public static String getRepositorySpec(ArchivalUnit au) {
Configuration auConfig = au.getConfiguration();
if (auConfig != null) { // can be null in unit tests
String repoSpec = auConfig.get(PluginManager.AU_PARAM_REPOSITORY);
if (repoSpec != null && repoSpec.startsWith("local:")) {
return repoSpec;
}
}
return "local:" + CurrentConfig.getParam(PARAM_CACHE_LOCATION);
}
public void process(ArchiveEntry ae) {
// By default the files have to go in the crawler's AU
ArchivalUnit au = crawlFacade.getAu();
// By default the path should start at the AU's base url.
Configuration config = au.getConfiguration();
String url = config.get(ConfigParamDescr.BASE_URL.getKey());
ae.setBaseUrl(url);
ae.setRestOfUrl(ae.getName());
CIProperties cip = new CIProperties();
ae.setHeaderFields(cip);
}
public BePressArticleIterator(
ArchivalUnit au, SubTreeArticleIterator.Spec spec, boolean isSection) {
super(au, spec);
String volumeAsString = au.getConfiguration().get(ConfigParamDescr.VOLUME_NUMBER.getKey());
String journalAbbr = au.getConfiguration().get(ConfigParamDescr.JOURNAL_ABBR.getKey());
if (isSection) {
journalAbbr = journalAbbr + "/" + au.getConfiguration().get("journal_section");
}
// pick up issue level and lower (make (art)?[0-9]+ optional because a few au's have article
// at issue level
this.pattern =
Pattern.compile(
String.format(
"/%s/((([^0-9]+/)?(vol)?%s/(iss)?[0-9]+(/(art)?[0-9]+)?)|(vol%s/(?-i:[A-Z])[0-9]+))$",
journalAbbr, volumeAsString, volumeAsString),
Pattern.CASE_INSENSITIVE);
this.TOC_pattern =
Pattern.compile(
String.format("/%s/([^0-9]+/)?(vol)?%s/(iss)?[0-9]+$", journalAbbr, volumeAsString),
Pattern.CASE_INSENSITIVE);
}