From 08f8e3b4a4879dd7c91c16fbce80c4f2bc5e357f Mon Sep 17 00:00:00 2001
From: Jacky Zhao <j.zhao2k19@gmail.com>
Date: Mon, 10 Jul 2023 02:32:24 +0000
Subject: [PATCH] docs + various polish
---
quartz/processors/parse.ts | 139 +++++++++++++++++++++++++++++++++++++++-------
1 files changed, 117 insertions(+), 22 deletions(-)
diff --git a/quartz/processors/parse.ts b/quartz/processors/parse.ts
index b7ccdd2..9947860 100644
--- a/quartz/processors/parse.ts
+++ b/quartz/processors/parse.ts
@@ -1,3 +1,4 @@
+import esbuild from 'esbuild'
import remarkParse from 'remark-parse'
import remarkRehype from 'remark-rehype'
import { Processor, unified } from "unified"
@@ -8,16 +9,20 @@
import { read } from 'to-vfile'
import { slugify } from '../path'
import path from 'path'
-import { QuartzTransformerPlugin } from '../plugins/types'
+import os from 'os'
+import workerpool, { Promise as WorkerPromise } from 'workerpool'
+import { QuartzTransformerPluginInstance } from '../plugins/types'
+import { QuartzLogger } from '../log'
+import chalk from 'chalk'
export type QuartzProcessor = Processor<MDRoot, HTMLRoot, void>
-export function createProcessor(transformers: QuartzTransformerPlugin[]): any {
+export function createProcessor(transformers: QuartzTransformerPluginInstance[]): QuartzProcessor {
// base Markdown -> MD AST
let processor = unified().use(remarkParse)
// MD AST -> MD AST transforms
- for (const plugin of transformers) {
- processor = processor.use(plugin.markdownPlugins())
+ for (const plugin of transformers.filter(p => p.markdownPlugins)) {
+ processor = processor.use(plugin.markdownPlugins!())
}
// MD AST -> HTML AST
@@ -25,34 +30,124 @@
// HTML AST -> HTML AST transforms
- for (const plugin of transformers) {
- processor = processor.use(plugin.htmlPlugins())
+ for (const plugin of transformers.filter(p => p.htmlPlugins)) {
+ processor = processor.use(plugin.htmlPlugins!())
}
return processor
}
-export async function parseMarkdown(processor: QuartzProcessor, baseDir: string, fps: string[], verbose: boolean): Promise<ProcessedContent[]> {
- const perf = new PerfTimer()
- const res: ProcessedContent[] = []
- for (const fp of fps) {
- const file = await read(fp)
+function* chunks<T>(arr: T[], n: number) {
+ for (let i = 0; i < arr.length; i += n) {
+ yield arr.slice(i, i + n)
+ }
+}
- // base data properties that plugins may use
- file.data.slug = slugify(path.relative(baseDir, file.path))
- file.data.filePath = fp
+async function transpileWorkerScript() {
+ // transpile worker script
+ const cacheFile = "./.quartz-cache/transpiled-worker.mjs"
+ const fp = "./quartz/worker.ts"
+ return esbuild.build({
+ entryPoints: [fp],
+ outfile: path.join("quartz", cacheFile),
+ bundle: true,
+ keepNames: true,
+ platform: "node",
+ format: "esm",
+ packages: "external",
+ plugins: [
+ {
+ name: 'css-and-scripts-as-text',
+ setup(build) {
+ build.onLoad({ filter: /\.scss$/ }, (_) => ({
+ contents: '',
+ loader: 'text'
+ }))
+ build.onLoad({ filter: /\.inline\.(ts|js)$/ }, (_) => ({
+ contents: '',
+ loader: 'text'
+ }))
+ }
+ }
+ ]
+ })
+}
- const ast = processor.parse(file)
- res.push([await processor.run(ast, file), file])
+export function createFileParser(transformers: QuartzTransformerPluginInstance[], baseDir: string, fps: string[], allSlugs: string[], verbose: boolean) {
+ return async (processor: QuartzProcessor) => {
+ const res: ProcessedContent[] = []
+ for (const fp of fps) {
+ try {
+ const file = await read(fp)
- if (verbose) {
- console.log(`[process] ${fp} -> ${file.data.slug}`)
+ // strip leading and trailing whitespace
+ file.value = file.value.toString().trim()
+
+ // Text -> Text transforms
+ for (const plugin of transformers.filter(p => p.textTransform)) {
+ file.value = plugin.textTransform!(file.value)
+ }
+
+ // base data properties that plugins may use
+ file.data.slug = slugify(path.relative(baseDir, file.path))
+ file.data.allSlugs = allSlugs
+ file.data.filePath = fp
+
+ const ast = processor.parse(file)
+ const newAst = await processor.run(ast, file)
+ res.push([newAst, file])
+
+ if (verbose) {
+ console.log(`[process] ${fp} -> ${file.data.slug}`)
+ }
+ } catch (err) {
+ console.log(chalk.red(`\nFailed to process \`${fp}\`: `) + err)
+ process.exit(1)
+ }
}
+
+ return res
+ }
+}
+
+export async function parseMarkdown(transformers: QuartzTransformerPluginInstance[], baseDir: string, fps: string[], verbose: boolean): Promise<ProcessedContent[]> {
+ const perf = new PerfTimer()
+ const log = new QuartzLogger(verbose)
+
+ const CHUNK_SIZE = 128
+ let concurrency = fps.length < CHUNK_SIZE ? 1 : os.availableParallelism()
+
+ // get all slugs ahead of time as each thread needs a copy
+ // const slugs: string[] = fps.map(fp => slugify(path))
+ const allSlugs = fps.map(fp => slugify(path.relative(baseDir, path.resolve(fp))))
+
+ let res: ProcessedContent[] = []
+ log.start(`Parsing input files using ${concurrency} threads`)
+ if (concurrency === 1) {
+ const processor = createProcessor(transformers)
+ const parse = createFileParser(transformers, baseDir, fps, allSlugs, verbose)
+ res = await parse(processor)
+ } else {
+ await transpileWorkerScript()
+ const pool = workerpool.pool(
+ './quartz/bootstrap-worker.mjs',
+ {
+ minWorkers: 'max',
+ maxWorkers: concurrency,
+ workerType: 'thread'
+ }
+ )
+
+ const childPromises: WorkerPromise<ProcessedContent[]>[] = []
+ for (const chunk of chunks(fps, CHUNK_SIZE)) {
+ childPromises.push(pool.exec('parseFiles', [baseDir, chunk, allSlugs, verbose]))
+ }
+
+ const results: ProcessedContent[][] = await WorkerPromise.all(childPromises)
+ res = results.flat()
+ await pool.terminate()
}
- if (verbose) {
- console.log(`Parsed and transformed ${res.length} Markdown files in ${perf.timeSince()}`)
- }
-
+ log.success(`Parsed ${res.length} Markdown files in ${perf.timeSince()}`)
return res
}
--
Gitblit v1.10.0