diff options
| author | Tim Groeneveld <tim@timg.ws> | 2015-03-10 11:50:15 +1100 |
|---|---|---|
| committer | Tim Groeneveld <tim@timg.ws> | 2015-03-10 11:50:15 +1100 |
| commit | 81bce8244b3d0e29918117a0fdcf40589e184ba9 (patch) | |
| tree | fa4e7f7123c3b074e008652bea52342ee360d436 | |
| parent | 45ff91859b99d8e132fca0afddb85a7581858044 (diff) | |
| download | ftr-site-config-81bce8244b3d0e29918117a0fdcf40589e184ba9.tar.gz ftr-site-config-81bce8244b3d0e29918117a0fdcf40589e184ba9.zip | |
Add some other missing news sites that do not always correctly work.
* business.time.com
* galwayindependent.com
* help.fivefilters.org
* ippmedia.com
* mitchellrepublic.com
* news.menshealth.com
* planetsave.com
* yourerie.com
| -rw-r--r-- | business.time.com.txt | 17 | ||||
| -rw-r--r-- | galwayindependent.com.txt | 3 | ||||
| -rw-r--r-- | help.fivefilters.org.txt | 2 | ||||
| -rw-r--r-- | ippmedia.com.txt | 4 | ||||
| -rw-r--r-- | mitchellrepublic.com.txt | 3 | ||||
| -rw-r--r-- | news.menshealth.com.txt | 1 | ||||
| -rw-r--r-- | planetsave.com.txt | 1 | ||||
| -rw-r--r-- | yourerie.com.txt | 2 |
8 files changed, 33 insertions, 0 deletions
diff --git a/business.time.com.txt b/business.time.com.txt new file mode 100644 index 0000000..5502bea --- /dev/null +++ b/business.time.com.txt @@ -0,0 +1,17 @@ +# 2011-10-25 - carlo@... - Initial setup. + +single_page_link: //li[@class='print']/a/@href + +title: //h1 +author: //meta[@name="byline"]/@content +date: //meta[@name="date"]/@content + +strip: //span[@class="see"] +strip: //div[@class="byline"] +strip: //div[@id="date2"] +strip: //h1 +strip: //div[@class='post-rail-ad'] +strip: //div[@class='post-rail-content'] +strip: //aside[@class='post-rail'] + +test_url: http://www.time.com/time/specials/packages/article/0,28804,2094921_2094923_2094924,00.html diff --git a/galwayindependent.com.txt b/galwayindependent.com.txt new file mode 100644 index 0000000..d45b7ac --- /dev/null +++ b/galwayindependent.com.txt @@ -0,0 +1,3 @@ +title: //div[@class='leftCol']/h1 + +prune: no diff --git a/help.fivefilters.org.txt b/help.fivefilters.org.txt new file mode 100644 index 0000000..70a7d15 --- /dev/null +++ b/help.fivefilters.org.txt @@ -0,0 +1,2 @@ +title: //div[@class="title"]/h3 +date: substring-after(//div[@class="meta"], ": ") diff --git a/ippmedia.com.txt b/ippmedia.com.txt new file mode 100644 index 0000000..99f25dc --- /dev/null +++ b/ippmedia.com.txt @@ -0,0 +1,4 @@ +title: //div[@class="content_title"]//h2 +author: substring-after(//div[@class="byline"], "By ") +date: //div[@class="publish_date"] +strip: //div[@class="read_image_box"] diff --git a/mitchellrepublic.com.txt b/mitchellrepublic.com.txt new file mode 100644 index 0000000..fae858a --- /dev/null +++ b/mitchellrepublic.com.txt @@ -0,0 +1,3 @@ +body: //div[@class='section'] +strip_id_or_class: mediumtxt +strip: //strong[contains diff --git a/news.menshealth.com.txt b/news.menshealth.com.txt new file mode 100644 index 0000000..a07fdac --- /dev/null +++ b/news.menshealth.com.txt @@ -0,0 +1 @@ +strip: //span[@style="color: #cf1206;"] diff --git a/planetsave.com.txt b/planetsave.com.txt new file mode 100644 index 0000000..d6f34e2 --- /dev/null +++ b/planetsave.com.txt @@ -0,0 +1 @@ +strip_id_or_class: author-bio-box diff --git a/yourerie.com.txt b/yourerie.com.txt new file mode 100644 index 0000000..b46b09e --- /dev/null +++ b/yourerie.com.txt @@ -0,0 +1,2 @@ +body: //div[@class="nxFullTextData"] +test_url: http://yourerie.com/fulltext?nxd_id=306552 |