Support local HTML files for conversion

Allow the CLI argument to be either a remote URL or a local HTML path, and document the new usage in README/help examples.

README.md

@@ -30,7 +30,7 @@ node src/index.js --help
 Usage: pdfy [options] <url>
 
 Arguments:
-  url                      URL of the article to convert
+  url                        URL of the article to convert, or path to a local HTML file
 
 Options:
   -t, --theme <theme>        Reader View theme (light, dark, sepia, groove-dark, solarized-light, solarized-dark, nord-light, nord-dark)
@@ -77,7 +77,9 @@ CLI flags always override config file values. Use `--config <path>` to specify a
 
 ```bash
 node src/index.js https://example.com/article
+node src/index.js /path/to/saved/page.html
 node src/index.js https://example.com/article --theme dark --css config/rules.css
+node src/index.js /path/to/saved/page.html --theme dark --highlight-style monokai
 node src/index.js https://example.com/article --prefs config/reader-view-preferences.json
 node src/index.js https://example.com/article --highlight-style monokai
 node src/index.js https://example.com/article --config .pdfyrc

src/cli.js

@@ -16,7 +16,7 @@ const program = new Command()
 program
   .name('pdfy')
   .description('Convert web articles to PDF using Reader View and Puppeteer')
-  .argument('<url>', 'URL of the article to convert')
+  .argument('<url>', 'URL of the article to convert, or path to a local HTML file')
   .option(
     '-t, --theme <theme>',
     `Reader View theme (${VALID_THEMES.join(', ')})`
@@ -42,6 +42,8 @@ program
 Examples:
   node src/index.js <URL>
   node src/index.js <URL> --theme dark --highlight-style monokai --css config/rules.css
+  node src/index.js <path/to/file.html> --theme dark --highlight-style monokai
+  node src/index.js <URL> --prefs config/reader-view-preferences.json
   npm start -- <URL> --theme dark --highlight-style monokai --css config/rules.css
 `
   )

src/index.js

@@ -1,6 +1,7 @@
 import fs from 'node:fs'
-import process from 'node:process'
 import path from 'node:path'
+import { pathToFileURL } from 'node:url'
+import process from 'node:process'
 import readline from 'node:readline/promises'
 import { parseArgs } from './cli.js'
 import { loadConfigFile, loadPreferences, readCustomCss } from './config.js'
@@ -31,7 +32,12 @@ async function main () {
   const customCss = cssPath ? readCustomCss(path.resolve(cssPath)) : null
 
   let html
-  const currentUrl = url
+  const filePath = path.resolve(url)
+  let currentUrl = url
+  if (fs.existsSync(filePath) && fs.statSync(filePath).isFile()) {
+    currentUrl = pathToFileURL(filePath).href
+    logger.info(`Detected local file, converting to file URL: ${currentUrl}`)
+  }
 
   for (let attempt = 0; attempt < 2; attempt++) {
     try {
@@ -66,6 +72,7 @@ async function main () {
 
         logger.info(`Loading article from local file: "${resolvedPath}"`)
         html = fs.readFileSync(resolvedPath, 'utf-8')
+        currentUrl = pathToFileURL(resolvedPath).href
         continue
       }
       throw err

src/pdf.js

@@ -1,5 +1,6 @@
 import fs from 'node:fs'
 import path from 'node:path'
+import { pathToFileURL } from 'node:url'
 import logger from './logger.js'
 
 const HIGHLIGHT_JS =
@@ -44,8 +45,19 @@ export function enhanceHtml (html, customCss, highlightStyle = 'github') {
   return enhanced
 }
 
-export async function generatePdf (page, html, outputPath, { timeout = 30000 } = {}) {
-  await page.setContent(html, { waitUntil: 'networkidle0', timeout })
+export async function generatePdf (page, html, outputPath, { timeout = 30000, url = null } = {}) {
+  const isFileUrl = url && url.startsWith('file://')
+
+  let tempFile = null
+  if (isFileUrl) {
+    const sourceDir = path.dirname(new URL(url).pathname)
+    tempFile = path.join(sourceDir, `.pdfy-temp-${Date.now()}.html`)
+    fs.writeFileSync(tempFile, html)
+    await page.goto(pathToFileURL(tempFile).href, { waitUntil: 'networkidle0', timeout })
+  } else {
+    await page.setContent(html, { waitUntil: 'networkidle0', timeout })
+  }
+
   await page.waitForFunction(
     () => Array.from(document.images).every(img => img.complete),
     { timeout, polling: 200 }
@@ -68,5 +80,8 @@ export async function generatePdf (page, html, outputPath, { timeout = 30000 } =
   if (outputPath) {
     logger.info(`PDF saved: "${outputPath}"`)
   }
+  if (tempFile) {
+    try { fs.unlinkSync(tempFile) } catch { }
+  }
   return buffer
 }

src/pdfy.js

@@ -124,7 +124,7 @@ export async function convert (url, options = {}) {
       fs.mkdirSync(path.dirname(outputPath), { recursive: true })
     }
 
-    const pdfBuffer = await generatePdf(pdfPage, enhancedHtml, outputPath, { timeout })
+    const pdfBuffer = await generatePdf(pdfPage, enhancedHtml, outputPath, { timeout, url })
     await pdfPage.close()
 
     return { title, pdfBuffer, outputPath }