import mwcli
router = mwcli.Router(
"mwtext",
"This script provides access to a set of utilities for text processing", {
'preprocess_text':
"Converts an XML dump to preprocessed plaintext. " +
"One line per chunk.",
'learn_vectors':
"Learn a set of word vectors from preprocessed " + "plaintext",
'word2vec2gensim':
"Converts word2vec format to gensim KeyedVector " + "binaries"
})
main = router.main
from __future__ import absolute_import
import mwcli
router = mwcli.Router(
u"mwxml",
u"This script provides access to a set of utilities for extracting " +
u"content from MediaWiki XML dumps.", {
u'dump2revdocs':
u"Converts XML dumps to revision documents (XML --> JSON)",
u'validate':
u"Compares a stream of revision documents against a schema",
u'normalize':
u"Converts a stream of old revision documents to documents " +
u"that validate against the current schema",
u'inflate':
u"Converts a stream of flat revision documents to standard " +
u"revision documents"
})
main = router.main
import mwcli
router = mwcli.Router(
"mwrefs",
"A set of utilities for extracting and processing <ref>s in " +
" MediaWiki projects.",
{'diffs': "Extracts changes to <ref>s from XML dumps",
'extract': "Extracts all <ref>s from XML dumps",
'fetch_references': "Gets the reference documents for a revision from " +
"the a MediaWiki API"}
)
main = router.main
import mwcli
router = mwcli.Router(
"mwviews",
"This script provides access to a set of utilities for processing view counts.",
{
'aggregate':
"Aggregate view counts from hourly view files",
'fetch_global_namespaces':
"Fetches a dataset of namespace names for all wikis"
})
main = router.main
import mwcli
router = mwcli.Router(
"mwtext",
"This script provides access to a set of utilities for text processing", {
'transform_content':
"Transforms an XML dump using a transformer",
'words2plaintext':
"Converts a 'words' type transformation into " +
"plaintext -- optionally with labels",
'learn_vectors':
"Learn a set of word vectors from preprocessed " + "plaintext",
'word2vec2gensim':
"Converts word2vec format to gensim KeyedVector " + "binaries"
})
main = router.main