🐛 fix valid Unicode Properties (fixes #6)

mysticatea · Jul 23, 2019 · 2bd358f · 2bd358f
1 parent 00abeab
commit 2bd358f
Showing 6 changed files with 418 additions and 478 deletions.
diff --git a/package.json b/package.json
@@ -13,11 +13,13 @@
   "devDependencies": {
     "@mysticatea/eslint-plugin": "^11.0.0",
     "@types/eslint": "^4.16.2",
+    "@types/jsdom": "^12.2.4",
     "@types/mocha": "^5.2.2",
     "@types/node": "^12.6.8",
     "coveralls": "^3.0.1",
     "dts-bundle": "^0.7.3",
     "eslint": "^6.1.0",
+    "jsdom": "^15.1.1",
     "mocha": "^6.2.0",
     "npm-run-all": "^4.1.5",
     "nyc": "^14.1.1",
@@ -40,7 +42,9 @@
     "pretest": "run-s build lint",
     "test": "nyc _mocha \"test/*.ts\" --reporter dot --timeout 10000",
     "update:test": "ts-node scripts/update-fixtures.ts",
-    "update:ids": "ts-node scripts/update-unicode-ids.ts",
+    "update:unicode": "run-s update:unicode:*",
+    "update:unicode:ids": "ts-node scripts/update-unicode-ids.ts",
+    "update:unicode:props": "ts-node scripts/update-unicode-properties.ts",
     "preversion": "npm test",
     "version": "npm run -s build",
     "postversion": "git push && git push --tags",

diff --git a/scripts/update-unicode-properties.ts b/scripts/update-unicode-properties.ts
@@ -0,0 +1,213 @@
+import fs from "fs"
+import { JSDOM, DOMWindow } from "jsdom"
+import { CLIEngine } from "eslint"
+
+const DataSources = [
+    {
+        url: "https://www.ecma-international.org/ecma-262/9.0/",
+        version: 2018,
+        binProperties: "#table-binary-unicode-properties",
+        gcValues: "#table-unicode-general-category-values",
+        scValues: "#table-unicode-script-values",
+    },
+    {
+        url: "https://www.ecma-international.org/ecma-262/10.0/",
+        version: 2019,
+        binProperties: "#table-binary-unicode-properties",
+        gcValues: "#table-unicode-general-category-values",
+        scValues: "#table-unicode-script-values",
+    },
+    {
+        url: "https://tc39.es/ecma262/",
+        version: 2020,
+        binProperties: "#table-binary-unicode-properties",
+        gcValues: "#table-unicode-general-category-values",
+        scValues: "#table-unicode-script-values",
+    },
+]
+const FILE_PATH = "src/unicode/properties.ts"
+const logger = console
+
+type Datum = {
+    binProperties: string[]
+    gcValues: string[]
+    scValues: string[]
+}
+
+// Main
+;(async () => {
+    const data: Record<number, Datum> = Object.create(null)
+    const existing = {
+        binProperties: new Set<string>(),
+        gcValues: new Set<string>(),
+        scValues: new Set<string>(),
+    }
+
+    for (const {
+        binProperties,
+        gcValues,
+        scValues,
+        url,
+        version,
+    } of DataSources) {
+        logger.log("---- ECMAScript %d ----", version)
+        const datum: Datum = {
+            binProperties: [],
+            gcValues: [],
+            scValues: [],
+        }
+        data[version] = datum
+
+        let window: DOMWindow | null = null
+        do {
+            try {
+                logger.log("Fetching data from %o", url)
+                ;({ window } = await JSDOM.fromURL(url))
+            } catch (error) {
+                if (!error || error.message !== "Error: socket hang up") {
+                    throw error
+                }
+                logger.log("Failed: %s", error)
+                await new Promise(resolve => setTimeout(resolve, 2000))
+            }
+        } while (window == null)
+
+        logger.log("Parsing tables")
+        datum.binProperties = collectValues(
+            window,
+            binProperties,
+            existing.binProperties,
+        )
+        datum.gcValues = collectValues(window, gcValues, existing.gcValues)
+        datum.scValues = collectValues(window, scValues, existing.scValues)
+
+        logger.log("Done")
+    }
+
+    logger.log("Generating code...")
+    let code = `/* This file was generated with ECMAScript specifications. */
+
+const gcNamePattern = /^(?:General_Category|gc)$/u
+const scNamePattern = /^(?:Script(?:_Extensions)?|scx?)$/u
+const gcValuePatterns = {
+    ${Array.from(
+        Object.keys(data),
+        version => `es${version}: null as RegExp | null,`,
+    ).join("\n")}
+}
+const scValuePatterns = {
+    ${Array.from(
+        Object.keys(data),
+        version => `es${version}: null as RegExp | null,`,
+    ).join("\n")}
+}
+const binPropertyPatterns = {
+    ${Array.from(
+        Object.keys(data),
+        version => `es${version}: null as RegExp | null,`,
+    ).join("\n")}
+}
+
+export function isValidUnicodeProperty(version: number, name: string, value: string): boolean {
+    if (gcNamePattern.test(name)) {
+        ${Array.from(Object.entries(data), ([version, { gcValues }]) =>
+            makeVerificationCode(version, "gcValuePatterns", gcValues, 52),
+        ).join("\n")}
+    }
+    if (scNamePattern.test(name)) {
+        ${Array.from(Object.entries(data), ([version, { scValues }]) =>
+            makeVerificationCode(version, "scValuePatterns", scValues, 52),
+        ).join("\n")}
+    }
+    return false
+}
+
+export function isValidLoneUnicodeProperty(version: number, value: string): boolean {
+    ${Array.from(Object.entries(data), ([version, { binProperties }]) =>
+        makeVerificationCode(version, "binPropertyPatterns", binProperties, 56),
+    ).join("\n")}
+    return false
+}
+`
+
+    logger.log("Formatting code...")
+    const engine = new CLIEngine({ fix: true })
+    const result = engine.executeOnText(code, "properties.ts").results[0]
+    code = result.output || code
+
+    logger.log("Writing '%s'...", FILE_PATH)
+    await save(code)
+
+    logger.log("Completed!")
+})().catch(error => {
+    logger.error(error.stack)
+    process.exitCode = 1
+})
+
+function collectValues(
+    window: Window,
+    id: string,
+    existingSet: Set<string>,
+): string[] {
+    return Array.from(
+        window.document.querySelectorAll(`${id} td:nth-child(1) code`),
+        node => node.textContent || "",
+    )
+        .filter(value => {
+            if (existingSet.has(value)) {
+                return false
+            }
+            existingSet.add(value)
+            return true
+        })
+        .sort(undefined)
+}
+
+function makeVerificationCode(
+    version: string,
+    patternVar: string,
+    values: string[],
+    maxLen: number,
+): string {
+    if (values.length === 0) {
+        return ""
+    }
+
+    return `
+        if (version >= ${version}) {
+            if (!${patternVar}.es${version}) {
+                ${patternVar}.es${version} = new RegExp(
+                    ${makeRegExpPatternCode(values, maxLen)},
+                    "u"
+                )
+            }
+            if (${patternVar}.es${version}.test(value)) {
+                return true
+            }
+        }
+    `
+}
+
+function makeRegExpPatternCode(names: string[], maxLen: number): string {
+    const lines = ["^(?:"]
+    for (const name of names) {
+        const line = lines[lines.length - 1]
+        const part = `${name}|`
+
+        if (line.length + part.length > maxLen) {
+            lines.push(part)
+        } else {
+            lines[lines.length - 1] += part
+        }
+    }
+    lines[lines.length - 1] = `${lines[lines.length - 1].replace(/\|$/u, "")})$`
+    return lines.map(line => `"${line}"`).join("+")
+}
+
+function save(content: string): Promise<void> {
+    return new Promise((resolve, reject) => {
+        fs.writeFile(FILE_PATH, content, error =>
+            error ? reject(error) : resolve(),
+        )
+    })
+}
diff --git a/src/unicode/index.ts b/src/unicode/index.ts
@@ -1,5 +1,8 @@
 export { isIdContinue, isIdStart } from "./ids"
-export { PropertyData } from "./property-data"
+export {
+    isValidLoneUnicodeProperty,
+    isValidUnicodeProperty,
+} from "./properties"
 
 export const Null = 0x00
 export const Backspace = 0x08