summaryrefslogtreecommitdiff
path: root/main.py
diff options
context:
space:
mode:
authorssstvinc2 <sstvinc2@gmail.com>2017-02-09 20:24:25 -0500
committerssstvinc2 <sstvinc2@gmail.com>2017-02-09 20:24:25 -0500
commitcbbd1122c78834b1beef349797d142ed6d23218a (patch)
tree2afab3496e02cc1744e413d82ce509a25e3465f1 /main.py
parent439b2cf7317af9444e99ba58188d7ce80a906af5 (diff)
Added/tweaked a few news sources, and added a source list to the page
Diffstat (limited to 'main.py')
-rwxr-xr-xmain.py65
1 files changed, 46 insertions, 19 deletions
diff --git a/main.py b/main.py
index 19fe8b0..edec792 100755
--- a/main.py
+++ b/main.py
@@ -13,10 +13,55 @@ def main():
def run():
sourceList=[]
+
+
+
+ sourceList.append(NewsSource('NBC News',
+ 'http://nbcnews.com',
+ ['top-stories-section', 'panel_hero', '<a href="'],
+ ['<div class="story-link', '<a href="'],
+ [],
+ None, None,
+ 'ad-content ad-xs mobilebox1', 'panel panel_default',
+ None, None))
+
+
+ sourceList.append(NewsSource('CBS News',
+ 'http://cbsnews.com',
+ ['<a href="'],
+ ['<li data-tb-region-item>', '<a href="'],
+ [],
+ 'Big News Area Side Assets', '</a>'
+ 'Big News Area Side Assets', '</ul></div>',
+ None, None))
+
+
+
+ sourceList.append(NewsSource('The Blaze',
+ 'http://theblaze.com',
+ ['<a class="gallery-link" href="'],
+ ['</figure>\n\n<figure class="gallery-item">', 'href="'],
+ [],
+ '<!-- home -->', '<!-- loop-home -->',
+ '<!-- home -->', '<!-- loop-home -->',
+ None, None))
+
+
+ sourceList.append(NewsSource('Weekly Standard',
+ 'http://www.weeklystandard.com/',
+ ['<div class="lead-photo">', 'href="'],
+ ['<div class="lead-photo">', 'href="'],
+ [],
+ '<div id="region_1"', '<div id="region_2"',
+ '<div class="widget lead-story layout-3col-feature" data-count="2">', '<div id="region_2"',
+ None, None))
+
+
+
sourceList.append(NewsSource('New York Times',
'http://nytimes.com',
['<a href="'],#'<h1 class="story-heading"><a href="'],#['"b-column column', 'h2 class="story-heading"><a href="'],
- ['article class="story theme-summary', 'h2 class="story-heading"><a href="'],
+ ['<article', '<a href="'],
['<hr class="single-rule"', 'article class="story theme-summary', 'h2 class="story-heading"><a href="'],
'<div class="b-column column">', '<!-- close photo-spot-region -->',
'section id="top-news" class="top-news"', '</div><!-- close a-column -->',
@@ -33,24 +78,6 @@ def run():
- sourceList.append(NewsSource('NBC News',
- 'http://nbcnews.com',
- ['top-stories-section', 'panel_hero', '<a href="'],
- ['panel panel_default', '<a href="'],
- [],
- None, None,
- 'row_no-clear ad-container ad-container_default ad-hide ad-container-mobilebox1', 'js-more-topstories',
- None, None))
-
-
- sourceList.append(NewsSource('CBS News',
- 'http://cbsnews.com',
- ['<h1 class="title"><a href="'],
- ['<li data-tb-region-item>', '<a href="'],
- [],
- None, None,
- 'Big News Area Side Assets', '</ul></div>',
- None, None))
#scrape all urls and build data structure
newsSourceArr=buildNewsSourceArr(sourceList)