/**
* Return the archive file type corresponding to the CU's MIME type or filename extension, or null
* if none.
*/
public String getFromCu(CachedUrl cu) throws MalformedURLException {
String res = getFromMime(cu.getContentType());
if (res == null) {
res = getFromUrl(cu.getUrl());
}
return res;
}
private void deleteBlock(CachedUrl cu) throws IOException {
log.info("deleting " + cu.getUrl());
CachedUrlSetSpec cuss = new SingleNodeCachedUrlSetSpec(cu.getUrl());
ArchivalUnit au = cu.getArchivalUnit();
CachedUrlSet cus = au.makeCachedUrlSet(cuss);
NodeManager nm = au.getPlugin().getDaemon().getNodeManager(au);
nm.deleteNode(cus);
}
/**
* Return true if, interpreting URLs as filenames, dirCu is a directory containing fileCu. Used to
* exclude directory content from output files, so they can be unpacked by standard utilities
* (e.g., unzip). Shouldn't be called with equal URLs, but return false in that case, as we
* wouldn't want to exclude the URL
*/
boolean isDirOf(CachedUrl dirCu, CachedUrl fileCu) {
String dir = dirCu.getUrl();
String file = fileCu.getUrl();
if (!dir.endsWith("/")) {
dir = dir + "/";
}
return file.startsWith(dir) && !file.equals(dir);
}
public void testFunctionalFromTarHierarchy() throws Exception {
log.debug3("in testFromTarHierarchy");
// load the tarballs
InputStream file_input = null;
try {
file_input = getResourceAsStream(realTARFile_A);
// UrlCacher uc = au.makeUrlCacher(TAR_A_BASE);
// uc.storeContent(file_input, tarHeader);
UrlCacher uc = tarAu.makeUrlCacher(new UrlData(file_input, tarHeader, TAR_A_BASE));
uc.storeContent();
IOUtil.safeClose(file_input);
file_input = getResourceAsStream(realTARFile_B);
// uc = au.makeUrlCacher(TAR_B_BASE);
// uc.storeContent(file_input, tarHeader);
uc = tarAu.makeUrlCacher(new UrlData(file_input, tarHeader, TAR_B_BASE));
uc.storeContent();
IOUtil.safeClose(file_input);
} catch (IOException e) {
// TODO Auto-generated catch block
e.printStackTrace();
} finally {
IOUtil.safeClose(file_input);
}
CachedUrlSet cus = tarAu.getAuCachedUrlSet();
for (CachedUrl cu : cus.getCuIterable()) {
log.debug3("AU - cu is: " + cu.getUrl());
cu.release();
}
// We need to start from the level of the ArticleMetadataExtractor
MyListEmitter emitter = new MyListEmitter();
ArticleMetadataExtractor amEx =
new ElsevierDeferredArticleMetadataExtractor(ArticleFiles.ROLE_ARTICLE_METADATA);
Iterator<ArticleFiles> it = tarAu.getArticleIterator(MetadataTarget.Any());
while (it.hasNext()) {
ArticleFiles af = it.next();
log.debug3("Metadata test - articlefiles " + af.toString());
// CachedUrl cu = af.getFullTextCu();
CachedUrl cu = af.getRoleCu(ArticleFiles.ROLE_ARTICLE_METADATA);
log.debug3("metadata cu is " + cu.getUrl());
// List<ArticleMetadata> mdlist = mle.extract(MetadataTarget.Any(), cu);
amEx.extract(MetadataTarget.Any(), af, emitter);
List<ArticleMetadata> returnList = emitter.getAmList();
assertNotNull(returnList);
log.debug3("size of returnList is " + returnList.size());
Iterator<ArticleMetadata> mdIt = returnList.iterator();
ArticleMetadata mdRecord = null;
while (mdIt.hasNext()) {
mdRecord = (ArticleMetadata) mdIt.next();
validateCompleteMetadataRecord(mdRecord);
}
}
}
public void testArticleCountAndType() throws Exception {
int expCount = 28;
PluginTestUtil.crawlSimAu(sau);
String pat1 = "branch(\\d+)/(\\d+file\\.html)";
String rep1 = "aps/journal/v123/n$1/full/$2";
PluginTestUtil.copyAu(sau, nau, ".*[^.][^p][^d][^f]$", pat1, rep1);
String pat2 = "branch(\\d+)/(\\d+file\\.pdf)";
String rep2 = "aps/journal/v123/n$1/pdf/$2";
PluginTestUtil.copyAu(sau, nau, ".*\\.pdf$", pat2, rep2);
// Remove some URLs
int deleted = 0;
for (Iterator it = nau.getAuCachedUrlSet().contentHashIterator(); it.hasNext(); ) {
CachedUrlSetNode cusn = (CachedUrlSetNode) it.next();
if (cusn instanceof CachedUrl) {
CachedUrl cu = (CachedUrl) cusn;
String url = cu.getUrl();
if (url.contains("/journal/")
&& (url.endsWith("1file.html") || url.endsWith("2file.pdf"))) {
deleteBlock(cu);
++deleted;
}
}
}
assertEquals(8, deleted);
Iterator<ArticleFiles> it = nau.getArticleIterator();
int count = 0;
int countHtmlOnly = 0;
int countPdfOnly = 0;
while (it.hasNext()) {
ArticleFiles af = it.next();
log.info(af.toString());
CachedUrl cu = af.getFullTextCu();
String url = cu.getUrl();
assertNotNull(cu);
String contentType = cu.getContentType();
log.debug("count " + count + " url " + url + " " + contentType);
count++;
if (af.getRoleUrl(ArticleFiles.ROLE_FULL_TEXT_PDF) == null) {
++countHtmlOnly;
}
if (af.getRoleUrl(ArticleFiles.ROLE_FULL_TEXT_PDF) == url) {
++countPdfOnly;
}
}
log.debug("Article count is " + count);
assertEquals(expCount, count);
assertEquals(4, countHtmlOnly);
assertEquals(4, countPdfOnly);
}
@Override
public void extract(MetadataTarget target, CachedUrl cu, Emitter emitter) throws IOException {
ArticleMetadata am = new SimpleHtmlMetaTagMetadataExtractor().extract(target, cu);
am.cook(tagMap);
String url = am.get(MetadataField.FIELD_ACCESS_URL);
ArchivalUnit au = cu.getArchivalUnit();
if (url == null || url.isEmpty() || !au.makeCachedUrl(url).hasContent()) {
url = cu.getUrl();
}
am.replace(
MetadataField.FIELD_ACCESS_URL,
HttpToHttpsUtil.AuUtil.normalizeHttpHttpsFromBaseUrl(au, url));
emitter.emitMetadata(cu, am);
}
@Override
public ArticleMetadata extract(MetadataTarget target, CachedUrl cu) throws IOException {
ArticleMetadata am = super.extract(target, cu);
am.cook(tagMap);
String url = am.get(MetadataField.FIELD_ACCESS_URL);
if (url != null && !url.isEmpty()) {
CachedUrl val = cu.getArchivalUnit().makeCachedUrl(url);
if (!val.hasContent()) {
am.replace(MetadataField.FIELD_ACCESS_URL, cu.getUrl());
}
} else {
am.replace(MetadataField.FIELD_ACCESS_URL, cu.getUrl());
}
return am;
}
protected void writeCu(CachedUrl cu) throws IOException {
String url = cu.getUrl();
long contentSize = cu.getContentSize();
CIProperties props = cu.getProperties();
long fetchTime = Long.parseLong(props.getProperty(CachedUrl.PROPERTY_FETCH_TIME));
InputStream contentIn = cu.getUnfilteredInputStream();
try {
if (isResponse) {
String hdrString = getHttpResponseString(cu);
long size = contentSize + hdrString.length();
InputStream headerIn = new ReaderInputStream(new StringReader(hdrString));
InputStream concat = new SequenceInputStream(headerIn, contentIn);
try {
aw.write(xlateFilename(url), cu.getContentType(), getHostIp(), fetchTime, size, concat);
} finally {
IOUtil.safeClose(concat);
}
} else {
aw.write(
xlateFilename(url),
cu.getContentType(),
getHostIp(),
fetchTime,
cu.getContentSize(),
contentIn);
}
} finally {
AuUtil.safeRelease(cu);
}
}
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
public ArticleMetadata extract(MetadataTarget target, CachedUrl cu) throws IOException {
log.debug3("Metadata - cachedurl cu:" + cu.getUrl());
ArticleMetadata am = super.extract(target, cu);
am.cook(tagMap);
return am;
} // extract
protected String getHttpResponseString(CachedUrl cu) {
Properties cuProps = cu.getProperties();
Properties filteredProps = filterResponseProps(cuProps);
String hdrString = PropUtil.toHeaderString(filteredProps);
StringBuilder sb = new StringBuilder(hdrString.length() + 30);
String line1 = inferHttpResponseCode(cu, cuProps);
sb.append(line1);
sb.append(Constants.CRLF);
sb.append(hdrString);
sb.append(Constants.CRLF);
return sb.toString();
}
/**
* Lookup the CU's archive file type in its AU's ArchiveFileTypes
*
* @return the file extension (including dot), or null if none found
*/
public static String getArchiveExtension(CachedUrl cu) {
ArchiveFileTypes aft = cu.getArchivalUnit().getArchiveFileTypes();
if (aft == null) {
return null;
}
try {
return aft.getFromCu(cu);
} catch (MalformedURLException e) {
log.warning("isArchive(" + cu + ")", e);
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();
}
private void writeFiles() {
PlatformUtil platutil = PlatformUtil.getInstance();
CuIterator iter = AuUtil.getCuIterator(au);
int errs = 0;
CachedUrl curCu = null;
CachedUrl nextCu = getNextCu(iter);
while (nextCu != null) {
curCu = nextCu;
nextCu = getNextCu(iter);
if (excludeDirNodes && nextCu != null && isDirOf(curCu, nextCu)) {
continue;
}
CachedUrl[] cuVersions =
curCu.getCuVersions(maxVersions > 0 ? maxVersions : Integer.MAX_VALUE);
for (CachedUrl cu : cuVersions) {
try {
log.debug2("Exporting " + cu.getUrl());
writeCu(cu);
} catch (IOException e) {
if (platutil.isDiskFullError(e)) {
recordError("Disk full, can't write export file.");
isDiskFull = true;
return;
}
} catch (Exception e) {
// XXX Would like to differentiate between errors opening or
// reading CU, which shouldn't cause abort, and errors writing
// to export file, which should.
recordError("Unable to copy " + cu.getUrl(), e);
if (errs++ >= maxErrors) {
recordError("Aborting after " + errs + " errors");
return;
}
}
}
}
}
/*
* 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;
}
/** Explode the archive into its constituent elements */
public void explode() throws CacheException {
CachedUrl cachedUrl = null;
int goodEntries = 0;
int badEntries = 0;
int ignoredEntries = 0;
int entriesBetweenSleep = 0;
ArchiveReader arcReader = null;
logger.info(
(storeArchive ? "Storing" : "Fetching") + " WARC file: " + origUrl + " will explode");
try {
if (storeArchive) {
UrlCacher uc = au.makeUrlCacher(new UrlData(arcStream, arcProps, fetchUrl));
BitSet bs = new BitSet();
bs.set(UrlCacher.DONT_CLOSE_INPUT_STREAM_FLAG);
uc.setFetchFlags(bs);
uc.storeContent();
archiveData.resetInputStream();
arcStream = archiveData.input;
}
// Wrap it in an ArchiveReader
logger.debug3("About to wrap stream");
arcReader = wrapStream(fetchUrl, arcStream);
logger.debug3("wrapStream() returns " + (arcReader == null ? "null" : "non-null"));
// Explode it
if (arcReader == null) {
throw new CacheException.ExploderException("no WarcReader for " + origUrl);
}
ArchivalUnit au = crawlFacade.getAu();
Set stemSet = new HashSet();
logger.debug("Exploding " + fetchUrl);
// Iterate through the elements in the WARC file, except the first
Iterator i = arcReader.iterator();
// Skip first record
for (i.next(); i.hasNext(); ) {
// XXX probably not necessary
helper.pokeWDog();
if ((++entriesBetweenSleep % sleepAfter) == 0) {
long pauseTime =
CurrentConfig.getTimeIntervalParam(PARAM_RETRY_PAUSE, DEFAULT_RETRY_PAUSE);
Deadline pause = Deadline.in(pauseTime);
logger.debug3("Sleeping for " + StringUtil.timeIntervalToString(pauseTime));
while (!pause.expired()) {
try {
pause.sleep();
} catch (InterruptedException ie) {
// no action
}
}
}
ArchiveRecord element = (ArchiveRecord) i.next();
// Each element is a URL to be cached in a suitable AU
ArchiveRecordHeader elementHeader = element.getHeader();
String elementUrl = elementHeader.getUrl();
String elementMimeType = elementHeader.getMimetype();
long elementLength = elementHeader.getLength();
logger.debug2("WARC url " + elementUrl + " mime " + elementMimeType);
if (elementUrl.startsWith("http:")) {
ArchiveEntry ae =
new ArchiveEntry(
elementUrl,
elementLength,
0, // XXX need to convert getDate string to long
element, // ArchiveRecord extends InputStream
this,
fetchUrl);
ae.setHeaderFields(makeCIProperties(elementHeader));
long bytesStored = elementLength;
logger.debug3("ArchiveEntry: " + ae.getName() + " bytes " + bytesStored);
try {
helper.process(ae);
} catch (PluginException ex) {
throw new CacheException.ExploderException("helper.process() threw", ex);
}
if (ae.getBaseUrl() != null) {
if (ae.getRestOfUrl() != null && ae.getHeaderFields() != null) {
storeEntry(ae);
handleAddText(ae);
goodEntries++;
crawlFacade.getCrawlerStatus().addContentBytesFetched(bytesStored);
} else {
ignoredEntries++;
}
} else {
badEntries++;
logger.debug2("Can't map " + elementUrl + " from " + archiveUrl);
}
}
}
} catch (IOException ex) {
throw new CacheException.ExploderException(ex);
} finally {
if (arcReader != null)
try {
arcReader.close();
arcReader = null;
} catch (IOException ex) {
throw new CacheException.ExploderException(ex);
}
if (cachedUrl != null) {
cachedUrl.release();
}
IOUtil.safeClose(arcStream);
}
if (badEntries == 0 && goodEntries > 0) {
// Make it look like a new crawl finished on each AU to which
// URLs were added.
for (Iterator it = touchedAus.iterator(); it.hasNext(); ) {
ArchivalUnit au = (ArchivalUnit) it.next();
logger.debug3(archiveUrl + " touching " + au.toString());
AuUtil.getDaemon(au).getNodeManager(au).newContentCrawlFinished();
}
} else {
ArchivalUnit au = crawlFacade.getAu();
String msg = archiveUrl + ": " + badEntries + "/" + goodEntries + " bad entries";
throw new CacheException.UnretryableException(msg);
}
}