Code refactoring

Author: thedoorbell

src/downloadFiles.js

@@ -8,27 +8,23 @@ import Listr from 'listr'
 const log = debug('page-loader')
 
 const tasks = ($, links, url, filesDirPath, selector, attr) => new Listr(links
-  .filter((link) => {
+  .map((link) => {
     const absLink = new URL(link, url)
     if (absLink.hostname === new URL(url).hostname) {
-      return link
+      return {
+        title: `${absLink.href}`,
+        task: () => downloadFile($, link, absLink, filesDirPath, selector, attr),
+      }
     }
+    else return null
   })
-  .map((link) => {
-    const absLink = new URL(link, url)
-    return {
-      title: `${absLink.href}`,
-      task: () => downloadFile($, link, absLink, filesDirPath, selector, attr),
-    }
-  }), {
+  .filter(link => link !== null && link !== undefined), {
   concurrent: true,
   exitOnError: false,
 })
 
 const downloadFile = ($, link, absLink, filesDirPath, selector, attr) => {
-  const fileName = path.extname(absLink.pathname)
-    ? generateName(absLink.href).replace(/-(?=[a-zA-Z0-9]+$)/, '.')
-    : generateName(absLink.href) + '.html'
+  const fileName = generateName(absLink.href)
   const filePath = path.join(filesDirPath, fileName)
 
   log('download file %s', absLink.href)

src/generateName.js

@@ -1,6 +1,11 @@
+import path from 'path'
+
 const generateName = (url) => {
   const href = new URL(url)
-  return (href.hostname + href.pathname).replace(/[^a-zA-Z0-9]/g, '-')
+  const baseName = (href.hostname + href.pathname).replace(/[^a-zA-Z0-9]/g, '-')
+  return path.extname(href.pathname)
+    ? baseName.replace(/-(?=[a-zA-Z0-9]+$)/, '.')
+    : baseName + '.html'
 }
 
 export default generateName

src/index.js

@@ -1,7 +1,6 @@
 import path from 'path'
 import fsp from 'fs/promises'
 import * as cheerio from 'cheerio'
-import * as prettier from 'prettier'
 import debug from 'debug'
 import getData from './getData.js'
 import downloadFiles from './downloadFiles.js'
@@ -15,9 +14,9 @@ const downloadPage = (url, outputDir = process.cwd()) => {
   log('download page %s to %s', url, outputDir)
 
   const absOutputDir = getAbsolutePath(outputDir)
-  const fileName = generateName(url) + '.html'
+  const fileName = generateName(url)
   const filePath = path.join(absOutputDir, fileName)
-  const filesDirName = generateName(url) + '_files'
+  const filesDirName = generateName(url).replace('.html', '_files')
   const filesDirPath = path.join(absOutputDir, filesDirName)
 
   return fsp.access(absOutputDir)
@@ -37,12 +36,8 @@ const downloadPage = (url, outputDir = process.cwd()) => {
         .then(() => $)
     })
     .then(($) => {
-      log('files downloaded, format HTML')
-      return prettier.format($.html(), { parser: 'html' })
-    })
-    .then((formattedHtml) => {
       log('write to file %s', filePath)
-      return fsp.writeFile(filePath, formattedHtml)
+      return fsp.writeFile(filePath, $.html())
     })
     .then(() => {
       log('done, output file is %s', filePath)
@@ -56,6 +51,9 @@ const downloadPage = (url, outputDir = process.cwd()) => {
       if (err.code === 'EACCES') {
         throw new Error(`ERROR: No access to output directory - ${outputDir}`)
       }
+      if (err.message.startsWith('Cant get ')) {
+        throw err
+      }
       throw new Error(`ERROR: ${err.message}`)
     })
 }