Merge pull request #48 from vbanos/configurable-warc-filename

Configurable WARC filenames

tests/test_writer.py

@@ -24,6 +24,7 @@ import fcntl
 from multiprocessing import Process, Queue
 from datetime import datetime
 import pytest
+import re
 from warcprox.mitmproxy import ProxyingRecorder
 from warcprox.warcproxy import RecordedUrl
 from warcprox.writer import WarcWriter
@@ -163,3 +164,21 @@ def test_special_dont_write_prefix():
             wwt.stop.set()
             wwt.join()
 
+
+def test_warc_writer_filename(tmpdir):
+    """Test if WarcWriter is writing WARC files with custom filenames.
+    """
+    recorder = ProxyingRecorder(None, None, 'sha1', url='http://example.com')
+    recorded_url = RecordedUrl(
+            url='http://example.com', content_type='text/plain', status=200,
+            client_ip='127.0.0.2', request_data=b'abc',
+            response_recorder=recorder, remote_ip='127.0.0.3',
+            timestamp=datetime.utcnow())
+
+    dirname = os.path.dirname(str(tmpdir.mkdir('test-warc-writer')))
+    wwriter = WarcWriter(Options(directory=dirname, prefix='foo',
+        warc_filename='{timestamp17}_{prefix}_{timestamp14}_{serialno}'))
+    wwriter.write_records(recorded_url)
+    warcs = [fn for fn in os.listdir(dirname)]
+    assert warcs
+    assert re.search('\d{17}_foo_\d{14}_00000.warc.open', wwriter._fpath)

warcprox/main.py

@@ -77,6 +77,9 @@ def _build_arg_parser(prog):
             help='where to store and load generated certificates')
     arg_parser.add_argument('-d', '--dir', dest='directory',
             default='./warcs', help='where to write warcs')
+    arg_parser.add_argument('--warc-filename', dest='warc_filename',
+            default='{prefix}-{timestamp17}-{serialno}-{randomtoken}',
+            help='define custom WARC filename with variables {prefix}, {timestamp14}, {timestamp17}, {serialno}, {randomtoken}, {hostname}, {shorthostname}')
     arg_parser.add_argument('-z', '--gzip', dest='gzip', action='store_true',
             help='write gzip-compressed warc records')
     arg_parser.add_argument('--no-warc-open-suffix', dest='no_warc_open_suffix',

warcprox/writer.py

@@ -28,6 +28,7 @@ import fcntl
 import time
 import warcprox
 import os
+import socket
 import string
 import random
 import threading
@@ -42,6 +43,8 @@ class WarcWriter:
         self._last_activity = time.time()
 
         self.gzip = options.gzip or False
+        self.warc_filename = options.warc_filename or \
+            '{prefix}-{timestamp17}-{randomtoken}-{serialno}'
         digest_algorithm = options.digest_algorithm or 'sha1'
         base32 = options.base32
         self.record_builder = warcprox.warc.WarcRecordBuilder(
@@ -68,6 +71,10 @@ class WarcWriter:
         now = datetime.utcnow()
         return '{:%Y%m%d%H%M%S}{:03d}'.format(now, now.microsecond//1000)
 
+    def timestamp14(self):
+        now = datetime.utcnow()
+        return '{:%Y%m%d%H%M%S}'.format(now)
+
     def close_writer(self):
         with self._lock:
             if self._fpath:
@@ -86,8 +93,32 @@ class WarcWriter:
                 self._fpath = None
                 self._f = None
 
+    def serial(self):
+        return '{:05d}'.format(self._serial)
+
     # h3 default <!-- <property name="template" value="${prefix}-${timestamp17}-${serialno}-${heritrix.pid}~${heritrix.hostname}~${heritrix.port}" /> -->
-    # ${prefix}-${timestamp17}-${randomtoken}-${serialno}.warc.gz"
+    def _warc_filename(self):
+        """WARC filename is configurable with CLI parameter --warc-filename.
+        Default: '{prefix}-{timestamp17}-{serialno}-{randomtoken}'
+        Available variables are: prefix, timestamp14, timestamp17, serialno,
+        randomtoken, hostname, shorthostname.
+        Extension ``.warc`` or ``.warc.gz`` is appended automatically.
+        """
+        hostname = socket.getfqdn()
+        shorthostname = hostname.split('.')[0]
+        fname = self.warc_filename.format(prefix=self.prefix,
+                                          timestamp14=self.timestamp14(),
+                                          timestamp17=self.timestamp17(),
+                                          serialno=self.serial(),
+                                          randomtoken=self._randomtoken,
+                                          hostname=hostname,
+                                          shorthostname=shorthostname)
+        if self.gzip:
+            fname = fname + '.warc.gz'
+        else:
+            fname = fname + '.warc'
+        return fname
+
     def _writer(self):
         with self._lock:
             if self._fpath and os.path.getsize(
@@ -95,9 +126,7 @@ class WarcWriter:
                 self.close_writer()
 
             if self._f == None:
-                self._f_finalname = '{}-{}-{:05d}-{}.warc{}'.format(
+                self._f_finalname = self._warc_filename()
-                        self.prefix, self.timestamp17(), self._serial,
-                        self._randomtoken, '.gz' if self.gzip else '')
                 self._fpath = os.path.sep.join([
                     self.directory, self._f_finalname + self._f_open_suffix])