/**
* Create a FilteredInputStream that excludes the the atyponBaseFilters and moreNodes
*
* @param au The archival unit
* @param in Incoming input stream
* @param encoding The encoding
* @param moreNodes An array of NodeFilters to be excluded with atyponBaseFilters
*/
public InputStream createFilteredInputStream(
ArchivalUnit au, InputStream in, String encoding, NodeFilter[] moreNodes)
throws PluginException {
NodeFilter[] bothFilters = addTo(moreNodes);
return new HtmlFilterInputStream(
in, encoding, HtmlNodeFilterTransform.exclude(new OrFilter(bothFilters)));
}
public InputStream createFilteredInputStream(ArchivalUnit au, InputStream in, String encoding) {
NodeFilter[] filters =
new NodeFilter[] {
// filter out script
new TagNameFilter("script"),
// Menu and related articles/other issues
HtmlNodeFilters.tagWithAttribute("div", "class", "A_Left_Column"),
// Footer menu that seems to currently be blank
HtmlNodeFilters.tagWithAttribute("div", "class", "A_Foot_Menu"),
// Copyright
HtmlNodeFilters.tagWithAttribute("div", "class", "A_Foot_Copy"),
// Lazy HTML these are inserted everywhere in an attempt to get the layout to work
HtmlNodeFilters.tagWithAttribute("div", "style", "clear:both;"),
// search box with broken layout. they may try to fix this
HtmlNodeFilters.tagWithAttribute("div", "id", "search"),
// another blank navigation div
HtmlNodeFilters.tagWithAttribute("div", "id", "Sub_Top_Nav"),
};
return new HtmlFilterInputStream(
in, encoding, HtmlNodeFilterTransform.exclude(new OrFilter(filters)));
}
/**
* Create a FilteredInputStream that excludes the the atyponBaseFilters
*
* @param au The archival unit
* @param in Incoming input stream
* @param encoding The encoding
*/
public InputStream createFilteredInputStream(ArchivalUnit au, InputStream in, String encoding)
throws PluginException {
return new HtmlFilterInputStream(
in, encoding, HtmlNodeFilterTransform.exclude(new OrFilter(baseAtyponFilters)));
}
