|
| 1 | +from collections import namedtuple |
| 2 | +from urllib.parse import urljoin |
| 3 | +import threading |
| 4 | +import requests |
| 5 | +import datetime |
| 6 | +import logging |
| 7 | +import queue |
| 8 | +import time |
| 9 | +import lzma |
| 10 | +import json |
| 11 | +import math |
| 12 | + |
| 13 | +# Allows you to retrieve the arguments passed to a function and |
| 14 | +# an arbitrary value by passing a 'store' or '_store' argument |
| 15 | +# as its return value. Good for use with concurrent.futures. |
| 16 | +def wrap_call(function, *args, **kwargs): |
| 17 | + store = kwargs.pop('_store', kwargs.pop('store', None)) |
| 18 | + Wrap = namedtuple('Wrap', ['result', 'store', 'args', 'kwargs']) |
| 19 | + return Wrap(function(*args, **kwargs), store, args, kwargs) |
| 20 | + |
| 21 | +class IndexServer(object): |
| 22 | + # Index metadata schema: |
| 23 | + # { name.str: { path.str, lzma.bool, info.str }, ... } |
| 24 | + def __init__(self, endpoint): |
| 25 | + self.endpoint = endpoint |
| 26 | + r = requests.get(urljoin(self.endpoint, 'meta.json')) |
| 27 | + r.raise_for_status() |
| 28 | + self.meta = r.json() |
| 29 | + |
| 30 | + def available_indexes(self): |
| 31 | + return self.meta['indexes'].keys() |
| 32 | + |
| 33 | + def get_latest(self): |
| 34 | + return self.meta['latest'] |
| 35 | + |
| 36 | + def get_anchor(self): |
| 37 | + return self.meta.get('anchor', None) |
| 38 | + |
| 39 | + def autodetect_anchor(self, anchor_hash): |
| 40 | + anchor = self.get_anchor() |
| 41 | + autodetect = dict() |
| 42 | + if anchor: |
| 43 | + autodetect = anchor['autodetect'] |
| 44 | + return autodetect.get(anchor_hash, None) |
| 45 | + |
| 46 | + def info(self, name): |
| 47 | + return self.meta['indexes'][name]['info'] |
| 48 | + |
| 49 | + def fetch(self, name, reporter, block_size=2048): |
| 50 | + index_meta = self.meta['indexes'][name] |
| 51 | + r = requests.get(urljoin(self.endpoint, index_meta['path']), stream=True) |
| 52 | + r.raise_for_status() |
| 53 | + data = bytearray() |
| 54 | + total_size = int(r.headers.get('content-length', 0)) |
| 55 | + for chunk, report in reporter.task_it('Fetching index %s' % name, r.iter_content(block_size), length=math.ceil(total_size / block_size)): |
| 56 | + report() |
| 57 | + data += chunk |
| 58 | + if index_meta['lzma']: |
| 59 | + data = lzma.decompress(data) |
| 60 | + return json.loads(data) |
| 61 | + |
| 62 | +class PoisonPill(object): |
| 63 | + pass |
| 64 | + |
| 65 | +class Task(object): |
| 66 | + def __init__(self, title, unit, length): |
| 67 | + self.title = title |
| 68 | + self.unit = unit |
| 69 | + self.length = length |
| 70 | + |
| 71 | +class ProgressReporter(object): |
| 72 | + def __init__(self, logger='reporter'): |
| 73 | + self._start = None |
| 74 | + self._stopped = False |
| 75 | + self._task = None |
| 76 | + self._report_event = threading.Event() |
| 77 | + self._step_queue = queue.Queue() |
| 78 | + self._task_queue = queue.Queue() |
| 79 | + self.logger = logging.getLogger(logger) |
| 80 | + |
| 81 | + def stop(self): |
| 82 | + self._stopped = True |
| 83 | + self._step_queue.put(PoisonPill()) |
| 84 | + self._task_queue.put(PoisonPill()) |
| 85 | + |
| 86 | + def is_stopped(self): |
| 87 | + return self._stopped |
| 88 | + |
| 89 | + def task(self, title, unit=None, length=None): |
| 90 | + if self._stopped: |
| 91 | + raise ValueError('operation on stopped reporter') |
| 92 | + self._task = Task(title, unit, length) |
| 93 | + self._task_queue.put(self._task) |
| 94 | + if not self._start: |
| 95 | + self._start = time.time() |
| 96 | + else: |
| 97 | + self._step_queue.put(PoisonPill()) |
| 98 | + |
| 99 | + def task_it(self, title, iterator, unit=None, length=None): |
| 100 | + if not length: |
| 101 | + length = len(iterator) if iterator else None |
| 102 | + self.task(title, unit=unit, length=length) |
| 103 | + for item in iterator: |
| 104 | + if self._stopped: |
| 105 | + raise ValueError('operation on stopped reporter') |
| 106 | + yield item, self.report |
| 107 | + if not self._report_event.isSet(): |
| 108 | + raise RuntimeError('report not called in previous iteration') |
| 109 | + self._report_event.clear() |
| 110 | + |
| 111 | + def report(self, payload=None): |
| 112 | + if not self._report_event.isSet(): |
| 113 | + self._step_queue.put(payload) |
| 114 | + self._report_event.set() |
| 115 | + |
| 116 | + def get_current_task(self): |
| 117 | + return self._task |
| 118 | + |
| 119 | + def tasks(self): |
| 120 | + while True: |
| 121 | + payload = self._task_queue.get() |
| 122 | + if isinstance(payload, PoisonPill): |
| 123 | + break |
| 124 | + self.logger.info('*** Task Start: %s ***' % payload.title) |
| 125 | + yield payload |
| 126 | + self.logger.info('*** Task End: %s ***' % payload.title) |
| 127 | + |
| 128 | + def steps(self): |
| 129 | + while True: |
| 130 | + payload = self._step_queue.get() |
| 131 | + if isinstance(payload, PoisonPill): |
| 132 | + break |
| 133 | + #self.logger.info('Step > %s' % payload) |
| 134 | + yield payload |
| 135 | + |
| 136 | + def step(self, payload=None): |
| 137 | + if self._stopped: |
| 138 | + raise ValueError('operation on stopped reporter') |
| 139 | + self._step_queue.put(payload) |
| 140 | + |
| 141 | + def elapsed(self): |
| 142 | + elapsed = 0 |
| 143 | + if self._start: |
| 144 | + elapsed = time.time() - self._start |
| 145 | + return str(datetime.timedelta(seconds=elapsed)) |
0 commit comments