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()
# Search in source or context
if source or context:
index = FULLTEXT_INDEX.source_searcher(not appsettings.OFFLOAD_INDEXING)
with index as searcher:
if source:
results = self.__search(searcher, "source", SOURCE_SCHEMA, query)
ret = ret.union(results)
if context:
results = self.__search(searcher, "context", SOURCE_SCHEMA, query)
ret = ret.union(results)
# Search in target
if translation:
sample = self.all()[0]
index = FULLTEXT_INDEX.target_searcher(sample.translation.language.code, not appsettings.OFFLOAD_INDEXING)
with index as searcher:
results = self.__search(searcher, "target", TARGET_SCHEMA, query)
ret = ret.union(results)
if checksums:
return ret
return self.filter(checksum__in=ret)
def fulltext(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()
# Search in source or context
if source or context:
index = FULLTEXT_INDEX.source_searcher(
not appsettings.OFFLOAD_INDEXING
)
with index as searcher:
if source:
results = self.__search(
searcher,
'source',
SOURCE_SCHEMA,
query
)
ret = ret.union(results)
if context:
results = self.__search(
searcher,
'context',
SOURCE_SCHEMA,
query
)
ret = ret.union(results)
# Search in target
if translation:
sample = self.all()[0]
index = FULLTEXT_INDEX.target_searcher(
sample.translation.language.code,
not appsettings.OFFLOAD_INDEXING
)
with index as searcher:
results = self.__search(
searcher,
'target',
TARGET_SCHEMA,
query
)
ret = ret.union(results)
if checksums:
return ret
return self.filter(checksum__in=ret)
def fulltext(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()
# Search in source or context
if source or context:
index = FULLTEXT_INDEX.source_searcher()
with index as searcher:
if source:
results = self.__search(
searcher,
'source',
SOURCE_SCHEMA,
query
)
ret = ret.union(results)
if context:
results = self.__search(
searcher,
'context',
SOURCE_SCHEMA,
query
)
ret = ret.union(results)
# Search in target
if translation:
sample = self.all()[0]
index = FULLTEXT_INDEX.target_searcher(
sample.translation.language.code,
)
with index as searcher:
results = self.__search(
searcher,
'target',
TARGET_SCHEMA,
query
)
ret = ret.union(results)
if checksums:
return ret
return self.filter(checksum__in=ret)
def more_like_this(self, unit, top=5):
'''
Finds closely similar units.
'''
index = FULLTEXT_INDEX.source_searcher()
source_string = unit.get_source_plurals()[0]
parser = qparser.QueryParser('source', SOURCE_SCHEMA)
parsed = parser.parse(source_string)
checksums = set()
with index as searcher:
# Search for same string
results = searcher.search(parsed)
if len(results) == 0:
return self.none()
first_hit = results[0]
# Find similar results to first one
more_results = first_hit.more_like_this(
'source',
source_string,
top
)
# Include all more like this results
for result in more_results:
checksums.add(result['checksum'])
# Remove all original matches
for result in results:
checksums.discard(result['checksum'])
return self.filter(
checksum__in=checksums,
translation__language=unit.translation.language,
translated=True
).exclude(
pk=unit.id
)
def add_to_index(self, unit, source=True):
"""
Updates/Adds to all indices given unit.
"""
if appsettings.OFFLOAD_INDEXING:
from trans.models import IndexUpdate
IndexUpdate.objects.get_or_create(unit=unit, source=source)
return
writer_target = FULLTEXT_INDEX.target_writer(unit.translation.language.code)
writer_source = FULLTEXT_INDEX.source_writer()
if source:
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, source=True):
'''
Updates/Adds to all indices given unit.
'''
if appsettings.OFFLOAD_INDEXING:
from trans.models.unitdata import IndexUpdate
IndexUpdate.objects.get_or_create(unit=unit, source=source)
return
writer_target = FULLTEXT_INDEX.target_writer(
unit.translation.language.code
)
writer_source = FULLTEXT_INDEX.source_writer()
if source:
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 same_source(self, unit):
'''
Finds units with same source.
'''
index = FULLTEXT_INDEX.source_searcher()
source_string = unit.get_source_plurals()[0]
parser = qparser.QueryParser('source', SOURCE_SCHEMA)
parsed = parser.parse(source_string)
checksums = set()
with index as searcher:
# Search for same string
results = searcher.search(parsed)
for result in results:
checksums.add(result['checksum'])
return self.filter(
checksum__in=checksums,
translation__language=unit.translation.language,
translated=True
).exclude(
pk=unit.id
)
def similar(self, unit):
"""
Finds similar units to current unit.
"""
ret = set([unit.checksum])
index = FULLTEXT_INDEX.source_searcher(not appsettings.OFFLOAD_INDEXING)
with index as searcher:
# Extract up to 10 terms from the source
key_terms = searcher.key_terms_from_text("source", unit.source, numterms=10)
terms = [kw[0] for kw in key_terms 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
project = unit.translation.subproject.project
return self.filter(
translation__subproject__project=project, translation__language=unit.translation.language, checksum__in=ret
).exclude(target__in=["", unit.target])
def more_like_this(self, unit):
'''
Finds closely similar units.
'''
index = FULLTEXT_INDEX.source_searcher(
not appsettings.OFFLOAD_INDEXING
)
source_string = unit.get_source_plurals()[0]
parser = qparser.QueryParser('source', SOURCE_SCHEMA)
parsed = parser.parse(source_string)
checksums = set()
with index as searcher:
# Search for same string
results = searcher.search(parsed)
if len(results) == 0:
return self.none()
first_hit = results[0]
# Find similar results to first one
more_results = first_hit.more_like_this(
'source',
source_string,
500
)
# Include all more like this results
for result in more_results:
checksums.add(result['checksum'])
# Remove all original matches
for result in results:
checksums.discard(result['checksum'])
return self.filter(
checksum__in=checksums,
translation__language=unit.translation.language,
translated=True
).exclude(
pk=unit.id
)
def same_source(self, unit):
'''
Finds units with same source.
'''
index = FULLTEXT_INDEX.source_searcher(
not appsettings.OFFLOAD_INDEXING
)
source_string = unit.get_source_plurals()[0]
parser = qparser.QueryParser('source', SOURCE_SCHEMA)
parsed = parser.parse(source_string)
checksums = set()
with index as searcher:
# Search for same string
results = searcher.search(parsed)
for result in results:
checksums.add(result['checksum'])
return self.filter(
checksum__in=checksums,
translation__language=unit.translation.language,
translated=True
).exclude(
pk=unit.id
)
