automate schema generation

.github/workflows/deploy.yml

@@ -34,7 +34,7 @@ jobs:
           go-version: '1.23'
           check-latest: true
       - run: make build
-      - run: mkdir public && cp index.html main.wasm wasm_exec.js xray-schema.json public
+      - run: mkdir public && cp index.html main.wasm wasm_exec.js xray.schema.json public
       - name: Upload artifact
         uses: actions/upload-pages-artifact@v1
         with:

.gitignore

@@ -1,3 +1,4 @@
 *.wasm
 *.js
 assets
+*.schema.json

Makefile

@@ -1,4 +1,4 @@
-build: main.wasm wasm_exec.js
+build: main.wasm wasm_exec.js xray.schema.json
 .PHONY: build
 
 dev-lite:
@@ -30,5 +30,12 @@ wasm_exec.js:
 main.wasm: assets/geoip.dat assets/geosite.dat assets/xray-patched main.go go.mod xray-wasm.patch
 	GOARCH=wasm GOOS=js go build -o main.wasm main.go
 
+xray.schema.json:
+	rm -rf assets/xray-docs-next
+	mkdir -p assets
+	cd assets && git clone --depth 1 https://github.com/xtls/xray-docs-next
+	rg '' assets/xray-docs-next/docs/en/config/ | cut -d: -f2- | python3 scrape-docs.py > xray.schema.json
+
+
 serve:
 	python3 -mhttp.server

index.html

@@ -174,7 +174,7 @@ <h1>Xray Config Validator</h1>
                 return error.message;
             }
         }
-        fetch('xray-schema.json')
+        fetch('xray.schema.json')
             .then(response => response.json())
             .then(schema => {
                 monaco.languages.json.jsonDefaults.setDiagnosticsOptions({

scrape-docs.py

@@ -0,0 +1,134 @@
+import os
+import json
+import sys
+
+from typing import Iterator, Sequence, TypedDict
+
+class RawType(TypedDict):
+    title: str
+    description: str
+    raw_properties: list[dict]
+
+class JsonschemaType(TypedDict):
+    title: str
+    description: str
+    properties: dict[str, dict]
+
+def parse(stdin: Iterator[str]) -> Iterator[JsonschemaType]:
+    current_obj: RawType | None = None
+
+    for line in stdin:
+        if not line.strip():
+            continue
+        elif line.startswith("## "):
+            if current_obj:
+                yield {
+                    "title": current_obj['title'],
+                    "description": current_obj['description'],
+                    "properties": {
+                        x['name']: x
+                        for x in current_obj['raw_properties']
+                    }
+                }
+
+            current_obj = {
+                "title": line[2:].strip(),
+                "description": "",
+                "raw_properties": []
+            }
+        elif line.startswith("> ") and ":" in line and current_obj:
+            name, ty = line[2:].split(":", 1)
+            if name == "Tony":
+                continue
+
+            name = name.strip(" `")
+
+            current_obj['raw_properties'].append({
+                "name": name,
+                "description": "",
+                **parse_type(ty),
+            })
+        elif current_obj:
+            if current_obj['raw_properties']:
+                current_obj['raw_properties'][-1]['description'] += line
+            else:
+                current_obj['description'] += line
+
+def parse_type(input: str) -> dict:
+    input = input \
+            .replace('<Badge text="WIP" type="warning"/>', '') \
+            .replace('<Badge text="BETA" type="warning"/>', '') \
+            .strip()
+
+    if not input:
+        return {}
+
+    if input.startswith("\\[") and input.endswith("\\]"):
+        return {
+            "type": "array",
+            "items": parse_type(input[2:-2])
+        }
+
+    if input.startswith("[") and input.endswith("]"):
+        return {
+            "type": "array",
+            "items": parse_type(input[1:-1])
+        }
+
+    if input.startswith("[") and input.endswith(")"):
+        name = input.split("]")[0].strip("[]")
+        return {
+            "$ref": f"#/definitions/{name}"
+        }
+
+    if input in ("true", "false", "true | false", "bool"):
+        return {"type": "bool"}
+
+    if " | " in input:
+        return {"anyOf": [parse_type(x) for x in input.split(" | ")]}
+
+    if input.endswith("Object"):
+        return {
+            "$ref": f"#/definitions/{input}"
+        }
+
+    if input in ("address", "address_port", "CIDR"):
+        return {"type": "string"}
+
+    if input in ("string", "number"):
+        return {"type": input}
+
+    if input == "int/string":
+        return parse_type("int | string")
+
+    if input == "int":
+        return {"type": "integer"}
+
+    if input.startswith("map"):
+        return {"type": "object"}
+
+    if input.startswith('"') and input.endswith('"'):
+        return {"const": input[1:-1]}
+
+    if input.startswith("a list of"):
+        return {}
+
+    if input == "string array":
+        return {"type": "array", "items": {"type": "string"}}
+
+    if input.startswith("string, any of"):
+        return {"type": "string"}
+
+    raise Exception(input)
+
+def main():
+    schema = {
+        "$schema": "http://json-schema.org/draft-07/schema#",
+        "$ref": "#/definitions/Basic Configuration Modules",
+        "definitions": {x['title']: x for x in parse(sys.stdin)},
+    }
+
+    print(json.dumps(schema, indent=2))
+
+if __name__ == "__main__":
+    main()