From 095fa7e06537feb7b92f2ef2335bf04855de3418 Mon Sep 17 00:00:00 2001 From: JJfutbol Date: Thu, 23 Jul 2009 00:31:49 +0000 Subject: git-svn-id: http://arc90labs-readability.googlecode.com/svn/trunk@72 d4e419ec-0920-11de-bbfd-a7c1bc4c261e --- js/readability.js | 12 ++++++++---- 1 file changed, 8 insertions(+), 4 deletions(-) (limited to 'js') diff --git a/js/readability.js b/js/readability.js index cde729d..92a9f5c 100755 --- a/js/readability.js +++ b/js/readability.js @@ -129,6 +129,8 @@ function determineContentScore(score, parent, element) if (element.tagName.toLowerCase() == "p" && getWordCount(element) > 20) //|| (score == 0 && getText(element).length > 10)) score++; + console.log(element.tagName.toLowerCase() + " " + getWordCount(element)); + //if (getWordCount(element) > 30) // score++; @@ -152,7 +154,7 @@ function determineContentScore(score, parent, element) function parseContent() { // replace all doubled-up
tags with

tags, and remove inline fonts document.body.innerHTML = document.body.innerHTML.replace(/]*>\s| *]*>/gi, "

").replace(/<\/?font[^>]*>/gi, ""); - console.log(document.body.innerHTML.substring(0, 600)); + var articleContent = document.createElement("DIV"); var paragraphs = document.getElementsByTagName("P"); var contentBlocks = []; @@ -248,11 +250,13 @@ function parseContent() { { var parentNode = paragraphs[i].parentNode; + // TODO: originally the if/continue statement below checked if the parent + // was the body tag and if it was continued on.. why? + // if the parent happens to be a form element, accessing properties // such as id or className don't work, or rather it attempts to access - // children so we need to make sure we only deal with string values, - // also if the parent element is the body then its ignored - if (parentNode.tagName.toLowerCase() == "body" || typeof parentNode.id != "string" || typeof parentNode.className != "string") + // children so we need to make sure we only deal with string values + if (typeof parentNode.id != "string" || typeof parentNode.className != "string") continue; // initialize readability score data -- cgit v1.2.3