Upload json_semval/schema_utils.py with huggingface_hub

json_semval/schema_utils.py

@@ -0,0 +1,59 @@
+from __future__ import annotations
+
+from typing import Any, Dict, List
+
+
+def collect_enums(schema: Dict[str, Any]) -> Dict[str, List[str]]:
+    enums: Dict[str, List[str]] = {}
+
+    def _walk(subschema: Dict[str, Any], path: List[str]) -> None:
+        if "enum" in subschema and isinstance(subschema["enum"], list):
+            enums["".join(path)] = [str(v) for v in subschema["enum"]]
+        if subschema.get("type") == "object":
+            props = subschema.get("properties", {})
+            for key, child in props.items():
+                _walk(child, path + [f".{key}"])
+        if subschema.get("type") == "array":
+            items = subschema.get("items")
+            if isinstance(items, dict):
+                _walk(items, path + ["[0]"])
+
+    _walk(schema, [])
+    return enums
+
+
+def collect_formats(schema: Dict[str, Any]) -> Dict[str, str]:
+    fmts: Dict[str, str] = {}
+
+    def _walk(subschema: Dict[str, Any], path: List[str]) -> None:
+        if "format" in subschema:
+            fmts["".join(path)] = subschema["format"]
+        if subschema.get("type") == "object":
+            for key, child in subschema.get("properties", {}).items():
+                _walk(child, path + [f".{key}"])
+        if subschema.get("type") == "array":
+            items = subschema.get("items")
+            if isinstance(items, dict):
+                _walk(items, path + ["[0]"])
+
+    _walk(schema, [])
+    return fmts
+
+
+def collect_ranges(schema: Dict[str, Any]) -> Dict[str, Dict[str, Any]]:
+    ranges: Dict[str, Dict[str, Any]] = {}
+
+    def _walk(subschema: Dict[str, Any], path: List[str]) -> None:
+        for k in ["minimum", "maximum", "minLength", "maxLength", "pattern"]:
+            if k in subschema:
+                ranges.setdefault("".join(path), {})[k] = subschema[k]
+        if subschema.get("type") == "object":
+            for key, child in subschema.get("properties", {}).items():
+                _walk(child, path + [f".{key}"])
+        if subschema.get("type") == "array":
+            items = subschema.get("items")
+            if isinstance(items, dict):
+                _walk(items, path + ["[0]"])
+
+    _walk(schema, [])
+    return ranges