def start(cdxjFilePath=INDEX_FILE, proxy=None):
hostPort = ipwbConfig.getIPWBReplayConfig()
app.proxy = proxy
if not hostPort:
ipwbConfig.setIPWBReplayConfig(IPWBREPLAY_IP, IPWBREPLAY_PORT)
hostPort = ipwbConfig.getIPWBReplayConfig()
if ipwbConfig.isDaemonAlive():
if cdxjFilePath == INDEX_FILE:
ipwbConfig.firstRun()
ipwbConfig.setIPWBReplayIndexPath(cdxjFilePath)
app.cdxjFilePath = cdxjFilePath
else:
print('Sample data not pulled from IPFS.')
print('Check that the IPFS daemon is running.')
try:
print('IPWB replay started on http://{0}:{1}'.format(
IPWBREPLAY_IP, IPWBREPLAY_PORT))
app.run(host='0.0.0.0', port=IPWBREPLAY_PORT)
except gaierror:
print('Detected no active Internet connection.')
print('Overriding to use default IP and port configuration.')
app.run()
except socketerror:
print('Address {0}:{1} already in use!'.format(IPWBREPLAY_IP,
IPWBREPLAY_PORT))
sys.exit()
def showAdmin():
ipfsEndpoint = '{0}:{1}'.format(IPFSAPI_HOST, IPFSAPI_PORT)
status = {
'ipwbVersion': ipwbVersion,
'ipfsEndpoint': ipfsEndpoint,
'ipfsAlive': ipwbUtils.isDaemonAlive(ipfsEndpoint)
}
# TODO: Calculate actual URI-R/M counts
indexes = [{
'path': ipwbUtils.getIPWBReplayIndexPath(),
'enabled': True,
'urimCount': '#URI-M',
'urirCount': '#URI-R'
}]
# TODO: Calculate actual values
summary = {
'urimCount': '#URI-M',
'urirCount': '#URI-R',
'htmlCount': '#HTML',
'earliest': 'Date1',
'latest': 'Date2'
}
return render_template('admin.html',
status=status,
indexes=indexes,
summary=summary)
def start(cdxjFilePath, proxy=None):
hostPort = ipwbUtils.getIPWBReplayConfig()
app.proxy = proxy
if not hostPort:
ipwbUtils.setIPWBReplayConfig(IPWBREPLAY_HOST, IPWBREPLAY_PORT)
hostPort = ipwbUtils.getIPWBReplayConfig()
if ipwbUtils.isDaemonAlive():
ipwbUtils.setIPWBReplayIndexPath(cdxjFilePath)
app.cdxjFilePath = cdxjFilePath
else:
print('Sample data not pulled from IPFS.')
print('Check that the IPFS daemon is running.')
# Perform checks for CDXJ file existence, TODO: reuse cached contents
app.cdxjFileContents = getIndexFileContents(cdxjFilePath)
try:
print('IPWB replay started on http://{0}:{1}'.format(
IPWBREPLAY_HOST, IPWBREPLAY_PORT))
app.run(host='0.0.0.0', port=IPWBREPLAY_PORT)
except gaierror:
print('Detected no active Internet connection.')
print('Overriding to use default IP and port configuration.')
app.run()
except socketerror:
print('Address {0}:{1} already in use!'.format(IPWBREPLAY_HOST,
IPWBREPLAY_PORT))
sys.exit()
def checkArgs_index(args):
if not util.isDaemonAlive():
sys.exit()
encKey = None
compressionLevel = None
if args.e:
encKey = ''
if args.c:
compressionLevel = 6 # Magic 6, TA-DA!
indexer.indexFileAt(args.warcPath,
encKey,
compressionLevel,
args.compressFirst,
debug=args.debug)
def generateDaemonStatusButton():
text = 'Not Running'
buttonText = 'Start'
if ipwbConfig.isDaemonAlive():
text = 'Running'
buttonText = 'Stop'
statusPageHTML = '<html id="status{0}" class="status">'.format(buttonText)
statusPageHTML += ('<head><base href="/webui/" /><link rel="stylesheet" '
'type="text/css" href="webui.css" />'
'<script src="webui.js"></script>'
'</head><body>')
buttonHTML = '{0}<button>{1}</button>'.format(text, buttonText)
footer = '<script>assignStatusButtonHandlers()</script></body></html>'
return Response('{0}{1}{2}'.format(statusPageHTML, buttonHTML, footer))
def show_uri(path, datetime=None):
global IPFS_API
if len(path) == 0:
return showWebUI('index.html')
# TODO: Use a better approach to serve static contents
# instead of using the same logic for every JS file as the SW script
localScripts = [
'serviceWorker.js', 'reconstructive.js', 'reconstructive-banner.js'
]
if path in localScripts:
return getServiceWorker(path)
daemonAddress = '{0}:{1}'.format(IPFSAPI_IP, IPFSAPI_PORT)
if not ipwbConfig.isDaemonAlive(daemonAddress):
errStr = ('IPFS daemon not running. '
'Start it using $ ipfs daemon on the command-line '
' or from the <a href="/">'
'IPWB replay homepage</a>.')
return Response(errStr, status=503)
path = getCompleteURI(path)
cdxjLine = ''
try:
surtedURI = surt.surt(path,
path_strip_trailing_slash_unless_empty=False)
indexPath = ipwbConfig.getIPWBReplayIndexPath()
searchString = surtedURI
if datetime is not None:
searchString = surtedURI + ' ' + datetime
cdxjLine = getCDXJLine_binarySearch(searchString, indexPath)
print('CDXJ Line: {0}'.format(cdxjLine))
except Exception as e:
print(sys.exc_info()[0])
respString = ('{0} not found :(' +
' <a href="http://{1}:{2}">Go home</a>').format(
path, IPWBREPLAY_IP, IPWBREPLAY_PORT)
return Response(respString)
if cdxjLine is None: # Resource not found in archives
return generateNoMementosInterface(path, datetime)
cdxjParts = cdxjLine.split(" ", 2)
jObj = json.loads(cdxjParts[2])
datetime = cdxjParts[1]
digests = jObj['locator'].split('/')
class HashNotFoundError(Exception):
pass
try:
def handler(signum, frame):
raise HashNotFoundError()
signal.signal(signal.SIGALRM, handler)
signal.alarm(10)
payload = IPFS_API.cat(digests[-1])
header = IPFS_API.cat(digests[-2])
signal.alarm(0)
except ipfsapi.exceptions.TimeoutError:
print("{0} not found at {1}".format(cdxjParts[0], digests[-1]))
respString = ('{0} not found in IPFS :(' +
' <a href="http://{1}:{2}">Go home</a>').format(
path, IPWBREPLAY_IP, IPWBREPLAY_PORT)
return Response(respString)
except TypeError:
print('A type error occurred')
print(traceback.format_exc())
print(sys.exc_info()[0])
except HashNotFoundError:
print("Hashes not found")
return '', 404
except Exception as e:
print('Unknown exception occurred while fetching from ipfs.')
print(sys.exc_info()[0])
sys.exit()
if 'encryption_method' in jObj:
keyString = None
while keyString is None:
if 'encryption_key' in jObj:
keyString = jObj['encryption_key']
else:
askForKey = ('Enter a path for file',
' containing decryption key: \n> ')
keyString = raw_input(askForKey)
encryptionMethod = None
if jObj['encryption_method'] == 'xor':
encryptionMethod = XOR
pKey = encryptionMethod.new(keyString)
payload = pKey.decrypt(base64.b64decode(payload))
hKey = encryptionMethod.new(keyString)
header = hKey.decrypt(base64.b64decode(header))
hLines = header.split('\n')
hLines.pop(0)
resp = Response(payload)
for idx, hLine in enumerate(hLines):
k, v = hLine.split(': ', 1)
if k.lower() == 'transfer-encoding' and v.lower() == 'chunked':
try:
unchunkedPayload = extractResponseFromChunkedData(payload)
except Exception as e:
continue # Data may have no actually been chunked
resp.set_data(unchunkedPayload)
if k.lower() not in ["content-type", "content-encoding"]:
k = "X-Archive-Orig-" + k
resp.headers[k] = v
# Add ipwb header for additional SW logic
newPayload = resp.get_data()
ipwbjsinject = """<script src="/webui/webui.js"></script>
<script>injectIPWBJS()</script>"""
newPayload = newPayload.replace('</html>', ipwbjsinject + '</html>')
resp.set_data(newPayload)
resp.headers['Memento-Datetime'] = ipwbConfig.datetimeToRFC1123(datetime)
# Get TimeMap for Link response header
respWithLinkHeader = getLinkHeaderAbbreviatedTimeMap(path, datetime)
resp.headers['Link'] = respWithLinkHeader.replace('\n', ' ')
return resp
def show_uri(path):
global IPFS_API
if len(path) == 0:
return showWebUI('index.html')
sys.exit()
daemonAddress = '{0}:{1}'.format(IPFSAPI_IP, IPFSAPI_PORT)
if not ipwbConfig.isDaemonAlive(daemonAddress):
errStr = ('IPFS daemon not running. '
'Start it using $ ipfs daemon on the command-line '
' or from the <a href="/">'
'IPWB replay homepage</a>.')
return Response(errStr)
# show the user profile for that user
cdxjLine = ''
try:
s = surt.surt(path, path_strip_trailing_slash_unless_empty=False)
indexPath = ipwbConfig.getIPWBReplayIndexPath()
cdxjLine = getCDXJLine_binarySearch(s, indexPath)
except:
print sys.exc_info()[0]
respString = ('{0} not found :(' +
' <a href="http://{1}:{2}">Go home</a>').format(
path, IPWBREPLAY_IP, IPWBREPLAY_PORT)
return Response(respString)
if cdxjLine is None: # Resource not found in archives
return Response(status=404)
cdxjParts = cdxjLine.split(" ", 2)
jObj = json.loads(cdxjParts[2])
digests = jObj['locator'].split('/')
try:
payload = IPFS_API.cat(digests[-1], timeout=1)
header = IPFS_API.cat(digests[-2])
except ipfsapi.exceptions.TimeoutError:
print "{0} not found at {1}".format(cdxjParts[0], digests[-1])
respString = ('{0} not found in IPFS :(' +
' <a href="http://{1}:{2}">Go home</a>').format(
path, IPWBREPLAY_IP, IPWBREPLAY_PORT)
return Response(respString)
except:
print sys.exc_info()[0]
print "general error"
sys.exit()
if 'encryption_method' in jObj:
pKey = XOR.new(jObj['encryption_key'])
payload = pKey.decrypt(base64.b64decode(payload))
hKey = XOR.new(jObj['encryption_key'])
header = hKey.decrypt(base64.b64decode(header))
hLines = header.split('\n')
hLines.pop(0)
resp = Response(payload)
for idx, hLine in enumerate(hLines):
k, v = hLine.split(': ', 1)
if k.lower() != "content-type":
k = "X-Archive-Orig-" + k
resp.headers[k] = v
return resp
def show_uri(path, datetime=None):
global IPFS_API
if len(path) == 0:
return showWebUI('index.html')
sys.exit()
if path == 'serviceWorker.js':
return getServiceWorker(path)
sys.exit()
daemonAddress = '{0}:{1}'.format(IPFSAPI_IP, IPFSAPI_PORT)
if not ipwbConfig.isDaemonAlive(daemonAddress):
errStr = ('IPFS daemon not running. '
'Start it using $ ipfs daemon on the command-line '
' or from the <a href="/">'
'IPWB replay homepage</a>.')
return Response(errStr)
cdxjLine = ''
try:
surtedURI = surt.surt( # Good ol' pep8 line length
path,
path_strip_trailing_slash_unless_empty=False)
indexPath = ipwbConfig.getIPWBReplayIndexPath()
searchString = surtedURI
if datetime is not None:
searchString = surtedURI + ' ' + datetime
cdxjLine = getCDXJLine_binarySearch(searchString, indexPath)
except:
print sys.exc_info()[0]
respString = ('{0} not found :(' +
' <a href="http://{1}:{2}">Go home</a>').format(
path, IPWBREPLAY_IP, IPWBREPLAY_PORT)
return Response(respString)
if cdxjLine is None: # Resource not found in archives
msg = '<h1>ERROR 404</h1>'
msg += 'No capture found for {0} at {1}.'.format(path, datetime)
linesWithSameURIR = getCDXJLinesWithURIR(path)
if linesWithSameURIR:
msg += '<p>{0} capture(s) available:</p><ul>'.format(
len(linesWithSameURIR))
for line in linesWithSameURIR:
fields = line.split(' ', 2)
msg += ('<li><a href="/{1}/{0}">{0} at {1}</a></li>'.format(
unsurt(fields[0]), fields[1]))
msg += '</ul>'
return Response(msg, status=404)
cdxjParts = cdxjLine.split(" ", 2)
jObj = json.loads(cdxjParts[2])
datetime = cdxjParts[1]
digests = jObj['locator'].split('/')
try:
payload = IPFS_API.cat(digests[-1], timeout=1)
header = IPFS_API.cat(digests[-2])
except ipfsapi.exceptions.TimeoutError:
print "{0} not found at {1}".format(cdxjParts[0], digests[-1])
respString = ('{0} not found in IPFS :(' +
' <a href="http://{1}:{2}">Go home</a>').format(
path, IPWBREPLAY_IP, IPWBREPLAY_PORT)
return Response(respString)
except:
print sys.exc_info()[0]
print "general error"
sys.exit()
if 'encryption_method' in jObj:
pKey = XOR.new(jObj['encryption_key'])
payload = pKey.decrypt(base64.b64decode(payload))
hKey = XOR.new(jObj['encryption_key'])
header = hKey.decrypt(base64.b64decode(header))
hLines = header.split('\n')
hLines.pop(0)
resp = Response(payload)
for idx, hLine in enumerate(hLines):
k, v = hLine.split(': ', 1)
if k.lower() == 'transfer-encoding' and v.lower() == 'chunked':
resp.set_data(extractResponseFromChunkedData(payload))
if k.lower() != "content-type":
k = "X-Archive-Orig-" + k
resp.headers[k] = v
resp.headers['Memento-Datetime'] = ipwbConfig.datetimeToRFC1123(datetime)
return resp
def show_uri(path, datetime=None):
global IPFS_API
if len(path) == 0:
return showWebUI('index.html')
# TODO: Use a better approach to serve static contents
# instead of using the same logic for every JS file as the SW script
localScripts = [
'serviceWorker.js', 'reconstructive.js', 'reconstructive-banner.js'
]
if path in localScripts:
return getServiceWorker(path)
daemonAddress = '{0}:{1}'.format(IPFSAPI_HOST, IPFSAPI_PORT)
if not ipwbUtils.isDaemonAlive(daemonAddress):
errStr = ('IPFS daemon not running. '
'Start it using $ ipfs daemon on the command-line '
' or from the <a href="/">'
'IPWB replay homepage</a>.')
return Response(errStr, status=503)
path = getCompleteURI(path)
cdxjLine = ''
try:
surtedURI = surt.surt(path,
path_strip_trailing_slash_unless_empty=False)
indexPath = ipwbUtils.getIPWBReplayIndexPath()
searchString = surtedURI
if datetime is not None:
searchString = surtedURI + ' ' + datetime
cdxjLine = getCDXJLine_binarySearch(searchString, indexPath)
print('CDXJ Line: {0}'.format(cdxjLine))
except Exception as e:
print(sys.exc_info()[0])
respString = ('{0} not found :(' +
' <a href="http://{1}:{2}">Go home</a>').format(
path, IPWBREPLAY_HOST, IPWBREPLAY_PORT)
return Response(respString)
if cdxjLine is None: # Resource not found in archives
return generateNoMementosInterface(path, datetime)
cdxjParts = cdxjLine.split(" ", 2)
jObj = json.loads(cdxjParts[2])
datetime = cdxjParts[1]
digests = jObj['locator'].split('/')
class HashNotFoundError(Exception):
pass
payload = None
header = None
try:
def handler(signum, frame):
raise HashNotFoundError()
if os.name != 'nt': # Bug #310
signal.signal(signal.SIGALRM, handler)
signal.alarm(10)
payload = IPFS_API.cat(digests[-1])
header = IPFS_API.cat(digests[-2])
if os.name != 'nt': # Bug #310
signal.alarm(0)
except ipfsapi.exceptions.TimeoutError:
print("{0} not found at {1}".format(cdxjParts[0], digests[-1]))
respString = ('{0} not found in IPFS :(' +
' <a href="http://{1}:{2}">Go home</a>').format(
path, IPWBREPLAY_HOST, IPWBREPLAY_PORT)
return Response(respString)
except TypeError as e:
print('A type error occurred')
print(e)
abort(500)
except HTTPError as e:
print("Fetching from the IPFS failed")
print(e)
abort(503)
except HashNotFoundError:
if payload is None:
print("Hashes not found:\n\t{0}\n\t{1}".format(
digests[-1], digests[-2]))
abort(404)
else: # payload found but not header, fabricate header
print("HTTP header not found, fabricating for resp replay")
header = ''
except Exception as e:
print('Unknown exception occurred while fetching from ipfs.')
print(e)
abort(500)
if 'encryption_method' in jObj:
keyString = None
while keyString is None:
if 'encryption_key' in jObj:
keyString = jObj['encryption_key']
else:
askForKey = ('Enter a path for file',
' containing decryption key: \n> ')
keyString = raw_input(askForKey)
paddedEncryptionKey = pad(keyString, AES.block_size)
key = base64.b64encode(paddedEncryptionKey)
nonce = b64decode(jObj['encryption_nonce'])
cipher = AES.new(key, AES.MODE_CTR, nonce=nonce)
header = cipher.decrypt(base64.b64decode(header))
payload = cipher.decrypt(base64.b64decode(payload))
hLines = header.split('\n')
hLines.pop(0)
status = 200
if 'status_code' in jObj:
status = jObj['status_code']
resp = Response(payload, status=status)
for idx, hLine in enumerate(hLines):
k, v = hLine.split(': ', 1)
if k.lower() == 'transfer-encoding' and v.lower() == 'chunked':
try:
unchunkedPayload = extractResponseFromChunkedData(payload)
except Exception as e:
continue # Data may have no actually been chunked
resp.set_data(unchunkedPayload)
if k.lower() not in ["content-type", "content-encoding", "location"]:
k = "X-Archive-Orig-" + k
resp.headers[k] = v
# Add ipwb header for additional SW logic
newPayload = resp.get_data()
ipwbjsinject = """<script src="/webui/webui.js"></script>
<script>injectIPWBJS()</script>"""
newPayload = newPayload.replace('</html>', ipwbjsinject + '</html>')
resp.set_data(newPayload)
resp.headers['Memento-Datetime'] = ipwbUtils.digits14ToRFC1123(datetime)
if header is None:
resp.headers['X-Headers-Generated-By'] = 'InterPlanetary Wayback'
# Get TimeMap for Link response header
# respWithLinkHeader = getLinkHeaderAbbreviatedTimeMap(path, datetime)
# resp.headers['Link'] = respWithLinkHeader.replace('\n', ' ')
if status[0] == '3' and isUri(resp.headers.get('Location')):
# Bad assumption that the URI-M will contain \d14 but works for now.
uriBeforeURIR = request.url[:re.search(r'/\d{14}/', request.url).end()]
newURIM = uriBeforeURIR + resp.headers['Location']
resp.headers['Location'] = newURIM
return resp
