Allow external definitions, fixes #10

adtl/__init__.py

@@ -385,14 +385,31 @@ def make_fields_optional(
     return _schema
 
 
+def relative_path(source_file, target_file):
+    return Path(source_file).parent / target_file
+
+
+def read_definition(file: Path) -> Dict[str, Any]:
+    "Reads definition from file into a dictionary"
+    if file.suffix == ".json":
+        with file.open() as fp:
+            return json.load(fp)
+    elif file.suffix == ".toml":
+        with file.open("rb") as fp:
+            return tomli.load(fp)
+    else:
+        raise ValueError(f"Unsupported file format: {file}")
+
+
 class Parser:
-    def __init__(self, spec: Union[str, Path, StrDict]):
+    def __init__(self, spec: Union[str, Path, StrDict], include_defs: List[str] = []):
         "Loads specification from spec in format (default json)"
 
         self.data: StrDict = {}
         self.defs: StrDict = {}
         self.fieldnames: Dict[str, List[str]] = {}
         self.specfile = None
+        self.include_defs = include_defs
         self.validators: StrDict = {}
         self.schemas: StrDict = {}
         self.date_fields = []
@@ -414,7 +431,15 @@ def __init__(self, spec: Union[str, Path, StrDict]):
         else:
             self.spec = spec
         self.header = self.spec.get("adtl", {})
+        if self.specfile:
+            self.include_defs = [
+                relative_path(self.specfile, definition_file)
+                for definition_file in self.header.get("include-def", [])
+            ] + self.include_defs
         self.defs = self.header.get("defs", {})
+        if self.include_defs:
+            for definition_file in self.include_defs:
+                self.defs.update(read_definition(definition_file))
         self.spec = expand_refs(self.spec, self.defs)
 
         self.validate_spec()
@@ -714,8 +739,14 @@ def main():
     cmd.add_argument(
         "--encoding", help="Encoding input file is in", default="utf-8-sig"
     )
+    cmd.add_argument(
+        "--include-def",
+        action="append",
+        help="Include external definition (TOML or JSON)",
+    )
     args = cmd.parse_args()
-    spec = Parser(args.spec)
+    include_defs = args.include_def or []
+    spec = Parser(args.spec, include_defs=include_defs)
     if output := spec.parse(args.file, encoding=args.encoding).save(
         args.output or spec.name
     ):

docs/specification.md

@@ -42,9 +42,9 @@ These metadata fields are defined under a header key `adtl`.
 
 ### Optional fields
 
-### Optional fields
-
 * **defs**: Definitions that can be referred to elsewhere in the schema
+* **include-def** (list): List of additional TOML or JSON files to import as
+  definitions
 * **defaultDateFormat**: Default source date format, applied to all fields
   with either "date_" / "_date" in the field name or that have format date
   set in the JSON schema
@@ -81,6 +81,42 @@ someTable = { groupBy = "subjid", aggregation = "lastNotNull" }
 someReference = { values = { 1 = true, 2 = false } }
 ```
 
+Often some definitions are repeated across files. adtl supports including
+definitions from external files using the *include-def* keyword under the
+`[adtl]` section. As an example, a mapping of country codes to country names
+could be stored in `countries.toml`:
+
+```toml
+[countryMap.values]
+1 = "ALB"
+2 = "ZZZ"
+# and so on
+```
+
+This could be included in adtl, and used as a reference just as if it was included in the TOML file directly:
+
+```toml
+[adtl]
+include-def = ["countries.toml"]
+
+# ...
+
+[cases.country_iso3]
+field = "country"
+ref = "countryMap
+```
+
+Definition files can also be included from the command line by passing the
+`--include-def` flag to adtl. This is useful when the included file can change
+from one run to another, or in cases where the definitions/mappings are located
+externally. The following would produce an equivalent result to the
+`include-def` assignment in the above example, assuming `data.csv` is the source
+data file:
+
+```shell
+adtl parser.toml data.csv --include-def countries.toml
+```
+
 ## Table mappings
 
 Each table has its associated field mappings under a key of the same

tests/parsers/groupBy-defs-include.toml

@@ -0,0 +1,30 @@
+[adtl]
+  name = "groupBy"
+  description = "Example using groupBy"
+  include-def = ["include-def.toml"]
+
+[subject]
+  dataset_id = "dataset-2020-03-23"
+  country_iso3 = "GBR"
+
+  [subject.sex_at_birth]
+    ref = "sexMapping"
+    field = "sex"
+    description = "Sex at Birth"
+
+  [subject.subject_id]
+    field = "subjid"
+    description = "Subject ID"
+
+  [subject.enrolment_date]
+    field = "dsstdat"
+    description = "Date of Enrolment"
+
+  [subject.admission_date]
+    field = "hostdat"
+    description = "Admission date at this facility"
+
+[adtl.tables.subject]
+  kind = "groupBy"
+  groupBy = "subject_id"
+  aggregation = "lastNotNull"

tests/parsers/groupBy-external-defs.toml

@@ -0,0 +1,29 @@
+[adtl]
+  name = "groupBy"
+  description = "Example using groupBy"
+
+[subject]
+  dataset_id = "dataset-2020-03-23"
+  country_iso3 = "GBR"
+
+  [subject.sex_at_birth]
+    ref = "sexMapping"
+    field = "sex"
+    description = "Sex at Birth"
+
+  [subject.subject_id]
+    field = "subjid"
+    description = "Subject ID"
+
+  [subject.enrolment_date]
+    field = "dsstdat"
+    description = "Date of Enrolment"
+
+  [subject.admission_date]
+    field = "hostdat"
+    description = "Admission date at this facility"
+
+[adtl.tables.subject]
+  kind = "groupBy"
+  groupBy = "subject_id"
+  aggregation = "lastNotNull"

tests/parsers/include-def.toml

@@ -0,0 +1,4 @@
+[sexMapping.values]
+  1 = "male"
+  2 = "female"
+  3 = "non_binary"

tests/test_parser.py

@@ -626,10 +626,26 @@ def test_reference_expansion():
     assert ps_ref.spec == ps_noref.spec
 
 
-def test_format_equivalence():
-    adtl_json = parser.Parser(TEST_PARSERS_PATH / "groupBy-defs.json")
-    adtl_toml = parser.Parser(TEST_PARSERS_PATH / "groupBy-defs.toml")
-    assert adtl_json.spec == adtl_toml.spec
+def test_reference_expansion_with_include():
+    ps_noinclude = parser.Parser(TEST_PARSERS_PATH / "groupBy-defs.toml")
+    ps_include = parser.Parser(TEST_PARSERS_PATH / "groupBy-defs-include.toml")
+    del ps_noinclude.spec["adtl"]["defs"]
+    del ps_include.spec["adtl"]["include-def"]
+    assert ps_noinclude.spec == ps_include.spec
+
+
+def test_external_definitions():
+    with pytest.raises(KeyError):
+        parser.Parser(TEST_PARSERS_PATH / "groupBy-external-defs.toml")
+    ps = parser.Parser(
+        TEST_PARSERS_PATH / "groupBy-external-defs.toml",
+        include_defs=[TEST_PARSERS_PATH / "include-def.toml"],
+    )
+    assert ps.defs["sexMapping"]["values"] == {
+        "1": "male",
+        "2": "female",
+        "3": "non_binary",
+    }
 
 
 FOR_PATTERN = [