def _read_document(self):
"""Must set the self._content and self._metadata variables."""
with io.open(self._document_path, 'r', encoding='utf-8', errors='ignore') as f:
text = f.read()
metadata, content = basic_tools.seperate_metadata_and_content(text)
self._metadata = basic_tools.metadata_to_dict(metadata)
for f in self._filters:
content = f(content)
self._content = content
def __init__(self, dataset_directory, is_recursive=True):
"""
dataset_directory -- a relative or absolute file path to the \
directory that contains the documents directory and the dataset_metadata.txt file.
is_recursive -- find documents recursively in the documents directory, \
by default documents will be found recursively.
Document metadata types will be infered from the document metadata.
"""
# create commonly used directory and file paths
self._dataset_directory = dataset_directory
self._abs_dataset_directory = os.path.abspath(dataset_directory)
self._metadata_file = os.path.join(self._abs_dataset_directory,
'dataset_metadata.txt')
self._documents_directory = os.path.join(self._abs_dataset_directory,
'documents')
self.is_recursive = is_recursive
self._filters = []
self.metadata = {}
# load the dataset metadata
with io.open(self._metadata_file,
'r',
encoding='utf-8',
errors='ignore') as meta_file:
content = meta_file.read()
metadata, __ = basic_tools.seperate_metadata_and_content(content)
self.metadata = basic_tools.metadata_to_dict(metadata)
self.metadata_types = {}
basic_tools.collect_types(self.metadata_types, self.metadata)
if not 'readable_name' in self.metadata:
identifier = self._dataset_directory.replace('_', ' ').replace(
'/', ' ').title()
else:
identifier = self.metadata['readable_name']
self.name = basic_tools.remove_punctuation(identifier)
# find and sort all file paths
self._list_of_documents = basic_tools.get_all_files_from_directory(
self._documents_directory, self.is_recursive)
self._list_of_documents.sort()
# find any bad documents and find document metadata
self.document_metadata_types = {}
bad_doc_indices = []
for doc_index, doc in enumerate(self):
try:
basic_tools.collect_types(self.document_metadata_types,
doc.metadata)
except Exception as e:
print("Bad document: ", self._list_of_documents[doc_index])
bad_doc_indices.append(doc_index)
while len(bad_doc_indices) != 0:
remove_index = bad_doc_indices.pop()
del self._list_of_documents[remove_index]
def _read_document(self):
"""Must set the self._content and self._metadata variables."""
with io.open(self._document_path,
'r',
encoding='utf-8',
errors='ignore') as f:
text = f.read()
metadata, content = basic_tools.seperate_metadata_and_content(text)
self._metadata = basic_tools.metadata_to_dict(metadata)
for f in self._filters:
content = f(content)
self._content = content
def __init__(self, dataset_directory, is_recursive=True):
"""
dataset_directory -- a relative or absolute file path to the \
directory that contains the documents directory and the dataset_metadata.txt file.
is_recursive -- find documents recursively in the documents directory, \
by default documents will be found recursively.
Document metadata types will be infered from the document metadata.
"""
# create commonly used directory and file paths
self._dataset_directory = dataset_directory
self._abs_dataset_directory = os.path.abspath(dataset_directory)
self._metadata_file = os.path.join(self._abs_dataset_directory,
'dataset_metadata.txt')
self._documents_directory = os.path.join(self._abs_dataset_directory,
'documents')
self.is_recursive = is_recursive
self._filters = []
self.metadata = {}
# load the dataset metadata
with io.open(self._metadata_file, 'r', encoding='utf-8', errors='ignore') as meta_file:
content = meta_file.read()
metadata, __ = basic_tools.seperate_metadata_and_content(content)
self.metadata = basic_tools.metadata_to_dict(metadata)
self.metadata_types = {}
basic_tools.collect_types(self.metadata_types, self.metadata)
if not 'readable_name' in self.metadata:
identifier = self._dataset_directory.replace('_', ' ').replace('/', ' ').title()
else:
identifier = self.metadata['readable_name']
self.name = basic_tools.remove_punctuation(identifier)
# find and sort all file paths
self._list_of_documents = basic_tools.get_all_files_from_directory(self._documents_directory, self.is_recursive)
self._list_of_documents.sort()
# find any bad documents and find document metadata
self.document_metadata_types = {}
bad_doc_indices = []
for doc_index, doc in enumerate(self):
try:
basic_tools.collect_types(self.document_metadata_types, doc.metadata)
except Exception as e:
print("Bad document: ", self._list_of_documents[doc_index])
bad_doc_indices.append(doc_index)
while len(bad_doc_indices) != 0:
remove_index = bad_doc_indices.pop()
del self._list_of_documents[remove_index]
