def __init__(self, storage, schema, segment):
self.storage = storage
self.schema = schema
self.segment = segment
if hasattr(self.segment, "uuid"):
self.uuid_string = str(self.segment.uuid)
else:
import uuid
self.uuid_string = str(uuid.uuid4())
# Term index
tf = storage.open_file(segment.termsindex_filename)
self.termsindex = TermIndexReader(tf)
# Term vector index, and vector postings: lazy load
self.vectorindex = None
self.vpostfile = None
# Stored fields file
sf = storage.open_file(segment.storedfields_filename, mapped=False)
self.storedfields = StoredFieldReader(sf)
# Field length file
self.fieldlengths = None
if self.schema.has_scorable_fields():
flf = storage.open_file(segment.fieldlengths_filename)
self.fieldlengths = LengthReader(flf, segment.doc_count_all())
# Copy info from underlying segment
self._has_deletions = segment.has_deletions()
self._doc_count = segment.doc_count()
# Postings file
self.postfile = self.storage.open_file(segment.termposts_filename,
mapped=False)
# Dawg file
self.dawg = None
if any(field.spelling for field in self.schema):
fname = segment.dawg_filename
if self.storage.file_exists(fname):
dawgfile = self.storage.open_file(fname, mapped=False)
self.dawg = DiskNode.load(dawgfile, expand=False)
self.dc = segment.doc_count_all()
assert self.dc == self.storedfields.length
self.set_caching_policy()
self.is_closed = False
self._sync_lock = Lock()
def test_stored_fields():
with TempStorage("storedfields") as st:
sf = st.create_file("test.sf")
sfw = StoredFieldWriter(sf, ["a", "b"])
sfw.append({"a": "hello", "b": "there"})
sfw.append({"a": "one", "b": "two"})
sfw.append({"a": "alfa", "b": "bravo"})
sfw.close()
sf = st.open_file("test.sf")
sfr = StoredFieldReader(sf)
assert_equal(sfr[0], {"a": "hello", "b": "there"})
assert_equal(sfr[2], {"a": "alfa", "b": "bravo"})
assert_equal(sfr[1], {"a": "one", "b": "two"})
sfr.close()
def __init__(self, storage, schema, segment):
self.storage = storage
self.schema = schema
self.segment = segment
if hasattr(self.segment, "uuid"):
self.uuid_string = str(self.segment.uuid)
else:
import uuid
self.uuid_string = str(uuid.uuid4())
# Term index
tf = storage.open_file(segment.termsindex_filename)
self.termsindex = TermIndexReader(tf)
# Term postings file, vector index, and vector postings: lazy load
self.postfile = None
self.vectorindex = None
self.vpostfile = None
# Stored fields file
sf = storage.open_file(segment.storedfields_filename, mapped=False)
self.storedfields = StoredFieldReader(sf)
# Field length file
self.fieldlengths = None
if self.schema.has_scorable_fields():
flf = storage.open_file(segment.fieldlengths_filename)
self.fieldlengths = LengthReader(flf, segment.doc_count_all())
# Copy methods from underlying segment
self.has_deletions = segment.has_deletions
self.is_deleted = segment.is_deleted
self.doc_count = segment.doc_count
# Postings file
self.postfile = self.storage.open_file(segment.termposts_filename,
mapped=False)
self.dc = segment.doc_count_all()
assert self.dc == self.storedfields.length
self.set_caching_policy()
self.is_closed = False
self._sync_lock = Lock()
