#!/usr/bin/env python3
"""
warc-extractor, a simple command line tool for expanding warc files.
Copyright (C) 2014 Ryan Chartier
Portions (C) 2012 Internet Archive
This program is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation, either version 3 of the License, or
(at your option) any later version.
This program is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with this program. If not, see .
warc.utils
~~~~~~~~~~
This file is part of warc
:copyright: (c) 2012 Internet Archive
warc.warc
~~~~~~~~~
Python library to work with WARC files.
:copyright: (c) 2012 Internet Archive
"""
from collections import MutableMapping
from http.client import HTTPMessage
from urllib.parse import urlparse, unquote
from pprint import pprint
import os
import re
import argparse
import sys
import mimetypes
import shutil
import email.parser
import gzip
import datetime
import uuid
import re
import io
import hashlib
#---------------------------------------------------
# warc.utils -
#---------------------------------------------------
SEP = re.compile("[;:=]")
class CaseInsensitiveDict(MutableMapping):
"""Almost like a dictionary, but keys are case-insensitive.
>>> d = CaseInsensitiveDict(foo=1, Bar=2)
>>> d['foo']
1
>>> d['bar']
2
>>> d['Foo'] = 11
>>> d['FOO']
11
>>> d.keys()
["foo", "bar"]
"""
def __init__(self, *args, **kwargs):
self._d = {}
self.update(dict(*args, **kwargs))
def __setitem__(self, name, value):
self._d[name.lower()] = value
def __getitem__(self, name):
return self._d[name.lower()]
def __delitem__(self, name):
del self._d[name.lower()]
def __eq__(self, other):
return isinstance(other, CaseInsensitiveDict) and other._d == self._d
def __iter__(self):
return iter(self._d)
def __len__(self):
return len(self._d)
class FilePart:
"""File interface over a part of file.
Takes a file and length to read from the file and returns a file-object
over that part of the file.
"""
def __init__(self, fileobj, length):
self.fileobj = fileobj
self.length = length
self.offset = 0
self.buf = b''
def read(self, size=-1):
if size == -1:
size = self.length
if len(self.buf) >= size:
content = self.buf[:size]
self.buf = self.buf[size:]
else:
size = min(size, self.length - self.offset)
content = self.buf + self.fileobj.read(size - len(self.buf))
self.buf = b''
self.offset += len(content)
return content
def _unread(self, content):
self.buf = content + self.buf
self.offset -= len(content)
def readline(self, size=1024):
chunks = []
chunk = self.read(size)
while chunk and b"\n" not in chunk:
chunks.append(chunk)
chunk = self.read(size)
if b"\n" in chunk:
index = chunk.index(b"\n")
self._unread(chunk[index+1:])
chunk = chunk[:index+1]
chunks.append(chunk)
return b"".join(chunks)
def __iter__(self):
line = self.readline()
while line:
yield line
line = self.readline()
class HTTPObject(CaseInsensitiveDict):
"""Small object to help with parsing HTTP warc entries"""
def __init__(self, request_file):
#Parse version line
id_str_raw = request_file.readline()
id_str = id_str_raw.decode("iso-8859-1")
if "HTTP" not in id_str:
#This is not an HTTP object.
request_file._unread(id_str_raw)
raise ValueError("Object is not HTTP.")
words = id_str.split()
command = path = status = error = version = None
#If length is not 3 it is a bad version line.
if len(words) >= 3:
if words[1].isdigit():
version = words[0]
error = words[1]
status = " ".join(words[2:])
else:
command, path, version = words
self._id = {
"vline": id_str_raw,
"command": command,
"path": path,
"status": status,
"error": error,
"version": version,
}
self._header, self.hstring = self._parse_headers(request_file)
super().__init__(self._header)
self.payload = request_file
self._content = None
@staticmethod
def _parse_headers(fp):
"""This is a modification of the python3 http.clint.parse_headers function."""
headers = []
while True:
line = fp.readline(65536)
headers.append(line)
if line in (b'\r\n', b'\n', b''):
break
hstring = b''.join(headers)
return email.parser.Parser(_class=HTTPMessage).parsestr(hstring.decode('iso-8859-1')), hstring
def __repr__(self):
return(self.vline + str(self._header))
def __getitem__(self, name):
try:
return super().__getitem__(name)
except KeyError:
value = name.lower()
if value == "content_type":
return self.content.type
elif value in self.content:
return self.content[value]
elif value in self._id:
return self._id[value]
else:
raise
def _reset(self):
self.payload._unread(self.hstring)
self.payload._unread(self._id['vline'])
def write_to(self, f):
f.write(self._id['vline'])
f.write(self.hstring)
f.write(self.payload.read())
f.write(b"\r\n\r\n")
f.flush()
@property
def content(self):
if self._content is None:
try:
string = self._d["content-type"]
except KeyError:
string = ''
self._content = ContentType(string)
return self._content
@property
def vline(self):
return self._id["vline"].decode("iso-8859-1")
@property
def version(self):
return self._id["version"]
def write_payload_to(self, fp):
encoding = self._header.get("Transfer-Encoding", "None")
if encoding == "chunked":
found = b''
length = int(str(self.payload.readline(), "iso-8859-1").rstrip(), 16)
while length > 0:
found += self.payload.read(length)
self.payload.readline()
length = int(str(self.payload.readline(), "iso-8859-1").rstrip(), 16)
else:
length = int(self._header.get("Content-Length", -1))
found = self.payload.read(length)
fp.write(found)
class ContentType(CaseInsensitiveDict):
def __init__(self, string):
data = {}
self.type = ''
if string:
_list = [i.strip() for i in string.lower().split(";")]
self.type = _list[0]
data["type"] = _list[0]
for i in _list[1:]:
test = [n.strip() for n in re.split(SEP, i)]
data[test[0]] = test[1]
super().__init__(data)
def __repr__(self):
return self.type
#---------------------------------------------------
# warc.warc -
#---------------------------------------------------
class WARCHeader(CaseInsensitiveDict):
"""The WARC Header object represents the headers of a WARC record.
It provides dictionary like interface for accessing the headers.
The following mandatory fields are accessible also as attributes.
* h.record_id == h['WARC-Record-ID']
* h.content_length == int(h['Content-Length'])
* h.date == h['WARC-Date']
* h.type == h['WARC-Type']
:params headers: dictionary of headers.
:params defaults: If True, important headers like WARC-Record-ID,
WARC-Date, Content-Type and Content-Length are
initialized to automatically if not already present.
"""
CONTENT_TYPES = dict(warcinfo='application/warc-fields',
response='application/http; msgtype=response',
request='application/http; msgtype=request',
metadata='application/warc-fields')
KNOWN_HEADERS = {
"type": "WARC-Type",
"date": "WARC-Date",
"record_id": "WARC-Record-ID",
"ip_address": "WARC-IP-Address",
"target_uri": "WARC-Target-URI",
"warcinfo_id": "WARC-Warcinfo-ID",
"request_uri": "WARC-Request-URI",
"content_type": "Content-Type",
"content_length": "Content-Length"
}
def __init__(self, headers, defaults=False):
self.version = "WARC/1.0"
super().__init__(headers)
if defaults:
self.init_defaults()
def __repr__(self):
return "".format(self.type, self.record_id)
def init_defaults(self):
"""Initializes important headers to default values, if not already specified.
The WARC-Record-ID header is set to a newly generated UUID.
The WARC-Date header is set to the current datetime.
The Content-Type is set based on the WARC-Type header.
The Content-Length is initialized to 0.
"""
if "WARC-Record-ID" not in self:
self['WARC-Record-ID'] = "" % uuid.uuid1()
if "WARC-Date" not in self:
self['WARC-Date'] = datetime.datetime.utcnow().strftime('%Y-%m-%dT%H:%M:%SZ')
if "Content-Type" not in self:
self['Content-Type'] = WARCHeader.CONTENT_TYPES.get(self.type, "application/octet-stream")
def write_to(self, f):
"""Writes this header to a file, in the format specified by WARC.
"""
f.write(self.version.encode() + b"\r\n")
for name, value in self.items():
name = name.title()
# Use standard forms for commonly used patterns
name = name.replace("Warc-", "WARC-").replace("-Ip-", "-IP-").replace("-Id", "-ID").replace("-Uri", "-URI")
entry = "{}: {}\r\n".format(str(name), str(value)).encode()
f.write(entry)
# Header ends with an extra CRLF
f.write(b"\r\n")
@property
def content_length(self):
"""The Content-Length header as int."""
return int(self['Content-Length'])
@property
def type(self):
"""The value of WARC-Type header."""
return self['WARC-Type']
@property
def record_id(self):
"""The value of WARC-Record-ID header."""
return self['WARC-Record-ID']
@property
def date(self):
"""The value of WARC-Date header."""
return self['WARC-Date']
class WARCRecord(object):
"""The WARCRecord object represents a WARC Record.
"""
def __init__(self, header=None, payload=None, headers={}, defaults=True):
"""Creates a new WARC record.
"""
if header is None and defaults is True:
headers.setdefault("WARC-Type", "response")
self.header = header or WARCHeader(headers, defaults=True)
if defaults is True and 'Content-Length' not in self.header:
if payload:
self.header['Content-Length'] = len(payload)
else:
self.header['Content-Length'] = "0"
if defaults is True and 'WARC-Payload-Digest' not in self.header:
self.header['WARC-Payload-Digest'] = self._compute_digest(payload)
if isinstance(payload, str):
payload = payload.encode()
if isinstance(payload, bytes):
payload = io.BytesIO(payload)
self.payload = payload
self._http = None
self._content = None
def _compute_digest(self, payload):
return "sha1:" + hashlib.sha1(payload).hexdigest()
def write_to(self, f):
self.header.write_to(f)
if self.http:
self.http._reset()
f.write(self.payload.read())
f.write(b"\r\n")
f.write(b"\r\n")
f.flush()
@property
def content(self):
if self._content is None:
try:
string = self.header["content-type"]
except KeyError:
string = ''
self._content = ContentType(string)
return self._content
@property
def http(self):
if self._http is None:
if 'application/http' in self.header['content-type']:
self._http = HTTPObject(self.payload)
else:
self._http = False
return self._http
@property
def type(self):
"""Record type"""
return self.header.type
@property
def url(self):
"""The value of the WARC-Target-URI header if the record is of type "response"."""
return self.header.get('WARC-Target-URI')
@property
def ip_address(self):
"""The IP address of the host contacted to retrieve the content of this record.
This value is available from the WARC-IP-Address header."""
return self.header.get('WARC-IP-Address')
@property
def date(self):
"""UTC timestamp of the record."""
return self.header.get("WARC-Date")
@property
def checksum(self):
return self.header.get('WARC-Payload-Digest')
def __getitem__(self, name):
try:
return self.header[name]
except KeyError:
if name == "content_type":
return self.content.type
elif name in self.content:
return self.content[name]
def __setitem__(self, name, value):
self.header[name] = value
def __contains__(self, name):
return name in self.header
def __repr__(self):
return "" % (self.type, self['WARC-Record-ID'])
@staticmethod
def from_response(response):
"""Creates a WARCRecord from given response object.
This must be called before reading the response. The response can be
read after this method is called.
:param response: An instance of :class:`requests.models.Response`.
"""
# Get the httplib.HTTPResponse object
http_response = response.raw._original_response
# HTTP status line, headers and body as strings
status_line = "HTTP/1.1 %d %s" % (http_response.status, http_response.reason)
headers = str(http_response.msg)
body = http_response.read()
# Monkey-patch the response object so that it is possible to read from it later.
response.raw._fp = io.BytesIO(body)
# Build the payload to create warc file.
payload = status_line + "\r\n" + headers + "\r\n" + body
headers = {
"WARC-Type": "response",
"WARC-Target-URI": response.request.url.encode('utf-8')
}
return WARCRecord(payload=payload, headers=headers)
class WARCFile:
def __init__(self, filename=None, mode=None, fileobj=None, compress=None):
if fileobj is None:
fileobj = open(filename, mode or "rb")
mode = fileobj.mode
# initiaize compress based on filename, if not already specified
if compress is None and filename and filename.endswith(".gz"):
compress = True
if compress:
fileobj = gzip.open(fileobj, mode)
self.fileobj = fileobj
self._reader = None
def __enter__(self):
return self
def __exit__(self, exc_type, exc_value, traceback):
self.close()
def __iter__(self):
return iter(self.reader)
@property
def reader(self):
if self._reader is None:
self._reader = WARCReader(self.fileobj)
return self._reader
def write_record(self, warc_record):
"""Adds a warc record to this WARC file.
"""
warc_record.write_to(self.fileobj)
def read_record(self):
"""Reads a warc record from this WARC file."""
return self.reader.read_record()
def close(self):
self.fileobj.close()
def tell(self):
"""Returns the file offset.
"""
return self.fileobj.tell()
class WARCReader:
RE_VERSION = re.compile("WARC/(\d+.\d+)\r\n")
RE_HEADER = re.compile(r"([a-zA-Z_\-]+): *(.*)\r\n")
SUPPORTED_VERSIONS = ["1.0"]
def __init__(self, fileobj):
self.fileobj = fileobj
self.current_payload = None
def read_header(self, fileobj):
version_line = fileobj.readline().decode("utf-8")
if not version_line:
return None
m = self.RE_VERSION.match(version_line)
if not m:
raise IOError("Bad version line: %r" % version_line)
version = m.group(1)
if version not in self.SUPPORTED_VERSIONS:
raise IOError("Unsupported WARC version: %s" % version)
headers = {}
while True:
line = fileobj.readline().decode("utf-8")
if line == "\r\n": # end of headers
break
m = self.RE_HEADER.match(line)
if not m:
raise IOError("Bad header line: %r" % line)
name, value = m.groups()
headers[name] = value
return WARCHeader(headers)
def expect(self, fileobj, expected_line, message=None):
line = fileobj.readline().decode("utf-8")
if line != expected_line:
message = message or "Expected %r, found %r" % (expected_line, line)
raise IOError(message)
def finish_reading_current_record(self):
# consume the footer from the previous record
if self.current_payload:
# consume all data from the current_payload before moving to next record
self.current_payload.read()
self.expect(self.current_payload.fileobj, "\r\n")
self.expect(self.current_payload.fileobj, "\r\n")
self.current_payload = None
def read_record(self):
self.finish_reading_current_record()
fileobj = self.fileobj
header = self.read_header(fileobj)
if header is None:
return None
self.current_payload = FilePart(fileobj, header.content_length)
record = WARCRecord(header, self.current_payload, defaults=False)
return record
def _read_payload(self, fileobj, content_length):
size = 0
while size < content_length:
chunk_size = min(1024, content_length-size)
chunk = fileobj.read(chunk_size)
size += chunk_size
yield chunk
def __iter__(self):
record = self.read_record()
while record is not None:
yield record
record = self.read_record()
#---------------------------------------------------
# Extractor -
#---------------------------------------------------
counts = {}
class filterObject:
"""Basic object for storing filters."""
def __init__(self, string):
self.result = True
if string[0] == "!":
self.result = False
string = string[1:]
_list = string.lower().split(":")
self.http = (_list[0] == 'http')
if self.http:
del _list[0]
self.k = _list[0]
self.v = _list[1]
def inc(obj, header=False, dic=False):
"""Short script for counting entries."""
if header:
try:
obj = obj[header]
except KeyError:
obj = None
holder = counts
if dic:
if dic not in counts:
counts[dic] = {}
holder = counts[dic]
if obj in holder:
holder[obj] += 1
else:
holder[obj] = 1
def warc_records(string, path):
"""Iterates over warc records in path."""
for filename in os.listdir(path):
if re.search(string, filename) and ".warc" in filename:
print("parsing", filename)
with WARCFile(path + filename) as warc_file:
for record in warc_file:
yield record
def checkFilter(filters, record):
"""Check record against filters."""
for i in filters:
if i.http:
if not record.http:
return False
value = record.http
else:
value = record.header
string = value.get(i.k, None)
if not string or (i.v in string) != i.result:
return False
return True
def parse(args):
#Clear output warc file.
if args.dump == "warc":
if args.silence:
print("Recording", args.dump, "to", args.output + ".")
with open(args.output_path + args.output, "wb"):
pass
for record in warc_records(args.string, args.path):
try:
#Filter out unwanted entries.
if not checkFilter(args.filter, record):
continue
#Increment Index counters.
if args.silence:
inc("records")
inc(record,"warc-type", "types")
inc(record, "content_type", "warc-content")
if record.http:
inc(record.http, "content_type", "http-content")
inc(record.http, "error", "status")
#Dump records to file.
if args.dump == "warc":
with open(args.output_path + args.output, "ab") as output:
record.write_to(output)
if args.dump == "content":
url = urlparse(unquote(record['WARC-Target-URI']))
#Set up folder
index = url.path.rfind("/") + 1
file = url.path[index:]
path = url.path[:index]
#Process filename
if "." not in file:
path += file
if not path.endswith("/"):
path += "/"
file = 'index.html'
#Final fixes.
path = path.replace(".", "-")
host = url.hostname.replace('www.', '', 1)
path = args.output_path + host + path
#Create new directories
if not os.path.exists(path):
try:
os.makedirs(path)
except OSError:
path = "/".join([i[:25] for i in path.split("/")])
os.makedirs(path)
#Test if file has a proper extension.
index = file.index(".")
suffix = file[index:]
content = record.http.get("content_type", "")
slist = mimetypes.guess_all_extensions(content)
if suffix not in slist:
#Correct suffix if we can.
suffix = mimetypes.guess_extension(content)
if suffix:
file = file[:index] + suffix
else:
inc(record.http, "content_type", "unknown mime type")
#Check for gzip compression.
if record.http.get("content-encoding", None) == "gzip":
file += ".gz"
path += file
#If Duplicate file then insert numbers
index = path.rfind(".")
temp = path
n = 0
while os.path.isfile(temp):
n +=1
temp = path[:index] + "("+ str(n) + ")" + path[index:]
path = temp
#Write file.
with open(path, 'wb') as fp:
record.http.write_payload_to(fp)
except Exception:
if args.error:
if args.silence:
print("Error in record. Recording to error.warc.")
with open(args.output_path + "error.warc", "ab") as fp:
record.write_to(fp)
else:
raise
#print results
if args.silence:
print("-----------------------------")
for i in counts:
print("\nCount of {}.".format(i))
pprint(counts[i])
if __name__ == "__main__":
parser = argparse.ArgumentParser(description='Extracts attributes from warc files.')
parser.add_argument("filter", nargs='*', help="Attributes to filter by. Entries that do not contain filtered elements are ignored. Example: warc-type:response, would ignore all warc entries that are not responses. Attributes in an HTTP object should be prefixed by 'http'. Example, http:error:200.")
parser.add_argument("-silence", action="store_false", help="Silences output of warc data.")
parser.add_argument("-error", action="store_true", help="Silences most errors and records problematic warc entries to error.warc.")
parser.add_argument("-string", default="", help="Regular expression to limit parsed warc files. Defaults to empty string.")
parser.add_argument("-path", default="./", help="Path to folder containing warc files. Defaults to current folder.")
parser.add_argument("-output_path", default="data/", help="Path to folder to dump content files. Defaults to data/ folder.")
parser.add_argument("-output", default="output.warc", help="File to output warc contents. Defaults to 'output.warc'.")
parser.add_argument("-dump", choices=['warc', 'content'], type=str, help="Dumps all entries that survived filter. 'warc' creates a filtered warc file. 'content' tries to reproduce file structure of archived websites.")
args = parser.parse_args()
if args.path[-1] != "/":
args.path += "/"
if args.output_path[-1] != "/":
args.output_path += "/"
if args.dump:
if not os.path.exists(args.output_path):
os.makedirs(args.output_path)
#Forced filters
if args.dump == "content":
args.filter.append("warc-type:response")
args.filter.append("content-type:application/http")
args.filter = [filterObject(i) for i in args.filter]
args.string = re.compile(args.string)
parse(args)