ep_readability - npm Package Compare versions

Comparing version 0.0.15 to 0.0.16

ep_readability.tmproj

package.json

		{
		"name": "ep_readability",
		"version": "0.0.15",
		"version": "0.0.16",
		"description": "Calculates the Flesch readability index in Etherpad, client based.",
		@@ -5,0 +5,0 @@ "author": {

README.md

		# Click button to calculate the Flesch readability index

		There'll be a button which calculates and displays the Flesch redability index.

		Calculating is triggered by entering !, ?, RETURN or BACKSPACE. Or by pushing the button.
		Calculation of the Flesch index is triggered by entering !, ?, RETURN or BACKSPACE. Or by pushing the button.

		@@ -10,6 +11,12 @@ The button shows a tooltip, which provides additional information.

		## Calculation details

		The original Flesch index is calculated for all languages. German umlauts are replaced (ä = a, ß = ss...).

		## Readability
		Numbers are counted as a single syllable (1000,56 = one syllable).

		Links are counted as a single syllable (http://aaa-bbb.bla?foo=bar = one syllable).

		## Readability scale

		* -1000: Dissertation, red
		@@ -16,0 +23,0 @@ * 10: Officialese, red

static/js/readability_button.js

		@@ -10,17 +10,21 @@ // ----------------------------------------------------------------------------

		// @TODO rething this regex orgy: the input text is pretty wild sometimes
		exports.cleanText = function(text) {

		//console.log(text);
		// 1. Remove HTML entities @TODO add more entities
		// 2. Remove Etherpad spans, headings (style plugin) and formatting
		// 3. Get an Etherpad paragraph, add dot if no terminator or comma is at the end
		// @TODO replace the divs by a more simple paragraph placeholder delimiter
		// @TODO rething this regex orgy: the input text is pretty wild sometimes
		// 2. Check for plain text links
		// 3. Remove Etherpad spans, headings (style plugin) and formatting
		// 4. Get an Etherpad paragraph, add dot if no terminator or comma is at the end

		text = text.replace(/( )+/gm,' ')
		.replace(/(https?\|ftp):[\w\.\/\?&=-]+/gm, 'link')
		.replace(/<(?!div\|\/div)[^>]*>/gm,'').replace(/<\/(?!div)>/gm,'')
		.replace(/<div[^>]>([^<]+[^:, !?.-])\s<\/div>/gm, "$1.");

		//console.log(text);
		// This is not Etherpad specific:
		// @TODO Add URL matcher
		text = text
		.replace(/<[^>]+>/gm, '') // Strip tags (should be DIVs only at this point)

		.replace(/\d+([\.,]?\d+)*/g, 'two') // Convert numbers with delimiters to one syllable @TODO lookahead for space: EUR 1000. bla
		@@ -27,0 +31,0 @@ .replace(/[!?]/gm, '.') // Unify terminators

test/test.js

		@@ -65,2 +65,10 @@ var readab = require('../static/js/readability_button.js');
		test.done();
		}
		}

		exports.cleanTextLinks = function(test) {
		var text = '<div id="magicdomid8" class=""><br></div><div id="magicdomid9" class=""><span class="author-a-z65zjckdup0xdseeity url"><a href="http://jaja.nono-nana.bla?nix=doh">http://jaja.nono-nana.bla?nix=doh</a></span></div>';
		var result = 'link.';
		test.equal(result, readab.cleanText(text));
		test.done();
		}

New alerts