Consume command

ckanext/dcat/cli.py

@@ -1,25 +1,80 @@
 # -*- coding: utf-8 -*-
+import json
 
 import click
+
 import ckan.plugins.toolkit as tk
+
 import ckanext.dcat.utils as utils
+from ckanext.dcat.processors import RDFParser, RDFSerializer, DEFAULT_RDF_PROFILES
 
-@click.group()
-def generate_static():
-    """Generates static files containing all datasets.
 
-    """
+@click.group()
+def dcat():
+    """DCAT utilities for CKAN"""
     pass
 
-@generate_static.command()
-@click.argument('output', type=click.File(mode="w"))
-def json(output):
-    """The generate command will generate a static file containing all of
-    the datasets in the catalog in JSON format.
 
+@dcat.command()
+@click.argument("output", type=click.File(mode="w"))
+def generate_static(output):
+    """[Deprecated] Generate a static datasets file in JSON format
+    (requires the dcat_json_interface plugin) .
     """
     utils.generate_static_json(output)
 
 
+@dcat.command(context_settings={"show_default": True})
+@click.argument("input", type=click.File(mode="r"))
+@click.option(
+    "-o",
+    "--output",
+    type=click.File(mode="w"),
+    default="-",
+    help="By default the command will output the result to stdin, "
+    "alternatively you can provide a file path with this option",
+)
+@click.option(
+    "-f", "--format", default="xml", help="Serialization format (eg ttl, jsonld)"
+)
+@click.option(
+    "-p",
+    "--profiles",
+    default=" ".join(DEFAULT_RDF_PROFILES),
+    help="RDF profiles to use",
+)
+@click.option(
+    "-P", "--pretty", default=False, help="Make the output more human readable"
+)
+@click.option(
+    "-m", "--compat_mode", default=False, help="Compatibility mode (deprecated)"
+)
+def consume(input, output, format, profiles, pretty, compat_mode):
+    """
+    Parses DCAT RDF graphs into CKAN dataset JSON objects.
+
+    The input serializations can be provided as a path to a file, e.g.:
+
+        ckan dcat consume examples/dataset.ttl
+
+    Or be read from stdin:
+
+        ckan dcat consume -
+    """
+    contents = input.read()
+
+    if profiles:
+        profiles = profiles.split()
+    parser = RDFParser(profiles=profiles, compatibility_mode=compat_mode)
+    parser.parse(contents, _format=format)
+
+    ckan_datasets = [d for d in parser.datasets()]
+
+    indent = 4 if pretty else None
+    out = json.dumps(ckan_datasets, indent=indent)
+
+    output.write(out)
+
+
 def get_commands():
-    return [generate_static]
+    return [dcat]