From 3ac6b42e16dca5a44ed3fec2c0314f1dbbc2322b Mon Sep 17 00:00:00 2001
From: Jacky Zhao <j.zhao2k19@gmail.com>
Date: Sun, 16 Jul 2023 06:02:12 +0000
Subject: [PATCH] finish path refactoring, add sourcemap + better trace support
---
quartz/processors/parse.ts | 98 ++++++++++++++++++++++++++++++++-----------------
1 files changed, 64 insertions(+), 34 deletions(-)
diff --git a/quartz/processors/parse.ts b/quartz/processors/parse.ts
index 715a4e9..a782475 100644
--- a/quartz/processors/parse.ts
+++ b/quartz/processors/parse.ts
@@ -7,20 +7,23 @@
import { ProcessedContent } from '../plugins/vfile'
import { PerfTimer } from '../perf'
import { read } from 'to-vfile'
-import { slugify } from '../path'
+import { FilePath, ServerSlug, slugifyFilePath } from '../path'
import path from 'path'
import os from 'os'
import workerpool, { Promise as WorkerPromise } from 'workerpool'
-import { QuartzTransformerPlugin } from '../plugins/types'
+import { QuartzTransformerPluginInstance } from '../plugins/types'
+import { QuartzLogger } from '../log'
+import chalk from 'chalk'
+import { trace } from '../trace'
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
@@ -28,8 +31,8 @@
// 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
@@ -41,15 +44,11 @@
}
}
-async function transpileWorkerScript(verbose: boolean) {
+async function transpileWorkerScript() {
// transpile worker script
const cacheFile = "./.quartz-cache/transpiled-worker.mjs"
const fp = "./quartz/worker.ts"
- if (verbose) {
- console.log("Transpiling worker script")
- }
-
- await esbuild.build({
+ return esbuild.build({
entryPoints: [fp],
outfile: path.join("quartz", cacheFile),
bundle: true,
@@ -75,31 +74,61 @@
})
}
-export async function parseMarkdown(transformers: QuartzTransformerPlugin[], baseDir: string, fps: string[], verbose: boolean): Promise<ProcessedContent[]> {
+export function createFileParser(transformers: QuartzTransformerPluginInstance[], baseDir: string, fps: FilePath[], allSlugs: ServerSlug[], verbose: boolean) {
+ return async (processor: QuartzProcessor) => {
+ const res: ProcessedContent[] = []
+ for (const fp of fps) {
+ try {
+ const file = await read(fp)
+
+ // 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 = slugifyFilePath(path.relative(baseDir, file.path) as FilePath)
+ 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} -> ${file.data.slug}`)
+ }
+ } catch (err) {
+ trace(`\nFailed to process \`${fp}\``, err as Error)
+ process.exit(1)
+ }
+ }
+
+ return res
+ }
+}
+
+export async function parseMarkdown(transformers: QuartzTransformerPluginInstance[], baseDir: string, fps: FilePath[], 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()
- const res: ProcessedContent[] = []
+
+ // get all slugs ahead of time as each thread needs a copy
+ const allSlugs = fps.map(fp => slugifyFilePath(path.relative(baseDir, path.resolve(fp)) as FilePath))
+
+ let res: ProcessedContent[] = []
+ log.start(`Parsing input files using ${concurrency} threads`)
if (concurrency === 1) {
- // single-thread
const processor = createProcessor(transformers)
- for (const fp of fps) {
- const file = await read(fp)
-
- // base data properties that plugins may use
- file.data.slug = slugify(path.relative(baseDir, file.path))
- file.data.filePath = fp
-
- const ast = processor.parse(file)
- res.push([await processor.run(ast, file), file])
-
- if (verbose) {
- console.log(`[process] ${fp} -> ${file.data.slug}`)
- }
- }
+ const parse = createFileParser(transformers, baseDir, fps, allSlugs, verbose)
+ res = await parse(processor)
} else {
- await transpileWorkerScript(verbose)
+ await transpileWorkerScript()
const pool = workerpool.pool(
'./quartz/bootstrap-worker.mjs',
{
@@ -111,13 +140,14 @@
const childPromises: WorkerPromise<ProcessedContent[]>[] = []
for (const chunk of chunks(fps, CHUNK_SIZE)) {
- childPromises.push(pool.exec('parseFiles', [baseDir, chunk, verbose]))
+ childPromises.push(pool.exec('parseFiles', [baseDir, chunk, allSlugs, verbose]))
}
+
const results: ProcessedContent[][] = await WorkerPromise.all(childPromises)
- res.push(...results.flat())
+ res = results.flat()
await pool.terminate()
}
- console.log(`Parsed and transformed ${res.length} Markdown files with ${concurrency} cores in ${perf.timeSince()}`)
+ log.success(`Parsed ${res.length} Markdown files in ${perf.timeSince()}`)
return res
}
--
Gitblit v1.10.0