def showMementosForURIRs(urir):
urir = getCompleteURI(urir)
if ipwbConfig.isLocalHosty(urir):
urir = urir.split('/', 4)[4]
s = surt.surt(urir, path_strip_trailing_slash_unless_empty=False)
indexPath = ipwbConfig.getIPWBReplayIndexPath()
print('Getting CDXJ Lines with the URI-R {0} from {1}'.format(
urir, indexPath))
cdxjLinesWithURIR = getCDXJLinesWithURIR(urir, indexPath)
if len(cdxjLinesWithURIR) == 1:
fields = cdxjLinesWithURIR[0].split(' ', 2)
redirectURI = '/{1}/{0}'.format(unsurt(fields[0]), fields[1])
return redirect(redirectURI, code=302)
msg = ''
if cdxjLinesWithURIR:
msg += '<p>{0} capture(s) available:</p><ul>'.format(
len(cdxjLinesWithURIR))
for line in cdxjLinesWithURIR:
fields = line.split(' ', 2)
dt14 = fields[1]
dtrfc1123 = ipwbConfig.datetimeToRFC1123(fields[1])
msg += ('<li><a href="/{1}/{0}">{0} at {2}</a></li>'.format(
unsurt(fields[0]), dt14, dtrfc1123))
msg += '</ul>'
return Response(msg)
def generateTimeMapFromCDXJLines(cdxjLines, original, tmself):
tmData = '<{0}>; rel="original",\n'.format(unsurt(original))
tmData += '<{0}>; rel="self"; '.format(tmself)
tmData += 'type="application/link-format",\n'
hostAndPort = tmself[0:tmself.index('timemap/')]
for line in cdxjLines:
(surtURI, datetime, json) = line.split(' ', 2)
dtRFC1123 = ipwbConfig.datetimeToRFC1123(datetime)
tmData += '<{0}{1}/{2}>; rel="memento"; datetime="{3}",\n'.format(
hostAndPort, datetime, unsurt(surtURI), dtRFC1123)
tmData = tmData[0:-2] # Trim final , and LF
return tmData
def generateCDXJTimeMapFromCDXJLines(cdxjLines, original, tmself):
tmurl = getProxiedURIT(tmself)
if app.proxy is not None:
tmself = urlunsplit(tmurl)
# unsurted URI will never have a scheme, add one
originalURI = 'http://{0}'.format(unsurt(original))
tmData = '!context ["http://tools.ietf.org/html/rfc7089"]\n'
tmData += '!id {{"uri": "{0}"}}\n'.format(tmself)
tmData += '!keys ["memento_datetime_YYYYMMDDhhmmss"]\n'
tmData += '!meta {{"original_uri": "{0}"}}\n'.format(originalURI)
linkTMURI = tmself.replace('/timemap/cdxj/', '/timemap/link/')
tmData += ('!meta {{"timemap_uri": {{'
'"link_format": "{0}", '
'"cdxj_format": "{1}"'
'}}}}\n').format(linkTMURI, tmself)
hostAndPort = tmself[0:tmself.index('timemap/')]
for i, line in enumerate(cdxjLines):
(surtURI, datetime, json) = line.split(' ', 2)
dtRFC1123 = ipwbConfig.datetimeToRFC1123(datetime)
firstLastStr = ''
if len(cdxjLines) > 1:
if i == 0:
firstLastStr = 'first '
elif i == len(cdxjLines) - 1:
firstLastStr = 'last '
elif len(cdxjLines) == 1:
firstLastStr = 'first last '
tmData += ('{1} {{'
'"uri": "{0}{1}/{2}", '
'"rel": "{3}memento", '
'"datetime"="{4}"}}\n').format(hostAndPort, datetime,
unsurt(surtURI),
firstLastStr, dtRFC1123)
tmData = tmData[0:-1] # Trim final , and LF
return tmData
def generateLinkTimeMapFromCDXJLines(cdxjLines, original, tmself):
tmurl = getProxiedURIT(tmself)
if app.proxy is not None:
tmself = urlunsplit(tmurl)
# Extract and trim for host:port prepending
tmurl[2] = '' # Clear TM path
hostAndPort = urlunsplit(tmurl) + '/'
# unsurted URI will never have a scheme, add one
originalURI = 'http://{0}'.format(unsurt(original))
tmData = '<{0}>; rel="original",\n'.format(originalURI)
tmData += '<{0}>; rel="self timemap"; '.format(tmself)
tmData += 'type="application/link-format",\n'
cdxjTMURI = tmself.replace('/timemap/link/', '/timemap/cdxj/')
tmData += '<{0}>; rel="timemap"; '.format(cdxjTMURI)
tmData += 'type="application/cdxj+ors",\n'
for i, line in enumerate(cdxjLines):
(surtURI, datetime, json) = line.split(' ', 2)
dtRFC1123 = ipwbConfig.datetimeToRFC1123(datetime)
firstLastStr = ''
if len(cdxjLines) > 1:
if i == 0:
firstLastStr = 'first '
elif i == len(cdxjLines) - 1:
firstLastStr = 'last '
elif len(cdxjLines) == 1:
firstLastStr = 'first last '
tmData += '<{0}{1}/{2}>; rel="{3}memento"; datetime="{4}",\n'.format(
hostAndPort, datetime, unsurt(surtURI), firstLastStr, dtRFC1123)
tmData = tmData[0:-2] # Trim final , and LF
return tmData
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, 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
