public BibtexEntry parseEntry(BibtexEntryType tp) throws IOException {
String id = Util.createNeutralId(); // createId(tp, _db);
BibtexEntry result = new BibtexEntry(id, tp);
skipWhitespace();
consume('{', '(');
int c = peek();
if ((c != '\n') && (c != '\r')) skipWhitespace();
String key = null;
boolean doAgain = true;
while (doAgain) {
doAgain = false;
try {
if (key != null) key = key + parseKey(); // parseTextToken(),
else key = parseKey();
} catch (NoLabelException ex) {
// This exception will be thrown if the entry lacks a key
// altogether, like in "@article{ author = { ...".
// It will also be thrown if a key contains =.
c = (char) peek();
if (Character.isWhitespace(c) || (c == '{') || (c == '\"')) {
String fieldName = ex.getMessage().trim().toLowerCase();
String cont = parseFieldContent(fieldName);
result.setField(fieldName, cont);
} else {
if (key != null) key = key + ex.getMessage() + "=";
else key = ex.getMessage() + "=";
doAgain = true;
}
}
}
if ((key != null) && key.equals("")) key = null;
result.setField(BibtexFields.KEY_FIELD, key);
skipWhitespace();
while (true) {
c = peek();
if ((c == '}') || (c == ')')) {
break;
}
if (c == ',') consume(',');
skipWhitespace();
c = peek();
if ((c == '}') || (c == ')')) {
break;
}
parseField(result);
}
consume('}', ')');
return result;
}
private void parseField(BibtexEntry entry) throws IOException {
String key = parseTextToken().toLowerCase();
// Util.pr("Field: _"+key+"_");
skipWhitespace();
consume('=');
String content = parseFieldContent(key);
// Now, if the field in question is set up to be fitted automatically
// with braces around
// capitals, we should remove those now when reading the field:
if (Globals.prefs.putBracesAroundCapitals(key)) {
content = Util.removeBracesAroundCapitals(content);
}
if (content.length() > 0) {
if (entry.getField(key) == null) entry.setField(key, content);
else {
// The following hack enables the parser to deal with multiple
// author or
// editor lines, stringing them together instead of getting just
// one of them.
// Multiple author or editor lines are not allowed by the bibtex
// format, but
// at least one online database exports bibtex like that, making
// it inconvenient
// for users if JabRef didn't accept it.
if (key.equals("author") || key.equals("editor"))
entry.setField(key, entry.getField(key) + " and " + content);
}
}
}
private StringBuffer parseBracketedText() throws IOException {
// Util.pr("Parse bracketed text");
StringBuffer value = new StringBuffer();
consume('{');
int brackets = 0;
while (!((peek() == '}') && (brackets == 0))) {
int j = read();
if ((j == -1) || (j == 65535)) {
throw new RuntimeException("Error in line " + line + ": EOF in mid-string");
} else if (j == '{') brackets++;
else if (j == '}') brackets--;
// If we encounter whitespace of any kind, read it as a
// simple space, and ignore any others that follow immediately.
/*
* if (j == '\n') { if (peek() == '\n') value.append('\n'); } else
*/
if (Character.isWhitespace((char) j)) {
String whs = skipAndRecordWhitespace(j);
// System.out.println(":"+whs+":");
if (!whs.equals("") && !whs.equals("\n\t")) { // &&
// !whs.equals("\n"))
whs = whs.replaceAll("\t", ""); // Remove tabulators.
// while (whs.endsWith("\t"))
// whs = whs.substring(0, whs.length()-1);
value.append(whs);
} else {
value.append(' ');
}
} else value.append((char) j);
}
consume('}');
return value;
}
public BibtexString parseString() throws IOException {
// Util.pr("Parsing string");
skipWhitespace();
consume('{', '(');
// while (read() != '}');
skipWhitespace();
// Util.pr("Parsing string name");
String name = parseTextToken();
// Util.pr("Parsed string name");
skipWhitespace();
// Util.pr("Now the contents");
consume('=');
String content = parseFieldContent(name);
// Util.pr("Now I'm going to consume a }");
consume('}', ')');
// Util.pr("Finished string parsing.");
String id = Util.createNeutralId();
return new BibtexString(id, name, content);
}
private StringBuffer parseQuotedFieldExactly() throws IOException {
StringBuffer value = new StringBuffer();
consume('"');
int brackets = 0;
while (!((peek() == '"') && (brackets == 0))) {
int j = read();
if ((j == -1) || (j == 65535)) {
throw new RuntimeException("Error in line " + line + ": EOF in mid-string");
} else if (j == '{') brackets++;
else if (j == '}') brackets--;
value.append((char) j);
}
consume('"');
return value;
}
private String parseFieldContent(String key) throws IOException {
skipWhitespace();
StringBuffer value = new StringBuffer();
int c = '.';
while (((c = peek()) != ',') && (c != '}') && (c != ')')) {
if (_eof) {
throw new RuntimeException("Error in line " + line + ": EOF in mid-string");
}
if (c == '"') {
StringBuffer text = parseQuotedFieldExactly();
value.append(fieldContentParser.format(text));
/*
*
* The following code doesn't handle {"} correctly: // value is
* a string consume('"');
*
* while (!((peek() == '"') && (j != '\\'))) { j = read(); if
* (_eof || (j == -1) || (j == 65535)) { throw new
* RuntimeException("Error in line "+line+ ": EOF in
* mid-string"); }
*
* value.append((char) j); }
*
* consume('"');
*/
} else if (c == '{') {
// Value is a string enclosed in brackets. There can be pairs
// of brackets inside of a field, so we need to count the
// brackets to know when the string is finished.
StringBuffer text = parseBracketedTextExactly();
value.append(fieldContentParser.format(text, key));
} else if (Character.isDigit((char) c)) { // value is a number
String numString = parseTextToken();
// Morten Alver 2007-07-04: I don't see the point of parsing the integer
// and converting it back to a string, so I'm removing the construct below
// the following line:
value.append(numString);
/*
try {
// Fixme: What is this for?
value.append(String.valueOf(Integer.parseInt(numString)));
} catch (NumberFormatException e) {
// If Integer could not be parsed then just add the text
// Used to fix [ 1594123 ] Failure to import big numbers
value.append(numString);
}
*/
} else if (c == '#') {
consume('#');
} else {
String textToken = parseTextToken();
if (textToken.length() == 0)
throw new IOException(
"Error in line "
+ line
+ " or above: "
+ "Empty text token.\nThis could be caused "
+ "by a missing comma between two fields.");
value.append("#").append(textToken).append("#");
// Util.pr(parseTextToken());
// throw new RuntimeException("Unknown field type");
}
skipWhitespace();
}
// Util.pr("Returning field content: "+value.toString());
// Check if we are to strip extra pairs of braces before returning:
if (Globals.prefs.getBoolean("autoDoubleBraces")) {
// Do it:
while ((value.length() > 1)
&& (value.charAt(0) == '{')
&& (value.charAt(value.length() - 1) == '}')) {
value.deleteCharAt(value.length() - 1);
value.deleteCharAt(0);
}
// Problem: if the field content is "{DNA} blahblah {EPA}", one pair
// too much will be removed.
// Check if this is the case, and re-add as many pairs as needed.
while (hasNegativeBraceCount(value.toString())) {
value.insert(0, '{');
value.append('}');
}
}
return value.toString();
}
