(🎁) infer data vs schema when --input-file-type is auto (#1249)

* infer data vs schema when `--input-file-type` is auto * Improve is_schema conditions * Fix coverage * Fix is_schema * Improve coverage * Improve coverage * Improve coverage --------- Co-authored-by: KotlinIsland <[email protected]> Co-authored-by: Koudai Aono <[email protected]>
koxudaxi · Apr 16, 2023 · 03eeddb · 03eeddb
1 parent 083691c
commit 03eeddb
Show file tree

Hide file tree

Showing 14 changed files with 99 additions and 12 deletions.
diff --git a/datamodel_code_generator/__init__.py b/datamodel_code_generator/__init__.py
@@ -168,6 +168,37 @@ def is_openapi(text: str) -> bool:
     return 'openapi' in load_yaml(text)
 
 
+JSON_SCHEMA_URLS: Tuple[str, ...] = (
+    'http://json-schema.org/',
+    'https://json-schema.org/',
+)
+
+
+def is_schema(text: str) -> bool:
+    data = load_yaml(text)
+    if not isinstance(data, dict):
+        return False
+    schema = data.get('$schema')
+    if isinstance(schema, str) and any(
+        schema.startswith(u) for u in JSON_SCHEMA_URLS
+    ):  # pragma: no cover
+        return True
+    if isinstance(data.get('type'), str):
+        return True
+    if any(
+        isinstance(data.get(o), list)
+        for o in (
+            'allOf',
+            'anyOf',
+            'oneOf',
+        )
+    ):
+        return True
+    if isinstance(data.get('properties'), dict):
+        return True
+    return False
+
+
 class InputFileType(Enum):
     Auto = 'auto'
     OpenAPI = 'openapi'
@@ -304,11 +335,8 @@ def generate(
                 if isinstance(input_, Path)
                 else input_text
             )
-            input_file_type = (
-                InputFileType.OpenAPI
-                if is_openapi(input_text_)  # type: ignore
-                else InputFileType.JsonSchema
-            )
+            assert isinstance(input_text_, str)
+            input_file_type = infer_input_type(input_text_)
             print(
                 inferred_message.format(input_file_type.value),
                 file=sys.stderr,
@@ -483,6 +511,14 @@ def get_header_and_first_line(csv_file: IO[str]) -> Dict[str, Any]:
             file.close()
 
 
+def infer_input_type(text: str) -> InputFileType:
+    if is_openapi(text):
+        return InputFileType.OpenAPI
+    elif is_schema(text):
+        return InputFileType.JsonSchema
+    return InputFileType.Json
+
+
 inferred_message = (
     'The input file type was determined to be: {}\nThis can be specificied explicitly with the '
     '`--input-file-type` option.'

diff --git a/tests/data/jsonschema/items_boolean.json b/tests/data/jsonschema/items_boolean.json
@@ -1,5 +1,5 @@
 {
-  "$schema ": "http://json-schema.org/draft-07/schema#",
+  "$schema": "http://json-schema.org/draft-07/schema#",
   "type": "object",
   "properties": {
     "example": {

diff --git a/tests/data/jsonschema/root_id.json b/tests/data/jsonschema/root_id.json
@@ -1,6 +1,6 @@
 {
   "$id": "https://example.com/root_id.json",
-  "$schema ": "http://json-schema.org/draft-07/schema#",
+  "$schema": "http://json-schema.org/draft-07/schema#",
   "definitions": {
     "Person": {
       "$ref": "person.json"

diff --git a/tests/data/jsonschema/root_id_absolute_url.json b/tests/data/jsonschema/root_id_absolute_url.json
@@ -1,6 +1,6 @@
 {
   "$id": "https://example.com/root_id.json",
-  "$schema ": "http://json-schema.org/draft-07/schema#",
+  "$schema": "http://json-schema.org/draft-07/schema#",
   "definitions": {
     "Person": {
       "$ref": "person.json"

diff --git a/tests/data/jsonschema/root_id_ref.json b/tests/data/jsonschema/root_id_ref.json
@@ -1,5 +1,5 @@
 {
-  "$schema ": "http://json-schema.org/draft-07/schema#",
+  "$schema": "http://json-schema.org/draft-07/schema#",
   "definitions": {
     "Person": {
       "$ref": "root_id.json#/definitions/Person"

diff --git a/tests/data/jsonschema/root_id_self_ref.json b/tests/data/jsonschema/root_id_self_ref.json
@@ -1,6 +1,6 @@
 {
   "$id": "https://example.com/root_id_self_ref.json",
-  "$schema ": "http://json-schema.org/draft-07/schema#",
+  "$schema": "http://json-schema.org/draft-07/schema#",
   "definitions": {
     "Person": {
       "$ref": "person.json"

diff --git a/tests/data/openapi/complex_reference.json b/tests/data/openapi/complex_reference.json
@@ -1,4 +1,5 @@
 {
+    "openapi": "3.0.0",
     "components": {
         "schemas": {
             "A": {

diff --git a/tests/data/openapi/datetime.yaml b/tests/data/openapi/datetime.yaml
@@ -1,3 +1,4 @@
+openapi: "3.0.0"
 components:
   schemas:
     InventoryItem:

diff --git a/tests/data/openapi/definitions.yaml b/tests/data/openapi/definitions.yaml
@@ -1,3 +1,4 @@
+openapi: "3.0.0"
 schemas:
   Problem:
     properties:

diff --git a/tests/data/openapi/discriminator.yaml b/tests/data/openapi/discriminator.yaml
@@ -1,3 +1,4 @@
+openapi: "3.0.0"
 components:
   schemas:
     ObjectBase:

diff --git a/tests/data/openapi/override_required_all_of.yaml b/tests/data/openapi/override_required_all_of.yaml
@@ -1,3 +1,4 @@
+openapi: "3.0.0"
 components:
   schemas:
     ObjectBase:

diff --git a/tests/data/openapi/x_enum_varnames.yaml b/tests/data/openapi/x_enum_varnames.yaml
@@ -1,4 +1,4 @@
-openapi: 3.0
+openapi: "3.0.0"
 components:
   schemas:
     string:

diff --git a/tests/root_id.json b/tests/root_id.json
@@ -1,6 +1,6 @@
 {
   "$id": "https://example.com/root_id.json",
-  "$schema ": "http://json-schema.org/draft-07/schema#",
+  "$schema": "http://json-schema.org/draft-07/schema#",
   "definitions": {
     "Person": {
       "$ref": "person.json"

diff --git a/tests/test_infer_input_type.py b/tests/test_infer_input_type.py
@@ -0,0 +1,46 @@
+from pathlib import Path
+
+from datamodel_code_generator import InputFileType, infer_input_type
+
+DATA_PATH: Path = Path(__file__).parent / 'data'
+
+
+def test_infer_input_type():
+    def assert_infer_input_type(file: Path, raw_data_type: InputFileType) -> None:
+        __tracebackhide__ = True
+        if file.is_dir():
+            return
+        if file.suffix not in ('.yaml', '.json'):
+            return
+        result = infer_input_type(file.read_text())
+        assert result == raw_data_type, f'{file} was the wrong type!'
+
+    for file in (DATA_PATH / 'json').rglob('*'):
+        if str(file).endswith('broken.json'):
+            continue
+        assert_infer_input_type(file, InputFileType.Json)
+    for file in (DATA_PATH / 'jsonschema').rglob('*'):
+        if str(file).endswith(('external_child.json', 'external_child.yaml')):
+            continue
+        if 'reference_same_hierarchy_directory' in str(file):
+            continue
+        assert_infer_input_type(file, InputFileType.JsonSchema)
+    for file in (DATA_PATH / 'openapi').rglob('*'):
+        if str(file).endswith(
+            (
+                'aliases.json',
+                'extra_data.json',
+                'invalid.yaml',
+                'list.json',
+                'empty_data.json',
+                'root_model.yaml',
+                'json_pointer.yaml',
+                'const.json',
+            )
+        ):
+            continue
+
+        if str(file).endswith('not.json'):
+            assert_infer_input_type(file, InputFileType.Json)
+            continue
+        assert_infer_input_type(file, InputFileType.OpenAPI)