@crawlee/utils 4.0.0-beta.176 → 4.0.0-beta.177

package/internals/cheerio.js

@@ -30,7 +30,7 @@ const BLOCK_TAGS_REGEX = /^(p|h1|h2|h3|h4|h5|h6|ol|ul|li|pre|address|blockquote|
  * @return Plain text
  */
 export async function htmlToText(htmlOrCheerioElement) {
-    const { load } = await import('cheerio');
+    const { load } = await import('cheerio/slim');
     if (!htmlOrCheerioElement)
         return '';
     const $ = typeof htmlOrCheerioElement === 'function' ? htmlOrCheerioElement : load(htmlOrCheerioElement);
@@ -42,8 +42,10 @@ export async function htmlToText(htmlOrCheerioElement) {
             if (elem.type === 'text') {
                 // Compress spaces, unless we're inside <pre> element
                 let compr;
-                if (elem.parent?.tagName === 'pre')
-                    compr = elem.data;
+                if (elem.parent?.tagName === 'pre') {
+                    // Strip single newline right after `<pre>` that htmlparser2 keeps
+                    compr = elem.parent.children?.[0] === elem ? elem.data.replace(/^\r?\n/, '') : elem.data;
+                }
                 else
                     compr = elem.data.replace(/\s+/g, ' ');
                 // If text is empty or ends with a whitespace, don't add the leading whitespace

package/internals/extract-microdata.js

@@ -1,7 +1,7 @@
 import { isTag } from 'domhandler';
 export async function extractMicrodata(htmlOrCheerioElement) {
     // Dynamic so that importing `@crawlee/utils` does not pull in cheerio - see #3836.
-    const { load } = await import('cheerio');
+    const { load } = await import('cheerio/slim');
     const $ = typeof htmlOrCheerioElement === 'string' ? load(htmlOrCheerioElement) : htmlOrCheerioElement;
     const context = { $ };
     return $('[itemscope]')

package/internals/open_graph_parser.js

@@ -366,7 +366,7 @@ const parseOpenGraphProperty = (property, $) => {
     };
 };
 export async function parseOpenGraph(item, additionalProperties) {
-    const { load } = await import('cheerio');
+    const { load } = await import('cheerio/slim');
     const $ = typeof item === 'string' ? load(item) : item;
     return [...(additionalProperties || []), ...OPEN_GRAPH_PROPERTIES].reduce((acc, curr) => {
         return {

package/internals/social.js

@@ -590,7 +590,7 @@ export const DISCORD_REGEX_GLOBAL = new RegExp(DISCORD_REGEX_STRING, 'ig');
  * @return An object with the social handles.
  */
 export async function parseHandlesFromHtml(html, data = null) {
-    const cheerio = await import('cheerio');
+    const cheerio = await import('cheerio/slim');
     const result = {
         emails: [],
         phones: [],

package/package.json

@@ -1,6 +1,6 @@
 {
     "name": "@crawlee/utils",
-    "version": "4.0.0-beta.176",
+    "version": "4.0.0-beta.177",
     "description": "A set of shared utilities that can be used by crawlers",
     "engines": {
         "node": ">=22.0.0"
@@ -44,8 +44,8 @@
     "dependencies": {
         "@apify/ps-tree": "^1.2.0",
         "@apify/validations": "^1.1.0",
-        "@crawlee/http-client": "4.0.0-beta.176",
-        "@crawlee/types": "4.0.0-beta.176",
+        "@crawlee/http-client": "4.0.0-beta.177",
+        "@crawlee/types": "4.0.0-beta.177",
         "@types/sax": "^1.2.7",
         "cheerio": "^1.0.0",
         "domhandler": "^5.0.3",
@@ -64,5 +64,5 @@
             }
         }
     },
-    "gitHead": "3e9e27d524dff8fe72aaf337e3d24fc2232eb241"
+    "gitHead": "bfd191ba71902fc682dd21fb9ab17590037a60d5"
 }