protected ArticleFiles processFullTextPdf(CachedUrl cu, Matcher mat) {
ArticleFiles af = new ArticleFiles();
af.setRoleCu(ArticleFiles.ROLE_FULL_TEXT_PDF, cu);
String abstractUrl = mat.replaceAll("$1aspx");
CachedUrl abstractCu = cu.getArchivalUnit().makeCachedUrl(abstractUrl);
if (abstractCu.hasContent()) af.setFullTextCu(abstractCu);
else af.setFullTextCu(cu);
log.debug3("returning full text: " + af.getFullTextUrl());
return af;
}
private String getUrlContent(CachedUrl url) throws IOException {
InputStream content = url.getUnfilteredInputStream();
ByteArrayOutputStream baos = new ByteArrayOutputStream();
StreamUtil.copy(content, baos);
content.close();
String contentStr = new String(baos.toByteArray());
baos.close();
return contentStr;
}
@Override
protected ArticleFiles createArticleFiles(CachedUrl cu) {
String url = cu.getUrl();
Matcher mat = ABSTRACT_PATTERN.matcher(url);
if (mat.find()) {
return processAbstract(cu, mat);
}
log.warning("Mismatch between article iterator factory and article iterator: " + url);
return null;
}
@Override
protected ArticleFiles createArticleFiles(CachedUrl cu) {
String url = cu.getUrl();
log.debug3("iterating url: " + url);
Matcher mat = PATTERN.matcher(url);
if (mat.find()) {
return processFullTextPdf(cu, mat);
}
log.warning("Mismatch between article iterator factory and article iterator: " + url);
return null;
}
protected void checkFilter(SimulatedArchivalUnit sau) throws Exception {
log.debug("checkFilter()");
CachedUrl cu = sau.makeCachedUrl(sau.getUrlRoot() + "/001file.html");
enableFilter(sau, true);
InputStream is = cu.openForHashing();
String expected = "001file.html This is file 1, depth 0, branch 0. foobar ";
assertEquals(expected, StringUtil.fromInputStream(is));
is.close();
enableFilter(sau, false);
cu = sau.makeCachedUrl(sau.getUrlRoot() + "/001file.html");
is = cu.openForHashing();
expected =
"<HTML><HEAD><TITLE>001file.html</TITLE></HEAD><BODY>\n"
+ "This is file 1, depth 0, branch 0.<br><!-- comment --> "
+ "Citation String foobar<br><script>"
+ "(defun fact (n) (cond ((= n 0) 1) (t (fact (sub1 n)))))</script>\n"
+ "</BODY></HTML>";
assertEquals(expected, StringUtil.fromInputStream(is));
is.close();
}
/*
* hasArticleMetadata(CachedUrl cu)
* Given the CachedUrl for the potential abstract file, using the existing
* SimpleHtmlMetaTagMetadataExtractor to parse the file and
* retrieve any contained metadata. If a doi or author exists, it's an article
* NOT defining the Metadata Extractor here!
*/
private boolean hasArticleMetadata(CachedUrl cu) {
MetadataTarget at = new MetadataTarget(MetadataTarget.PURPOSE_ARTICLE);
ArticleMetadata am;
SimpleHtmlMetaTagMetadataExtractor ext = new SimpleHtmlMetaTagMetadataExtractor();
if (cu != null && cu.hasContent()) {
try {
at.setFormat("text/html");
am = ext.extract(at, cu);
if ((am.containsRawKey("bepress_citation_journal_title"))
|| (am.containsRawKey("bepress_citation_abstract_html_url"))
|| (am.containsRawKey("bepress_citation_doi"))
|| (am.containsRawKey("bepress_citation_author"))) {
return true;
}
} catch (IOException e) {
e.printStackTrace();
}
}
return false; // no reasonable metadata, probably a toc
}
/*
* This is comlicated. MOST AUs have articles that live below and issue level TOC
* that is,
* <blah>/<journal_id>/vol#/iss#/ is a toc with no relevant metadata
* <blah>/<journal_id>/vol#/iss#/xxx is an article with metadata
* (eg Economist Voice V1)
* BUT
* in some AUs there are issues with only 1 article, in which case
* <blah>/<journal_id>/vol#/iss#/ is an abstract with metadata
* (eg Rhodes Cook V4)
* and a few AUs with a mixture
* (eg Forum for Health Economics V5)
* So to identify ALL articles, we'll also have to capture issue level items and then look
* at the html and if it has article metadata in it, count it as an article.
*
*/
@Override
protected ArticleFiles createArticleFiles(CachedUrl cu) {
String url = cu.getUrl();
Matcher mat = pattern.matcher(url);
if (mat.find()) {
// we matched, but could this pattern potentially be a toc?
Matcher tocmat = TOC_pattern.matcher(url);
// if we could be a TOC then we must have metadata to be considered an article
if (tocmat.find()) {
if (hasArticleMetadata(cu)) {
return processUrl(cu, mat);
}
} else {
// we're not a potential TOC, so treat this as an article without checking
return processUrl(cu, mat);
}
return null; // this was a TOC, not an article
}
log.warning("Mismatch between article iterator factory and article iterator: " + url);
return null;
}
/*
* In order to find full text PDF you need to find the citation_pdf_url meta tag in the
* abstract html pull out the pdf url normalize it (reorder params...) and find the matching
* cached URL
*/
protected ArticleFiles processAbstract(CachedUrl absCu, Matcher absMat) {
NodeList nl = null;
ArticleFiles af = new ArticleFiles();
if (absCu != null && absCu.hasContent()) {
// TEMPORARY: set absCU as default full text CU in case there is
// no PDF CU with content; the current metadata manager currently
// uses only the full text CU, but this will change with the new
// metadata schema that can have multiple CUs for an article.
af.setFullTextCu(absCu);
af.setRoleCu(ArticleFiles.ROLE_ABSTRACT, absCu);
try {
InputStreamSource is =
new InputStreamSource(new Stream(absCu.getUnfilteredInputStream()));
Page pg = new Page(is);
Lexer lx = new Lexer(pg);
Parser parser = new Parser(lx);
Lexer.STRICT_REMARKS = false;
NodeFilter nf =
new NodeFilter() {
public boolean accept(Node node) {
if (!(node instanceof MetaTag)) return false;
MetaTag meta = (MetaTag) node;
if (!"citation_pdf_url".equalsIgnoreCase(meta.getMetaTagName())) return false;
return true;
}
};
nl = parser.extractAllNodesThatMatch(nf);
} catch (ParserException e) {
log.debug("Unable to parse abstract page html", e);
} catch (UnsupportedEncodingException e) {
log.debug("Bad encoding in abstact page html", e);
} finally {
absCu.release();
}
}
try {
if (nl != null) {
if (nl.size() > 0) {
// minimally encode URL to prevent URL constructor
// from stripping trailing spaces
String pdfUrlStr = ((MetaTag) nl.elementAt(0)).getMetaContent();
URL pdfUrl = new URL(UrlUtil.minimallyEncodeUrl(pdfUrlStr));
List<String> paramList = new ArrayList<String>();
paramList.add("fileType");
paramList.add("fileId");
paramList.add("fileName");
pdfUrl = reArrangeUrlParams(pdfUrl, paramList);
if (!pdfUrl.getHost().startsWith("www.")) {
pdfUrl = new URL(pdfUrl.getProtocol(), "www." + pdfUrl.getHost(), pdfUrl.getFile());
}
// note: must leave URL encoded because that's how we store URLs
CachedUrl pdfCu = au.makeCachedUrl(pdfUrl.toString());
if (pdfCu != null && pdfCu.hasContent()) {
// replace absCU with pdfCU if exists and has content
af.setFullTextCu(pdfCu);
af.setRoleCu(ArticleFiles.ROLE_FULL_TEXT_PDF, pdfCu);
}
}
}
} catch (MalformedURLException e) {
log.debug("Badly formatted pdf url link", e);
} catch (IllegalArgumentException e) {
log.debug("Badly formatted pdf url link", e);
}
return af;
}
