fix: Correct nesting and other possible issues in child nodes of A tag (

closes #25)
crosstype · Aug 23, 2021 · dcd6b20 · dcd6b20
1 parent 422bcf9
commit dcd6b20
Show file tree

Hide file tree

Showing 5 changed files with 26 additions and 7 deletions.
diff --git a/src/config.ts b/src/config.ts
@@ -194,7 +194,8 @@ export const defaultTranslators: TranslatorConfigObject = {
     if (node.textContent === href) return { content: `<${href}>` };
 
     return {
-      postprocess: ({ content }) => content.replace(/(?:\r?\n)+/g, ''),
+      postprocess: ({ content }) => content.replace(/(?:\r?\n)+/g, ' '),
+      childTranslators: visitor.instance.aTagTranslators,
       prefix: '[',
       postfix: ']' + (!options.useLinkReferenceDefinitions
                ? `(${href}${title ? ` "${title}"` : ''})`
@@ -226,6 +227,16 @@ export const defaultCodeBlockTranslators: TranslatorConfigObject = {
   'img': { recurse: false }
 }
 
+export const aTagTranslatorConfig: TranslatorConfigObject = {
+  'br': { content: '\n', recurse: false },
+  'hr': { content: '\n', recurse: false },
+  'pre': defaultTranslators['pre'],
+  'strong,b': defaultTranslators['strong,b'],
+  'del,s,strike': defaultTranslators['del,s,strike'],
+  'em,i': defaultTranslators['em,i'],
+  'img': defaultTranslators['img']
+}
+
 // endregion
 
 

diff --git a/src/main.ts b/src/main.ts
@@ -1,7 +1,8 @@
 import { NodeHtmlMarkdownOptions } from './options';
 import { TranslatorCollection, TranslatorConfigObject } from './translator';
 import {
-  defaultBlockElements, defaultCodeBlockTranslators, defaultIgnoreElements, defaultOptions, defaultTranslators
+  aTagTranslatorConfig, defaultBlockElements, defaultCodeBlockTranslators, defaultIgnoreElements, defaultOptions,
+  defaultTranslators
 } from './config';
 import { parseHTML } from './utilities';
 import { getMarkdownForHtmlNodes } from './visitor';
@@ -23,6 +24,7 @@ type Options = Partial<NodeHtmlMarkdownOptions>
 
 export class NodeHtmlMarkdown {
   public translators = new TranslatorCollection();
+  public aTagTranslators = new TranslatorCollection();
   public codeBlockTranslators = new TranslatorCollection();
   public readonly options: NodeHtmlMarkdownOptions
 
@@ -50,6 +52,9 @@ export class NodeHtmlMarkdown {
     for (const [ elems, cfg ] of Object.entries({ ...defaultCodeBlockTranslators, ...customCodeBlockTranslators }))
       this.codeBlockTranslators.set(elems, cfg, true);
 
+    for (const [ elems, cfg ] of Object.entries(aTagTranslatorConfig))
+      this.aTagTranslators.set(elems, cfg, true);
+
     // TODO - Workaround for upstream issue (may not be fixed) - https://github.com/taoqf/node-html-parser/issues/78
     if (!this.options.textReplace) this.options.textReplace = [];
     this.options.textReplace.push([ /^<!DOCTYPE.*>/gmi, '' ]);

diff --git a/src/visitor.ts b/src/visitor.ts
@@ -161,8 +161,8 @@ export class Visitor {
     if (textOnly || !isElementNode(node)) return;
 
     /* Handle element node */
-    const translators = metadata?.translators ?? this.instance.translators;
-    const translatorCfgOrFactory = translators[node.tagName] as TranslatorConfig | TranslatorConfigFactory;
+    const translatorCfgOrFactory: TranslatorConfig | TranslatorConfigFactory | undefined =
+      metadata?.translators ? metadata.translators[node.tagName] : this.instance.translators[node.tagName];
 
     /* Update metadata with list detail */
     switch (node.tagName) {
@@ -183,6 +183,7 @@ export class Visitor {
           ...metadata,
           preserveWhitespace: true
         }
+        break;
     }
     if (metadata) this.nodeMetadata.set(node, metadata);
 

diff --git a/test/default-tags.test.ts b/test/default-tags.test.ts
@@ -48,8 +48,10 @@ describe(`Default Tags`, () => {
         <a href="${url}">a<br><br>b<strong>c</strong></a>
         <a>a<strong>b</strong></a> <!-- This node is treated as text due to no href -->
         <a href="${url}">${url}</a>
+        <!-- see: https://github.com/crosstype/node-html-markdown/issues/25 -->
+        <a href="${url}">a<a href="2">nested</a><img src="${url}">b</a>
     `);
-    expect(res).toBe(`[a    b**c**](${url}) a**b** <${url}> `);
+    expect(res).toBe(`[a b**c**](${url}) a**b** <${url}> [anested![](${url})b](${url}) `);
   });
 
   test(`Image (img)`, () => {

diff --git a/test/options.test.ts b/test/options.test.ts
@@ -265,13 +265,13 @@ text`);
     instance.options.useLinkReferenceDefinitions = false;
     let res = translate(html);
     expect(res).toBe(
-      `Hello: [a    b**c**](${url}) a**b** [link2](${url}/other) [repeat link](${url}) <${url}> Goodbye!`
+      `Hello: [a b**c**](${url}) a**b** [link2](${url}/other) [repeat link](${url}) <${url}> Goodbye!`
     );
 
     instance.options.useLinkReferenceDefinitions = true;
     res = translate(html);
     expect(res).toBe(
-      `Hello: [a    b**c**][1] a**b** [link2][2] [repeat link][1] <${url}> Goodbye!\n\n[1]: ${url}\n[2]: ${url}/other`
+      `Hello: [a b**c**][1] a**b** [link2][2] [repeat link][1] <${url}> Goodbye!\n\n[1]: ${url}\n[2]: ${url}/other`
     );
 
     instance.options.useLinkReferenceDefinitions = originalUseLinkReferenceDefinitions;