def test_get_filelist_10000_files_in_2_layers_of_subfolders_99_files_each(
self):
"""List the files of a directory with 10000 files inside it, spread across 2 layer of subfolders of 99 files each.
Expected output: list with 10000 strings, equal to the filepaths of the temporary files created
"""
files_created = []
# Python 2.x
if sys.version_info[0] < 3:
# Work in a temporary directory
indir = tempfile.mkdtemp()
# Create ten temporary subdirectories inside the temporary directory
for _ in range(10):
subdir = tempfile.mkdtemp(dir=indir)
# Create ten temporary files inside each temporary subdirectory
for _ in range(10):
file_handler = tempfile.NamedTemporaryFile(dir=subdir,
delete=False)
files_created.append(file_handler.name)
# Create ten temporary subsubdirectories
for _ in range(10):
subsubdir = tempfile.mkdtemp(dir=subdir)
# Create 99 temporary files inside each temporary subsubdirectory
for _ in range(99):
file_handler = tempfile.NamedTemporaryFile(
dir=subsubdir, delete=False)
files_created.append(file_handler.name)
filelist = fs.get_filelist(indir)
# Remove the temporary files manually as a precaution
for filepath in files_created:
os.remove(filepath)
# Remove the temporary directory (mdktemp must be manually deleted)
# shutil.rmtree is used instead of os.remove to avoid OSError
shutil.rmtree(indir, ignore_errors=True)
# Python 3.x
else:
# Work in a temporary directory
with tempfile.TemporaryDirectory() as indir:
# Create ten temporary subdirectories inside the temporary directory
for _ in range(10):
subdir = tempfile.mkdtemp(dir=indir)
# Create ten temporary files inside each temporary subdirectory
for _ in range(10):
file_handler = tempfile.NamedTemporaryFile(
dir=subdir, delete=False)
files_created.append(file_handler.name)
# Create ten temporary subsubdirectories
for _ in range(10):
subsubdir = tempfile.mkdtemp(dir=subdir)
# Create 99 temporary files inside each temporary subsubdirectory
for _ in range(99):
file_handler = tempfile.NamedTemporaryFile(
dir=subsubdir, delete=False)
files_created.append(file_handler.name)
filelist = fs.get_filelist(indir)
# Remove the temporary file manually as a precaution
# (the with statement automatically deletes a TemporaryDirectory type of folder)
for filepath in files_created:
os.remove(filepath)
# Typecast both lists to sets to make an unordered comparison
self.assertEqual(set(files_created), set(filelist))
def test_get_filelist_1_file_in_folder(self):
"""List the files of a directory with only one file inside it.
Expected output: list with one string, equal to the filepath of the temporary file created
"""
files_created = []
# Python 2.x
if sys.version_info[0] < 3:
# Work in a temporary directory
indir = tempfile.mkdtemp()
# delete is set to False to avoid OSError at the end of the test
file1 = tempfile.NamedTemporaryFile(dir=indir, delete=False)
files_created.append(file1.name)
filelist = fs.get_filelist(indir)
# Remove the temporary file manually as a precaution
os.remove(file1.name)
# Remove the temporary directory (mdktemp must be manually deleted)
# shutil.rmtree is used instead of os.remove to avoid OSError
shutil.rmtree(indir, ignore_errors=True)
# Python 3.x
else:
# Work in a temporary directory
with tempfile.TemporaryDirectory() as indir:
# delete is set to False to avoid FileNotFoundError at the end of the test
file1 = tempfile.NamedTemporaryFile(dir=indir, delete=False)
files_created.append(file1.name)
filelist = fs.get_filelist(indir)
# Remove the temporary file manually as a precaution
# (the with statement automatically deletes the folder)
os.remove(file1.name)
self.assertEqual(files_created, filelist)
def test_get_filelist_empty_folder(self):
"""List the files of an empty directory.
It is necessary to make a temporary directory inside the folder obtained via tempfile.gettempdir(),
otherwise files that are already in the temp folder (created by other programs in the system) will pollute
the test and fill the list.
Expected output: empty list []
"""
# Python 2.x
if sys.version_info[0] < 3:
# Work in a temporary directory
indir = tempfile.mkdtemp()
filelist = fs.get_filelist(indir)
# Remove the temporary directory (mdktemp must be manually deleted)
# shutil.rmtree is used instead of os.remove to avoid OSError
shutil.rmtree(indir, ignore_errors=True)
# Python 3.x
else:
# Work in a temporary directory
with tempfile.TemporaryDirectory() as indir:
filelist = fs.get_filelist(indir)
# Empty lists evaluate to false
self.assertFalse(filelist)
def test_get_filelist_1000_files_in_folder(self):
"""List the files of a directory with 1000 files inside it.
Expected output: list with 1000 strings, equal to the filepaths of the temporary files created
"""
files_created = []
# Python 2.x
if sys.version_info[0] < 3:
# Work in a temporary directory
indir = tempfile.mkdtemp()
for _ in range(1000):
# delete is set to False to avoid OSError at the end of the test
file_handler = tempfile.NamedTemporaryFile(dir=indir,
delete=False)
files_created.append(file_handler.name)
filelist = fs.get_filelist(indir)
# Remove the temporary files manually as a precaution
for filepath in files_created:
os.remove(filepath)
# Remove the temporary directory (mdktemp must be manually deleted)
# shutil.rmtree is used instead of os.remove to avoid OSError
shutil.rmtree(indir, ignore_errors=True)
# Python 3.x
else:
# Work in a temporary directory
with tempfile.TemporaryDirectory() as indir:
for _ in range(1000):
# delete is set to False to avoid OSError at the end of the test
file_handler = tempfile.NamedTemporaryFile(dir=indir,
delete=False)
files_created.append(file_handler.name)
filelist = fs.get_filelist(indir)
# Remove the temporary file manually as a precaution
# (the with statement automatically deletes the folder)
for filepath in files_created:
os.remove(filepath)
# Typecast both lists to sets to make an unordered comparison
self.assertEqual(set(files_created), set(filelist))
def main():
# Get CLI arguments
args = cli.parse_arguments()
# If input is a file
if os.path.isfile(args.input):
# Set IO variables
infile = args.input
indir = os.path.dirname(infile)
outdir = fs.set_outdir(args.output, indir)
# Extract text
text = txt.get_text(infile, disable_no_ext_prompt=args.noprompt)
# If text has been extracted successfully (and infile was not empty)
if text:
outfile = fs.compose_unique_filepath(infile, outdir)
try:
fs.write_str_to_file(text, outfile)
except OSError as e:
print(e)
# If input is a folder
elif os.path.isdir(args.input):
# Set IO variables
indir = args.input
outdir = fs.set_outdir(args.output, indir, input_isdir=True)
filelist = fs.get_filelist(indir)
for infile in filelist:
text = txt.get_text(infile, disable_no_ext_prompt=args.noprompt)
if text:
outfile = fs.compose_unique_filepath(infile, outdir)
try:
fs.write_str_to_file(text, outfile)
except OSError as e:
print(e)
else:
exit("Error: input must be an existing file or directory")
