html-to-text - npm Package Compare versions

Comparing version 2.1.0 to 2.1.1

CHANGELOG.md

 # Changelog
 
+## Version 2.1.1
+
+ * Extra space ploblem fixed. #88
+
 ## Version 2.1.0
@@ -4,0 +8,0 @@

lib/formatter.js

@@ -8,3 +8,3 @@ var _ = require('underscore');
 function formatText(elem, options) {
-	var text = (options.isInPre ? elem.raw : _s.strip(elem.raw));
+	var text = elem.raw;
 	text = he.decode(text, options.decodeOptions);
@@ -15,3 +15,3 @@
 	} else {
-		return helper.wordwrap(elem.needsSpace ? ' ' + text : text, options);
+		return helper.wordwrap(elem.trimLeadingSpace ? _s.lstrip(text) : text, options);
 	}
@@ -63,7 +63,9 @@ }
 	var storedCharCount = options.lineCharCount;
-	var result = _s.strip(fn(elem.children || [], options));
-	if (!result) {
-		result = '';
+	var text = fn(elem.children || [], options);
+	if (!text) {
+		text = '';
 	}
 
+	var result = elem.trimLeadingSpace ? _s.lstrip(text) : text;
+
 	if (!options.ignoreHref) {
@@ -78,3 +80,3 @@ // Get the href, if present
 			}
-			if (!options.hideLinkHrefIfSameAsText || href != result.replace('\n', '')) {
+			if (!options.hideLinkHrefIfSameAsText || href != _s.replaceAll(result, '\n', '')) {
 				result += ' [' + href + ']';
@@ -87,3 +89,3 @@ }
 
-	return formatText({ raw: result || href, needsSpace: elem.needsSpace }, options);
+	return formatText({ raw: result || href, trimLeadingSpace: elem.trimLeadingSpace }, options);
 }
@@ -90,0 +92,0 @@

lib/helper.js

@@ -69,3 +69,3 @@ var _ = require('underscore');
         // Add buffer to result if we can't fit any more words in the buffer.
-        if ((max || max === 0) &&
+        if ((max || max === 0) && length > 0 &&
             ((length + word.length > max) || (length + word.indexOf('\n') > max)))
@@ -105,3 +105,11 @@ {
     result += buffer.join(' ');
-    return _s.rstrip(result);
+
+    // Preserve trailing space
+    if (!_s.endsWith(text, ' ')) {
+      result = _s.rtrim(result);
+    } else if (!_s.endsWith(result, ' ')) {
+      result = result + ' ';
+    }
+
+    return result;
 };
@@ -113,6 +121,2 @@
 
-exports.isArray = function isArray(x) {
-    return Object.prototype.toString.call(x) === "[object Array]";
-};
-
 exports.splitCssSearchTag = function splitCssSearchTag(tagString) {
@@ -119,0 +123,0 @@ function getParams(re, string) {

lib/html-to-text.js

@@ -49,3 +49,3 @@ var fs = require('fs');
 	var result = '';
-	var baseElements = helper.isArray(options.baseElement) ? options.baseElement : [options.baseElement];
+	var baseElements = Array.isArray(options.baseElement) ? options.baseElement : [options.baseElement];
 	for (var idx = 0; idx < baseElements.length; ++idx) {
@@ -109,3 +109,3 @@ result += walk(filterBody(handler.dom, options, baseElements[idx]), options);
 	}
-	var whiteSpaceRegex = /\S$/;
+	var whiteSpaceRegex = /\s$/;
 	_.each(dom, function(elem) {
@@ -119,5 +119,5 @@ switch(elem.type) {
 					case 'a':
-						// Inline element needs a leading space if `result` currently
-						// doesn't end with whitespace
-						elem.needsSpace = whiteSpaceRegex.test(result);
+						// Inline element needs its leading space to be trimmed if `result`
+						// currently ends with whitespace
+						elem.trimLeadingSpace = whiteSpaceRegex.test(result);
 						result += format.anchor(elem, walk, options);
@@ -164,5 +164,5 @@ break;
 				if (elem.raw !== '\r\n') {
-					// Text needs a leading space if `result` currently
-					// doesn't end with whitespace
-					elem.needsSpace = whiteSpaceRegex.test(result);
+					// Text needs its leading space to be trimmed if `result`
+					// currently ends with whitespace
+					elem.trimLeadingSpace = whiteSpaceRegex.test(result);
 					result += format.text(elem, options);
@@ -169,0 +169,0 @@ }

package.json

 {
   "name": "html-to-text",
-  "version": "2.1.0",
+  "version": "2.1.1",
   "description": "Advanced html to plain text converter",
@@ -24,3 +24,3 @@ "main": "index.js",
   "dependencies": {
-    "he": "^0.5.0",
+    "he": "^1.0.0",
     "htmlparser": "^1.7.7",
@@ -27,0 +27,0 @@ "optimist": "^0.6.1",

README.md

@@ -31,3 +31,3 @@ # node-html-to-text
 
-```
+```javascript
 var htmlToText = require('html-to-text');
@@ -37,3 +37,3 @@
 	tables: ['#invoice', '.address']
-}, function(err, text) {
+}, (err, text) => {
 	if (err) return console.error(err);
@@ -46,3 +46,3 @@ console.log(text);
 
-```
+```javascript
 var htmlToText = require('html-to-text');
@@ -95,3 +95,3 @@
 
-```
+```html
 <html>
@@ -291,3 +291,3 @@ <head>
 
-Copyright (c) 2015 werk85 &lt;[e-mail]&gt;
+Copyright (c) 2016 werk85 &lt;[e-mail]&gt;
 
@@ -312,3 +312,1 @@ Permission is hereby granted, free of charge, to any person obtaining
 SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
-
-[![githalytics.com alpha](https://cruel-carlota.pagodabox.com/cc03826a7e68e1bb680cf2226276e031 "githalytics.com")](http://githalytics.com/werk85/node-html-to-text)

test/html-to-text.js

@@ -55,3 +55,3 @@ var expect = require('chai').expect;
         var testString = "<p>We appreciate your business. And we hope you'll check out our <a href=\"http://example.com/\">new products</a>!</p>";
-        expect(htmlToText.fromString(testString, {} )).to.equal('We appreciate your business. And we hope you\'ll check out our new products\n[http://example.com/] !');
+        expect(htmlToText.fromString(testString, {} )).to.equal('We appreciate your business. And we hope you\'ll check out our new products\n[http://example.com/]!');
       });
@@ -149,3 +149,3 @@
       ';
-      var resultExpected = 'Good morning Jacob,Lorem ipsum dolor sit amet\n\nLorem ipsum dolor sit amet.\n\n * run in the park (in progress)';
+      var resultExpected = 'Good morning Jacob, Lorem ipsum dolor sit amet\n\nLorem ipsum dolor sit amet.\n\n * run in the park (in progress)';
       var result = htmlToText.fromString(html, { wordwrap: false });
@@ -169,3 +169,3 @@ expect(result).to.equal(resultExpected);
       ';
-      var resultExpected = 'Good morning Jacob,Lorem ipsum dolor sit amet.';
+      var resultExpected = 'Good morning Jacob, Lorem ipsum dolor sit amet.';
       var result = htmlToText.fromString(html, { tables: true });
@@ -172,0 +172,0 @@ expect(result).to.equal(resultExpected);

test/test.txt

@@ -61,2 +61,2 @@ PARAGRAPHS
 We appreciate your business. And we hope you'll check out our new products
-[http://example.com/] !
+[http://example.com/]!

No alert changes

Improved metrics

Total package byte prevSize

68525

0.11%

Lines of code

885

0.45%

Worsened metrics

Number of lines in readme file

307

-0.65%

Dependency changes

• + Added

  he@1.2.0
  (transitive)

• - Removed

  he@0.5.0
  (transitive)

• Updated

  he@^1.0.0