/**
* 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)
throws PluginException {
HtmlTransform[] transforms =
new HtmlTransform[] {
// Filter out <a target="_blank">...</a>
HtmlNodeFilterTransform.exclude(
HtmlNodeFilters.tagWithAttribute("a", "target", "_blank")),
};
return new HtmlFilterInputStream(in, encoding, new HtmlCompoundTransform(transforms));
}
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)));
}
public InputStream createFilteredInputStream(ArchivalUnit au, InputStream in, String encoding)
throws PluginException {
NodeFilter[] filters =
new NodeFilter[] {
// Filter out <div id="footer">...</div>
HtmlNodeFilters.tagWithAttribute("div", "id", "footer"),
// Filter out <div id="top-ad-alignment">...</div>
HtmlNodeFilters.tagWithAttribute("div", "id", "top-ad-alignment"),
// Filter out <div id="top-ad">...</div>
HtmlNodeFilters.tagWithAttribute("div", "id", "top-ad"),
// Filter out <div id="ident">...</div>
HtmlNodeFilters.tagWithAttribute("div", "id", "ident"),
// Filter out <div id="ad">...</div>
HtmlNodeFilters.tagWithAttribute("div", "id", "ad"),
// Filter out <div id="vertical-ad">...</div>
HtmlNodeFilters.tagWithAttribute("div", "id", "vertical-ad"),
// Filter out <div class="right-col-download">...</div>
HtmlNodeFilters.tagWithAttribute("div", "class", "right-col-download"),
// Filter out <div id="cart-navbar">...</div>
HtmlNodeFilters.tagWithAttribute("div", "id", "cart-navbar"),
// // Filter out <div class="heading-macfix article-access-options">...</div>
// HtmlNodeFilters.tagWithAttribute("div", "class", "heading-macfix
// article-access-options"),
// Filter out <div id="baynote-recommendations">...</div>
HtmlNodeFilters.tagWithAttribute("div", "id", "baynote-recommendations"),
// Filter out <div id="bookmarks-container">...</div>
HtmlNodeFilters.tagWithAttribute("div", "id", "bookmarks-container"),
// Filter out <div id="llb">...</div>
HtmlNodeFilters.tagWithAttribute("div", "id", "llb"),
// Filter out <a href="...">...</a> where the href value includes "exitTargetId" as a
// parameter
HtmlNodeFilters.tagWithAttributeRegex("a", "href", "[\\?&]exitTargetId="),
// Filter out <input name="exitTargetId">
HtmlNodeFilters.tagWithAttribute("input", "name", "exitTargetId"),
};
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)));
}