Added support for archive.rebeccablacktech.com.HEADmaster

Patched regex generator to now match any URL ending with a downloaded
file for replacement in the HTML. This is due to some sites hosting the
same file across multiple CDN servers, thus creating different links to
the same file within the page.

1 files changed, 5 insertions, 4 deletions

diff --git a/4car b/4car
index 0417d21..727a6e2 100755
--- a/4car
+++ b/4car
@@ -24,9 +24,10 @@ split_pipe() {
 }
 
 filter_image_urls() {
-  grep -oe 'http[s]*:\/\/i\.4cdn\.org\/[a-z0-9]\+\/[0-9]\+\.[a-zA-Z0-9]\+' \
-    -e 'http[s]*:\/\/is[0-9]\+\.4chan\.org\/[a-z0-9]\+\/[0-9]\+\.[a-zA-Z0-9]\+' \
-    -e 'http[s]*:\/\/img\.fireden\.net\/[a-z0-9]\+\/image\/[0-9]\+\/[0-9]\+\/[0-9]\+\.[a-zA-Z0-9]\+'
+  grep -oe 'http[s]\?:\/\/i\.4cdn\.org\/[a-z0-9]\+\/[0-9]\+\.[a-zA-Z0-9]\+' \
+    -e 'http[s]\?:\/\/is[0-9]\+\.4chan\.org\/[a-z0-9]\+\/[0-9]\+\.[a-zA-Z0-9]\+' \
+    -e 'http[s]\?:\/\/img\.fireden\.net\/[a-z0-9]\+\/image\/[0-9]\+\/[0-9]\+\/[0-9]\+\.[a-zA-Z0-9]\+' \
+    -e 'http[s]\?:\/\/s[0-9]\+\.desu-usergeneratedcontent\.xyz\/[a-z0-9]\+\/image\/[0-9]\+\/[0-9]\+\/[0-9]\+\.[a-zA-Z0-9]\+'
 }
 
 # Check dependencies
@@ -59,7 +60,7 @@ for i in "$@"; do
     wget -i - -P "$dir" -nv |&
     split_pipe |
     filter_image_urls |
-    sed "s/\(.*\)\/\(.*\)/s;\1\/\2;.\/$rdir\/\2;g/" |
+    sed "s/\(.*\)\/\(.*\)/s;http[s]\\\\{0,1\\\\}:\/\/[^ \"]*\/\2;.\/$rdir\/\2;g/" |
     sed -f - "$file" |
     sponge "$file"
 done