summaryrefslogblamecommitdiffstats
path: root/migratetags/matchers/defaultmatcher.h
blob: f2b6e9073fd3c901ac6f13f3e2f020db5b578fcf (plain) (tree)
1
2
3
4
5
6
7
8
9
10
                    
                             




                                       
                                                   

       

                                                                

         



                                                                       

                                    















                                                                             
 


















                                                                                                                                                  

















                                                                                      




                                               

      
#include "matcher.h"
#include <utf8greekaccents.h>

#ifndef defaultmatcher_h
#define defaultmatcher_h

class DefaultMatcher : public Matcher {
	UTF8GreekAccents sanitizeGreekAccentFilter;
public:

	DefaultMatcher() : sanitizeGreekAccentFilter() {
		sanitizeGreekAccentFilter.setOptionValue("off");
	}

// Compares 2 words and tries to give a percentage assurance of a match
// TODO: could use more smarts here
//
virtual int compare(const SWBuf &s1, const SWBuf &s2) {
	SWBuf t1 = sanitizeWord(s1);
	SWBuf t2 = sanitizeWord(s2);

	int retVal = 0;
	SWBuf largest  = (t1.length() > t2.length()) ? t1 : t2;
	SWBuf smallest = (t1.length() > t2.length()) ? t2 : t1;
	int matches = 0;
	int j = 0;
	for (int i = 0; i < smallest.length() && j < largest.length(); i++) {
		while (j < largest.length()) {
			if (smallest[i] == largest[j++]) {
				matches++;
				break;
			}
		}
	}
	return (((float)matches) / largest.length()) * 100;
}

// 
// This is where the magic happens
//
// we must point each targetMod word to an XMLTag
//
// when the magic is done, and your guess is made
// populate targetWordTags with the integer offset
// into wordTags for which XMLTag you think it should
// be.
//

virtual void matchWords(vector<int> &targetWordTags, const vector<SWBuf> &targetWords, const vector<SWBuf> &fromWords, vector<int> fromWordTags) {

	// initialize our results to all -1 so we can pop around and set
	// words as we find them, and know which ones we haven't yet set
	for (int i = 0; i < targetWords.size(); i++) targetWordTags.push_back(-1);


	// poor effort attempt
	for (int i = 0; i < targetWords.size(); ++i) {
		for (int j = 0; j < fromWords.size(); ++j) {
			if (fromWordTags[j] == -1) continue;
			int match = compare(targetWords[i], fromWords[j]);
			// if we have a better than XX% match of sequencial characters
			// then we'll say we have a match
			if (match > 49) {
				targetWordTags[i] = fromWordTags[j];
				fromWordTags[j] = -1;
				break;
			}
			// TOTRY: maybe check one word before and after?
			//
			// be creative!
			//
		}
	}
}
virtual SWBuf sanitizeWord(const SWBuf &word) {
	SWBuf t1 = word;
	t1.toUpper();
	return t1;
}
};
#endif