/**
* @param desc
* @return
*/
private String extractContents(String desc) {
Pattern p = Pattern.compile("((?:an?|the) (.*?)(?: of (.*))?)(?:,|\\s*located)");
Pattern p2 = Pattern.compile("((incomplete .*?)(?: of (.*))?)(?:,|\\s*located)");
String result = desc;
Matcher mat = p.matcher(desc);
Matcher mat2 = p2.matcher(desc);
mat = mat.find() ? mat : mat2.find() ? mat2 : null;
if (mat != null) {
String matchedText = mat.group(1);
int start = mat.start();
int end =
(matchedText.indexOf("located") >= 0)
? end = mat.start(1) + matchedText.indexOf("located")
: mat.end(1);
String contents = desc.substring(mat.start(1), end);
result = desc.substring(0, start) + desc.substring(end);
contents = StringUtils.trimToEmpty(contents);
if (contents.lastIndexOf(",") == contents.length() - 1) {
contents = contents.substring(0, contents.length() - 1);
}
ms.setContents(contents);
LOGGER.debug(" Contents: " + contents);
}
return result;
}
private void parseDescription() {
String desc = this.description;
ms.setDescription(desc);
desc = desc.replaceFirst("-+see German footnotes?;?-*\\s*", "");
if (StringUtils.isBlank(desc)) {
return;
} else if (desc.equals("paraphrase")) {
return;
}
if (debug) System.out.println("Parsing: " + desc);
desc = stripSeeAlso(desc);
desc = extractDate(desc);
desc = extractCodexName(desc);
desc = extractContents(desc);
desc = extractLocation(desc);
desc = extractRuinedMs(desc);
desc = extractClassification(desc);
if (!debug && !desc.matches("\\W*")) {
if (desc.matches("\\W*(from( the)?)? ?same.*")) {
// TODO parse related MS info.
} else {
// System.out.println(" Result: " + desc + "\t:: " + this.description );
unparsedDateCt++;
}
}
if (debug) System.out.println(" Result: " + desc);
if (debug) System.out.println();
}
/**
* @param desc
* @return
*/
private String extractLocation(String desc) {
Pattern locationPattern =
Pattern.compile(
"(?:, located, )?"
+ "lo?cate?d "
+ "(formerly|possibly)? ?"
+ "(?:at|on|in(?: the)?|with(?: the)?)? ?"
+ "((?:[^,]|(?:, [A-Z]{2}))*),? ?");
String result = desc;
Matcher mat = locationPattern.matcher(desc);
while (mat.find()) {
String location = mat.group(2);
Institution inst = institutionRepo.findOrCreate(location);
// em.merge(inst);
ms.setCurrentInstitution(inst);
result = desc.substring(0, mat.start()) + desc.substring(mat.end());
LOGGER.debug(" Location: " + location);
}
Pattern oldLocationPattern =
Pattern.compile("formerly " + "(?:at|in(?: the)?|with(?: the)?) " + "(.*)?");
mat = oldLocationPattern.matcher(result);
while (mat.find()) {
String institution = mat.group(1);
Institution inst = institutionRepo.findOrCreate(institution);
System.out.println(institution);
// em.merge(inst);
ms.addPreviousInstitution(inst);
result = result.substring(0, mat.start()) + result.substring(mat.end());
LOGGER.debug("Former Location: " + institution);
}
return result;
}
/**
* @param desc
* @return
*/
public String extractDate(String desc) {
String result = desc;
Matcher mat = datePattern.matcher(desc);
if (mat.find() && (mat.start() == 0)) {
HistoricalDate date = new HistoricalDate(mat.group(1));
em.persist(date);
ms.setDate(date);
result = desc.substring(mat.end());
if (date.getText().matches(UNKNOWN_DATE_RE)) {
return result;
}
String prefix = StringUtils.trimToNull(mat.group(2));
String firstDate = StringUtils.trimToNull(mat.group(3));
String separator = StringUtils.trimToNull(mat.group(4));
String lastDate = StringUtils.trimToNull(mat.group(5));
String suffix = StringUtils.trimToNull(mat.group(6));
if (suffix != null) {
if (suffix.matches("c\\.?")) {
date.setPrecision(Precision.CENTURY);
} else {
// Doesn't seem to happen
}
}
if (prefix != null) {
if (prefix.equalsIgnoreCase("ca.")) {
// mark date as approximate
date.setCertainty(Certainty.APPROXIMATE);
} else if (prefix.equalsIgnoreCase("possibly")) {
// mark date as uncertain
date.setCertainty(Certainty.POSSIBLE);
} else {
// doesen't occur
}
}
setStartDate(firstDate, date);
setEndDate(firstDate, lastDate, separator, date);
LOGGER.debug(" Date: " + date.getText());
} else {
// System.out.println(desc);
}
return result;
}
