def remove_asynchronous_hash_request(host, username, password, request_id):
"""Performs a removeAsynchronousHashRequest operation on the given host and returns
a hasherWsAsynchronousResult record with these fields:
- requestId (string)
- requestTime (long)
- blockFileDataHandler (base64Binary)
- blockFileName (string)
- bytesHashed (long)
- elapsedTime (long)
- errorMessage (string)
- filesHashed (int)
- hashResult (base64Binary)
- recordFileDataHandler (base64Binary)
- recordFileName (string)
- startTime (long)
- status (string)
Parameters:
:param host: a host:port pair (string)
:param username: a username for the host (string)
:param password: a password for the host (string)
:param request_id: a request id (string)
:return:
"""
client = make_client(host, username, password, _service)
return client.service.removeAsynchronousHashRequest(requestId=request_id)
def _hash_asynchronously(host, username, password, hasher_params):
"""Performs a hashAsynchronously operation on the given host with the given hasher parameters and returns
a hasherWsAsynchronousResult record with these fields:
- requestId (string)
- requestTime (long)
- blockFileDataHandler (base64Binary)
- blockFileName (string)
- bytesHashed (long)
- elapsedTime (long)
- errorMessage (string)
- filesHashed (int)
- hashResult (base64Binary)
- recordFileDataHandler (base64Binary)
- recordFileName (string)
- startTime (long)
- status (string)
Parameters:
:param host: a host:port pair (string)
:param username: a username for the host (string)
:param password: a password for the host (string)
:param hasher_params: hasherWsParams
:return:
"""
client = make_client(host, username, password, _service)
return client.service.hashAsynchronously(hasherParams=hasher_params)
def get_all_asynchronous_hash_results(host, username, password):
"""Performs a getAllAsynchronousHashResults operation on the givne host and returns
an array of hasherWsAsynchronousResult with these fields:
- requestId (string)
- requestTime (long)
- blockFileDataHandler (base64Binary)
- blockFileName (string)
- bytesHashed (long)
- elapsedTime (long)
- errorMessage (string)
- filesHashed (int)
- hashResult (base64Binary)
- recordFileDataHandler (base64Binary)
- recordFileName (string)
- startTime (long)
- status (string)
Parameters:
:param host: a host:port pair (string)
:param username: a username for the host (string)
:param password: a password for the host (string)
:return:
"""
client = make_client(host, username, password, _service)
return client.service.getAllAsynchronousHashResults()
def is_daemon_ready(host, username, password):
'''Performs an isDaemonReady operation on the given host and returns True or
False.
Parameters:
:param host: a host:port pair (string)
:param username: a username for the host (string)
:param password: a password for the host (string)
'''
client = make_client(host, username, password, _service)
return client.service.isDaemonReady()
def get_au_status(host, username, password, auid):
'''Performs a getAuStatus operation on the given host for the given AUID, and
returns a record with these fields (or None if zeep.exceptions.Fault with
'No Archival Unit with provided identifier' is raised):
- accessType (string)
- availableFromPublisher (boolean)
- contentSize (numeric)
- crawlPool (string)
- crawlProxy (string)
- crawlWindow (string)
- creationTime (numeric)
- currentlyCrawling (boolean)
- currentlyPolling (boolean)
- diskUsage (numeric)
- journalTitle (string)
- lastCompletedCrawl (numeric)
- lastCompletedPoll (numeric)
- lastCrawl (numeric)
- lastCrawlResult (string)
- lastCompletedDeepCrawl (numeric)
- lastDeepCrawl (numeric)
- lastDeepCrawlResult (string)
- lastCompletedDeepCrawlDepth (numeric)
- lastPoll (numeric)
- lastPollResult (string)
- lastMetadataIndex (numeric)
- pluginName (string)
- provider (string)
- publisher (string)
- publishingPlatform (string)
- recentPollAgreement (floating point)
- repository (string)
- status (string)
- subscriptionStatus (string)
- substanceState (string)
- volume (string) (the AU name)
- year (string)
Parameters:
:param host: a host:port pair (string)
:param username: a username for the host (string)
:param password: a password for the host (string)
:param auid: an auid to hash (string)
'''
client = make_client(host, username, password, _service)
try:
ret = client.service.getAuStatus(auId=auid)
return zeep.helpers.serialize_object(ret)
except zeep.exceptions.Fault as e:
if e.message == 'No Archival Unit with provided identifier':
return None
else:
raise
def query_crawls(host, username, password, select, where=None):
'''Performs a queryCrawls operation on the given host, using the given field
names to build a SELECT clause, optionally using the given string to build a
WHERE clause, and returns a list of records with these fields (populated or
not depending on the SELECT clause):
- auId (string)
- auName (string)
- bytesFetchedCount (long)
- crawlKey (string)
- crawlStatus (string)
- crawlType (string)
- duration (long)
- linkDepth (int)
- mimeTypeCount (int)
- mimeTypes (list of strings)
- offSiteUrlsExcludedCount (int)
- pagesExcluded (list of strings)
- pagesExcludedCount (int)
- pagesFetched (list of strings)
- pagesFetchedCount (int)
- pagesNotModified (list of strings)
- pagesNotModifiedCount (int)
- pagesParsed (list of strings)
- pagesParsedCount (int)
- pagesPending (list of strings)
- pagesPendingCount (int)
- pagesWithErrors, a list of records with these fields:
- message (string)
- severity (string)
- url (string)
- pagesWithErrorsCount (int)
- refetchDepth (int)
- sources (list of strings)
- startTime (long)
- startingUrls (list of strings)
Parameters:
:param host: a host:port pair (string)
:param username: a username for the host (string)
:param password: a password for the host (string)
:param select: if a list of strings, the field names to
be used in the SELECT clause; if a string, the single field name to be used in
the SELECT clause (string or list of strings)
:param where : optional statement for the WHERE clause (string, default: None)
Raises:
- ValueError if select is not of the right type
'''
query = construct_query(select, where)
client = make_client(host, username, password, _service)
ret = client.service.queryCrawls(crawlQuery=query)
return zeep.helpers.serialize_object(ret)
def get_auids(host, username, password):
'''Performs a getAuIds operation on the given host, which really produces a
sequence of all AUIDs with the AU names, and returns a list of records with
these fields:
- id (string)
- name (string)
Parameters:
:param host: a host:port pair (string)
:param username: a username for the host (string)
:param password: a password for the host (string)
'''
client = make_client(host, username, password, _service)
ret = client.service.getAuIds()
return zeep.helpers.serialize_object(ret)
def get_au_urls(host, username, password, auid, prefix=None):
'''Performs a getAuUrls operation on the given host for the given AUID and
returns a list of URLs (strings) in the AU. If the optional prefix argument is
given, limits the results to URLs with that prefix (including the URL itself).
Parameters:
:param host: a host:port pair (string)
:param username: a username for the host (string)
:param password: a password for the host (string)
:param auid: an auid to hash (string)
:param prefix: a URL prefix (string, default: None)
'''
client = make_client(host, username, password, _service)
ret = client.service.getAuUrls(auId=auid, url=prefix)
return ret
def add_aus_by_id_list(host, username, password, auids):
'''
Performs an addAusByIdList operation (which adds all given AUs on a single
host, by AUID), and returns a list of records with these fields:
- id (string): the AUID
- isSuccess (boolean): a success flag
- message (string): an error message
- name (string): the AU name
Parameters:
- host (string): a host:port pair
- username (string): a username for the host
- password (string): a password for the host
- auids (list of strings): a list of AUIDs
'''
client = make_client(host, username, password, _service)
return client.service.addAusByIdList(auIds=auids)
def deactivate_au_by_id(host, username, password, auid):
'''
Performs a deactivateAuById operation (which deactivates a single AU on a
single host, by AUID), and returns a record with these fields:
- id (string): the AUID
- isSuccess (boolean): a success flag
- message (string): an error message
- name (string): the AU name
Parameters:
- host (string): a host:port pair
- username (string): a username for the host
- password (string): a password for the host
- auid (string): an AUID
'''
client = make_client(host, username, password, _service)
return client.service.deactivateAuById(auId=auid)
def create_export_files(host, username, password, auid, options):
'''Performs a createExportFiles operation on the given host for the given AUID, and
returns a record with the files.
Parameters:
:param host: a host:port pair (string)
:param username: a username for the host (string)
:param password: a password for the host (string)
:param auid: an auid to hash (string)
Returns:
- ret (dict):
{
'auId': '<auid>',
'dataHandlerWrappers': [
{
'dataHandler' (base64Binary): zipped AU
'name' (string): '<prefix>-<timestamp>-<5 digit id>.zip',
'size' (long): size of dataHandler in bytes
}
]
}
'''
req = {
'auid': auid,
'compress': options.compress,
'excludeDirNodes': options.exclude_dir,
'filePrefix': options.output_prefix,
'fileType': options.file_type,
'maxSize': options.max_size,
'maxVersions': options.max_vers,
'xlateFilenames': options.translate
}
client = make_client(host, username, password, _service)
try:
ret = client.service.createExportFiles(req)
return zeep.helpers.serialize_object(ret)
except zeep.exceptions.Fault as e:
if e.message == 'No Archival Unit with provided identifier':
return None
else:
raise
def get_platform_configuration(host, username, password):
'''Performs a getPlatformConfiguration operation on the given host and returns
a record with these fields:
- adminEmail (string)
- buildHost (string)
- buildTimestamp (numeric)
- currentTime (numeric)
- currentWorkingDirectory (string)
- daemonVersion, a record with these fields:
- buildVersion (numeric)
- fullVersion (string)
- majorVersion (numeric)
- minorVersion (numeric)
- disks (list of strings)
- groups (list of strings)
- hostName (string)
- ipAddress (string)
- javaVersion, a record with these fields:
- runtimeName (string)
- runtimeVersion (string)
- specificationVersion (string)
- version (string)
- mailRelay (string)
- platform, a record with these fields:
- name (string)
- suffix (string)
- version (string)
- project (string)
- properties (list of strings)
- uptime (numeric)
- v3Identity (string)
Parameters:
:param host: a host:port pair (string)
:param username: a username for the host (string)
:param password: a password for the host (string)
'''
client = make_client(host, username, password, _service)
return zeep.helpers.serialize_object(
client.service.getPlatformConfiguration())
def query_tdb_titles_by_auid(host, username, password, auid, select="*"):
'''Performs a queryTdbTitles operation on the given host for the given auid,
and returns an list of records with these fields (populated or
not depending on the SELECT clause):
- EIssn (string)
- id (string)
- issn (string)
- issnL (string)
- issns (string)
- name (string)
- printIssn (string)
- proprietaryId (string)
- proprietaryIds (string)
- publicationType (string)
- tdbPublisher (tns:tdbPublisherWsResult)
Parameters:
:param host: a host:port pair (string)
:param username: a username for the host (string)
:param password: a password for the host (string)
:param auid: an auid to hash (string)
:param select: if a list of strings, the field names to
be used in the SELECT clause; if a string, the single field name to be used in
the SELECT clause (string or list of strings)
Raises:
- ValueError if auid or select is not of the right type
'''
client = make_client(host, username, password, _service)
query = construct_query(select)
try:
# first we get the publisher given the auid and filter the query
au_journal = client.service.getAuStatus(auId=auid)['journalTitle']
query += " WHERE name={}".format(repr(au_journal))
ret = client.service.queryTdbTitles(tdbTitleQuery=query)
return zeep.helpers.serialize_object(ret)
except zeep.exceptions.Fault as e:
if e.message == 'No Archival Unit with provided identifier':
return None
else:
raise
def query_aus(host, username, password, select, where=None):
'''Performs a queryAus operation on the given host, using the given field
names to build a SELECT clause, optionally using the given string to build a
WHERE clause, and returns a list of records with these fields (populated or
not depending on the SELECT clause):
- accessType (string)
- articleUrls (list of strings)
- auConfiguration, a record with these fields:
- defParams, a list of records with these fields:
- key (string)
- value (string)
- nonDefParams, a list of records with these fields:
- key (string)
- value (string)
- auId (string)
- availableFromPublisher (boolean)
- contentSize (numeric)
- crawlPool (string)
- crawlProxy (string)
- crawlWindow (string)
- creationTime (numeric)
- currentlyCrawling (boolean)
- currentlyPolling (boolean)
- diskUsage (numeric)
- highestPollAgreement (numeric)
- isBulkContent (boolean)
- journalTitle (string)
- lastCompletedCrawl (numeric)
- lastCompletedPoll (numeric)
- lastCrawl (numeric)
- lastCrawlResult (string)
- lastCompletedDeepCrawl (numeric)
- lastDeepCrawl (numeric)
- lastDeepCrawlResult (string)
- lastCompletedDeepCrawlDepth (numeric)
- lastPoll (numeric)
- lastPollResult (string)
- lastMetadataIndex (numeric)
- name (string)
- newContentCrawlUrls (list of strings)
- peerAgreements, a list of records with these fields:
- agreements, a record with these fields:
- entry, a list of records with these fields:
- key, a string among:
- "POR"
- "POP"
- "SYMMETRIC_POR"
- "SYMMETRIC_POP"
- "POR_HINT"
- "POP_HINT"
- "SYMMETRIC_POR_HINT"
- "SYMMETRIC_POP_HINT"
- "W_POR"
- "W_POP"
- "W_SYMMETRIC_POR"
- "W_SYMMETRIC_POP"
- "W_POR_HINT"
- "W_POP_HINT"
- "W_SYMMETRIC_POR_HINT"
- "W_SYMMETRIC_POP_HINT"
- value, a record with these fields:
- HighestPercentAgreement (floating point)
- HighestPercentAgreementTimestamp (numeric)
- PercentAgreement (floating point)
- PercentAgreementTimestamp (numeric)
- peerId (string)
- pluginName (string)
- publishingPlatform (string)
- recentPollAgreement (numeric)
- repositoryPath (string)
- subscriptionStatus (string)
- substanceState (string)
- tdbProvider (string)
- tdbPublisher (string)
- tdbYear (string)
- urlStems (list of strings)
- urls, a list of records with these fields:
- currentVersionSize (numeric)
- url (string)
- versionCount (numeric)
- volume (string)
Parameters:
:param host: a host:port pair (string)
:param username: a username for the host (string)
:param password: a password for the host (string)
:param select: if a list of strings, the field names to
be used in the SELECT clause; if a string, the single field name to be used in
the SELECT clause (string or list of strings)
:param where : optional statement for the WHERE clause (string, default: None)
Raises:
- ValueError if select is not of the right type
'''
query = construct_query(select, where)
client = make_client(host, username, password, _service)
ret = client.service.queryAus(auQuery=query)
return zeep.helpers.serialize_object(ret)
def request_crawl_by_id(host, username, password, auid, priority, force):
client = make_client(host, username, password, _service)
return zeep.helpers.serialize_object(
client.service.requestCrawlById(auId=auid, priority=priority, force=force))
def request_crawl_by_id_list(host, username, password, auids, priority, force):
client = make_client(host, username, password, _service)
return client.service.requestCrawlByIdList(auIds=auids, priority=priority, force=force)
def request_deep_crawl_by_id_list(host, username, password, auids, refetch_depth, priority, force):
client = make_client(host, username, password, _service)
return client.service.requestDeepCrawlByIdList(
auIds=auids, refetchDepth=refetch_depth, priority=priority, force=force)
