dumping to geojson

datapackage_pipelines/lib/dump/dumper_base.py

@@ -16,7 +16,7 @@
 from ...utilities.extended_json import json
 from ...wrapper import ingest, spew
 
-from .file_formats import CSVFormat, JSONFormat
+from .file_formats import CSVFormat, JSONFormat, GeoJSONFormat
 
 
 class DumperBase(object):
@@ -34,7 +34,8 @@ def __init__(self, debug=False):
         self.add_filehash_to_path = self.__params.get('add-filehash-to-path', False)
         self.file_format_handlers = {
             'csv': CSVFormat,
-            'json': JSONFormat
+            'json': JSONFormat,
+            'geojson': GeoJSONFormat
         }
         self.file_format_handlers.update(**self.__params.get('file-formatters', {}))
 
@@ -253,6 +254,7 @@ def write_file_to_output(self, filename, path):
         raise NotImplementedError()
 
     def rows_processor(self, resource, spec, temp_file, writer, fields, datapackage):
+        print(fields)
         file_formatter = self.file_formatters[spec['name']]
         for row in resource:
             file_formatter.write_row(writer, row, fields)

datapackage_pipelines/lib/dump/file_formats.py

@@ -152,3 +152,40 @@ def write_transformed_row(self, writer, transformed_row, fields):
 
     def finalize_file(self, writer):
         writer.write(']')
+
+class GeoJSONFormat(JSONFormat):
+
+    def initialize_file(self, file, headers):
+        writer = file
+        writer.write('{"type": "FeatureCollection","features":')
+        writer.__first = True
+        return writer
+
+
+    def write_transformed_row(self, writer, transformed_row, fields):
+        if not writer.__first:
+            writer.write(',')
+        else:
+            writer.__first = False
+        try:
+            geometry = {}
+            for k, v in transformed_row.items():
+                if fields[k]["type"] == "geopoint":
+                    geometry = {"type": "Point",
+                                "coordinates" : self._FileFormat__transform_value(v, fields[k]["type"])}
+                    break
+                elif fields[k]["type"] == "geojson":
+                    geometry = self._FileFormat__transform_value(v, fields[k]["type"])
+                    break
+        except Exception:
+            logging.exception('Missing point geometry in row %r', transformed_row)
+            raise
+        properties = dict((k, self._FileFormat__transform_value(v, fields[k]["type"]))
+                        for k, v in transformed_row.items() if fields[k]["type"] not in ["geopoint", "geojson"])
+        feature = {"geometry": geometry, "type": "Feature", "properties": properties}
+
+        writer.write(json.dumps(feature, sort_keys=True, ensure_ascii=True))
+
+    def finalize_file(self, writer):
+        writer.write(']}')
+