use click for marc2bf cli

.gitignore

@@ -42,3 +42,6 @@ nosetests.xml
 .mr.developer.cfg
 .project
 .pydevproject
+
+venv
+.venv

exec/marc2bf

@@ -1,94 +1,8 @@
 #!/usr/bin/env python
-#-*- mode: python -*-
+# -*- mode: python -*-
 
-import sys
-import json
-import logging
-import argparse
-
-from bibframe.reader import bfconvert
-from amara3.inputsource import inputsourcetype
-
-
-def run(inputs=None, base=None, out=None, limit=None, rdfttl=None, rdfxml=None, xml=None,
-        config=None, verbose=False, mods=None, modfiles=None, canonical=False, lax=False):
-    '''
-    Basically takes parameters typical for command line invocation and adapts them for use in the API
-
-    '''
-    if config is None:
-        config = {}
-    else:
-        config = json.load(config)
-
-    logger = logging.getLogger('marc2bf')
-    if verbose:
-        logger.setLevel(logging.DEBUG)
-
-    for mod in mods:
-        __import__(mod, globals(), locals(), [])
-
-    for modfile in modfiles:
-        with open(modfile) as f:
-            code = compile(f.read(), modfile, 'exec')
-            exec(code, globals(), locals())
-
-    bfconvert(inputs=inputs, entbase=base, out=out, limit=limit, rdfttl=rdfttl, rdfxml=rdfxml,
-                xml=xml, config=config, verbose=verbose, canonical=canonical, logger=logger,
-                lax=lax, defaultsourcetype=inputsourcetype.filename)
-    return
 
+from bibframe.cli import marc2bf
 
 if __name__ == '__main__':
-    #marc2bf -v test/resource/700t.mrx
-    #marc2bf -v -o /dev/null --rdfttl /tmp/foo.ttl test/resource/700t.mrx
-    #parser = argparse.ArgumentParser(prog="bootstrap", add_help=False)
-    parser = argparse.ArgumentParser()
-    parser.add_argument('inputs', metavar='inputs', nargs='*',
-                        help='One or more MARC/XML files to be parsed and converted to BIBFRAME RDF.')
-    parser.add_argument('-o', '--out', type=argparse.FileType('w'), default=sys.stdout,
-        help='File where raw Versa JSON output should be written'
-             '(default: write to stdout)')
-    parser.add_argument('-p', '--postout', metavar="IRI",
-        help='HTTP endpoint for pushing or posting raw Versa JSON output'
-             '(default: write to stdout)')
-    parser.add_argument('--rdfttl', type=argparse.FileType('wb'),
-        help='File where RDF Turtle output should be written')
-    parser.add_argument('--rdfxml', type=argparse.FileType('wb'),
-        help='File where RDF XML output should be written')
-    parser.add_argument('--xml', type=argparse.FileType('w'),
-        help='File where MicroXML output should be written')
-    parser.add_argument('-c', '--config', type=argparse.FileType('r'),
-        help='File containing config in JSON format')
-    #parser.add_argument('-s', '--stats', type=argparse.FileType('w'),
-    #    help='file where statistics output should be written in JSOn format')
-    parser.add_argument('-l', '--limit', metavar="NUMBER",
-        help='Limit the number of records processed to this number. If omitted, all records will be processed.')
-    parser.add_argument('-b', '--base', metavar="IRI", #dest="base",
-        help='Base IRI to be used for creating resources.')
-    parser.add_argument('--mod', metavar="PYMODULE", nargs="*", action='append',
-        help='Python module to be imported in order to register plugins (can be specified multiple times.')
-    parser.add_argument('--modfile', metavar="FILEPATH", nargs="*", action='append',
-        help='Python file to be executed as a module in order to register plugins (can be specified multiple times.')
-    parser.add_argument('-v', '--verbose', action='store_true',
-        help='Show additional messages and information')
-    parser.add_argument('--canonical', action='store_true',
-        help='Use Versa\'s canonical form for output. Warning: memory inefficient')
-    parser.add_argument('--lax', action='store_true',
-        help='Parse less strictly, e.g. accepting MARC/XML with bad namespace declarations')
-    #XXX: Any way to get generalized archive support using shutil? Perhaps along with tempfile?
-    #https://docs.python.org/3/library/shutil.html#archiving-operations
-    #parser.add_argument('-z', '--zipcheck', action='store_true',
-    #    help='Check for zip files among the inputs')
-    #
-    args = parser.parse_args()
-    args.mod = [i for items in args.mod or [] for i in items]
-    args.modfile = [i for items in args.modfile or [] for i in items]
-
-    run(inputs=args.inputs, base=args.base, out=args.out, limit=args.limit, rdfttl=args.rdfttl,
-        rdfxml=args.rdfxml, xml=args.xml, config=args.config, verbose=args.verbose,
-        mods=args.mod, modfiles=args.modfile, canonical=args.canonical, lax=args.lax)
-    #for f in args.inputs: f.close()
-    if args.rdfttl: args.rdfttl.close()
-    if args.rdfxml: args.rdfxml.close()
-    args.out.close()
+    marc2bf()

lib/cli.py

@@ -0,0 +1,111 @@
+import click
+import sys
+
+from bibframe.main import run
+
+
+@click.command()
+@click.argument(
+    "inputs",
+    type=click.Path(exists=True, dir_okay=False),
+    nargs=-1,  # unlimited number of arguments,
+    metavar="inputs",
+    # help="One or more MARC/XML files to be parsed and converted to BIBFRAME RDF.", # no help for argument in Click
+)
+@click.option(
+    "-o",
+    "--out",
+    type=click.File("w"),
+    default=sys.stdout,
+    help="File where raw Versa JSON output should be written"
+         "(default: write to stdout)",
+)
+# @click.option("-p", "--postout", metavar="IRI")
+@click.option(
+    "--rdfttl",
+    type=click.File("wb"),
+    help="File where RDF Turtle output should be written",
+)
+@click.option(
+    "--rdfxml",
+    type=click.File("wb"),
+    help="File where RDF XML output should be written",
+)
+@click.option(
+    "--xml", type=click.File("w"), help="File where MicroXML output should be written"
+)
+@click.option(
+    "-c", "--config", type=click.File("r"), help="File containing config in JSON format"
+)
+# @click.option(
+#     "-s", "--stats", type=click.File("w"),
+#     help='file where statistics output should be written in JSOn format'
+# )
+@click.option(
+    "-l",
+    "--limit",
+    type=click.INT,
+    metavar="NUMBER",
+    help="Limit the number of records processed to this number. If omitted, all records will be processed.",
+)
+@click.option(
+    "-b",
+    "--base",
+    metavar="IRI",
+    help="Base IRI to be used for creating resources.",
+)
+@click.option(
+    "--mod",
+    multiple=True,
+    metavar="PYMODULE",
+    help="Python module to be imported in order to register plugins (can be specified multiple times).",
+)
+@click.option(
+    "--modfile",
+    multiple=True,
+    metavar="FILEPATH",
+    help="Python file to be executed as a module in order to register plugins (can be specified multiple times).",
+)
+@click.option(
+    "-v", "--verbose", is_flag=True, help="Show additional messages and information"
+)
+@click.option(
+    "--canonical",
+    is_flag=True,
+    help="Use Versa's canonical form for output. Warning: memory inefficient",
+)
+@click.option(
+    "--lax",
+    is_flag=True,
+    help="Parse less strictly, e.g. accepting MARC/XML with bad namespace declarations",
+)
+def marc2bf(
+        inputs,
+        out,
+        rdfttl,
+        rdfxml,
+        xml,
+        config,
+        limit,
+        base,
+        mod,
+        modfile,
+        verbose,
+        canonical,
+        lax,
+):
+    run(
+        inputs=inputs,
+        base=base,
+        out=out,
+        limit=limit,
+        rdfttl=rdfttl,
+        rdfxml=rdfxml,
+        xml=xml,
+        config=config,
+        verbose=verbose,
+        mods=mod,
+        modfiles=modfile,
+        canonical=canonical,
+        lax=lax,
+    )

lib/main.py

@@ -0,0 +1,34 @@
+import json
+import logging
+
+from bibframe.reader import bfconvert
+from amara3.inputsource import inputsourcetype
+
+
+def run(inputs=None, base=None, out=None, limit=None, rdfttl=None, rdfxml=None, xml=None,
+        config=None, verbose=False, mods=None, modfiles=None, canonical=False, lax=False):
+    '''
+    Basically takes parameters typical for command line invocation and adapts them for use in the API
+
+    '''
+    if config is None:
+        config = {}
+    else:
+        config = json.load(config)
+
+    logger = logging.getLogger('marc2bf')
+    if verbose:
+        logger.setLevel(logging.DEBUG)
+
+    for mod in mods:
+        __import__(mod, globals(), locals(), [])
+
+    for modfile in modfiles:
+        with open(modfile) as f:
+            code = compile(f.read(), modfile, 'exec')
+            exec(code, globals(), locals())
+
+    bfconvert(inputs=inputs, entbase=base, out=out, limit=limit, rdfttl=rdfttl, rdfxml=rdfxml,
+              xml=xml, config=config, verbose=verbose, canonical=canonical, logger=logger,
+              lax=lax, defaultsourcetype=inputsourcetype.filename)
+    return

requirements.txt

@@ -5,3 +5,4 @@ mmh3
 pytest
 versa==0.5.2
 amara3.xml
+click