add config entries for application handlers

lib/argconfig_parse.py

@@ -33,8 +33,8 @@ def __init__(self, arglist):
                      to the current directory "."')
         self.parser.add_argument('--document-type', '-d', type=str, nargs='?', 
                 default='grant',
-                help='Choose whether your files are patent grants or applications. \
-                     Defaults to patent grants.')
+                help='Set the type of patent document to be parsed: grant (default) \
+                or application')
 
         # parse arguments and assign values
         args = self.parser.parse_args(self.arglist)

lib/config_parser.py

@@ -70,15 +70,17 @@ def get_dates(yearstring):
     return years
 
 
-def get_xml_handlers(configfile):
+def get_xml_handlers(configfile, document_type='grant'):
     """
     Called by parse.py to generate a lookup dictionary for which parser should
     be used for a given file
     """
     handler = ConfigParser()
     handler.read(configfile)
     xmlhandlers = {}
-    for yearrange, handler in handler.items('xml-handlers'):
+    config_item = 'grant-xml-handlers' if document_type == 'grant' \
+                   else 'application-xml-handlers'
+    for yearrange, handler in handler.items(config_item):
         for year in get_dates(yearrange):
             try:
                 xmlhandlers[year] = importlib.import_module(handler)

parse.py

@@ -10,7 +10,6 @@
 import shutil
 from lib.config_parser import get_xml_handlers
 
-xmlhandlers = get_xml_handlers('process.cfg')
 logfile = "./" + 'xml-parsing.log'
 logging.basicConfig(filename=logfile, level=logging.DEBUG)
 commit_frequency = alchemy.get_config().get('parse').get('commit_frequency')
@@ -173,5 +172,7 @@ def main(patentroot, xmlregex, verbosity, output_directory='.'):
     PATENTROOT = args.get_patentroot()
     VERBOSITY = args.get_verbosity()
     PATENTOUTPUTDIR = args.get_output_directory()
+    DOCUMENTTYPE = args.get_document_type()
+    xmlhandlers = get_xml_handlers('process.cfg', DOCUMENTTYPE)
 
     main(PATENTROOT, XMLREGEX, VERBOSITY, PATENTOUTPUTDIR)

process.cfg

@@ -70,7 +70,10 @@ datadir=/Users/gabe
 # format `ipgYYMMDD` is assumed), then the default parser is used.
 # The dates in the ranges are either YYYY or YYYYMMDD. If only one date is provided,
 # then the corresponding handler is assumed for all subsequent patents
-[xml-handlers]
+[grant-xml-handlers]
 2005-20130108=lib.handlers.grant_handler_v42
 20130115=lib.handlers.grant_handler_v44
 default=lib.handlers.grant_handler_v42
+
+[application-xml-handlers]
+default=lib.handlers.application_handler_v1