diff options
| -rw-r--r-- | .fok.nl.txt | 2 | ||||
| -rw-r--r-- | .wikipedia.org.txt | 3 | ||||
| -rw-r--r-- | deadspin.com.txt | 1 | ||||
| -rw-r--r-- | gawker.com.txt | 2 | ||||
| -rw-r--r-- | getpocket.com.txt | 1 | ||||
| -rw-r--r-- | gizmodo.com.txt | 2 | ||||
| -rw-r--r-- | io9.com.txt | 1 | ||||
| -rw-r--r-- | jalopnik.com.txt | 3 | ||||
| -rw-r--r-- | jezebel.com.txt | 1 | ||||
| -rw-r--r-- | kotaku.com.txt | 3 | ||||
| -rw-r--r-- | lifehacker.com.txt | 3 |
11 files changed, 22 insertions, 0 deletions
diff --git a/.fok.nl.txt b/.fok.nl.txt index 1d2bdf0..e6ad7d5 100644 --- a/.fok.nl.txt +++ b/.fok.nl.txt @@ -8,4 +8,6 @@ replace_string(90%;">Lees ook): 0%;"></h3></div></body></html> tidy: no prune: no +http_header(user-agent): Googlebot/2.1 + test_url: http://sport.fok.nl/nieuws/548725/1/1/50/rosicky-traint-weer-mee-bij-tsjechie.html
\ No newline at end of file diff --git a/.wikipedia.org.txt b/.wikipedia.org.txt index 13bccad..1da400b 100644 --- a/.wikipedia.org.txt +++ b/.wikipedia.org.txt @@ -14,6 +14,9 @@ strip: //div[@id='contentSub'] strip: //table[contains(@class, 'metadata')] strip: //*[contains(@class, 'noprint')] strip: //span[@class='noexcerpt'] + +http_header(user-agent): Mozilla/5.2 + prune: no tidy: no test_url: http://en.wikipedia.org/wiki/Christopher_Lloyd diff --git a/deadspin.com.txt b/deadspin.com.txt new file mode 100644 index 0000000..e6ca16a --- /dev/null +++ b/deadspin.com.txt @@ -0,0 +1 @@ +http_header(user-agent): PHP/5.3
\ No newline at end of file diff --git a/gawker.com.txt b/gawker.com.txt index 9bc5613..27e4b4b 100644 --- a/gawker.com.txt +++ b/gawker.com.txt @@ -3,4 +3,6 @@ body: //div[@class="post-body"] # Remove 'content is restricted' strip: //div[@id='agegate_IDHERE'] +http_header(user-agent): PHP/5.3 + test_url: http://gawker.com/#!5782070/russian-bomb-squad-successfully-defuses-sex-toy
\ No newline at end of file diff --git a/getpocket.com.txt b/getpocket.com.txt new file mode 100644 index 0000000..e6ca16a --- /dev/null +++ b/getpocket.com.txt @@ -0,0 +1 @@ +http_header(user-agent): PHP/5.3
\ No newline at end of file diff --git a/gizmodo.com.txt b/gizmodo.com.txt index e73ec9d..535041c 100644 --- a/gizmodo.com.txt +++ b/gizmodo.com.txt @@ -6,6 +6,8 @@ date: //span[@class="date"] prune: no +http_header(user-agent): PHP/5.3 + test_url: http://gizmodo.com/5880147/kuhn-rikon-improves-their-spice-grinder-with-grade-school-science test_url: http://gizmodo.com/what-van-goghs-paintings-would-look-like-if-they-came-874035680 test_url: http://gizmodo.com/vip.xml
\ No newline at end of file diff --git a/io9.com.txt b/io9.com.txt new file mode 100644 index 0000000..e6ca16a --- /dev/null +++ b/io9.com.txt @@ -0,0 +1 @@ +http_header(user-agent): PHP/5.3
\ No newline at end of file diff --git a/jalopnik.com.txt b/jalopnik.com.txt index fc2eef8..7823dbd 100644 --- a/jalopnik.com.txt +++ b/jalopnik.com.txt @@ -1,2 +1,5 @@ author: //span[@class='plus-icon'] + +http_header(user-agent): PHP/5.3 + test_url: http://jalopnik.com/5892124/1955-porsche-550-spyder-sells-for-record-3685-million/
\ No newline at end of file diff --git a/jezebel.com.txt b/jezebel.com.txt new file mode 100644 index 0000000..e6ca16a --- /dev/null +++ b/jezebel.com.txt @@ -0,0 +1 @@ +http_header(user-agent): PHP/5.3
\ No newline at end of file diff --git a/kotaku.com.txt b/kotaku.com.txt index be439d7..61ccbc4 100644 --- a/kotaku.com.txt +++ b/kotaku.com.txt @@ -1,2 +1,5 @@ author: //span[@class="plus-icon"] + +http_header(user-agent): PHP/5.3 + test_url: http://kotaku.com/5920211/save-the-furries-on-your-wii-in-this-weeks-nintendo-download
\ No newline at end of file diff --git a/lifehacker.com.txt b/lifehacker.com.txt index ec97f06..2777722 100644 --- a/lifehacker.com.txt +++ b/lifehacker.com.txt @@ -42,6 +42,9 @@ strip: //p[@class="arrow"] # Remove "track" image from article body strip: //img[@alt="track"] + +http_header(user-agent): PHP/5.3 + test_url: http://lifehacker.com/5925801/how-can-i-turn-vague-goals-into-actionable-to+dos test_url: http://lifehacker.com/5941600/hack-an-old-computer-mouse-into-a-wireless-bluetooth-mouse test_url: http://lifehacker.com/what-happens-to-the-brain-when-you-meditate-and-how-it-1202533314
\ No newline at end of file |