def __init__(
self,
path,
name=None,
extensions=(".txt",),
texts_path=None,
metadata_path=None,
output_path=None
):
# Main Corpus Path
if type(path) is not str:
raise ValueError("Invalid path argument provided.")
# If we didn't get an absolute path, assume it's a path relative to Ity.corpus_root.
if not os.path.isabs(path):
path = os.path.join(Ity.corpus_root, path)
# This call to os.path.abspath(), among other things, removes trailing
# slashes from the path.
self.path = os.path.abspath(path)
# Okay, does the path actually exist?
if not os.path.exists(self.path):
raise IOError("Corpus at path '%s' does not exist." % self.path)
# Texts Path
self.texts_path = get_valid_path(
path=texts_path,
relative_path_base=self.path,
fallback_path=self.path
)
# It's NOT okay if this path doesn't exist.
if type(self.texts_path) is not str or not os.path.exists(self.texts_path):
raise ValueError("Path to texts ('%s') doesn't exist." % self.texts_path)
# Corpus Name
if name is None or type(name) is not str:
name = os.path.basename(self.path)
self.name = name
# Metadata Path
self.metadata_path = get_valid_path(
path=metadata_path,
relative_path_base=self.path,
fallback_path=os.path.join(Ity.metadata_root, self.name)
)
# Output Path
self.output_path = get_valid_path(
path=output_path,
relative_path_base=self.path,
fallback_path=os.path.join(Ity.output_root, self.name)
)
# Extensions
if extensions is None or type(extensions) is str or len(extensions) == 0:
raise ValueError("Invalid extensions argument provided.")
self.extensions = extensions
self._texts = None
self.metadata = {}
self.batch_format_data = {}
def __init__(self,
path,
name=None,
extensions=(".txt", ),
texts_path=None,
metadata_path=None,
output_path=None):
# Main Corpus Path
if type(path) is not str:
raise ValueError("Invalid path argument provided.")
# If we didn't get an absolute path, assume it's a path relative to Ity.corpus_root.
if not os.path.isabs(path):
path = os.path.join(Ity.corpus_root, path)
# This call to os.path.abspath(), among other things, removes trailing
# slashes from the path.
self.path = os.path.abspath(path)
# Okay, does the path actually exist?
if not os.path.exists(self.path):
raise IOError("Corpus at path '%s' does not exist." % self.path)
# Texts Path
self.texts_path = get_valid_path(path=texts_path,
relative_path_base=self.path,
fallback_path=self.path)
# It's NOT okay if this path doesn't exist.
if type(self.texts_path) is not str or not os.path.exists(
self.texts_path):
raise ValueError("Path to texts ('%s') doesn't exist." %
self.texts_path)
# Corpus Name
if name is None or type(name) is not str:
name = os.path.basename(self.path)
self.name = name
# Metadata Path
self.metadata_path = get_valid_path(path=metadata_path,
relative_path_base=self.path,
fallback_path=os.path.join(
Ity.metadata_root, self.name))
# Output Path
self.output_path = get_valid_path(path=output_path,
relative_path_base=self.path,
fallback_path=os.path.join(
Ity.output_root, self.name))
# Extensions
if extensions is None or type(extensions) is str or len(
extensions) == 0:
raise ValueError("Invalid extensions argument provided.")
self.extensions = extensions
self._texts = None
self.metadata = {}
self.batch_format_data = {}
def __init__(self, path, name=None, corpus=None, output_path=None):
# Text Path
if type(path) is not str:
raise ValueError("Invalid path argument provided.")
# Is the path absolute?
if not os.path.isabs(path):
# No? Can we figure out where it is based on the corpus argument?
if corpus is None and hasattr(corpus, "texts_path") and type(corpus.texts_path) is str:
path = os.path.join(corpus.texts_path, path)
else:
raise ValueError("Given a relative path to a text without a corpus argument.")
# This call to os.path.abspath(), among other things, removes trailing
# slashes from the path.
self.path = os.path.abspath(path)
# Okay, does the path actually exist?
if not os.path.exists(self.path):
raise ValueError("Text file at path '%s' does not exist." % self.path)
# Text Name
if name is None or type(name) is not str:
name = os.path.splitext(os.path.basename(self.path))[0]
self.name = name
# Text Corpus (may be None)
self.corpus = corpus
# Output Path
self.output_path = get_valid_path(path=output_path, fallback_path=os.path.join(Ity.output_root, self.name))
self.metadata = None
self._text_str = None
self.tokens = []
self.tag_data = {}
self.format_data = {}
def __init__(self, path, name=None, corpus=None, output_path=None):
# Text Path
if type(path) is not str:
raise ValueError("Invalid path argument provided.")
# Is the path absolute?
if not os.path.isabs(path):
# No? Can we figure out where it is based on the corpus argument?
if (corpus is None and hasattr(corpus, "texts_path")
and type(corpus.texts_path) is str):
path = os.path.join(corpus.texts_path, path)
else:
raise ValueError(
"Given a relative path to a text without a corpus argument."
)
# This call to os.path.abspath(), among other things, removes trailing
# slashes from the path.
self.path = os.path.abspath(path)
# Okay, does the path actually exist?
if not os.path.exists(self.path):
raise ValueError("Text file at path '%s' does not exist." %
self.path)
# Text Name
if name is None or type(name) is not str:
name = os.path.splitext(os.path.basename(self.path))[0]
self.name = name
# Text Corpus (may be None)
self.corpus = corpus
# Output Path
self.output_path = get_valid_path(path=output_path,
fallback_path=os.path.join(
Ity.output_root, self.name))
self.metadata = None
self._text_str = None
self.tokens = []
self.tag_data = {}
self.format_data = {}
