private static void execFindConcordancesDoeCorpus(String[] args) {
int limit = 1000000;
if (args.length > 2) limit = 100;
String searchString = args[1];
execParseDoeCorpus(false /* doSave */);
ArrayList<DOECorpusLine> allLines = new ArrayList<DOECorpusLine>();
HashMap<String, ArrayList<DOECorpusLine>> matches = corpus.getConcordances(searchString);
for (ArrayList<DOECorpusLine> list : matches.values()) allLines.addAll(list);
if (allLines.size() > limit) allLines = getRandomSelectionFromList(allLines, limit);
System.out.println("Concordances for search term " + searchString + " with limit " + limit);
for (DOECorpusLine line : allLines) {
System.out.println(
"Doc ID: "
+ line.getShortTitle()
+ "; Line ID: "
+ line.getLineID()
+ "; "
+ line.getLine());
}
}
/**
* @param searchString
* @param matches
*/
private static void printSearchResults(
String searchString, HashMap<String, ArrayList<DOECorpusLine>> matches) {
// Print results
System.out.println(
"Found the search string "
+ searchString
+ " in the following "
+ matches.size()
+ " docs and lines: ");
for (String key : matches.keySet()) {
System.out.println("Doc ID: " + key);
int i = 0;
for (DOECorpusLine line : matches.get(key)) {
System.out.println(
"\t"
+ i
+ ") "
+ line.getLineID()
+ " - "
+ line.getLine().replace(searchString, "@@" + searchString + "@@"));
i++;
}
}
}