feat: Add schema validation and data combination script

.gitignore

@@ -0,0 +1,4 @@
+node_modules
+wpt-repo
+surveys
+web-features-mappings.combined.json

README.md

@@ -45,7 +45,7 @@ Examples of data sources which already map to web-features IDs and for which we
 
 ## The scripts folder
 
-The `/script/` folder contains the JavaScript files that are responsible for updating the mapping files.
+The `/scripts/` folder contains the JavaScript files that are responsible for updating the mapping files.
 
 To run these scripts:
 
@@ -63,6 +63,20 @@ The mappings are JSON files that are formatted as follows:
 }
 ```
 
+## Combined data
+
+The `combine` script generates a `web-features-mappings.combined.json` file in the root of the repository. This file contains all the mapping data from the `mappings` folder, combined into a single file.
+
+The format of the combined file is as follows:
+
+```json
+{
+  "chrome-use-counters": { ... },
+  "interop": { ... },
+  ...
+}
+```
+
 ## TODO
 
 * Add mapping to origin trials.

schemas.json

@@ -0,0 +1,303 @@
+{
+  "$schema": "https://json-schema.org/draft/2020-12/schema",
+  "title": "Web Feature Mappings Data",
+  "description": "A collection of schemas for each of the web feature mapping files.",
+  "type": "object",
+  "properties": {
+    "chrome-use-counters": {
+      "$ref": "#/$defs/chromeUseCountersFile"
+    },
+    "interop": {
+      "$ref": "#/$defs/interopFile"
+    },
+    "mdn-docs": {
+      "$ref": "#/$defs/mdnDocsFile"
+    },
+    "standards-positions": {
+      "$ref": "#/$defs/standardsPositionsFile"
+    },
+    "state-of-surveys": {
+      "$ref": "#/$defs/stateOfSurveysFile"
+    },
+    "wpt": {
+      "$ref": "#/$defs/wptFile"
+    }
+  },
+  "$comment": "The properties are the file names of the files in the mapping directory",
+  "additionalProperties": false,
+  "$defs": {
+    "chromeUseCountersFile": {
+      "description": "Schema for chrome-use-counters.json. An object where keys are feature IDs and values are usage metrics.",
+      "type": "object",
+      "patternProperties": {
+        ".*": {
+          "$ref": "#/$defs/useCounterValue"
+        }
+      },
+      "additionalProperties": false
+    },
+    "interopFile": {
+      "description": "Schema for interop.json. An object where keys are feature IDs and values are an array of interop records.",
+      "type": "object",
+      "patternProperties": {
+        ".*": {
+          "$ref": "#/$defs/interopHistory"
+        }
+      },
+      "additionalProperties": false
+    },
+    "mdnDocsFile": {
+      "description": "Schema for mdn-docs.json. An object where keys are feature IDs and values are an array of MDN documentation links.",
+      "type": "object",
+      "patternProperties": {
+        ".*": {
+          "$ref": "#/$defs/mdnLinkArray"
+        }
+      },
+      "additionalProperties": false
+    },
+    "standardsPositionsFile": {
+      "description": "Schema for standards-positions.json. An object where keys are feature IDs and values are an array of standards position records.",
+      "type": "object",
+      "patternProperties": {
+        ".*": {
+          "$ref": "#/$defs/standardsPositionArray"
+        }
+      },
+      "additionalProperties": false
+    },
+    "stateOfSurveysFile": {
+      "description": "Schema for state-of-surveys.json. An object where keys are feature IDs and values are an array of survey records.",
+      "type": "object",
+      "patternProperties": {
+        ".*": {
+          "$ref": "#/$defs/surveyRecordArray"
+        }
+      },
+      "additionalProperties": false
+    },
+    "wptFile": {
+      "description": "Schema for wpt.json. An object where keys are feature IDs and values are a record of Web Platform Tests.",
+      "type": "object",
+      "patternProperties": {
+        ".*": {
+          "$ref": "#/$defs/wptRecord"
+        }
+      },
+      "additionalProperties": false
+    },
+    "useCounterValue": {
+      "type": "object",
+      "description": "The usage metrics for a single web feature from chrome-use-counters.json.",
+      "properties": {
+        "percentageOfPageLoad": {
+          "type": "number"
+        },
+        "url": {
+          "type": "string",
+          "format": "uri"
+        }
+      },
+      "required": [
+        "percentageOfPageLoad",
+        "url"
+      ],
+      "additionalProperties": false
+    },
+    "interopHistory": {
+      "description": "An array of Interop records from interop.json.",
+      "type": "array",
+      "items": {
+        "$ref": "#/$defs/interopRecord"
+      },
+      "minItems": 1
+    },
+    "interopRecord": {
+      "type": "object",
+      "properties": {
+        "year": {
+          "type": "integer"
+        },
+        "label": {
+          "type": "string"
+        },
+        "url": {
+          "type": "string",
+          "format": "uri"
+        }
+      },
+      "required": [
+        "year",
+        "label",
+        "url"
+      ],
+      "additionalProperties": false
+    },
+    "mdnLinkArray": {
+      "description": "An array of MDN documentation links from mdn-docs.json.",
+      "type": "array",
+      "items": {
+        "$ref": "#/$defs/mdnLink"
+      },
+      "minItems": 1
+    },
+    "mdnLink": {
+      "type": "object",
+      "properties": {
+        "slug": {
+          "type": "string"
+        },
+        "title": {
+          "type": "string"
+        },
+        "anchor": {
+          "type": [
+            "string",
+            "null"
+          ]
+        },
+        "url": {
+          "type": "string",
+          "format": "uri"
+        }
+      },
+      "required": [
+        "slug",
+        "title",
+        "anchor",
+        "url"
+      ],
+      "additionalProperties": false
+    },
+    "standardsPositionArray": {
+      "description": "An array of standards position records from standards-positions.json.",
+      "type": "array",
+      "items": {
+        "$ref": "#/$defs/standardsPositionRecord"
+      },
+      "minItems": 1
+    },
+    "standardsPositionRecord": {
+      "type": "object",
+      "properties": {
+        "vendor": {
+          "type": "string",
+          "enum": [
+            "mozilla",
+            "webkit"
+          ]
+        },
+        "url": {
+          "type": "string",
+          "format": "uri"
+        },
+        "position": {
+          "type": "string",
+          "enum": [
+            "",
+            "blocked",
+            "defer",
+            "negative",
+            "neutral",
+            "oppose",
+            "positive",
+            "support"
+          ]
+        },
+        "concerns": {
+          "type": "array",
+          "items": {
+            "type": "string",
+            "enum": [
+              "API design",
+              "accessibility",
+              "annoyance",
+              "compatibility",
+              "complexity",
+              "dependencies",
+              "device independence",
+              "duplication",
+              "integration",
+              "internationalization",
+              "interoperability",
+              "maintenance",
+              "performance",
+              "portability",
+              "power",
+              "privacy",
+              "security",
+              "use cases",
+              "venue"
+            ]
+          }
+        }
+      },
+      "required": [
+        "vendor",
+        "url",
+        "position",
+        "concerns"
+      ],
+      "additionalProperties": false
+    },
+    "surveyRecordArray": {
+      "description": "An array of survey records from state-of-surveys.json.",
+      "type": "array",
+      "items": {
+        "$ref": "#/$defs/surveyRecord"
+      },
+      "minItems": 1
+    },
+    "surveyRecord": {
+      "type": "object",
+      "properties": {
+        "name": {
+          "type": "string"
+        },
+        "url": {
+          "type": "string",
+          "format": "uri"
+        },
+        "question": {
+          "type": "string"
+        },
+        "subQuestion": {
+          "type": "string"
+        },
+        "path": {
+          "type": "string"
+        }
+      },
+      "required": [
+        "name",
+        "url",
+        "question",
+        "subQuestion",
+        "path"
+      ],
+      "additionalProperties": false
+    },
+    "wptRecord": {
+      "type": "object",
+      "description": "A record of Web Platform Tests from wpt.json.",
+      "properties": {
+        "url": {
+          "type": "string",
+          "format": "uri"
+        },
+        "tests": {
+          "type": "array",
+          "items": {
+            "type": "string"
+          },
+          "minItems": 1
+        }
+      },
+      "required": [
+        "url",
+        "tests"
+      ],
+      "additionalProperties": false
+    }
+  }
+}

scripts/combine-mappings.js

@@ -0,0 +1,30 @@
+import fs from "fs/promises";
+import path from "path";
+import { glob } from "glob";
+import { fileURLToPath } from "url";
+
+const __dirname = path.dirname(fileURLToPath(import.meta.url));
+
+async function main() {
+  console.log("Combining mapping files...");
+
+  const mappingFiles = await glob(path.join(__dirname, "../mappings/*.json"));
+
+  const combinedData = {};
+
+  for (const file of mappingFiles) {
+    const fileKey = path.basename(file, ".json");
+    const data = JSON.parse(await fs.readFile(file, "utf-8"));
+    combinedData[fileKey] = data;
+  }
+
+  const outputFile = path.join(__dirname, "../web-features-mappings.combined.json");
+  await fs.writeFile(outputFile, JSON.stringify(combinedData, null, 2));
+
+  console.log(`Combined data written to ${outputFile}`);
+}
+
+main().catch((err) => {
+  console.error(err);
+  process.exit(1);
+});