def get_file(self, key, hashsum=True):
"""
@param key: Specifies which data to look at, e. g. "NOUNS"
(str).
@param hashsum: If True (default) return not only file's
contents, but also hashsum, otherwise return
only file content.
@return: Hashsum (str) of this document's data in some context
specified by param key. If no file available returns
None values.
"""
sha512sum = None
content = None
folder = self._get_folder_by_key(key)
try:
f = open(folder + self[self.DOC_ID], "r", get_def_enc())
content = f.readlines()
sha512sum = sha512("".join(content). \
encode(get_def_enc())).hexdigest()
f.close()
except IOError:
pass
if hashsum == False:
return content
# Otherwise: Return a tuple
return content, sha512sum
def write_file(self, key, hashsum=True):
"""
Writes a file (name: document name) to a specified folder.
@param key: Specifies which data to write, based on the
key of the data stored in in this object,
e. g. "WORDS" or "NOUNS".
@param hashsum: Defaults to True and is used to write an
hashsum of the file to an hashfile.
"""
folder = self._get_folder_by_key(key)
doc_id = self[self.DOC_ID]
doc_as_str = "\n".join(self[key]).encode(get_def_enc())
sha512sum = ""
if (hashsum == True):
hashdict = HashDict()
sha512sum = sha512(doc_as_str).hexdigest()
hashdict[folder + doc_id] = sha512sum
hashdict.save()
if not exists(folder):
print "Folder " + folder + " doesn't exist."
try:
makedirs(folder)
print "Folder " + folder + " created."
except Exception, e:
print str(e)
def write_symbols(self):
"""
Writes a file with all the symbols (i. e. this class's keys)
in order.
"""
f = open(get_symbols_file(), "w", encoding=get_def_enc())
for symbol in sorted(self.keys(), reverse=True):
f.write(symbol)
f.close()
def write_content(self, content_format="line", content_type="raw"):
"""
Write files in different possible formats and content types to
a folder on disk.
@param content_format: If "line" writes content unit line per
line.
@param content_type: If "raw", write tokens as is; other values
like "words" may (become) possible.
"""
d = get_mailfolder(content_format=content_format)
if not exists(d):
print "Folder " + d + " not availabe. Create it."
makedirs(d)
# For now only "raw" content_type exists, and "line"
# content_format
f = open(d + self[self.DOC_ID], "w", get_def_enc())
for t in self.get_tokens():
f.write(t + "\n")
f.close()
def _create_symbols_dict(self):
"""
Opens a file (for now) with all text available and stores its
symbols as keys (for each of whom a determined attribute is stored
as value). Meant as internal method.
"""
f = open(get_raw_file(), "r", encoding=get_def_enc())
# Get unique symbols first
symset = set()
for sym in f.read():
symset.add(sym)
# For each symbol (key) determine an attribute to store in this
# instance
for sym in symset:
symobj = self._classify_symbol(sym)
self.__setitem__(sym, symobj.get(sym))
f.close()
