wp-epub-gen 0.4.1 → 0.4.2

package/build/index.cjs

@@ -363,20 +363,24 @@ function loadAndProcessHtml(data) {
   }
   try {
     let $ = cheerio__namespace.load(trimmedData, {
-      xml: {
-        lowerCaseTags: true,
-        recognizeSelfClosing: true
-      }
+      xmlMode: true,
+      // @ts-ignore
+      decodeEntities: false,
+      lowerCaseTags: true,
+      recognizeSelfClosing: true,
+      lowerCaseAttributeNames: true
     });
     const body = $("body");
     if (body.length) {
       const html = body.html();
       if (html) {
         $ = cheerio__namespace.load(html, {
-          xml: {
-            lowerCaseTags: true,
-            recognizeSelfClosing: true
-          }
+          xmlMode: true,
+          // @ts-ignore
+          decodeEntities: false,
+          lowerCaseTags: true,
+          recognizeSelfClosing: true,
+          lowerCaseAttributeNames: true
         });
       }
     }
@@ -436,7 +440,7 @@ function processImages($, chapter, epubConfigs) {
     }
     const trimmedUrl = url2.trim();
     try {
-      if (!trimmedUrl.match(/^(https?:\/\/|data:|\.\/|\/)/)) {
+      if (!trimmedUrl.match(/^(https?:\/\/|file:\/\/|data:|\.\/|\/)/)) {
         logger.warn(`Image URL "${trimmedUrl}" appears to be invalid, but processing anyway`);
       }
     } catch (error) {
@@ -493,54 +497,20 @@ function processImages($, chapter, epubConfigs) {
     }
   });
 }
-function extractAndCleanHtmlContent($, originalData) {
+function extractAndCleanHtmlContent($) {
   let data;
   if ($("body").length) {
     data = $("body").html() || "";
   } else {
     data = $.root().html() || "";
   }
-  if (!originalData) {
-    return data.replace(
-      /<(br|hr|img|input|meta|area|base|col|embed|link|source|track|wbr)([^>]*?)><\/\1>/gi,
-      "<$1$2/>"
-    ).replace(
-      new RegExp("<(br|hr|img|input|meta|area|base|col|embed|link|source|track|wbr)([^>]*?)(?<!\\/)>", "gi"),
-      "<$1$2/>"
-    );
-  }
-  const entityMap = /* @__PURE__ */ new Map();
-  const entityRegex = /&[a-zA-Z][a-zA-Z0-9]*;|&#[0-9]+;|&#x[0-9a-fA-F]+;/g;
-  const matches = Array.from(originalData.matchAll(entityRegex));
-  let processedOriginal = originalData;
-  const timestamp = Date.now();
-  const randomId = Math.random().toString(36).substring(2, 8);
-  const placeholderPrefix = `__ENTITY_${timestamp}_${randomId}_`;
-  for (let i = matches.length - 1; i >= 0; i--) {
-    const match = matches[i];
-    const placeholder = `${placeholderPrefix}${i}__`;
-    entityMap.set(placeholder, match[0]);
-    processedOriginal = processedOriginal.substring(0, match.index) + placeholder + processedOriginal.substring(match.index + match[0].length);
-  }
-  const $temp = cheerio__namespace.load(processedOriginal, {
-    xmlMode: false
-  });
-  let tempData;
-  if ($temp("body").length) {
-    tempData = $temp("body").html() || "";
-  } else {
-    tempData = $temp.root().html() || "";
-  }
-  for (const [placeholder, entity] of entityMap) {
-    tempData = tempData.replace(new RegExp(placeholder, "g"), entity);
-  }
-  return tempData.replace(
+  return data.replace(
     /<(br|hr|img|input|meta|area|base|col|embed|link|source|track|wbr)([^>]*?)><\/\1>/gi,
     "<$1$2/>"
   ).replace(
     new RegExp("<(br|hr|img|input|meta|area|base|col|embed|link|source|track|wbr)([^>]*?)(?<!\\/)>", "gi"),
     "<$1$2/>"
-  );
+  ).replace(/<\/img\s*>/gi, "");
 }
 function processChildrenChapters(chapter, index2, epubConfigs) {
   if (Array.isArray(chapter.children)) {
@@ -574,7 +544,7 @@ function parseContent(content, index2, epubConfigs) {
     }
     processHtmlElements($, allowedAttributes, allowedXhtml11Tags, epubConfigs, index2);
     processImages($, chapter, epubConfigs);
-    chapter.data = extractAndCleanHtmlContent($, content.data);
+    chapter.data = extractAndCleanHtmlContent($);
   }
   processChildrenChapters(chapter, index2, epubConfigs);
   return chapter;
@@ -1245,6 +1215,7 @@ async function epubGen(options, configs) {
   if (configs?.logger) {
     logger.setLogger(configs.logger);
   }
+  logger.info("EpubGen started 101...");
   options = { ...options };
   const o = check(options);
   const verbose = options.verbose !== false;

package/build/index.js

@@ -342,20 +342,24 @@ function loadAndProcessHtml(data) {
   }
   try {
     let $ = cheerio.load(trimmedData, {
-      xml: {
-        lowerCaseTags: true,
-        recognizeSelfClosing: true
-      }
+      xmlMode: true,
+      // @ts-ignore
+      decodeEntities: false,
+      lowerCaseTags: true,
+      recognizeSelfClosing: true,
+      lowerCaseAttributeNames: true
     });
     const body = $("body");
     if (body.length) {
       const html = body.html();
       if (html) {
         $ = cheerio.load(html, {
-          xml: {
-            lowerCaseTags: true,
-            recognizeSelfClosing: true
-          }
+          xmlMode: true,
+          // @ts-ignore
+          decodeEntities: false,
+          lowerCaseTags: true,
+          recognizeSelfClosing: true,
+          lowerCaseAttributeNames: true
         });
       }
     }
@@ -415,7 +419,7 @@ function processImages($, chapter, epubConfigs) {
     }
     const trimmedUrl = url.trim();
     try {
-      if (!trimmedUrl.match(/^(https?:\/\/|data:|\.\/|\/)/)) {
+      if (!trimmedUrl.match(/^(https?:\/\/|file:\/\/|data:|\.\/|\/)/)) {
         logger.warn(`Image URL "${trimmedUrl}" appears to be invalid, but processing anyway`);
       }
     } catch (error) {
@@ -472,54 +476,20 @@ function processImages($, chapter, epubConfigs) {
     }
   });
 }
-function extractAndCleanHtmlContent($, originalData) {
+function extractAndCleanHtmlContent($) {
   let data;
   if ($("body").length) {
     data = $("body").html() || "";
   } else {
     data = $.root().html() || "";
   }
-  if (!originalData) {
-    return data.replace(
-      /<(br|hr|img|input|meta|area|base|col|embed|link|source|track|wbr)([^>]*?)><\/\1>/gi,
-      "<$1$2/>"
-    ).replace(
-      new RegExp("<(br|hr|img|input|meta|area|base|col|embed|link|source|track|wbr)([^>]*?)(?<!\\/)>", "gi"),
-      "<$1$2/>"
-    );
-  }
-  const entityMap = /* @__PURE__ */ new Map();
-  const entityRegex = /&[a-zA-Z][a-zA-Z0-9]*;|&#[0-9]+;|&#x[0-9a-fA-F]+;/g;
-  const matches = Array.from(originalData.matchAll(entityRegex));
-  let processedOriginal = originalData;
-  const timestamp = Date.now();
-  const randomId = Math.random().toString(36).substring(2, 8);
-  const placeholderPrefix = `__ENTITY_${timestamp}_${randomId}_`;
-  for (let i = matches.length - 1; i >= 0; i--) {
-    const match = matches[i];
-    const placeholder = `${placeholderPrefix}${i}__`;
-    entityMap.set(placeholder, match[0]);
-    processedOriginal = processedOriginal.substring(0, match.index) + placeholder + processedOriginal.substring(match.index + match[0].length);
-  }
-  const $temp = cheerio.load(processedOriginal, {
-    xmlMode: false
-  });
-  let tempData;
-  if ($temp("body").length) {
-    tempData = $temp("body").html() || "";
-  } else {
-    tempData = $temp.root().html() || "";
-  }
-  for (const [placeholder, entity] of entityMap) {
-    tempData = tempData.replace(new RegExp(placeholder, "g"), entity);
-  }
-  return tempData.replace(
+  return data.replace(
     /<(br|hr|img|input|meta|area|base|col|embed|link|source|track|wbr)([^>]*?)><\/\1>/gi,
     "<$1$2/>"
   ).replace(
     new RegExp("<(br|hr|img|input|meta|area|base|col|embed|link|source|track|wbr)([^>]*?)(?<!\\/)>", "gi"),
     "<$1$2/>"
-  );
+  ).replace(/<\/img\s*>/gi, "");
 }
 function processChildrenChapters(chapter, index2, epubConfigs) {
   if (Array.isArray(chapter.children)) {
@@ -553,7 +523,7 @@ function parseContent(content, index2, epubConfigs) {
     }
     processHtmlElements($, allowedAttributes, allowedXhtml11Tags, epubConfigs, index2);
     processImages($, chapter, epubConfigs);
-    chapter.data = extractAndCleanHtmlContent($, content.data);
+    chapter.data = extractAndCleanHtmlContent($);
   }
   processChildrenChapters(chapter, index2, epubConfigs);
   return chapter;
@@ -1224,6 +1194,7 @@ async function epubGen(options, configs) {
   if (configs?.logger) {
     logger.setLogger(configs.logger);
   }
+  logger.info("EpubGen started 101...");
   options = { ...options };
   const o = check(options);
   const verbose = options.verbose !== false;

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "wp-epub-gen",
-  "version": "0.4.1",
+  "version": "0.4.2",
   "description": "Epub generator.",
   "type": "module",
   "main": "build/index.cjs",