private BibtexEntry cleanup(BibtexEntry entry) {
if (entry == null) {
return null;
}
// clean up title
String title = entry.getField("title");
if (title != null) {
// USe the alt-text and replace image links
title = title.replaceAll("[ ]?img src=[^ ]+ alt=\"([^\"]+)\">[ ]?", "\\$$1\\$");
// Try to sort out most of the /spl / conversions
// Deal with this specific nested type first
title = title.replaceAll("/sub /spl infin//", "\\$_\\\\infty\\$");
title = title.replaceAll("/sup /spl infin//", "\\$\\^\\\\infty\\$");
// Replace general expressions
title = title.replaceAll("/[sS]pl ([^/]+)/", "\\$\\\\$1\\$");
// Deal with subscripts and superscripts
if (Globals.prefs.getBoolean(JabRefPreferences.USE_CONVERT_TO_EQUATION)) {
title = title.replaceAll("/sup ([^/]+)/", "\\$\\^\\{$1\\}\\$");
title = title.replaceAll("/sub ([^/]+)/", "\\$_\\{$1\\}\\$");
title = title.replaceAll("\\(sup\\)([^(]+)\\(/sup\\)", "\\$\\^\\{$1\\}\\$");
title = title.replaceAll("\\(sub\\)([^(]+)\\(/sub\\)", "\\_\\{$1\\}\\$");
} else {
title = title.replaceAll("/sup ([^/]+)/", "\\\\textsuperscript\\{$1\\}");
title = title.replaceAll("/sub ([^/]+)/", "\\\\textsubscript\\{$1\\}");
title = title.replaceAll("\\(sup\\)([^(]+)\\(/sup\\)", "\\\\textsuperscript\\{$1\\}");
title = title.replaceAll("\\(sub\\)([^(]+)\\(/sub\\)", "\\\\textsubscript\\{$1\\}");
}
// Replace \infin with \infty
title = title.replaceAll("\\\\infin", "\\\\infty");
// Unit formatting
if (Globals.prefs.getBoolean(JabRefPreferences.USE_UNIT_FORMATTER_ON_SEARCH)) {
title = unitFormatter.format(title);
}
// Automatic case keeping
if (Globals.prefs.getBoolean(JabRefPreferences.USE_CASE_KEEPER_ON_SEARCH)) {
title = caseKeeper.format(title);
}
// Write back
entry.setField("title", title);
}
// clean up author
/* String author = (String)entry.getField("author");
if (author != null) {
if (author.indexOf("a href=") >= 0) { // Author parsing failed because it was empty
entry.setField("author",""); // Maybe not needed anymore due to another change
} else {
author = author.replaceAll("\\s+", " ");
author = author.replaceAll("\\.", ". ");
author = author.replaceAll("([^;]+),([^;]+),([^;]+)","$1,$3,$2"); // Change order in case of Jr. etc
author = author.replaceAll(" ", " ");
author = author.replaceAll("\\. -", ".-");
author = author.replaceAll("; ", " and ");
author = author.replaceAll(" ,", ",");
author = author.replaceAll(" ", " ");
author = author.replaceAll("[ ,;]+$", "");
entry.setField("author", author);
}
}*/
// clean up month
String month = entry.getField("month");
if (month != null && !month.isEmpty()) {
month = month.replaceAll("\\.", "");
month = month.toLowerCase();
Pattern monthPattern = Pattern.compile("(\\d*+)\\s*([a-z]*+)-*(\\d*+)\\s*([a-z]*+)");
Matcher mm = monthPattern.matcher(month);
String date = month;
if (mm.find()) {
if (mm.group(3).isEmpty()) {
if (!mm.group(2).isEmpty()) {
date = "#" + mm.group(2).substring(0, 3) + "#";
if (!mm.group(1).isEmpty()) {
date += " " + mm.group(1) + ",";
}
} else {
date = mm.group(1) + ",";
}
} else if (mm.group(2).isEmpty()) {
if (!mm.group(4).isEmpty()) {
date =
"#" + mm.group(4).substring(0, 3) + "# " + mm.group(1) + "--" + mm.group(3) + ",";
} else {
date += ",";
}
} else {
date =
"#"
+ mm.group(2).substring(0, 3)
+ "# "
+ mm.group(1)
+ "--#"
+ mm.group(4).substring(0, 3)
+ "# "
+ mm.group(3)
+ ",";
}
}
// date = date.trim();
// if (!date.isEmpty()) {
entry.setField("month", date);
// }
}
// clean up pages
String field = "pages";
String pages = entry.getField(field);
if (pages != null) {
String[] pageNumbers = pages.split("-");
if (pageNumbers.length == 2) {
if (pageNumbers[0].equals(pageNumbers[1])) { // single page
entry.setField(field, pageNumbers[0]);
} else {
entry.setField(field, pages.replaceAll("-", "--"));
}
}
}
// clean up publication field
BibtexEntryType type = entry.getType();
String sourceField = "";
if (type.getName().equals("Article")) {
sourceField = "journal";
entry.clearField("booktitle");
} else if (type.getName().equals("Inproceedings")) {
sourceField = "booktitle";
}
String fullName = entry.getField(sourceField);
if (fullName != null) {
if (type.getName().equals("Article")) {
int ind = fullName.indexOf(": Accepted for future publication");
if (ind > 0) {
fullName = fullName.substring(0, ind);
entry.setField("year", "to be published");
entry.clearField("month");
entry.clearField("pages");
entry.clearField("number");
}
String[] parts = fullName.split("[\\[\\]]"); // [see also...], [legacy...]
fullName = parts[0];
if (parts.length == 3) {
fullName += parts[2];
}
if (entry.getField("note").equals("Early Access")) {
entry.setField("year", "to be published");
entry.clearField("month");
entry.clearField("pages");
entry.clearField("number");
}
} else {
fullName =
fullName
.replace("Conference Proceedings", "Proceedings")
.replace("Proceedings of", "Proceedings")
.replace("Proceedings.", "Proceedings");
fullName = fullName.replaceAll("International", "Int.");
fullName = fullName.replaceAll("Symposium", "Symp.");
fullName = fullName.replaceAll("Conference", "Conf.");
fullName = fullName.replaceAll(" on", " ").replace(" ", " ");
}
Matcher m1 = publicationPattern.matcher(fullName);
String abrvPattern = ".*[^,] '?\\d+\\)?";
if (m1.find()) {
String prefix = m1.group(2).trim();
String postfix = m1.group(1).trim();
String abrv = "";
String[] parts = prefix.split("\\. ", 2);
if (parts.length == 2) {
if (parts[0].matches(abrvPattern)) {
prefix = parts[1];
abrv = parts[0];
} else {
prefix = parts[0];
abrv = parts[1];
}
}
if (!prefix.matches(abrvPattern)) {
fullName = prefix + " " + postfix + " " + abrv;
fullName = fullName.trim();
} else {
fullName = postfix + " " + prefix;
}
}
if (type.getName().equals("Article")) {
fullName = fullName.replace(" - ", "-"); // IEE Proceedings-
fullName = fullName.trim();
if (Globals.prefs.getBoolean(JabRefPreferences.USE_IEEE_ABRV)) {
fullName = Globals.journalAbbrev.getMedlineAbbreviation(fullName).orElse(fullName);
}
}
if (type.getName().equals("Inproceedings")) {
Matcher m2 = proceedingPattern.matcher(fullName);
if (m2.find()) {
String prefix = m2.group(2);
String postfix = m2.group(1).replaceAll("\\.$", "");
if (!prefix.matches(abrvPattern)) {
String abrv = "";
String[] parts = postfix.split("\\. ", 2);
if (parts.length == 2) {
if (parts[0].matches(abrvPattern)) {
postfix = parts[1];
abrv = parts[0];
} else {
postfix = parts[0];
abrv = parts[1];
}
}
fullName = prefix.trim() + " " + postfix.trim() + " " + abrv;
} else {
fullName = postfix.trim() + " " + prefix.trim();
}
}
fullName = fullName.trim();
fullName =
fullName.replaceAll("^[tT]he ", "").replaceAll("^\\d{4} ", "").replaceAll("[,.]$", "");
String year = entry.getField("year");
fullName = fullName.replaceAll(", " + year + "\\.?", "");
if (!fullName.contains("Abstract")
&& !fullName.contains("Summaries")
&& !fullName.contains("Conference Record")) {
fullName = "Proc. " + fullName;
}
}
entry.setField(sourceField, fullName);
}
// clean up abstract
String abstr = entry.getField("abstract");
if (abstr != null) {
// Try to sort out most of the /spl / conversions
// Deal with this specific nested type first
abstr = abstr.replaceAll("/sub /spl infin//", "\\$_\\\\infty\\$");
abstr = abstr.replaceAll("/sup /spl infin//", "\\$\\^\\\\infty\\$");
// Replace general expressions
abstr = abstr.replaceAll("/[sS]pl ([^/]+)/", "\\$\\\\$1\\$");
// Deal with subscripts and superscripts
if (Globals.prefs.getBoolean(JabRefPreferences.USE_CONVERT_TO_EQUATION)) {
abstr = abstr.replaceAll("/sup ([^/]+)/", "\\$\\^\\{$1\\}\\$");
abstr = abstr.replaceAll("/sub ([^/]+)/", "\\$_\\{$1\\}\\$");
abstr = abstr.replaceAll("\\(sup\\)([^(]+)\\(/sup\\)", "\\$\\^\\{$1\\}\\$");
abstr = abstr.replaceAll("\\(sub\\)([^(]+)\\(/sub\\)", "\\_\\{$1\\}\\$");
} else {
abstr = abstr.replaceAll("/sup ([^/]+)/", "\\\\textsuperscript\\{$1\\}");
abstr = abstr.replaceAll("/sub ([^/]+)/", "\\\\textsubscript\\{$1\\}");
abstr = abstr.replaceAll("\\(sup\\)([^(]+)\\(/sup\\)", "\\\\textsuperscript\\{$1\\}");
abstr = abstr.replaceAll("\\(sub\\)([^(]+)\\(/sub\\)", "\\\\textsubscript\\{$1\\}");
}
// Replace \infin with \infty
abstr = abstr.replaceAll("\\\\infin", "\\\\infty");
// Write back
entry.setField("abstract", abstr);
}
// Clean up url
String url = entry.getField("url");
if (url != null) {
entry.setField("url", "http://ieeexplore.ieee.org" + url.replace("tp=&", ""));
}
return entry;
}