/**
* Process the identifiers.
*
* @param type The type of thing these identifiers are.
* @param input The creation input.
* @param ids The identifiers.
* @param unmatchedIds A collector for unmatched identifiers.
* @param tempBag The temporary bag to add results to.
* @throws IOException If we can't from the request.
* @throws ClassNotFoundException if the type is not valid.
* @throws InterMineException If something goes wrong building the bag.
* @throws ObjectStoreException If there is a problem on the database level.
*/
protected void processIdentifiers(
final String type,
ListCreationInput input,
final Set<String> ids,
final Set<String> unmatchedIds,
final InterMineBag tempBag)
throws IOException, ClassNotFoundException, InterMineException, ObjectStoreException {
final Collection<String> addIssues = input.getAddIssues();
String line;
final StrMatcher matcher = getMatcher();
final BufferedReader r = getReader(request);
try {
while ((line = r.readLine()) != null) {
final StrTokenizer st = new StrTokenizer(line, matcher, StrMatcher.doubleQuoteMatcher());
while (st.hasNext()) {
final String token = st.nextToken();
ids.add(token);
}
if (ids.size() >= BAG_QUERY_MAX_BATCH_SIZE) {
addIdsToList(ids, tempBag, type, input.getExtraValue(), unmatchedIds, addIssues);
ids.clear();
}
}
} finally {
if (r != null) {
r.close();
}
}
if (ids.size() > 0) {
addIdsToList(ids, tempBag, type, input.getExtraValue(), unmatchedIds, addIssues);
}
}
/**
* Parses the pattern string, which should be a comma separated list of regular expressions, each
* of which may be surrounded with double quotes.
*
* @param inStr pattern string to parse
* @return list of pattern regular expressions
*/
private ArrayList<Pattern> parsePatternString(String inStr) {
ArrayList<Pattern> result = new ArrayList<Pattern>();
StrTokenizer tokenizer = new StrTokenizer(inStr, ',', '"');
tokenizer.setIgnoreEmptyTokens(true);
tokenizer.setIgnoredMatcher(StrMatcher.charSetMatcher(" \t\n\r"));
while (tokenizer.hasNext()) {
String tok = (String) tokenizer.next();
Pattern pat = Pattern.compile(tok);
result.add(pat);
}
return result;
}
private ImportLine processLine(long lineNumber, String line, Map<Integer, String> labelsMap) {
ImportLine toReturn = new ImportLine(this, lineNumber);
String[] columns =
new StrTokenizer(line, this.delimitor)
.setIgnoreEmptyTokens(false)
.setTrimmerMatcher(StrMatcher.trimMatcher())
.getTokenArray();
for (int i = 0; i < columns.length; i++) {
String value = StringUtils.strip(columns[i], "\"");
if (this.labels) {
toReturn.addColumn(labelsMap.get(i), value);
} else {
toReturn.addColumn(String.valueOf(i), value);
}
}
return toReturn;
}
/**
* Get the String Matcher for parsing the list of identifiers.
*
* @return The matcher to use.
*/
protected StrMatcher getMatcher() {
final String bagUploadDelims = getProperty("list.upload.delimiters") + " ";
final StrMatcher matcher = StrMatcher.charSetMatcher(bagUploadDelims);
return matcher;
}
@Override
protected List<ImportLine> buildLignesSpecifique(String fileName) throws ImportException {
List<ImportLine> toReturn = new ArrayList<ImportLine>();
long lineNumber = 0;
try {
FileInputStream f = new FileInputStream(fileName);
BufferedReader bufferedReader = null;
try {
// le fichier est encode en UCS-2 Little Endian
if (this.littleEndian) {
bufferedReader = new BufferedReader(new UCSReader(f, UCSReader.UCS2LE));
} else {
bufferedReader = new BufferedReader(new InputStreamReader(f));
}
lineNumber = skipLines(bufferedReader, lineNumber);
// on lit les lignes suivantes jusqu'à baliseFin
boolean labelsRead = labels ? false : true;
Map<Integer, String> labelsMap = null;
String line;
while ((line = bufferedReader.readLine()) != null
&& (StringUtils.isBlank(endTag) || !StringUtils.equals(line, endTag))) {
lineNumber++;
if (!labelsRead) {
// On lit les intitulés
String[] labelsString =
new StrTokenizer(line, this.delimitor)
.setIgnoreEmptyTokens(false)
.setTrimmerMatcher(StrMatcher.trimMatcher())
.getTokenArray();
labelsMap = new HashMap<Integer, String>();
for (int i = 0; i < labelsString.length; i++) {
labelsMap.put(i, labelsString[i]);
}
} else {
if (StringUtils.isNotBlank(line)) {
toReturn.add(processLine(lineNumber, line, labelsMap));
}
}
}
if (!StringUtils.isBlank(endTag) && !StringUtils.equals(line, endTag)) {
throw new ParseFileImportException(
"Impossible de trouver la balise de fin : " + endTag, this);
}
} finally {
bufferedReader.close();
f.close();
}
} catch (IOException e) {
String message = "Erreur d'E/S lors de la lecture du fichier : " + fileName;
throw new ImportException(message, e);
}
return toReturn;
}
