public HashSet<NGramSet> findCommonNGrams(
String string1, String string2, int min, int max, boolean maximizePrimaryWindowSize) {
ordered_scores = new TreeMap<Double, Integer>();
errors = new ArrayList<Error>();
HashSet<NGramSet> NGramsWithMatches = new HashSet<NGramSet>();
// ensure that min <= max
if (min > max) {
int temp = max;
max = min;
min = temp;
logError("Min greater than max; assuming the opposite parameterization");
}
char[] chars1 = string1.toCharArray();
char[] chars2 = string2.toCharArray();
List<String> words1 = scanForWords(chars1);
List<String> words2 = scanForWords(chars2);
// when testing, restrict the length of documents to be small
if (isTesting) {
int maxSub = 1000;
words1 =
words1.subList(0, maxSizeOutOfRangeForSource(maxSub, words1) ? words1.size() : maxSub);
words2 =
words2.subList(0, maxSizeOutOfRangeForSource(maxSub, words2) ? words2.size() : maxSub);
}
NGramSetImpl.setMatchCase(matchCase);
NGramSetImpl.setUseStopWords(USESTOPWORDS);
NGramSetImpl.setStrictness(STRICT);
NGramSetImpl.setMinSize(min);
int leftMax = (words1.size() <= max || maximizePrimaryWindowSize) ? words1.size() : max;
int rightMax = (words2.size() <= max) ? words2.size() - 1 : max;
HashMap<String, List<NGramSet>> map = new HashMap<String, List<NGramSet>>();
if (rightMax < max) {
logError(
"Window size greater than number of length of secondary text; decreasing secondary window size to: "
+ rightMax);
}
if (leftMax < max && maximizePrimaryWindowSize) {
logError("Maximizing primary window");
} else if (leftMax < max) {
logError("Max out of range for primary source. Scaling down to: " + leftMax);
}
ArrayList<NGramSet> nGrams1 = null;
nGrams1 = getAllNGramsOfSize(words1, leftMax, null);
// ArrayList<NGramSet> nGrams2 =
getAllNGramsOfSize(words2, rightMax, map);
findAllCommon(NGramsWithMatches, nGrams1, map);
return NGramsWithMatches;
}
