html-to-text - npm Package Compare versions

Comparing version 2.1.3 to 3.0.0

CHANGELOG.md

		# Changelog

		## Version 3.0.0

		* Switched from `htmlparser` to `htmlparser2` #113
		* Treat non-numeric colspans as zero and handle them gracefully #105

		## Version 2.1.1
		@@ -4,0 +9,0 @@

lib/formatter.js

		@@ -8,3 +8,3 @@ var _ = require('underscore');
		function formatText(elem, options) {
		var text = elem.raw;
		var text = elem.data \|\| "";
		text = he.decode(text, options.decodeOptions);
		@@ -86,3 +86,3 @@

		return formatText({ raw: result \|\| href, trimLeadingSpace: elem.trimLeadingSpace }, options);
		return formatText({ data: result \|\| href, trimLeadingSpace: elem.trimLeadingSpace }, options);
		}
		@@ -113,3 +113,3 @@
		var nonWhiteSpaceChildren = (elem.children \|\| []).filter(function(child) {
		return child.type !== 'text' \|\| !whiteSpaceRegex.test(child.raw);
		return child.type !== 'text' \|\| !whiteSpaceRegex.test(child.data);
		});
		@@ -125,3 +125,3 @@ _.each(nonWhiteSpaceChildren, function(elem) {
		var nonWhiteSpaceChildren = (elem.children \|\| []).filter(function(child) {
		return child.type !== 'text' \|\| !whiteSpaceRegex.test(child.raw);
		return child.type !== 'text' \|\| !whiteSpaceRegex.test(child.data);
		});
		@@ -203,3 +203,3 @@ // Make sure there are list items present
		if (elem.attribs && elem.attribs.colspan) {
		times = elem.attribs.colspan - 1;
		times = elem.attribs.colspan - 1 \|\| 0;
		_.times(times, function() {
		@@ -206,0 +206,0 @@ rows.push(['']);

lib/html-to-text.js

		@@ -6,3 +6,3 @@ var fs = require('fs');
		var _s = require('underscore.string');
		var htmlparser = require('htmlparser');
		var htmlparser = require('htmlparser2');

		@@ -160,3 +160,3 @@ var helper = require('./helper');
		case 'text':
		if (elem.raw !== '\r\n') {
		if (elem.data !== '\r\n') {
		// Text needs its leading space to be trimmed if `result`
		@@ -163,0 +163,0 @@ // currently ends with whitespace

package.json

		{
		"name": "html-to-text",
		"version": "2.1.3",
		"version": "3.0.0",
		"description": "Advanced html to plain text converter",
		@@ -25,3 +25,3 @@ "main": "index.js",
		"he": "^1.0.0",
		"htmlparser": "^1.7.7",
		"htmlparser2": "^3.9.2",
		"optimist": "^0.6.1",
		@@ -47,4 +47,4 @@ "underscore": "^1.8.3",
		"chai": "^3.5.0",
		"mocha": "^2.4.5"
		"mocha": "^3.0.0"
		}
		}

test/html-to-text.js

		@@ -171,2 +171,19 @@ var expect = require('chai').expect;
		});

		it('does handle non-integer colspan on td element gracefully', function () {
		var html = 'Good morning Jacob, \
		<TABLE> \
		<CENTER> \
		<TBODY> \
		<TR> \
		<TD colspan="abc">Lorem ipsum dolor sit amet.</TD> \
		</TR> \
		</CENTER> \
		</TBODY> \
		</TABLE> \
		';
		var resultExpected = 'Good morning Jacob, Lorem ipsum dolor sit amet.';
		var result = htmlToText.fromString(html, { tables: true });
		expect(result).to.equal(resultExpected);
		});
		});
		@@ -218,3 +235,3 @@
		});


		});
		@@ -347,3 +364,3 @@
		});


		it('should not wrap a string if not wrapCharacters are found and forceWrapOnLimit is not set', function() {
		@@ -440,2 +457,18 @@ var testString = '<p>_This_string_is_meant_to_test_if_a_string_is_split_properly_across_anewlineandlong\nword_with_following_text.</p>';
		});

		describe('wbr', function() {
		it('should handle a large number of wbr tags w/o stack overflow', function() {
		var testString = "<!DOCTYPE html><html><head></head><body>\n";
		var expectedResult = "";
		for (var i = 0; i < 1000; i++){
		if (i !== 0 && i % 80 === 0) {
		expectedResult += "\n";
		}
		expectedResult += "n";
		testString += "<wbr>n";
		}
		testString += "</body></html>";
		expect(htmlToText.fromString(testString)).to.equal(expectedResult);
		});
		})
		});

New alerts