1 +
import fs from "fs"
2 +
import { JSDOM, DOMWindow } from "jsdom"
3 +
import { CLIEngine } from "eslint"
4 + 5 +
const DataSources = [
6 +
{
7 +
url: "https://www.ecma-international.org/ecma-262/9.0/",
8 +
version: 2018,
9 +
binProperties: "#table-binary-unicode-properties",
10 +
gcValues: "#table-unicode-general-category-values",
11 +
scValues: "#table-unicode-script-values",
12 +
},
13 +
{
14 +
url: "https://www.ecma-international.org/ecma-262/10.0/",
15 +
version: 2019,
16 +
binProperties: "#table-binary-unicode-properties",
17 +
gcValues: "#table-unicode-general-category-values",
18 +
scValues: "#table-unicode-script-values",
19 +
},
20 +
{
21 +
url: "https://tc39.es/ecma262/",
22 +
version: 2020,
23 +
binProperties: "#table-binary-unicode-properties",
24 +
gcValues: "#table-unicode-general-category-values",
25 +
scValues: "#table-unicode-script-values",
26 +
},
27 +
]
28 +
const FILE_PATH = "src/unicode/properties.ts"
29 +
const logger = console
30 + 31 +
type Datum = {
32 +
binProperties: string[]
33 +
gcValues: string[]
34 +
scValues: string[]
35 +
}
36 + 37 +
// Main
38 +
;(async () => {
39 +
const data: Record<number, Datum> = Object.create(null)
40 +
const existing = {
41 +
binProperties: new Set<string>(),
42 +
gcValues: new Set<string>(),
43 +
scValues: new Set<string>(),
44 +
}
45 + 46 +
for (const {
47 +
binProperties,
48 +
gcValues,
49 +
scValues,
50 +
url,
51 +
version,
52 +
} of DataSources) {
53 +
logger.log("---- ECMAScript %d ----", version)
54 +
const datum: Datum = {
55 +
binProperties: [],
56 +
gcValues: [],
57 +
scValues: [],
58 +
}
59 +
data[version] = datum
60 + 61 +
let window: DOMWindow | null = null
62 +
do {
63 +
try {
64 +
logger.log("Fetching data from %o", url)
65 +
;({ window } = await JSDOM.fromURL(url))
66 +
} catch (error) {
67 +
if (!error || error.message !== "Error: socket hang up") {
68 +
throw error
69 +
}
70 +
logger.log("Failed: %s", error)
71 +
await new Promise(resolve => setTimeout(resolve, 2000))
72 +
}
73 +
} while (window == null)
74 + 75 +
logger.log("Parsing tables")
76 +
datum.binProperties = collectValues(
77 +
window,
78 +
binProperties,
79 +
existing.binProperties,
80 +
)
81 +
datum.gcValues = collectValues(window, gcValues, existing.gcValues)
82 +
datum.scValues = collectValues(window, scValues, existing.scValues)
83 + 84 +
logger.log("Done")
85 +
}
86 + 87 +
logger.log("Generating code...")
88 +
let code = `/* This file was generated with ECMAScript specifications. */
89 + 90 +
const gcNamePattern = /^(?:General_Category|gc)$/u
91 +
const scNamePattern = /^(?:Script(?:_Extensions)?|scx?)$/u
92 +
const gcValuePatterns = {
93 +
${Array.from(
94 +
Object.keys(data),
95 +
version => `es${version}: null as RegExp | null,`,
96 +
).join("\n")}
97 +
}
98 +
const scValuePatterns = {
99 +
${Array.from(
100 +
Object.keys(data),
101 +
version => `es${version}: null as RegExp | null,`,
102 +
).join("\n")}
103 +
}
104 +
const binPropertyPatterns = {
105 +
${Array.from(
106 +
Object.keys(data),
107 +
version => `es${version}: null as RegExp | null,`,
108 +
).join("\n")}
109 +
}
110 + 111 +
export function isValidUnicodeProperty(version: number, name: string, value: string): boolean {
112 +
if (gcNamePattern.test(name)) {
113 +
${Array.from(Object.entries(data), ([version, { gcValues }]) =>
114 +
makeVerificationCode(version, "gcValuePatterns", gcValues, 52),
115 +
).join("\n")}
116 +
}
117 +
if (scNamePattern.test(name)) {
118 +
${Array.from(Object.entries(data), ([version, { scValues }]) =>
119 +
makeVerificationCode(version, "scValuePatterns", scValues, 52),
120 +
).join("\n")}
121 +
}
122 +
return false
123 +
}
124 + 125 +
export function isValidLoneUnicodeProperty(version: number, value: string): boolean {
126 +
${Array.from(Object.entries(data), ([version, { binProperties }]) =>
127 +
makeVerificationCode(version, "binPropertyPatterns", binProperties, 56),
128 +
).join("\n")}
129 +
return false
130 +
}
131 +
`
132 + 133 +
logger.log("Formatting code...")
134 +
const engine = new CLIEngine({ fix: true })
135 +
const result = engine.executeOnText(code, "properties.ts").results[0]
136 +
code = result.output || code
137 + 138 +
logger.log("Writing '%s'...", FILE_PATH)
139 +
await save(code)
140 + 141 +
logger.log("Completed!")
142 +
})().catch(error => {
143 +
logger.error(error.stack)
144 +
process.exitCode = 1
145 +
})
146 + 147 +
function collectValues(
148 +
window: Window,
149 +
id: string,
150 +
existingSet: Set<string>,
151 +
): string[] {
152 +
return Array.from(
153 +
window.document.querySelectorAll(`${id} td:nth-child(1) code`),
154 +
node => node.textContent || "",
155 +
)
156 +
.filter(value => {
157 +
if (existingSet.has(value)) {
158 +
return false
159 +
}
160 +
existingSet.add(value)
161 +
return true
162 +
})
163 +
.sort(undefined)
164 +
}
165 + 166 +
function makeVerificationCode(
167 +
version: string,
168 +
patternVar: string,
169 +
values: string[],
170 +
maxLen: number,
171 +
): string {
172 +
if (values.length === 0) {
173 +
return ""
174 +
}
175 + 176 +
return `
177 +
if (version >= ${version}) {
178 +
if (!${patternVar}.es${version}) {
179 +
${patternVar}.es${version} = new RegExp(
180 +
${makeRegExpPatternCode(values, maxLen)},
181 +
"u"
182 +
)
183 +
}
184 +
if (${patternVar}.es${version}.test(value)) {
185 +
return true
186 +
}
187 +
}
188 +
`
189 +
}
190 + 191 +
function makeRegExpPatternCode(names: string[], maxLen: number): string {
192 +
const lines = ["^(?:"]
193 +
for (const name of names) {
194 +
const line = lines[lines.length - 1]
195 +
const part = `${name}|`
196 + 197 +
if (line.length + part.length > maxLen) {
198 +
lines.push(part)
199 +
} else {
200 +
lines[lines.length - 1] += part
201 +
}
202 +
}
203 +
lines[lines.length - 1] = `${lines[lines.length - 1].replace(/\|$/u, "")})$`
204 +
return lines.map(line => `"${line}"`).join("+")
205 +
}
206 + 207 +
function save(content: string): Promise<void> {
208 +
return new Promise((resolve, reject) => {
209 +
fs.writeFile(FILE_PATH, content, error =>
210 +
error ? reject(error) : resolve(),
211 +
)
212 +
})
213 +
}
RetroSearch is an open source project built by @garambo | Open a GitHub Issue
Search and Browse the WWW like it's 1997 | Search results from DuckDuckGo
HTML:
3.2
| Encoding:
UTF-8
| Version:
0.7.4