datex2-xml2json 0.0.1__py3-none-any.whl

datex2_xml2json/json_validator.py

@@ -0,0 +1,36 @@
+from check_jsonschema.checker import SchemaChecker
+from check_jsonschema.file_wrapper import BinaryFileInput
+from check_jsonschema.formats import FormatOptions
+from check_jsonschema.instance_loader import InstanceLoader
+from check_jsonschema.regex_variants import RegexImplementation, RegexVariantName
+from check_jsonschema.reporter import REPORTER_BY_NAME, Reporter
+from check_jsonschema.schema_loader import SchemaLoader
+
+
+class JsonValidator:
+    def __init__(self, schema_path):
+        self.schema_loader = SchemaLoader(
+            schema_path,
+            disable_cache=False,
+            base_uri=None,
+            validator_class=None,
+        )
+
+    def build_reporter(self, output_format: str, verbosity: int) -> Reporter:
+        cls = REPORTER_BY_NAME[output_format]
+        return cls(verbosity=verbosity)
+
+    def validate(self, filename):
+        checker = SchemaChecker(
+            schema_loader=self.schema_loader,
+            instance_loader=InstanceLoader([BinaryFileInput(filename)]),
+            reporter=self.build_reporter('text', 2),
+            format_opts=FormatOptions(
+                regex_impl=RegexImplementation(RegexVariantName.default),
+                enabled=True,
+                disabled_formats=[],
+            ),
+            regex_impl=RegexImplementation(RegexVariantName.default),
+        )
+
+        return checker.run()

datex2_xml2json/merge.py

@@ -0,0 +1,39 @@
+from argparse import ArgumentParser
+
+import geoparquet_io as gpio
+
+
+def merge_partition(partion: str, outfile: str) -> None:
+    """
+    merge_partition reads all parquet files from a given partition (e.g. directory), merges them
+    and writes them to geoparquet file `outfile`.
+
+    Note: OGC_FID and index column are removed
+    """
+
+    # Read all partitions
+    table = gpio.read_partition(partion)
+    column_name = 'OGC_FID'
+    index = table.schema.get_field_index(column_name)
+
+    if index != -1:
+        table._table = table._table.remove_column(index)
+    table.write(outfile)
+
+
+if __name__ == '__main__':
+    parser = ArgumentParser()
+    parser.add_argument('partition', help='Partition (e.g. directory) of parquet files to merge')
+    parser.add_argument(
+        '-o',
+        '--outfile',
+        help='Path of merged parquet file',
+        default='merged.parquet',
+    )
+
+    args = parser.parse_args()
+
+    merge_partition(
+        args.partition,
+        args.outfile,
+    )

datex2_xml2json/schema/Baustellen-4-00-00.sch

@@ -0,0 +1,7 @@
+<schema xmlns="http://purl.oclc.org/dsdl/schematron">
+    <pattern>
+        <rule context="comment/values">
+            <assert test="value[count(*) &gt; 1]">More than one vlues</assert>
+        </rule>
+    </pattern>
+</schema>