diff options
Diffstat (limited to 'firewatch')
| -rw-r--r-- | firewatch/__init__.py | 0 | ||||
| -rw-r--r-- | firewatch/firewatch.py | 239 |
2 files changed, 239 insertions, 0 deletions
diff --git a/firewatch/__init__.py b/firewatch/__init__.py new file mode 100644 index 0000000..e69de29 --- /dev/null +++ b/firewatch/__init__.py diff --git a/firewatch/firewatch.py b/firewatch/firewatch.py new file mode 100644 index 0000000..df17e8a --- /dev/null +++ b/firewatch/firewatch.py @@ -0,0 +1,239 @@ +import errno +import json +import platform as PLATFORM +import requests +import sys +import time +from datetime import datetime, timedelta + + +conda_channel_pool = [ + f'https://repo.continuum.io/pkgs/main', + # f'https://repo.continuum.io/pkgs/free', # deprecated: conda<4.3.30 +] + +time_units = dict( + s=1, # second + m=60, # minute + h=3600, # hour + d=86400, # day + w=604800, # week + M=2.628e+6, # month + y=3.154e+7, # year + D=3.154e+8, # decade + c=3.154e+9, # century +) + +system_map = dict( + Linux='linux', + Darwin='osx', + Windows='win' +) + +machine_map = dict( + i386='32', + x86_64='64' +) + + +def extract_channel_platform(url): + """Returns last two elements in URL: (channel/platform-arch) + """ + parts = [x for x in url.split('/')] + result = '/'.join(parts[-2:]) + return result + + +def convert_human_timespan(t): + """Convert timespan to seconds to generate datetime.timedelta objects + """ + value, unit = int(t[:-1]), t[-1] + if unit not in time_units.keys(): + raise ValueError(f'Invalid time unit: "{unit}" (expected: [' + f'{"|".join([x for x in time_units.keys()])}])') + return value * time_units[unit] + + +def get_packages(channels): + packages = list() + for channel in channels: + repodata = f'{channel}/repodata.json' + data = dict( + packages=list(), + channel=extract_channel_platform(channel), + ) + + try: + with requests.get(repodata) as r: + r.raise_for_status() + data['packages'] = json.loads(r.text)['packages'] + packages.append(data) + + except requests.exceptions.RequestException as e: + print(f'Error {e.response.status_code}/{e.response.reason}:' + f' {channel}', file=sys.stderr) + except Exception as e: + print(e) + + return packages + + +def get_timestamps(data, brute_force=False): + """ Extract and convert package timestamps to datetime objects + """ + rt_fmt = '%a, %d %b %Y %H:%M:%S %Z' + + for base in data: + for pkg_name, pkg_info in base['packages'].items(): + result = dict() + result['name'] = pkg_name + result['channel'] = base['channel'] + + timestamp = datetime(1970, 1, 1) + # Continuum used 'date' for tracking some time ago + if 'date' in pkg_info: + date_str = [int(x) for x in pkg_info['date'].split('-')] + timestamp = datetime(*date_str) + + # Newer packages use 'timestamp', but depending on the direction + # of the wind, the unix epoch is stored in microseconds rather + # than seconds. So adjust for former case... + elif 'timestamp' in pkg_info: + timestamp = datetime.fromtimestamp(pkg_info['timestamp'] // 1000) + if timestamp < datetime(2000, 1, 1): + timestamp = datetime.fromtimestamp(pkg_info['timestamp']) + + # Scan remote server for 'last-modified' timestamp + # Don't do this unless you own the server you're spamming. + elif brute_force: + url = f'{result["channel"]}/{pkg_name}' + try: + modified = requests.head(url).headers['last-modified'] + except requests.exceptions.RequestException as e: + print(f'Error {e.response.status_code}/{e.response.reason}:' + f' {result["channel"]}', file=sys.stderr) + continue + except Exception as e: + print(e) + continue + + timestamp = datetime.strptime(modified, rt_fmt) + + result['timestamp'] = timestamp + yield result + + +def noarch_channel(channel, platform): + channel = channel.replace(f'{platform}', 'noarch') + return channel + + +def convert_channel(channel, platform, noarch=False): + # Strip trailing slash + if channel.endswith('/'): + channel = channel[:-1] + + # Sanitize URL by stripping out part we will adjust dynamically + if f'/{platform}' in channel: + pos = channel.find(f'/{platform}') + channel = channel[:pos] + + if '://' not in channel: + channel = f'https://conda.anaconda.org/{channel}/{platform}' + else: + channel = f'{channel}/{platform}' + + if noarch: + channel = noarch_channel(channel, platform) + + return channel + + +def get_platform(): + """Generate a conda compatible platform-arch string + """ + system = PLATFORM.system() + machine = PLATFORM.machine() + + result = None + try: + result = '-'.join([system_map[system], machine_map[machine]]) + except KeyError: + print(f'Unknown platform/arch combination: {system}/{machine}', + file=sys.stderr) + + return result + + +def main(): + from argparse import ArgumentParser + + parser = ArgumentParser() + parser.add_argument('--benchmark', action='store_true', + help='Display total time to parse and sort channel data') + + parser.add_argument('--brute-force', action='store_true', + help='Derive timestamps from HTTP header: "last-modified"') + + parser.add_argument('--channel', '-c', default=conda_channel_pool, + action='append', dest='channels', help='Conda channel') + + parser.add_argument('--order', '-o', default='asc', help='[asc|dsc]') + + parser.add_argument('--platform', '-p', default=[get_platform()], + action='append', dest='platforms', + help=f'[{"|".join(system_map.values())}]' + f'-[{"|".join(machine_map.values())}]') + + parser.add_argument('--time-span', '-t', default='1c', + help=f'i[{"|".join([x for x in time_units.keys()])}]' + ' (120s, 12h, 1d, 2w, 3m, 4y)') + + args = parser.parse_args() + + order = False # Ascending + if args.order != 'asc': + order = True # Descending + + if args.benchmark: + timer_start = time.time() + + channels = list() + for platform in set(args.platforms): + channels.extend([convert_channel(x, platform) for x in args.channels]) + channels.extend([convert_channel(x, platform, noarch=True) + for x in args.channels]) + channels = sorted(set(channels)) + + today = datetime.now() + span_delta = today - timedelta(seconds=convert_human_timespan(args.time_span)) + packages = get_packages(channels) + timestamps = sorted(list(get_timestamps(packages, args.brute_force)), + reverse=order, key=lambda x: x['timestamp']) + + if args.benchmark: + timer_stop = time.time() + print('#benchmark: {:.02f}s'.format(timer_stop - timer_start)) + + channel_width = max([len(extract_channel_platform(x)) for x in channels]) + 1 + print('#{:<20s} {:<{channel_width}s} {:<40s}'.format( + 'date', 'channel', 'package', channel_width=channel_width)) + + try: + for info in timestamps: + name = info['name'] + ts = info['timestamp'] + chn = info['channel'] + + tstr = ts.isoformat() + if span_delta < ts: + print(f'{tstr:<20s}: {chn:<{channel_width}s}: {name:<40s}') + except IOError as e: + # Broken pipe on '|head' + # TODO: Figure out why + if e.errno == errno.EPIPE: + pass + + +if __name__ == '__main__': + main() |