def main(argv=None):
args = docopt(__doc__,
argv=argv,
version='OpenAccess_EPUB v.' + __version__,
options_first=True)
if args['--epub3']:
epub_version = 3
elif args['--epub2']:
epub_version = 2
else:
epub_version = None
#Basic logging configuration
oae_logging.config_logging(args['--no-log-file'],
args['--log-to'],
args['--log-level'],
args['--silent'],
args['--verbosity'])
#Get a logger, the 'openaccess_epub' logger was set up above
command_log = logging.getLogger('openaccess_epub.commands.convert')
#Load the config module, we do this after logging configuration
config = openaccess_epub.utils.load_config_module()
current_dir = os.getcwd()
#Our basic flow is to iterate over the args['INPUT'] list
for inpt in args['INPUT']:
#We have to temporarily re-base our log while input utils do some work
if not args['--no-log-file'] and not args['--log-to']:
oae_logging.replace_filehandler(logname='openaccess_epub',
new_file='temp.log',
level=args['--log-level'],
frmt=oae_logging.STANDARD_FORMAT)
command_log.info('Processing input: {0}'.format(inpt))
#First we need to know the name of the file and where it is
if inpt.lower().endswith('.xml'): # This is direct XML file
root_name = openaccess_epub.utils.file_root_name(inpt)
abs_input_path = openaccess_epub.utils.get_absolute_path(inpt)
elif inpt.lower().startswith('doi:'): # This is a DOI
root_name = input_utils.doi_input(inpt)
abs_input_path = os.path.join(current_dir, root_name + '.xml')
elif any(inpt.lower().startswith(i) for i in ['http:', 'https:']):
root_name = input_utils.url_input(inpt)
abs_input_path = os.path.join(current_dir, root_name + '.xml')
else:
sys.exit('{0} not recognized as XML, DOI, or URL'.format(inpt))
if not args['--no-log-file'] and not args['--log-to']:
log_name = root_name + '.log'
log_path = os.path.join(os.path.dirname(abs_input_path), log_name)
#Re-base the log file to the new file location
oae_logging.replace_filehandler(logname='openaccess_epub',
new_file=log_path,
level=args['--log-level'],
frmt=oae_logging.STANDARD_FORMAT)
#Now we move over to the new log file
shutil.copy2('temp.log', log_path)
os.remove('temp.log')
#Now that we should be done configuring logging, let's parse the article
parsed_article = Article(abs_input_path,
validation=not args['--no-validate'])
if parsed_article.publisher is None:
command_log.critical('Publisher support was not established, aborting')
sys.exit(1)
#Get the output directory
if args['--output'] is not None:
output_directory = openaccess_epub.utils.get_absolute_path(args['--output'])
else:
if os.path.isabs(config.default_output): # Absolute remains so
output_directory = config.default_output
else: # Else rendered relative to input
abs_dirname = os.path.dirname(abs_input_path)
output_directory = os.path.normpath(os.path.join(abs_dirname, config.default_output))
#The root name must be added on for output
output_directory = os.path.join(output_directory, root_name)
#Make the call to make_EPUB
success = make_EPUB(parsed_article,
output_directory,
abs_input_path,
args['--images'],
config_module=config,
epub_version=epub_version)
#Cleanup removes the produced output directory, keeps the EPUB
if not args['--no-cleanup']:
command_log.info('Removing {0}'.format(output_directory))
shutil.rmtree(output_directory)
#Running epubcheck on the output verifies the validity of the EPUB,
#requires a local installation of java and epubcheck.
if not args['--no-epubcheck'] and success:
epub_name = '{0}.epub'.format(output_directory)
openaccess_epub.utils.epubcheck(epub_name, config)
def get_output_directory(args):
"""
Determination of the directory for output placement involves possibilities
for explicit user instruction (absolute path or relative to execution) and
implicit default configuration (absolute path or relative to input) from
the system global configuration file. This function is responsible for
reliably returning the appropriate output directory which will contain any
log(s), ePub(s), and unzipped output of OpenAccess_EPUB.
It utilizes the parsed args, passed as an object, and is self-sufficient in
accessing the config file.
All paths returned by this function are absolute.
"""
#Import the global config file as a module
import imp
config_path = os.path.join(cache_location(), 'config.py')
try:
config = imp.load_source('config', config_path)
except IOError:
print(('Could not find {0}, please run oae-quickstart'.format(
config_path)))
sys.exit()
#args.output is the explicit user instruction, None if unspecified
if args.output:
#args.output may be an absolute path
if os.path.isabs(args.output):
return args.output # return as is
#or args.output may be a relative path, relative to cwd
else:
return evaluate_relative_path(relative=args.output)
#config.default_output for default behavior without explicit instruction
else:
#config.default_output may be an absolute_path
if os.path.isabs(config.default_output):
return config.default_output
#or config.default_output may be a relative path, relative to input
else:
if args.input: # The case of single input
if 'http://www' in args.input:
#Fetched from internet by URL
raw_name = url_input(args.input, download=False)
abs_input_path = os.path.join(os.getcwd(),
raw_name + '.xml')
elif args.input[:4] == 'doi:':
#Fetched from internet by DOI
raw_name = doi_input(args.input, download=False)
abs_input_path = os.path.join(os.getcwd(),
raw_name + '.xml')
else:
#Local option, could be anywhere
abs_input_path = get_absolute_path(args.input)
abs_input_parent = os.path.split(abs_input_path)[0]
return evaluate_relative_path(abs_input_parent,
config.default_output)
elif args.batch: # The case of Batch Mode
#Batch should only work on a supplied directory
abs_batch_path = get_absolute_path(args.batch)
return abs_batch_path
elif args.zip:
#Zip is a local-only option, behaves just like local xml
abs_input_path = get_absolute_path(args.zip)
abs_input_parent = os.path.split(abs_input_path)[0]
return evaluate_relative_path(abs_input_parent,
config.default_output)
elif args.collection:
return os.getcwd()
else: # Un-handled or currently unsupported options
print('The output location could not be determined...')
sys.exit()
def get_output_directory(args):
"""
Determination of the directory for output placement involves possibilities
for explicit user instruction (absolute path or relative to execution) and
implicit default configuration (absolute path or relative to input) from
the system global configuration file. This function is responsible for
reliably returning the appropriate output directory which will contain any
log(s), ePub(s), and unzipped output of OpenAccess_EPUB.
It utilizes the parsed args, passed as an object, and is self-sufficient in
accessing the config file.
All paths returned by this function are absolute.
"""
#Import the global config file as a module
import imp
config_path = os.path.join(cache_location(), 'config.py')
try:
config = imp.load_source('config', config_path)
except IOError:
print('Could not find {0}, please run oae-quickstart'.format(config_path))
sys.exit()
#args.output is the explicit user instruction, None if unspecified
if args.output:
#args.output may be an absolute path
if os.path.isabs(args.output):
return args.output # return as is
#or args.output may be a relative path, relative to cwd
else:
return evaluate_relative_path(relative=args.output)
#config.default_output for default behavior without explicit instruction
else:
#config.default_output may be an absolute_path
if os.path.isabs(config.default_output):
return config.default_output
#or config.default_output may be a relative path, relative to input
else:
if args.input: # The case of single input
if 'http://www' in args.input:
#Fetched from internet by URL
raw_name = url_input(args.input, download=False)
abs_input_path = os.path.join(os.getcwd(), raw_name+'.xml')
elif args.input[:4] == 'doi:':
#Fetched from internet by DOI
raw_name = doi_input(args.input, download=False)
abs_input_path = os.path.join(os.getcwd(), raw_name+'.xml')
else:
#Local option, could be anywhere
abs_input_path = get_absolute_path(args.input)
abs_input_parent = os.path.split(abs_input_path)[0]
return evaluate_relative_path(abs_input_parent, config.default_output)
elif args.batch: # The case of Batch Mode
#Batch should only work on a supplied directory
abs_batch_path = get_absolute_path(args.batch)
return abs_batch_path
elif args.zip:
#Zip is a local-only option, behaves just like local xml
abs_input_path = get_absolute_path(args.zip)
abs_input_parent = os.path.split(abs_input_path)[0]
return evaluate_relative_path(abs_input_parent, config.default_output)
elif args.collection:
return os.getcwd()
else: # Un-handled or currently unsupported options
print('The output location could not be determined...')
sys.exit()
def main(argv=None):
args = docopt(__doc__,
argv=argv,
version='OpenAccess_EPUB v.' + __version__,
options_first=True)
if args['--epub3']:
epub_version = 3
elif args['--epub2']:
epub_version = 2
else:
epub_version = None
#Basic logging configuration
oae_logging.config_logging(args['--no-log-file'], args['--log-to'],
args['--log-level'], args['--silent'],
args['--verbosity'])
#Get a logger, the 'openaccess_epub' logger was set up above
command_log = logging.getLogger('openaccess_epub.commands.convert')
#Load the config module, we do this after logging configuration
config = openaccess_epub.utils.load_config_module()
current_dir = os.getcwd()
#Our basic flow is to iterate over the args['INPUT'] list
for inpt in args['INPUT']:
#We have to temporarily re-base our log while input utils do some work
if not args['--no-log-file'] and not args['--log-to']:
oae_logging.replace_filehandler(logname='openaccess_epub',
new_file='temp.log',
level=args['--log-level'],
frmt=oae_logging.STANDARD_FORMAT)
command_log.info('Processing input: {0}'.format(inpt))
#First we need to know the name of the file and where it is
if inpt.lower().endswith('.xml'): # This is direct XML file
root_name = openaccess_epub.utils.file_root_name(inpt)
abs_input_path = openaccess_epub.utils.get_absolute_path(inpt)
elif inpt.lower().startswith('doi:'): # This is a DOI
root_name = input_utils.doi_input(inpt)
abs_input_path = os.path.join(current_dir, root_name + '.xml')
elif any(inpt.lower().startswith(i) for i in ['http:', 'https:']):
root_name = input_utils.url_input(inpt)
abs_input_path = os.path.join(current_dir, root_name + '.xml')
else:
sys.exit('{0} not recognized as XML, DOI, or URL'.format(inpt))
if not args['--no-log-file'] and not args['--log-to']:
log_name = root_name + '.log'
log_path = os.path.join(os.path.dirname(abs_input_path), log_name)
#Re-base the log file to the new file location
oae_logging.replace_filehandler(logname='openaccess_epub',
new_file=log_path,
level=args['--log-level'],
frmt=oae_logging.STANDARD_FORMAT)
#Now we move over to the new log file
shutil.copy2('temp.log', log_path)
os.remove('temp.log')
#Now that we should be done configuring logging, let's parse the article
parsed_article = Article(abs_input_path,
validation=not args['--no-validate'])
if parsed_article.publisher is None:
command_log.critical(
'Publisher support was not established, aborting')
sys.exit(1)
#Get the output directory
if args['--output'] is not None:
output_directory = openaccess_epub.utils.get_absolute_path(
args['--output'])
else:
if os.path.isabs(config.default_output): # Absolute remains so
output_directory = config.default_output
else: # Else rendered relative to input
abs_dirname = os.path.dirname(abs_input_path)
output_directory = os.path.normpath(
os.path.join(abs_dirname, config.default_output))
#The root name must be added on for output
output_directory = os.path.join(output_directory, root_name)
#Make the call to make_EPUB
success = make_EPUB(parsed_article,
output_directory,
abs_input_path,
args['--images'],
config_module=config,
epub_version=epub_version)
#Cleanup removes the produced output directory, keeps the EPUB
if not args['--no-cleanup']:
command_log.info('Removing {0}'.format(output_directory))
shutil.rmtree(output_directory)
#Running epubcheck on the output verifies the validity of the EPUB,
#requires a local installation of java and epubcheck.
if not args['--no-epubcheck'] and success:
epub_name = '{0}.epub'.format(output_directory)
openaccess_epub.utils.epubcheck(epub_name, config)
