From c538c151c7462ad0395ff2c15c5e11e89e362aa8 Mon Sep 17 00:00:00 2001
From: Striven <sg.striven@cutecat.club>
Date: Sat, 04 Apr 2026 19:47:16 +0000
Subject: [PATCH] Initial commit
---
quartz/processors/parse.ts | 235 +++++++++++++++++++++++++++++++++++++++-------------------
1 files changed, 158 insertions(+), 77 deletions(-)
diff --git a/quartz/processors/parse.ts b/quartz/processors/parse.ts
index 4fc13f8..1099cd9 100644
--- a/quartz/processors/parse.ts
+++ b/quartz/processors/parse.ts
@@ -1,40 +1,47 @@
-import esbuild from 'esbuild'
-import remarkParse from 'remark-parse'
-import remarkRehype from 'remark-rehype'
+import esbuild from "esbuild"
+import remarkParse from "remark-parse"
+import remarkRehype from "remark-rehype"
import { Processor, unified } from "unified"
-import { Root as MDRoot } from 'remark-parse/lib'
-import { Root as HTMLRoot } from 'hast'
-import { ProcessedContent } from '../plugins/vfile'
-import { PerfTimer } from '../perf'
-import { read } from 'to-vfile'
-import { slugify } from '../path'
-import path from 'path'
-import os from 'os'
-import workerpool, { Promise as WorkerPromise } from 'workerpool'
-import { QuartzTransformerPlugin } from '../plugins/types'
-import { QuartzLogger } from '../log'
-import chalk from 'chalk'
+import { Root as MDRoot } from "remark-parse/lib"
+import { Root as HTMLRoot } from "hast"
+import { MarkdownContent, ProcessedContent } from "../plugins/vfile"
+import { PerfTimer } from "../util/perf"
+import { read } from "to-vfile"
+import { FilePath, QUARTZ, slugifyFilePath } from "../util/path"
+import path from "path"
+import workerpool, { Promise as WorkerPromise } from "workerpool"
+import { QuartzLogger } from "../util/log"
+import { trace } from "../util/trace"
+import { BuildCtx, WorkerSerializableBuildCtx } from "../util/ctx"
+import { styleText } from "util"
-export type QuartzProcessor = Processor<MDRoot, HTMLRoot, void>
-export function createProcessor(transformers: QuartzTransformerPlugin[]): QuartzProcessor {
- // base Markdown -> MD AST
- let processor = unified().use(remarkParse)
+export type QuartzMdProcessor = Processor<MDRoot, MDRoot, MDRoot>
+export type QuartzHtmlProcessor = Processor<undefined, MDRoot, HTMLRoot>
- // MD AST -> MD AST transforms
- for (const plugin of transformers) {
- processor = processor.use(plugin.markdownPlugins())
- }
+export function createMdProcessor(ctx: BuildCtx): QuartzMdProcessor {
+ const transformers = ctx.cfg.plugins.transformers
- // MD AST -> HTML AST
- processor = processor.use(remarkRehype, { allowDangerousHtml: true })
+ return (
+ unified()
+ // base Markdown -> MD AST
+ .use(remarkParse)
+ // MD AST -> MD AST transforms
+ .use(
+ transformers.flatMap((plugin) => plugin.markdownPlugins?.(ctx) ?? []),
+ ) as unknown as QuartzMdProcessor
+ // ^ sadly the typing of `use` is not smart enough to infer the correct type from our plugin list
+ )
+}
-
- // HTML AST -> HTML AST transforms
- for (const plugin of transformers) {
- processor = processor.use(plugin.htmlPlugins())
- }
-
- return processor
+export function createHtmlProcessor(ctx: BuildCtx): QuartzHtmlProcessor {
+ const transformers = ctx.cfg.plugins.transformers
+ return (
+ unified()
+ // MD AST -> HTML AST
+ .use(remarkRehype, { allowDangerousHtml: true })
+ // HTML AST -> HTML AST transforms
+ .use(transformers.flatMap((plugin) => plugin.htmlPlugins?.(ctx) ?? []))
+ )
}
function* chunks<T>(arr: T[], n: number) {
@@ -49,51 +56,63 @@
const fp = "./quartz/worker.ts"
return esbuild.build({
entryPoints: [fp],
- outfile: path.join("quartz", cacheFile),
+ outfile: path.join(QUARTZ, cacheFile),
bundle: true,
keepNames: true,
platform: "node",
format: "esm",
packages: "external",
+ sourcemap: true,
+ sourcesContent: false,
plugins: [
{
- name: 'css-and-scripts-as-text',
+ name: "css-and-scripts-as-text",
setup(build) {
build.onLoad({ filter: /\.scss$/ }, (_) => ({
- contents: '',
- loader: 'text'
+ contents: "",
+ loader: "text",
}))
build.onLoad({ filter: /\.inline\.(ts|js)$/ }, (_) => ({
- contents: '',
- loader: 'text'
+ contents: "",
+ loader: "text",
}))
- }
- }
- ]
+ },
+ },
+ ],
})
}
-export function createFileParser(baseDir: string, fps: string[], verbose: boolean) {
- return async (processor: QuartzProcessor) => {
- const res: ProcessedContent[] = []
+export function createFileParser(ctx: BuildCtx, fps: FilePath[]) {
+ const { argv, cfg } = ctx
+ return async (processor: QuartzMdProcessor) => {
+ const res: MarkdownContent[] = []
for (const fp of fps) {
try {
+ const perf = new PerfTimer()
const file = await read(fp)
+ // strip leading and trailing whitespace
+ file.value = file.value.toString().trim()
+
+ // Text -> Text transforms
+ for (const plugin of cfg.plugins.transformers.filter((p) => p.textTransform)) {
+ file.value = plugin.textTransform!(ctx, file.value.toString())
+ }
+
// base data properties that plugins may use
- file.data.slug = slugify(path.relative(baseDir, file.path))
- file.data.filePath = fp
+ file.data.filePath = file.path as FilePath
+ file.data.relativePath = path.posix.relative(argv.directory, file.path) as FilePath
+ file.data.slug = slugifyFilePath(file.data.relativePath)
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}`)
+ if (argv.verbose) {
+ console.log(`[markdown] ${fp} -> ${file.data.slug} (${perf.timeSince()})`)
}
} catch (err) {
- console.log(chalk.red(`\nFailed to process \`${fp}\`: `) + err)
- process.exit(1)
+ trace(`\nFailed to process markdown \`${fp}\``, err as Error)
}
}
@@ -101,41 +120,103 @@
}
}
-export async function parseMarkdown(transformers: QuartzTransformerPlugin[], baseDir: string, fps: string[], verbose: boolean): Promise<ProcessedContent[]> {
- const perf = new PerfTimer()
- const log = new QuartzLogger(verbose)
+export function createMarkdownParser(ctx: BuildCtx, mdContent: MarkdownContent[]) {
+ return async (processor: QuartzHtmlProcessor) => {
+ const res: ProcessedContent[] = []
+ for (const [ast, file] of mdContent) {
+ try {
+ const perf = new PerfTimer()
- const CHUNK_SIZE = 128
- let concurrency = fps.length < CHUNK_SIZE ? 1 : os.availableParallelism()
- let res: ProcessedContent[] = []
+ const newAst = await processor.run(ast as MDRoot, file)
+ res.push([newAst, file])
- log.start(`Parsing input files using ${concurrency} threads`)
- if (concurrency === 1) {
- // single-thread
- const processor = createProcessor(transformers)
- const parse = createFileParser(baseDir, fps, verbose)
- res = await parse(processor)
- } else {
- await transpileWorkerScript()
- const pool = workerpool.pool(
- './quartz/bootstrap-worker.mjs',
- {
- minWorkers: 'max',
- maxWorkers: concurrency,
- workerType: 'thread'
+ if (ctx.argv.verbose) {
+ console.log(`[html] ${file.data.slug} (${perf.timeSince()})`)
+ }
+ } catch (err) {
+ trace(`\nFailed to process html \`${file.data.filePath}\``, err as Error)
}
- )
-
- const childPromises: WorkerPromise<ProcessedContent[]>[] = []
- for (const chunk of chunks(fps, CHUNK_SIZE)) {
- childPromises.push(pool.exec('parseFiles', [baseDir, chunk, verbose]))
}
- const results: ProcessedContent[][] = await WorkerPromise.all(childPromises)
+ return res
+ }
+}
+
+const clamp = (num: number, min: number, max: number) =>
+ Math.min(Math.max(Math.round(num), min), max)
+
+export async function parseMarkdown(ctx: BuildCtx, fps: FilePath[]): Promise<ProcessedContent[]> {
+ const { argv } = ctx
+ const perf = new PerfTimer()
+ const log = new QuartzLogger(argv.verbose)
+
+ // rough heuristics: 128 gives enough time for v8 to JIT and optimize parsing code paths
+ const CHUNK_SIZE = 128
+ const concurrency = ctx.argv.concurrency ?? clamp(fps.length / CHUNK_SIZE, 1, 4)
+
+ let res: ProcessedContent[] = []
+ log.start(`Parsing input files using ${concurrency} threads`)
+ if (concurrency === 1) {
+ try {
+ const mdRes = await createFileParser(ctx, fps)(createMdProcessor(ctx))
+ res = await createMarkdownParser(ctx, mdRes)(createHtmlProcessor(ctx))
+ } catch (error) {
+ log.end()
+ throw error
+ }
+ } else {
+ await transpileWorkerScript()
+ const pool = workerpool.pool("./quartz/bootstrap-worker.mjs", {
+ minWorkers: "max",
+ maxWorkers: concurrency,
+ workerType: "thread",
+ })
+ const errorHandler = (err: any) => {
+ console.error(err)
+ process.exit(1)
+ }
+
+ const serializableCtx: WorkerSerializableBuildCtx = {
+ buildId: ctx.buildId,
+ argv: ctx.argv,
+ allSlugs: ctx.allSlugs,
+ allFiles: ctx.allFiles,
+ incremental: ctx.incremental,
+ }
+
+ const textToMarkdownPromises: WorkerPromise<MarkdownContent[]>[] = []
+ let processedFiles = 0
+ for (const chunk of chunks(fps, CHUNK_SIZE)) {
+ textToMarkdownPromises.push(pool.exec("parseMarkdown", [serializableCtx, chunk]))
+ }
+
+ const mdResults: Array<MarkdownContent[]> = await Promise.all(
+ textToMarkdownPromises.map(async (promise) => {
+ const result = await promise
+ processedFiles += result.length
+ log.updateText(`text->markdown ${styleText("gray", `${processedFiles}/${fps.length}`)}`)
+ return result
+ }),
+ ).catch(errorHandler)
+
+ const markdownToHtmlPromises: WorkerPromise<ProcessedContent[]>[] = []
+ processedFiles = 0
+ for (const mdChunk of mdResults) {
+ markdownToHtmlPromises.push(pool.exec("processHtml", [serializableCtx, mdChunk]))
+ }
+ const results: ProcessedContent[][] = await Promise.all(
+ markdownToHtmlPromises.map(async (promise) => {
+ const result = await promise
+ processedFiles += result.length
+ log.updateText(`markdown->html ${styleText("gray", `${processedFiles}/${fps.length}`)}`)
+ return result
+ }),
+ ).catch(errorHandler)
+
res = results.flat()
await pool.terminate()
}
- log.success(`Parsed ${res.length} Markdown files in ${perf.timeSince()}`)
+ log.end(`Parsed ${res.length} Markdown files in ${perf.timeSince()}`)
return res
}
--
Gitblit v1.10.0