def handle(self, *args, **options):
languages = Language.objects.all()
base = IndexUpdate.objects.all()
if base.count() == 0:
return
with FULLTEXT_INDEX.source_writer(buffered = False) as writer:
for update in base.iterator():
Unit.objects.add_to_source_index(
update.unit.checksum,
update.unit.source,
update.unit.context,
writer)
for lang in languages:
with FULLTEXT_INDEX.target_writer(lang = lang.code, buffered = False) as writer:
for update in base.filter(unit__translation__language =
lang).exclude(unit__target = '').iterator():
Unit.objects.add_to_target_index(
update.unit.checksum,
update.unit.target,
writer)
base.delete()
def handle(self, *args, **options):
languages = Language.objects.all()
if options['clean']:
create_source_index()
for lang in languages:
create_target_index(lang = lang.code)
base = self.get_units(*args, **options)
if base.count() == 0:
return
with FULLTEXT_INDEX.source_writer(buffered = False) as writer:
for unit in base.values('checksum', 'source', 'context').iterator():
Unit.objects.add_to_source_index(
unit['checksum'],
unit['source'],
unit['context'],
writer)
for lang in languages:
with FULLTEXT_INDEX.target_writer(lang = lang.code, buffered = False) as writer:
for unit in base.filter(translation__language =
lang).exclude(target = '').values('checksum', 'target').iterator():
Unit.objects.add_to_target_index(
unit['checksum'],
unit['target'],
writer)
def handle(self, *args, **options):
languages = Language.objects.all()
base = IndexUpdate.objects.all()
if base.count() == 0:
return
with FULLTEXT_INDEX.source_writer(buffered=False) as writer:
for update in base.iterator():
Unit.objects.add_to_source_index(update.unit.checksum,
update.unit.source,
update.unit.context, writer)
for lang in languages:
with FULLTEXT_INDEX.target_writer(lang=lang.code,
buffered=False) as writer:
for update in base.filter(
unit__translation__language=lang).exclude(
unit__target='').iterator():
Unit.objects.add_to_target_index(update.unit.checksum,
update.unit.target,
writer)
base.delete()
def search(self, query, source=True, context=True, translation=True, checksums=False):
"""
Performs full text search on defined set of fields.
Returns queryset unless checksums is set.
"""
ret = set()
if source or context:
with FULLTEXT_INDEX.source_searcher(not settings.OFFLOAD_INDEXING) as searcher:
if source:
ret = ret.union(self.__search(searcher, "source", SOURCE_SCHEMA, query))
if context:
ret = ret.union(self.__search(searcher, "context", SOURCE_SCHEMA, query))
if translation:
sample = self.all()[0]
with FULLTEXT_INDEX.target_searcher(
sample.translation.language.code, not settings.OFFLOAD_INDEXING
) as searcher:
ret = ret.union(self.__search(searcher, "target", TARGET_SCHEMA, query))
if checksums:
return ret
return self.filter(checksum__in=ret)
def handle(self, *args, **options):
languages = Language.objects.all()
if options['clean']:
create_source_index()
for lang in languages:
create_target_index(lang=lang.code)
base = self.get_units(*args, **options)
if base.count() == 0:
return
with FULLTEXT_INDEX.source_writer(buffered=False) as writer:
for unit in base.values('checksum', 'source',
'context').iterator():
Unit.objects.add_to_source_index(unit['checksum'],
unit['source'],
unit['context'], writer)
for lang in languages:
with FULLTEXT_INDEX.target_writer(lang=lang.code,
buffered=False) as writer:
for unit in base.filter(translation__language=lang).exclude(
target='').values('checksum', 'target').iterator():
Unit.objects.add_to_target_index(unit['checksum'],
unit['target'], writer)
def add_to_index(self, unit, writer_target=None, writer_source=None):
"""
Updates/Adds to all indices given unit.
"""
if writer_target is None:
writer_target = FULLTEXT_INDEX.target_writer(unit.translation.language.code)
if writer_source is None:
writer_source = FULLTEXT_INDEX.source_writer()
self.add_to_source_index(unit.checksum, unit.source, unit.context, writer_source)
self.add_to_target_index(unit.checksum, unit.target, writer_target)
def add_to_index(self, unit):
"""
Updates/Adds to all indices given unit.
"""
if settings.OFFLOAD_INDEXING:
from weblate.trans.models import IndexUpdate
IndexUpdate.objects.get_or_create(unit=unit)
return
writer_target = FULLTEXT_INDEX.target_writer(unit.translation.language.code)
writer_source = FULLTEXT_INDEX.source_writer()
self.add_to_source_index(unit.checksum, unit.source, unit.context, writer_source)
self.add_to_target_index(unit.checksum, unit.target, writer_target)
def similar(self, unit):
'''
Finds similar units to current unit.
'''
ret = set([unit.checksum])
with FULLTEXT_INDEX.source_searcher(not appsettings.OFFLOAD_INDEXING) as searcher:
# Extract up to 10 terms from the source
terms = [kw for kw, score in searcher.key_terms_from_text('source', unit.source, numterms=10) if not kw in IGNORE_SIMILAR]
cnt = len(terms)
# Try to find at least configured number of similar strings, remove up to 4 words
while len(ret) < appsettings.SIMILAR_MESSAGES and cnt > 0 and len(terms) - cnt < 4:
for search in itertools.combinations(terms, cnt):
results = self.search(
' '.join(search),
True,
False,
False,
True
)
ret = ret.union(results)
cnt -= 1
return self.filter(
translation__subproject__project=unit.translation.subproject.project,
translation__language=unit.translation.language,
checksum__in=ret
).exclude(
target__in=['', unit.target]
)
def similar(self, unit):
'''
Finds similar units to current unit.
'''
ret = set([unit.checksum])
with FULLTEXT_INDEX.source_searcher(
not settings.OFFLOAD_INDEXING) as searcher:
# Extract up to 10 terms from the source
terms = [
kw for kw, score in searcher.key_terms_from_text(
'source', unit.source, numterms=10)
if not kw in IGNORE_SIMILAR
]
cnt = len(terms)
# Try to find at least configured number of similar strings, remove up to 4 words
while len(ret) < settings.SIMILAR_MESSAGES and cnt > 0 and len(
terms) - cnt < 4:
for search in itertools.combinations(terms, cnt):
ret = ret.union(
self.search(' '.join(search), True, False, False,
True))
cnt -= 1
return self.filter(
translation__subproject__project=unit.translation.subproject.
project,
translation__language=unit.translation.language,
checksum__in=ret).exclude(target__in=['', unit.target])
def add_to_index(self, unit):
'''
Updates/Adds to all indices given unit.
'''
if settings.OFFLOAD_INDEXING:
from weblate.trans.models import IndexUpdate
IndexUpdate.objects.get_or_create(unit=unit)
return
writer_target = FULLTEXT_INDEX.target_writer(
unit.translation.language.code)
writer_source = FULLTEXT_INDEX.source_writer()
self.add_to_source_index(unit.checksum, unit.source, unit.context,
writer_source)
self.add_to_target_index(unit.checksum, unit.target, writer_target)
def handle(self, *args, **options):
languages = Language.objects.all()
# Optionally rebuild indices from scratch
if options['clean']:
create_source_index()
for lang in languages:
create_target_index(lang=lang.code)
units = self.get_units(*args, **options)
# Update source index
with FULLTEXT_INDEX.source_writer(buffered=False) as writer:
checksums = units.values('checksum', 'source', 'context')
for unit in checksums.iterator():
Unit.objects.add_to_source_index(
unit['checksum'],
unit['source'],
unit['context'],
writer
)
# Update per language indices
for lang in languages:
with FULLTEXT_INDEX.target_writer(lang=lang.code, buffered=False) as writer:
language_units = units.filter(
translation__language=lang
).exclude(
target=''
).values(
'checksum', 'target'
)
for unit in language_units.iterator():
Unit.objects.add_to_target_index(
unit['checksum'],
unit['target'],
writer
)
def search(self,
query,
source=True,
context=True,
translation=True,
checksums=False):
'''
Performs full text search on defined set of fields.
Returns queryset unless checksums is set.
'''
ret = set()
if source or context:
with FULLTEXT_INDEX.source_searcher(
not settings.OFFLOAD_INDEXING) as searcher:
if source:
ret = ret.union(
self.__search(searcher, 'source', SOURCE_SCHEMA,
query))
if context:
ret = ret.union(
self.__search(searcher, 'context', SOURCE_SCHEMA,
query))
if translation:
sample = self.all()[0]
with FULLTEXT_INDEX.target_searcher(
sample.translation.language.code,
not settings.OFFLOAD_INDEXING) as searcher:
ret = ret.union(
self.__search(searcher, 'target', TARGET_SCHEMA, query))
if checksums:
return ret
return self.filter(checksum__in=ret)
