Fix util.load_yaml_or_json failures when loading json files

Line number information are not available through the json API, but
providing null line number is enough to make it compatible with other
parts.

Author: serge-sans-paille

glean_parser/util.py

@@ -10,7 +10,7 @@
 from pathlib import Path
 import sys
 import textwrap
-from typing import Any, Callable, Iterable, Sequence, Tuple, Union, Optional
+from typing import Any, Callable, Iterable, Sequence, Tuple, Union, Optional, TextIO
 import urllib.request
 
 import appdirs  # type: ignore
@@ -97,6 +97,18 @@ def _construct_mapping_adding_line(loader, node):
     return yaml.load(stream, SafeLineLoader)
 
 
+def json_load(stream: TextIO):
+
+    # The Python json parser doesn't give us access to the line number
+    def _construct_mapping_adding_empty_line(obj):
+        mapping = DictWrapper(obj)
+        mapping.defined_in = {"line": 0}
+        return mapping
+
+    return json.load(stream, object_hook=_construct_mapping_adding_empty_line)
+
+
+
 def ordered_yaml_dump(data, **kwargs):
     class OrderedDumper(yaml.Dumper):
         pass
@@ -126,7 +138,7 @@ def load_yaml_or_json(path: Path):
 
     if path.suffix == ".json":
         with path.open("r", encoding="utf-8") as fd:
-            return json.load(fd)
+            return json_load(fd)
     elif path.suffix in (".yml", ".yaml", ".yamlx"):
         with path.open("r", encoding="utf-8") as fd:
             return yaml_load(fd)

tests/test_parser.py

@@ -7,11 +7,13 @@
 import json
 import re
 import textwrap
+import tempfile
 
 import pytest
 
 from glean_parser import metrics
 from glean_parser import parser
+from glean_parser.util import load_yaml_or_json
 
 import util
 
@@ -1181,6 +1183,19 @@ def test_no_internal_fields_exposed():
     assert expected_json == out_json
 
 
+def test_json_loader():
+    yaml_input = ROOT / "data" / "all_metrics.yaml"
+    yaml_obj = load_yaml_or_json(yaml_input)
+
+    with tempfile.NamedTemporaryFile(delete_on_close=False) as fd:
+        json.dump(yaml_obj, fd)
+        json_input = fd.name
+
+    json_obj = load_yaml_or_json(json_input)
+
+    assert yaml_obj == json_obj
+
+
 def test_object():
     structure = {"type": "array", "items": {"type": "number"}}
     contents = [{"category": {"metric": {"type": "object", "structure": structure}}}]