def gettermmatcher(self):
"""returns the terminology matcher"""
if self.is_terminology_project:
if self.non_db_state.termmatcher is None:
self.require_units()
newmtime = self.get_mtime()
if newmtime != self.non_db_state.termmatchermtime:
self.non_db_state.termmatcher = match.terminologymatcher(self.stores.iterator())
self.non_db_state.termmatchermtime = newmtime
else:
try:
termfilename = "pootle-terminology." + self.project.localfiletype
store = self.stores.get(name=termfilename)
newmtime = store.get_mtime()
if newmtime != self.non_db_state.termmatchermtime:
self.non_db_state.termmatcher = match.terminologymatcher(store)
self.non_db_state.termmatchermtime = newmtime
except Store.DoesNotExist:
try:
termproject = TranslationProject.objects.get(language=self.language_id, project__code="terminology")
self.non_db_state.termmatcher = termproject.gettermmatcher()
self.non_db_state.termmatchermtime = termproject.non_db_state.termmatchermtime
except TranslationProject.DoesNotExist:
pass
return self.non_db_state.termmatcher
def test_hyphen_mismatch(self):
"""Tests that we can match with some spacing mismatch"""
csvfile = self.buildcsv(["pre-order"])
matcher = match.terminologymatcher(csvfile)
candidates = self.candidatestrings(matcher.matches("You can preorder"))
assert candidates == ["preorder"]
candidates = self.candidatestrings(matcher.matches("You can pre order"))
assert candidates == ["pre order"]
csvfile = self.buildcsv(["pre order"])
matcher = match.terminologymatcher(csvfile)
candidates = self.candidatestrings(matcher.matches("You can preorder"))
assert candidates == ["preorder"]
candidates = self.candidatestrings(matcher.matches("You can pre order"))
assert candidates == ["pre order"]
def test_hyphen_mismatch(self):
"""Tests that we can match with some spacing mismatch"""
csvfile = self.buildcsv(["pre-order"])
matcher = match.terminologymatcher(csvfile)
candidates = self.candidatestrings(matcher.matches("You can preorder"))
assert candidates == ["preorder"]
candidates = self.candidatestrings(matcher.matches("You can pre order"))
assert candidates == ["pre order"]
csvfile = self.buildcsv(["pre order"])
matcher = match.terminologymatcher(csvfile)
candidates = self.candidatestrings(matcher.matches("You can preorder"))
assert candidates == ["preorder"]
candidates = self.candidatestrings(matcher.matches("You can pre order"))
assert candidates == ["pre order"]
def test_terminology(self):
csvfile = self.buildcsv(["file", "computer", "directory"])
matcher = match.terminologymatcher(csvfile)
candidates = self.candidatestrings(
matcher.matches("Copy the files from your computer"))
candidates.sort()
assert candidates == ["computer", "file"]
def test_space_mismatch(self):
"""Tests that we can match with some spacing mismatch"""
csvfile = self.buildcsv(["down time"])
matcher = match.terminologymatcher(csvfile)
candidates = self.candidatestrings(
matcher.matches("%d minutes downtime"))
assert candidates == ["downtime"]
def test_past_tences(self):
"""Tests matching of some past tenses"""
csvfile = self.buildcsv(["submit", "certify"])
matcher = match.terminologymatcher(csvfile)
candidates = self.candidatestrings(matcher.matches("The bug was submitted"))
assert candidates == ["submit"]
candidates = self.candidatestrings(matcher.matches("The site is certified"))
def gettermmatcher(self):
"""returns the terminology matcher"""
terminology_stores = Store.objects.none()
mtime = None
if self.is_terminology_project:
terminology_stores = self.stores.all()
mtime = self.get_mtime()
else:
# Get global terminology first
try:
termproject = TranslationProject.objects.get(
language=self.language_id, project__code='terminology')
mtime = termproject.get_mtime()
terminology_stores = termproject.stores.all()
except TranslationProject.DoesNotExist:
pass
local_terminology = self.stores.filter(
name__startswith='pootle-terminology')
for store in local_terminology.iterator():
mtime = max(mtime, store.get_mtime())
terminology_stores = terminology_stores | local_terminology
if mtime is None:
return
if mtime != self.non_db_state.termmatchermtime:
self.non_db_state.termmatcher = match.terminologymatcher(
terminology_stores.iterator())
self.non_db_state.termmatchermtime = mtime
return self.non_db_state.termmatcher
def test_past_tences(self):
"""Tests matching of some past tenses"""
csvfile = self.buildcsv(["submit", "certify"])
matcher = match.terminologymatcher(csvfile)
candidates = self.candidatestrings(matcher.matches("The bug was submitted"))
assert candidates == ["submit"]
candidates = self.candidatestrings(matcher.matches("The site is certified"))
def gettermmatcher(self):
"""returns the terminology matcher"""
terminology_stores = Store.objects.none()
mtime = None
if self.is_terminology_project:
terminology_stores = self.stores.all()
mtime = self.get_mtime()
else:
# Get global terminology first
try:
termproject = TranslationProject.objects.get(language=self.language_id, project__code="terminology")
mtime = termproject.get_mtime()
terminology_stores = termproject.stores.all()
except TranslationProject.DoesNotExist:
pass
local_terminology = self.stores.filter(name__startswith="pootle-terminology")
for store in local_terminology.iterator():
mtime = max(mtime, store.get_mtime())
terminology_stores = terminology_stores | local_terminology
if mtime is None:
return
if mtime != self.non_db_state.termmatchermtime:
self.non_db_state.termmatcher = match.terminologymatcher(terminology_stores.iterator())
self.non_db_state.termmatchermtime = mtime
return self.non_db_state.termmatcher
def test_brackets(self):
"""Tests that brackets at the end of a term are ignored"""
csvfile = self.buildcsv(["file (noun)", "ISP (Internet Service Provider)"])
matcher = match.terminologymatcher(csvfile)
candidates = self.candidatestrings(matcher.matches("Open File"))
assert candidates == ["file"]
candidates = self.candidatestrings(matcher.matches("Contact your ISP"))
# we lowercase everything - that is why we get it back differently.
# we don't change the target text, though
assert candidates == ["isp"]
def test_brackets(self):
"""Tests that brackets at the end of a term are ignored"""
csvfile = self.buildcsv(["file (noun)", "ISP (Internet Service Provider)"])
matcher = match.terminologymatcher(csvfile)
candidates = self.candidatestrings(matcher.matches("Open File"))
assert candidates == ["file"]
candidates = self.candidatestrings(matcher.matches("Contact your ISP"))
# we lowercase everything - that is why we get it back differerntly.
# we don't change the target text, though
assert candidates == ["isp"]
def _init_matcher(self):
"""
Initialize the matcher to be used by the C{TerminologyPlaceable} parser.
"""
if self.matcher in TerminologyPlaceable.matchers:
TerminologyPlaceable.matchers.remove(self.matcher)
self.store = TranslationStore()
self.store.makeindex()
self.matcher = terminologymatcher(self.store)
TerminologyPlaceable.matchers.append(self.matcher)
def _init_matcher(self):
"""
Initialize the matcher to be used by the C{TerminologyPlaceable} parser.
"""
if self.matcher in TerminologyPlaceable.matchers:
TerminologyPlaceable.matchers.remove(self.matcher)
self.store = TranslationStore()
self.store.makeindex()
self.matcher = terminologymatcher(self.store)
TerminologyPlaceable.matchers.append(self.matcher)
def load_files(self):
if self.matcher in TerminologyPlaceable.matchers:
TerminologyPlaceable.matchers.remove(self.matcher)
self.stores = []
for filename in self.config['files']:
if not filename:
continue
if not os.path.isfile(filename):
logging.debug('Not a file: "%s"' % (filename))
self.stores.append(factory.getobject(filename))
self.matcher = terminologymatcher(self.stores)
TerminologyPlaceable.matchers.append(self.matcher)
def load_files(self):
if self.matcher in TerminologyPlaceable.matchers:
TerminologyPlaceable.matchers.remove(self.matcher)
self.stores = []
for filename in self.config['files']:
if not filename:
continue
if not os.path.isfile(filename):
logging.debug('Not a file: "%s"' % (filename))
self.stores.append(factory.getobject(filename))
self.matcher = terminologymatcher(self.stores)
TerminologyPlaceable.matchers.append(self.matcher)
def _init_matcher(self):
"""
Initialize the matcher to be used by the C{TerminologyPlaceable} parser.
"""
if self.matcher in TerminologyPlaceable.matchers:
TerminologyPlaceable.matchers.remove(self.matcher)
from translate.storage.base import TranslationStore
self.store = TranslationStore()
self.store.makeindex()
from translate.search.match import terminologymatcher
self.matcher = terminologymatcher(self.store)
TerminologyPlaceable.matchers.append(self.matcher)
def _init_matcher(self):
"""
Initialize the matcher to be used by the C{TerminologyPlaceable} parser.
"""
if self.matcher in TerminologyPlaceable.matchers:
TerminologyPlaceable.matchers.remove(self.matcher)
from translate.storage.base import TranslationStore
self.store = TranslationStore()
self.store.makeindex()
from translate.search.match import terminologymatcher
self.matcher = terminologymatcher(self.store)
TerminologyPlaceable.matchers.append(self.matcher)
def init_matcher(self, filename=''):
"""
Initialize the matcher to be used by the C{TerminologyPlaceable} parser.
"""
if self.matcher in TerminologyPlaceable.matchers:
TerminologyPlaceable.matchers.remove(self.matcher)
if os.path.isfile(filename):
logging.debug('Loading terminology from %s' % (filename))
self.store = factory.getobject(filename)
else:
logging.debug('Creating empty terminology store')
self.store = TranslationStore()
self.store.makeindex()
self.matcher = terminologymatcher(self.store)
TerminologyPlaceable.matchers.append(self.matcher)
def gettermmatcher(self):
"""Returns the terminology matcher."""
terminology_stores = Store.objects.none()
mtime = None
if not self.is_terminology_project:
# Get global terminology first
try:
termproject = TranslationProject.objects \
.get_terminology_project(self.language_id)
mtime = termproject.get_mtime()
terminology_stores = termproject.stores.all()
except TranslationProject.DoesNotExist:
pass
local_terminology = self.stores.filter(
name__startswith='pootle-terminology')
for store in local_terminology.iterator():
if mtime is None:
mtime = store.get_mtime()
else:
mtime = max(mtime, store.get_mtime())
terminology_stores = terminology_stores | local_terminology
if mtime is None:
return
if mtime != self.non_db_state.termmatchermtime:
from translate.search import match
self.non_db_state.termmatcher = match.terminologymatcher(
terminology_stores.iterator(),
)
self.non_db_state.termmatchermtime = mtime
return self.non_db_state.termmatcher
def test_terminology(self):
csvfile = self.buildcsv(["file", "computer", "directory"])
matcher = match.terminologymatcher(csvfile)
candidates = self.candidatestrings(matcher.matches("Copy the files from your computer"))
candidates.sort()
assert candidates == ["computer", "file"]
def test_space_mismatch(self):
"""Tests that we can match with some spacing mismatch"""
csvfile = self.buildcsv(["down time"])
matcher = match.terminologymatcher(csvfile)
candidates = self.candidatestrings(matcher.matches("%d minutes downtime"))
assert candidates == ["downtime"]
def setup_method(self, method):
self.term_po = pofile(BytesIO(self.TERMINOLOGY.encode("utf-8")))
self.matcher = terminologymatcher(self.term_po)
self.test_string = u"<b>Inpüt</b> file name thingy."
def setup_method(self, method):
self.term_po = pofile(BytesIO(self.TERMINOLOGY.encode('utf-8')))
self.matcher = terminologymatcher(self.term_po)
self.test_string = u'<b>Inpüt</b> file name thingy.'
def setup_method(self, method):
self.term_po = pofile(StringIO(self.TERMINOLOGY))
self.matcher = terminologymatcher(self.term_po)
self.test_string = u'<b>Inpüt</b> file name thingy.'