/**
* Inspects candidate group to determine if it comprises instance of Format
*
* @param jhove2 framework configured with SourceFactory
* @return FormatIdentification for this group if it comprises instance of a Format; otherwise
* returns null;
* @throws JHOVE2Exception
*/
protected ClumpSource recognizeGroupedSource(GlobPathMatchInfoGroup fileGroup, JHOVE2 jhove2)
throws JHOVE2Exception {
FormatIdentification fi = null;
ClumpSource clumpSource = null;
if (fileGroup.getMustHaveCount() >= this.minMustHavesToIdentify) {
fi =
new FormatIdentification(
this.format, GLOB_PATH_CONFIDENCE, this.getReportableIdentifier());
if (jhove2.getSourceFactory() == null) {
throw new JHOVE2Exception("JHOVE2 SourceFactory is null");
}
clumpSource = jhove2.getSourceFactory().getClumpSource(jhove2);
clumpSource = (ClumpSource) clumpSource.addPresumptiveFormat(fi);
for (GlobPathMatchInfo sourceInfo : fileGroup.getSourceMatchInfoList()) {
if ((sourceInfo.isMustHave() || sourceInfo.isMayHave())
|| (this.includeUnmatchedFromGroup)) {
Source sourceInfoSource = sourceInfo.getSource();
sourceInfoSource = clumpSource.addChildSource(sourceInfoSource);
sourceInfo.setSource(sourceInfoSource);
}
}
}
return clumpSource;
}
/**
* Constructs candidate instances of an aggregate format by grouping together all children of
* source parameter that match fileGroupingToken
*
* @param source Source object whose child Sources are to be explored for groups constituting
* instances of a format
* @return Collection of GlobPathMatchInfoGroup objects, each of which contains a list of likely
* related Sources comprising an instance of a Format, and indications as to whether or not
* each Source in the group is a required or optional or unspecified component of that Format
* instances
* @throws JHOVE2Exception
*/
protected Collection<GlobPathMatchInfoGroup> groupSources(Source source) throws JHOVE2Exception {
HashMap<String, GlobPathMatchInfoGroup> groupMap =
new HashMap<String, GlobPathMatchInfoGroup>();
for (Source childSource : source.getChildSources()) {
File sourceFile = childSource.getFile();
if (sourceFile != null) {
String filePath = childSource.getFile().getPath();
// does the Source file path match the pattern that indicates a related file?
Matcher m = this.fileGroupingPattern.matcher(filePath);
if (m.matches()) {
// might have more than one instance of a format in the Source, so
// we have to group related files together
String groupString = null;
String mustHaveString = null;
String mayHaveString = null;
try {
// get the value of the capture group which is the key to a format instance
// (group of files)
groupString = m.group(this.fileGroupingCaptureGroupIndex);
// get the value of the capture group that indicates a file in the group
// is one of the files required by the format definition
mustHaveString = m.group(this.mustHaveCaptureGroupIndex);
// get the value of the capture group that indicates a file in the group
// is one of the files considered optional by the format definition
mayHaveString = m.group(this.mayHaveCaptureGroupIndex);
} catch (IllegalStateException ise) {
// should not occur, we were inside if (m.matches()) statement
throw new JHOVE2Exception(
"Exception thrown grouping patterns: Check configuration", ise);
} catch (IndexOutOfBoundsException iob) {
throw new JHOVE2Exception(
"Exception thrown grouping patterns: Check configuration", iob);
}
GlobPathMatchInfo fileInfo = new GlobPathMatchInfo(childSource);
boolean matchesMustHaves = false;
Matcher m2 = null;
if (this.mustHavePattern != null) {
m2 = this.mustHavePattern.matcher(mustHaveString);
matchesMustHaves = m2.matches();
}
fileInfo.setMustHave(matchesMustHaves);
boolean matchesMayHaves = false;
if (this.mayHavePattern != null) {
m2 = this.mayHavePattern.matcher(mayHaveString);
matchesMayHaves = m2.matches();
}
fileInfo.setMayHave(matchesMayHaves);
GlobPathMatchInfoGroup infoGroup;
// is this the first occurrence of grouping key?
if (!(groupMap.containsKey(groupString))) {
// if so, add grouping key and new GlobPathMatchInfoGroup to groupMaP
infoGroup = new GlobPathMatchInfoGroup();
infoGroup.setGroupKey(groupString);
groupMap.put(groupString, infoGroup);
} else {
// otherwise just retrieve
infoGroup = groupMap.get(groupString);
}
// add information about current Source to list associated with this grouping key
infoGroup.getSourceMatchInfoList().add(fileInfo);
// increment counter information associated with this grouping key
if (matchesMustHaves) {
infoGroup.setMustHaveCount(infoGroup.getMustHaveCount() + 1);
}
if (matchesMayHaves) {
infoGroup.setMayHaveCount(infoGroup.getMayHaveCount() + 1);
}
if (!matchesMustHaves && !matchesMayHaves) {
infoGroup.setUnmatchedCount(infoGroup.getUnmatchedCount() + 1);
}
} // end if (m.matches()){
} // end if sourceFile != null
} // end for (Source childSource:source.getChildSources())
// we don't need the keys to the map any more; just return the values
return groupMap.values();
}
