|
| 1 | +import fs from 'fs'; |
| 2 | +import { readFile, rename } from 'fs/promises'; |
| 3 | +import pathlib from 'path'; |
| 4 | +import { pipeline } from 'stream/promises'; |
| 5 | +import { parseArgs } from 'util'; |
| 6 | +import { JSDOM } from 'jsdom'; |
| 7 | +import { RewritingStream } from 'parse5-html-rewriting-stream'; |
| 8 | +import tmp from 'tmp'; |
| 9 | + |
| 10 | +const { positionals: cliArgs } = parseArgs({ |
| 11 | + allowPositionals: true, |
| 12 | + options: {}, |
| 13 | +}); |
| 14 | +if (cliArgs.length < 3) { |
| 15 | + const self = pathlib.relative(process.cwd(), process.argv[1]); |
| 16 | + console.error(`Usage: node ${self} <template.html> <data.json> <file.html>... |
| 17 | +
|
| 18 | +{{identifier}} substrings in template.html are replaced from data.json, then |
| 19 | +the result is inserted into each file.html.`); |
| 20 | + process.exit(64); |
| 21 | +} |
| 22 | + |
| 23 | +const main = async ([templateFile, dataFile, ...files]) => { |
| 24 | + // Substitute data into the template. |
| 25 | + const [ |
| 26 | + template, |
| 27 | + data, |
| 28 | + ] = await Promise.all([ |
| 29 | + readFile(templateFile, 'utf8'), |
| 30 | + readFile(dataFile, 'utf8').then(JSON.parse), |
| 31 | + ]) |
| 32 | + const formatErrors = []; |
| 33 | + const placeholderPatt = /[{][{](?:([\p{ID_Start}$_][\p{ID_Continue}$]*)[}][}]|.*?(?:[}][}]|(?=[{][{])|$))/gsu; |
| 34 | + const resolved = template.replaceAll(placeholderPatt, (m, name, i) => { |
| 35 | + if (!name) { |
| 36 | + const trunc = m.replace(/([^\n]{29}(?!$)|[^\n]{,29}(?=\n)).*/s, '$1…'); |
| 37 | + formatErrors.push(new SyntaxError(`bad placeholder at index ${i}: ${trunc}`)); |
| 38 | + } else if (!Object.hasOwn(data, name)) { |
| 39 | + formatErrors.push(new ReferenceError(`no data for ${m}`)); |
| 40 | + } |
| 41 | + return data[name]; |
| 42 | + }); |
| 43 | + if (formatErrors.length > 0) { |
| 44 | + throw new AggregateError(formatErrors); |
| 45 | + } |
| 46 | + |
| 47 | + // Parse the template into DOM nodes for appending to page head (metadata such |
| 48 | + // as <style> elements) or prepending to page body (everything else). |
| 49 | + const jsdomOpts = { contentType: 'text/html; charset=utf-8' }; |
| 50 | + const { document } = new JSDOM(resolved, jsdomOpts).window; |
| 51 | + const headHTML = document.head.innerHTML; |
| 52 | + const bodyHTML = document.body.innerHTML; |
| 53 | + |
| 54 | + // Perform the insertions. |
| 55 | + const work = files.map(async (file) => { |
| 56 | + await null; |
| 57 | + const { name: tmpName, fd, removeCallback } = tmp.fileSync({ |
| 58 | + tmpdir: pathlib.dirname(file), |
| 59 | + prefix: pathlib.basename(file), |
| 60 | + postfix: '.tmp', |
| 61 | + detachDescriptor: true, |
| 62 | + }); |
| 63 | + try { |
| 64 | + // Make a pipeline: fileReader -> inserter -> finisher -> fileWriter |
| 65 | + const fileReader = fs.createReadStream(file, 'utf8'); |
| 66 | + const fileWriter = fs.createWriteStream('', { fd, flush: true }); |
| 67 | + |
| 68 | + // Insert headHTML at the end of a possibly implied head, and bodyHTML at |
| 69 | + // the beginning of a possibly implied body. |
| 70 | + // https://html.spec.whatwg.org/multipage/parsing.html#parsing-main-inhtml |
| 71 | + let mode = 'before html'; // | 'before head' | 'in head' | 'after head' | '$DONE' |
| 72 | + const stayInHead = new Set([ |
| 73 | + 'base', |
| 74 | + 'basefont', |
| 75 | + 'bgsound', |
| 76 | + 'head', |
| 77 | + 'link', |
| 78 | + 'meta', |
| 79 | + 'noframes', |
| 80 | + 'noscript', |
| 81 | + 'script', |
| 82 | + 'style', |
| 83 | + 'template', |
| 84 | + 'title', |
| 85 | + ]); |
| 86 | + const inserter = new RewritingStream(); |
| 87 | + const onEndTag = function (tag) { |
| 88 | + if (tag.tagName === 'head') { |
| 89 | + this.emitRaw(headHTML); |
| 90 | + mode = 'after head'; |
| 91 | + } |
| 92 | + this.emitEndTag(tag); |
| 93 | + }; |
| 94 | + const onStartTag = function (tag) { |
| 95 | + const echoTag = () => this.emitStartTag(tag); |
| 96 | + if (mode === 'before html' && tag.tagName === 'html') { |
| 97 | + mode = 'before head'; |
| 98 | + } else if (mode !== 'after head' && stayInHead.has(tag.tagName)) { |
| 99 | + mode = 'in head'; |
| 100 | + } else { |
| 101 | + if (mode !== 'after head') { this.emitRaw(headHTML); } |
| 102 | + // Emit either `${bodyTag}${bodyHTML}` or `${bodyHTML}${otherTag}`. |
| 103 | + const emits = [echoTag, () => this.emitRaw(bodyHTML)]; |
| 104 | + if (tag.tagName !== 'body') { emits.reverse(); } |
| 105 | + for (const emit of emits) { emit(); } |
| 106 | + mode = '$DONE'; |
| 107 | + this.off('endTag', onEndTag).off('startTag', onStartTag); |
| 108 | + return; |
| 109 | + } |
| 110 | + echoTag(); |
| 111 | + }; |
| 112 | + inserter.on('endTag', onEndTag).on('startTag', onStartTag); |
| 113 | + |
| 114 | + // Ensure headHTML/bodyHTML insertion before EOF. |
| 115 | + const finisher = async function* (source) { |
| 116 | + for await (const chunk of source) { yield chunk; } |
| 117 | + if (mode === '$DONE') { return; } |
| 118 | + if (mode !== 'after head') { yield headHTML; } |
| 119 | + yield bodyHTML; |
| 120 | + }; |
| 121 | + |
| 122 | + await pipeline(fileReader, inserter, finisher, fileWriter); |
| 123 | + |
| 124 | + // Now that the temp file is complete, overwrite the source file. |
| 125 | + await rename(tmpName, file); |
| 126 | + } finally { |
| 127 | + removeCallback(); |
| 128 | + } |
| 129 | + }); |
| 130 | + const results = await Promise.allSettled(work); |
| 131 | + |
| 132 | + const failures = results.filter((result) => result.status !== 'fulfilled'); |
| 133 | + if (failures.length > 0) { |
| 134 | + throw new AggregateError(failures.map((r) => r.reason)); |
| 135 | + } |
| 136 | +}; |
| 137 | + |
| 138 | +main(cliArgs).catch((err) => { |
| 139 | + console.error(err); |
| 140 | + process.exit(1); |
| 141 | +}); |
0 commit comments