From f7029012dfb73ce04405bfe44e4e4d984818bf5f Mon Sep 17 00:00:00 2001
From: Ben Schlegel <ben5.schlegel@gmail.com>
Date: Sat, 16 Sep 2023 19:58:38 +0000
Subject: [PATCH] feat: black magic

---
 quartz/processors/parse.ts |  128 ++++++++++++++++++++++--------------------
 1 files changed, 68 insertions(+), 60 deletions(-)

diff --git a/quartz/processors/parse.ts b/quartz/processors/parse.ts
index 9947860..29f92fc 100644
--- a/quartz/processors/parse.ts
+++ b/quartz/processors/parse.ts
@@ -1,37 +1,37 @@
-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 { QuartzTransformerPluginInstance } 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 { 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 } from "../util/ctx"
 
 export type QuartzProcessor = Processor<MDRoot, HTMLRoot, void>
-export function createProcessor(transformers: QuartzTransformerPluginInstance[]): QuartzProcessor {
+export function createProcessor(ctx: BuildCtx): QuartzProcessor {
+  const transformers = ctx.cfg.plugins.transformers
+
   // base Markdown -> MD AST
   let processor = unified().use(remarkParse)
 
   // MD AST -> MD AST transforms
-  for (const plugin of transformers.filter(p => p.markdownPlugins)) {
-    processor = processor.use(plugin.markdownPlugins!())
+  for (const plugin of transformers.filter((p) => p.markdownPlugins)) {
+    processor = processor.use(plugin.markdownPlugins!(ctx))
   }
 
   // MD AST -> HTML AST
   processor = processor.use(remarkRehype, { allowDangerousHtml: true })
 
-
   // HTML AST -> HTML AST transforms
-  for (const plugin of transformers.filter(p => p.htmlPlugins)) {
-    processor = processor.use(plugin.htmlPlugins!())
+  for (const plugin of transformers.filter((p) => p.htmlPlugins)) {
+    processor = processor.use(plugin.htmlPlugins!(ctx))
   }
 
   return processor
@@ -49,60 +49,62 @@
   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(transformers: QuartzTransformerPluginInstance[], baseDir: string, fps: string[], allSlugs: string[], verbose: boolean) {
+export function createFileParser(ctx: BuildCtx, fps: FilePath[]) {
+  const { argv, cfg } = ctx
   return async (processor: QuartzProcessor) => {
     const res: ProcessedContent[] = []
     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 transformers.filter(p => p.textTransform)) {
-          file.value = plugin.textTransform!(file.value)
+        for (const plugin of cfg.plugins.transformers.filter((p) => p.textTransform)) {
+          file.value = plugin.textTransform!(ctx, file.value)
         }
 
         // base data properties that plugins may use
-        file.data.slug = slugify(path.relative(baseDir, file.path))
-        file.data.allSlugs = allSlugs
+        file.data.slug = slugifyFilePath(path.posix.relative(argv.directory, file.path) as FilePath)
         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}`)
+        if (argv.verbose) {
+          console.log(`[process] ${fp} -> ${file.data.slug} (${perf.timeSince()})`)
         }
       } catch (err) {
-        console.log(chalk.red(`\nFailed to process \`${fp}\`: `) + err)
-        process.exit(1)
+        trace(`\nFailed to process \`${fp}\``, err as Error)
       }
     }
 
@@ -110,44 +112,50 @@
   }
 }
 
-export async function parseMarkdown(transformers: QuartzTransformerPluginInstance[], baseDir: string, fps: string[], verbose: boolean): Promise<ProcessedContent[]> {
+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(verbose)
+  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
-  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))))
+  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) {
-    const processor = createProcessor(transformers)
-    const parse = createFileParser(transformers, baseDir, fps, allSlugs, verbose)
-    res = await parse(processor)
+    try {
+      const processor = createProcessor(ctx)
+      const parse = createFileParser(ctx, fps)
+      res = await parse(processor)
+    } catch (error) {
+      log.end()
+      throw error
+    }
   } else {
     await transpileWorkerScript()
-    const pool = workerpool.pool(
-      './quartz/bootstrap-worker.mjs',
-      {
-        minWorkers: 'max',
-        maxWorkers: concurrency,
-        workerType: 'thread'
-      }
-    )
+    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]))
+      childPromises.push(pool.exec("parseFiles", [argv, chunk, ctx.allSlugs]))
     }
 
-    const results: ProcessedContent[][] = await WorkerPromise.all(childPromises)
+    const results: ProcessedContent[][] = await WorkerPromise.all(childPromises).catch((err) => {
+      const errString = err.toString().slice("Error:".length)
+      console.error(errString)
+      process.exit(1)
+    })
     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