
by only moving the files/folders to be deleted out of the library in the
main thread, they are deleted to recycle bin in a separate worker thread.
This has the added advantage that doing a restore from the recycle bin wont
cause metadata.db and the file system to get out of sync. Also, deleting
becomes much faster, since in the common case, the move is done by a simple
os.rename(). The downside is that if the user quits calibre while a long
move to recycle bin is happening, the files may not all be deleted.'''
daemon = True
def __init__(self):
Thread.__init__(self)
self.requests = Queue()
def shutdown(self, timeout=20):
self.requests.put(None)
self.join(timeout)
def create_staging(self, library_path):
base_path = os.path.dirname(library_path)
base = os.path.basename(library_path)
try:
ans = tempfile.mkdtemp(prefix=base+' deleted ', dir=base_path)
except OSError:
ans = tempfile.mkdtemp(prefix=base+' deleted ')
atexit.register(remove_dir, ans)
return ans
def remove_dir_if_empty(self, path):
try:
os.rmdir(path)
except OSError as e:
if e.errno == errno.ENOTEMPTY or len(os.listdir(path)) > 0:
# Some linux systems appear to raise an EPERM instead of an
# ENOTEMPTY, see https://bugs.launchpad.net/bugs/1240797
return
raise
def delete_books(self, paths, library_path):
tdir = self.create_staging(library_path)
self.queue_paths(tdir, paths, delete_empty_parent=True)
def queue_paths(self, tdir, paths, delete_empty_parent=True):
try:
self._queue_paths(tdir, paths, delete_empty_parent=delete_empty_parent)
except:
if os.path.exists(tdir):
shutil.rmtree(tdir, ignore_errors=True)
raise
def _queue_paths(self, tdir, paths, delete_empty_parent=True):
requests = []
for path in paths:
if os.path.exists(path):
basename = os.path.basename(path)
c = 0
while True:
dest = os.path.join(tdir, basename)
if not os.path.exists(dest):
break
c += 1
basename = '%d - %s' % (c, os.path.basename(path))
try:
shutil.move(path, dest)
except EnvironmentError:
if os.path.isdir(path):
# shutil.move may have partially copied the directory,
# so the subsequent call to move() will fail as the
# destination directory already exists
raise
# Wait a little in case something has locked a file
time.sleep(1)
shutil.move(path, dest)
if delete_empty_parent:
remove_dir_if_empty(os.path.dirname(path), ignore_metadata_caches=True)
requests.append(dest)
if not requests:
remove_dir_if_empty(tdir)
else:
self.requests.put(tdir)
def delete_files(self, paths, library_path):
tdir = self.create_staging(library_path)
self.queue_paths(tdir, paths, delete_empty_parent=False)
def run(self):
while True:
x = self.requests.get()
try:
if x is None:
break
try:
self.do_delete(x)
except:
import traceback
traceback.print_exc()
finally:
self.requests.task_done()
def wait(self):
'Blocks until all pending deletes have completed'
self.requests.join()
def do_delete(self, tdir):
if os.path.exists(tdir):
try:
for x in os.listdir(tdir):
x = os.path.join(tdir, x)
if os.path.isdir(x):
delete_tree(x)
else:
delete_file(x)
finally:
shutil.rmtree(tdir)
class Pool(Thread):
daemon = True
def __init__(self, max_workers=None, name=None):
Thread.__init__(self, name=name)
self.max_workers = max_workers or detect_ncpus()
self.available_workers = []
self.busy_workers = {}
self.pending_jobs = []
self.events = Queue()
self.results = Queue()
self.tracker = Queue()
self.terminal_failure = None
self.common_data = pickle_dumps(None)
self.shutting_down = False
self.start()
def set_common_data(self, data=None):
''' Set some data that will be passed to all subsequent jobs without
needing to be transmitted every time. You must call this method before
queueing any jobs, otherwise the behavior is undefined. You can call it
after all jobs are done, then it will be used for the new round of
jobs. Can raise the :class:`Failure` exception is data could not be
sent to workers.'''
if self.failed:
raise Failure(self.terminal_failure)
self.events.put(data)
def __call__(self, job_id, module, func, *args, **kwargs):
'''
Schedule a job. The job will be run in a worker process, with the
result placed in self.results. If a terminal failure has occurred
previously, this method will raise the :class:`Failure` exception.
:param job_id: A unique id for the job. The result will have this id.
:param module: Either a fully qualified python module name or python
source code which will be executed as a module.
Source code is detected by the presence of newlines in module.
:param func: Name of the function from ``module`` that will be
executed. ``args`` and ``kwargs`` will be passed to the function.
'''
if self.failed:
raise Failure(self.terminal_failure)
job = Job(job_id, module, func, args, kwargs)
self.tracker.put(None)
self.events.put(job)
def wait_for_tasks(self, timeout=None):
''' Wait for all queued jobs to be completed, if timeout is not None,
will raise a RuntimeError if jobs are not completed in the specified
time. Will raise a :class:`Failure` exception if a terminal failure has
occurred previously. '''
if self.failed:
raise Failure(self.terminal_failure)
if timeout is None:
self.tracker.join()
else:
join_with_timeout(self.tracker, timeout)
def shutdown(self, wait_time=0.1):
''' Shutdown this pool, terminating all worker process. The pool cannot
be used after a shutdown. '''
self.shutting_down = True
self.events.put(None)
self.shutdown_workers(wait_time=wait_time)
def create_worker(self):
a, b = Pipe()
with a:
cmd = 'from {0} import run_main, {1}; run_main({2!r}, {1})'.format(
self.__class__.__module__, 'worker_main', a.fileno())
p = start_worker(cmd, (a.fileno(), ))
sys.stdout.flush()
p.stdin.close()
w = Worker(p, b, self.events, self.name)
if self.common_data != pickle_dumps(None):
w.set_common_data(self.common_data)
return w
def start_worker(self):
try:
w = self.create_worker()
if not self.shutting_down:
self.available_workers.append(w)
except Exception:
import traceback
self.terminal_failure = TerminalFailure(
'Failed to start worker process', traceback.format_exc(), None)
self.terminal_error()
return False
def run(self):
if self.start_worker() is False:
return
while True:
event = self.events.get()
if event is None or self.shutting_down:
break
if self.handle_event(event) is False:
break
def handle_event(self, event):
if isinstance(event, Job):
job = event
if not self.available_workers:
if len(self.busy_workers) >= self.max_workers:
self.pending_jobs.append(job)
return
if self.start_worker() is False:
return False
return self.run_job(job)
elif isinstance(event, WorkerResult):
worker_result = event
self.busy_workers.pop(worker_result.worker, None)
self.available_workers.append(worker_result.worker)
self.tracker.task_done()
if worker_result.is_terminal_failure:
self.terminal_failure = TerminalFailure(
'Worker process crashed while executing job',
worker_result.result.traceback, worker_result.id)
self.terminal_error()
return False
self.results.put(worker_result)
else:
self.common_data = pickle_dumps(event)
if len(self.common_data) > MAX_SIZE:
self.cd_file = PersistentTemporaryFile('pool_common_data')
with self.cd_file as f:
f.write(self.common_data)
self.common_data = pickle_dumps(File(f.name))
for worker in self.available_workers:
try:
worker.set_common_data(self.common_data)
except Exception:
import traceback
self.terminal_failure = TerminalFailure(
'Worker process crashed while sending common data',
traceback.format_exc(), None)
self.terminal_error()
return False
while self.pending_jobs and self.available_workers:
if self.run_job(self.pending_jobs.pop()) is False:
return False
def run_job(self, job):
worker = self.available_workers.pop()
try:
worker(job)
except Exception:
import traceback
self.terminal_failure = TerminalFailure(
'Worker process crashed while sending job',
traceback.format_exc(), job.id)
self.terminal_error()
return False
self.busy_workers[worker] = job
@property
def failed(self):
return self.terminal_failure is not None
def terminal_error(self):
if self.shutting_down:
return
for worker, job in iteritems(self.busy_workers):
self.results.put(
WorkerResult(job.id, Result(None, None, None), True, worker))
self.tracker.task_done()
while self.pending_jobs:
job = self.pending_jobs.pop()
self.results.put(
WorkerResult(job.id, Result(None, None, None), True, None))
self.tracker.task_done()
self.shutdown()
def shutdown_workers(self, wait_time=0.1):
self.worker_data = self.common_data = None
for worker in self.busy_workers:
if worker.process.poll() is None:
try:
worker.process.terminate()
except OSError:
pass # If the process has already been killed
workers = [
w.process for w in self.available_workers + list(self.busy_workers)
]
aw = list(self.available_workers)
def join():
for w in aw:
try:
w(None)
except Exception:
pass
for w in workers:
try:
w.wait()
except Exception:
pass
reaper = Thread(target=join, name='ReapPoolWorkers')
reaper.daemon = True
reaper.start()
reaper.join(wait_time)
for w in self.available_workers + list(self.busy_workers):
try:
w.conn.close()
except Exception:
pass
for w in workers:
if w.poll() is None:
try:
w.kill()
except OSError:
pass
del self.available_workers[:]
self.busy_workers.clear()
if hasattr(self, 'cd_file'):
try:
os.remove(self.cd_file.name)
except OSError:
pass
