fix: Whitespace between some nodes not properly handled (fixes #16)

crosstype · Jul 13, 2021 · a7abf81 · a7abf81
1 parent 19f3177
commit a7abf81
Show file tree

Hide file tree

Showing 7 changed files with 21 additions and 15 deletions.
diff --git a/package.json b/package.json
@@ -48,7 +48,7 @@
     "node": ">=10.0.0"
   },
   "dependencies": {
-    "node-html-parser": "^4.0.0"
+    "node-html-parser": "^4.1.0"
   },
   "devDependencies": {
     "@types/jest": "^26.0.23",

diff --git a/src/utilities.ts b/src/utilities.ts
@@ -183,7 +183,7 @@ export function parseHTML(html: string, options: NodeHtmlMarkdownOptions): Eleme
     }
   } else nodeHtmlParse = getNodeHtmlParser();
 
-  if (!el) el = nodeHtmlParse!(html, nodeHtmlParserConfig).removeWhitespace();
+  if (!el) el = nodeHtmlParse!(html, nodeHtmlParserConfig);
   perfStop('parse');
 
   return el;

diff --git a/src/visitor.ts b/src/visitor.ts
@@ -137,8 +137,8 @@ export class Visitor {
     /* Handle text node */
     if (isTextNode(node))
       return node.isWhitespace
-             ? void 0
-             : this.appendResult(metadata?.noEscape ? node.text : this.processText(node.text));
+             ? (!result.text.length || result.trailingNewlineStats.whitespace > 0) ? void 0 : this.appendResult(' ')
+             : this.appendResult(metadata?.noEscape ? node.trimmedText : this.processText(node.trimmedText));
 
     if (textOnly || !isElementNode(node)) return;
 

diff --git a/test/default-tags.test.ts b/test/default-tags.test.ts
@@ -49,7 +49,7 @@ describe(`Default Tags`, () => {
         <a>a<strong>b</strong></a> <!-- This node is treated as text due to no href -->
         <a href="${url}">${url}</a>
     `);
-    expect(res).toBe(`[a    b**c**](${url})a**b**<${url}>`);
+    expect(res).toBe(`[a    b**c**](${url}) a**b** <${url}> `);
   });
 
   test(`Image (img)`, () => {
@@ -61,7 +61,7 @@ describe(`Default Tags`, () => {
         <img src="${url}4" title="t4" alt="a4">
         <img src="data:image/gif;base64,R/"><!-- This node is elided due to default option keepDataImages = false -->
     `);
-    expect(res).toBe(`![](${url}1)` + `![](${url}3 "t3")` + `![a4](${url}4 "t4")`);
+    expect(res).toBe(`![](${url}1)` + ` ![](${url}3 "t3")` + ` ![a4](${url}4 "t4") `);
   });
 
   test(`Pre-formatted Text (pre)`, () => {

diff --git a/test/options.test.ts b/test/options.test.ts
@@ -240,12 +240,12 @@ text`);
     instance.options.keepDataImages = true;
     const resKeep = translate(`<img alt="normal" src="normal_img.jpg">
       <img src="data:image/gif;base64,R0lGODlhEA"/>`);
-    expect(resKeep).toBe(`![normal](normal_img.jpg)![](data:image/gif;base64,R0lGODlhEA)`);
+    expect(resKeep).toBe(`![normal](normal_img.jpg) ![](data:image/gif;base64,R0lGODlhEA)`);
 
     instance.options.keepDataImages = false;
     const resNoKeep = translate(`<img alt="normal" src="normal_img.jpg">
       <img src="data:image/gif;base64,R0lGODlhEA"/>`);
-    expect(resNoKeep).toBe(`![normal](normal_img.jpg)`);
+    expect(resNoKeep).toBe(`![normal](normal_img.jpg) `);
 
     instance.options.keepDataImages = originalKeepDataImages;
   });
@@ -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;

diff --git a/test/special-cases.test.ts b/test/special-cases.test.ts
@@ -47,7 +47,7 @@ describe(`Special Cases`, () => {
     test.each(textFormatTags)(`%s`, tag => {
       const delim = delims[tag];
 
-      expect(translate(`<p><${tag}> &nbsp;Label:&nbsp; </${tag}>Value</p>`)).toBe(` ${delim}Label:${delim} Value`);
+      expect(translate(`<p><${tag}> &nbsp;Label:&nbsp; </${tag}>Value</p>`)).toBe(` ${delim}Label:${delim} Value`);
       expect(translate(`<p><${tag}>&nbsp; Label: &nbsp;</${tag}>Value</p>`)).toBe(` ${delim}Label:${delim} Value`);
     });
   });
@@ -62,4 +62,10 @@ describe(`Special Cases`, () => {
       );
     });
   });
+
+  // See: https://github.com/crosstype/node-html-markdown/issues/16
+  test(`Handles whitespace with single space`, () => {
+    const res = translate(`<span>test</span>  <span>test2 </span>\n<span>test3</span>\r\n\r\n\t\t\t<span>test4</span>`);
+    expect(res).toBe(`test test2 test3 test4`);
+  });
 });
diff --git a/yarn.lock b/yarn.lock
@@ -2963,10 +2963,10 @@ neo-async@^2.6.0:
   resolved "https://registry.yarnpkg.com/neo-async/-/neo-async-2.6.2.tgz#b4aafb93e3aeb2d8174ca53cf163ab7d7308305f"
   integrity sha512-Yd3UES5mWCSqR+qNT93S3UoYUkqAZ9lLg8a7g9rimsWmYGK8cVToA4/sF3RrshdyV3sAGMXVUmpMYOw+dLpOuw==
 
-node-html-parser@^4.0.0:
-  version "4.0.0"
-  resolved "https://registry.yarnpkg.com/node-html-parser/-/node-html-parser-4.0.0.tgz#95e4fec010c48425821d5659eaf39e27b0781b6d"
-  integrity sha512-vuOcp3u4GrfYOcqe+FpUffQKnfVBq781MbtFEcTR6fJYnYRE2qUPdDaDk7TGdq6H9r2B3TbyU6K9Rah6/C7qvg==
+node-html-parser@^4.1.0:
+  version "4.1.0"
+  resolved "https://registry.yarnpkg.com/node-html-parser/-/node-html-parser-4.1.0.tgz#9c23130609e203030113255c519ff2d3fbbfc439"
+  integrity sha512-sqSaqyrg1mqsRTl0r0h4UVAHBoW605rSpanOz0Yr7NzERl0fJ1eh6CRoZFcIXEIyzD3QJfGqsIMP+kzkViTiAA==
   dependencies:
     css-select "^4.1.3"
     he "1.2.0"