[ViewVC] Diff of: proj/trunk/Search/Ranker.cpp

Comparing trunk/Search/Ranker.cpp (file contents):
Revision 364 by Douglas Thrift, 2004-06-04T04:08:28-07:00 vs.
Revision 365 by douglas, 2008-08-11T15:16:35-07:00

#	Line 140 \| Line 140 \| void Ranker::rank(vector<string> query)
140
141		void Ranker::setSample()
142		{
143	<	map<unsigned, unsigned>::iterator itor;
144	<	multimap<unsigned, map<unsigned, unsigned>::iterator> distances;
143	>	map<size_t, size_t>::iterator itor;
144	>	multimap<size_t, map<size_t, size_t>::iterator> distances;
145
146		for (itor = occurrencesText.begin(); itor != occurrencesText.end(); itor++)
147		{
148	<	unsigned distance;
148	>	size_t distance;
149
150		if (++itor != occurrencesText.end())
151		{
152	<	unsigned next(itor->first);
152	>	size_t next(itor->first);
153
154		itor--;
155
#	Line 162 \| Line 162 \| void Ranker::setSample()
162		itor--;
163		}
164
165	<	distances.insert(pair<unsigned, map<unsigned,
166	<	unsigned>::iterator>(distance, itor));
165	>	distances.insert(pair<size_t, map<size_t,
166	>	size_t>::iterator>(distance, itor));
167		}
168
169		if (distances.begin() != distances.end())
#	Line 176 \| Line 176 \| void Ranker::setSample()
176
177		while (sampleLength < sampleMax && itor != occurrencesText.end())
178		{
179	<	unsigned found(itor->first), length(itor->second);
179	>	size_t found(itor->first), length(itor->second);
180
181	<	for (unsigned index(found); index > begin; index--)
181	>	for (size_t index(found); index > begin; index--)
182		{
183		if (found - index >= sampleMax - sampleLength - length)
184		{
#	Line 328 \| Line 328 \| string Ranker::getTitle()
328		string title, portion;
329		size_t begin(0);
330
331	<	for (map<unsigned, unsigned>::iterator itor = occurrencesTitle.begin();
331	>	for (map<size_t, size_t>::iterator itor = occurrencesTitle.begin();
332		itor != occurrencesTitle.end(); itor++)
333		{
334	<	unsigned found(itor->first), length(itor->second);
334	>	size_t found(itor->first), length(itor->second);
335
336		portion = Page::getTitle().substr(begin, found - begin);
337
#	Line 369 \| Line 369 \| string Ranker::getTitle()
369		string Ranker::getDescription()
370		{
371		string description, portion;
372	<	unsigned begin(0);
372	>	size_t begin(0);
373
374	<	for (map<unsigned, unsigned>::iterator itor =
374	>	for (map<size_t, size_t>::iterator itor =
375		occurrencesDescription.begin(); itor != occurrencesDescription.end();
376		itor++)
377		{
378	<	unsigned found(itor->first), length(itor->second);
378	>	size_t found(itor->first), length(itor->second);
379
380		portion = Page::getDescription().substr(begin, found - begin);
381
#	Line 410 \| Line 410 \| string Ranker::getDescription()
410		return description;
411		}
412
413	<	bool Ranker::operator==(const unsigned number) const
413	>	bool Ranker::operator==(const size_t number) const
414		{
415		return value == number;
416		}
#	Line 420 \| Line 420 \| bool Ranker::operator==(const Ranker& ra
420		return value == ranker.value;
421		}
422
423	<	bool Ranker::operator!=(const unsigned number) const
423	>	bool Ranker::operator!=(const size_t number) const
424		{
425		return value != number;
426		}
#	Line 430 \| Line 430 \| bool Ranker::operator!=(const Ranker& ra
430		return value != ranker.value;
431		}
432
433	<	bool Ranker::operator<(const unsigned number) const
433	>	bool Ranker::operator<(const size_t number) const
434		{
435		return value < number;
436		}
#	Line 440 \| Line 440 \| bool Ranker::operator<(const Ranker& ran
440		return value < ranker.value;
441		}
442
443	<	bool Ranker::operator>(const unsigned number) const
443	>	bool Ranker::operator>(const size_t number) const
444		{
445		return value > number;
446		}
#	Line 503 \| Line 503 \| void Ranker::rank()
503		{
504		string lowerDescription(tolower(Page::getDescription()));
505
506	<	for (unsigned index(0); index < required.size(); index++)
506	>	for (size_t index(0); index < required.size(); index++)
507		{
508		if (required[index].find("URL ") == 0)
509		{
#	Line 527 \| Line 527 \| void Ranker::rank()
527		}
528		}
529
530	<	for (unsigned index1(0); index1 < eitherOr.size(); index1++)
530	>	for (size_t index1(0); index1 < eitherOr.size(); index1++)
531		{
532		vector<string> words;
533	<	unsigned begin(0), found;
533	>	size_t begin(0), found;
534
535		do
536		{
#	Line 551 \| Line 551 \| void Ranker::rank()
551		while (begin < eitherOr[index1].length() && found !=
552		string::npos);
553
554	<	for (unsigned number(0); number < words.size(); number++)
554	>	for (size_t number(0); number < words.size(); number++)
555		{
556		if (words[index1].find("URL ") == 0)
557		{
#	Line 576 \| Line 576 \| void Ranker::rank()
576		}
577		}
578
579	<	for (unsigned index2(0); index2 < getHeadings().size(); index2++)
579	>	for (size_t index2(0); index2 < getHeadings().size(); index2++)
580		{
581		string lowerHeading = string(getHeadings()[index2].length(),
582		' ');
583
584	<	for (unsigned number(0); number <
584	>	for (size_t number(0); number <
585		getHeadings()[index2].length(); number++)
586		{
587		lowerHeading[number] = tolower(
588		getHeadings()[index2][number]);
589		}
590
591	<	for (unsigned number0(0); number0 < required.size(); number0++)
591	>	for (size_t number0(0); number0 < required.size(); number0++)
592		{
593		if (required[number0].find("URL ") == 0)
594		{
#	Line 611 \| Line 611 \| void Ranker::rank()
611		}
612		}
613
614	<	for (unsigned number1(0); number1 < eitherOr.size(); number1++)
614	>	for (size_t number1(0); number1 < eitherOr.size(); number1++)
615		{
616		vector<string> words;
617	<	unsigned begin(0), found;
617	>	size_t begin(0), found;
618
619		do
620		{
#	Line 635 \| Line 635 \| void Ranker::rank()
635		while (begin < eitherOr[number1].length() && found !=
636		string::npos);
637
638	<	for (unsigned number(0); number < words.size(); number++)
638	>	for (size_t number(0); number < words.size(); number++)
639		{
640		if (words[number].find("URL ") == 0)
641		{
#	Line 665 \| Line 665 \| void Ranker::rank()
665
666		void Ranker::checkRequired()
667		{
668	<	vector<unsigned> inURLs, inTitles, inTexts;
668	>	vector<size_t> inURLs, inTitles, inTexts;
669
670	<	for (unsigned index(0); index < required.size(); index++)
670	>	for (size_t index(0); index < required.size(); index++)
671		{
672	<	unsigned inURL(0), inTitle(0), inText(0);
672	>	size_t inURL(0), inTitle(0), inText(0);
673
674		if (required[index].find("URL ") == 0)
675		{
#	Line 728 \| Line 728 \| void Ranker::checkRequired()
728		inTexts.push_back(inText);
729		}
730
731	<	unsigned inURL(evaluate(inURLs)), inTitle(evaluate(inTitles)),
731	>	size_t inURL(evaluate(inURLs)), inTitle(evaluate(inTitles)),
732		inText(evaluate(inTexts));
733
734		requiredValue += (inURL && (allIn == url)) \|\| (inTitle && (allIn == title))
#	Line 738 \| Line 738 \| void Ranker::checkRequired()
738
739		void Ranker::checkExcluded()
740		{
741	<	vector<unsigned> inURLs, inTitles, inTexts;
741	>	vector<size_t> inURLs, inTitles, inTexts;
742
743	<	for (unsigned index(0); index < excluded.size(); index++)
743	>	for (size_t index(0); index < excluded.size(); index++)
744		{
745	<	unsigned inURL(0), inTitle(0), inText(0);
745	>	size_t inURL(0), inTitle(0), inText(0);
746
747		inURL = find(excluded[index], lowerURL.substr(7));
748		inTitle = find(excluded[index], lowerTitle);
#	Line 753 \| Line 753 \| void Ranker::checkExcluded()
753		inTexts.push_back(inText);
754		}
755
756	<	unsigned inURL(evaluate(inURLs)), inTitle = evaluate(inTitles),
756	>	size_t inURL(evaluate(inURLs)), inTitle = evaluate(inTitles),
757		inText(evaluate(inTexts));
758
759		excludedValue += (inURL && (allIn == url)) \|\| (inTitle && (allIn == title))
#	Line 763 \| Line 763 \| void Ranker::checkExcluded()
763
764		void Ranker::checkEitherOr()
765		{
766	<	vector<unsigned> inURLs, inTitles, inTexts;
766	>	vector<size_t> inURLs, inTitles, inTexts;
767
768	<	for (unsigned index(0); index < eitherOr.size(); index++)
768	>	for (size_t index(0); index < eitherOr.size(); index++)
769		{
770	<	vector<unsigned> inURLz, inTitlez, inTextz;
771	<	unsigned inURL(0), inTitle(0), inText(0);
770	>	vector<size_t> inURLz, inTitlez, inTextz;
771	>	size_t inURL(0), inTitle(0), inText(0);
772		vector<string> words;
773	<	unsigned begin(0), found;
773	>	size_t begin(0), found;
774
775		do
776		{
#	Line 789 \| Line 789 \| void Ranker::checkEitherOr()
789		}
790		while (begin < eitherOr[index].length() && found != string::npos);
791
792	<	for (unsigned number(0); number < words.size(); number++)
792	>	for (size_t number(0); number < words.size(); number++)
793		{
794	<	unsigned inURL(0), inTitle(0), inText(0);
794	>	size_t inURL(0), inTitle(0), inText(0);
795
796		if (words[number].find("URL ") == 0)
797		{
#	Line 850 \| Line 850 \| void Ranker::checkEitherOr()
850		inTextz.push_back(inText);
851		}
852
853	<	for (unsigned number0(0); number0 < inURLz.size(); number0++)
853	>	for (size_t number0(0); number0 < inURLz.size(); number0++)
854		{
855		inURL += inURLz[number0];
856		}
857
858	<	for (unsigned number1(0); number1 < inTitlez.size(); number1++)
858	>	for (size_t number1(0); number1 < inTitlez.size(); number1++)
859		{
860		inTitle += inTitlez[number1];
861		}
862
863	<	for (unsigned number2(0); number2 < inTextz.size(); number2++)
863	>	for (size_t number2(0); number2 < inTextz.size(); number2++)
864		{
865		inText += inTextz[number2];
866		}
#	Line 875 \| Line 875 \| void Ranker::checkEitherOr()
875		words.clear();
876		}
877
878	<	unsigned inURL(evaluate(inURLs)), inTitle = evaluate(inTitles),
878	>	size_t inURL(evaluate(inURLs)), inTitle = evaluate(inTitles),
879		inText(evaluate(inTexts));
880
881		eitherOrValue += (inURL && (allIn == url)) \|\| (inTitle && (allIn == title))
#	Line 883 \| Line 883 \| void Ranker::checkEitherOr()
883		inText : 0;
884		}
885
886	<	unsigned Ranker::find(string word, const string& where)
886	>	size_t Ranker::find(string word, const string& where)
887		{
888	<	unsigned value(0);
888	>	size_t value(0);
889
890		decrap(word);
891
#	Line 896 \| Line 896 \| unsigned Ranker::find(string word, const
896		}
897		else if (word.find_first_of(" \n\t") == string::npos)
898		{
899	<	unsigned begin(0), found;
899	>	size_t begin(0), found;
900
901		do
902		{
#	Line 930 \| Line 930 \| unsigned Ranker::find(string word, const
930		return value;
931		}
932
933	<	unsigned Ranker::find(string word, const string& where, map<unsigned,
934	<	unsigned>& occurrences)
933	>	size_t Ranker::find(string word, const string& where, map<size_t,
934	>	size_t>& occurrences)
935		{
936	<	unsigned value(0);
936	>	size_t value(0);
937
938		decrap(word);
939
#	Line 944 \| Line 944 \| unsigned Ranker::find(string word, const
944		}
945		else if (word.find_first_of(" \n ") == string::npos)
946		{
947	<	unsigned begin(0), found;
947	>	size_t begin(0), found;
948
949		do
950		{
#	Line 964 \| Line 964 \| unsigned Ranker::find(string word, const
964		{
965		value++;
966
967	<	occurrences.insert(pair<unsigned, unsigned>(found,
967	>	occurrences.insert(pair<size_t, size_t>(found,
968		word.length()));
969		}
970		}
#	Line 981 \| Line 981 \| unsigned Ranker::find(string word, const
981		return value;
982		}
983
984	<	unsigned Ranker::phrase(const string& phrase, const string& where)
984	>	size_t Ranker::phrase(const string& phrase, const string& where)
985		{
986	<	unsigned value(0);
986	>	size_t value(0);
987		vector<string> words;
988	<	unsigned begin(0), space;
988	>	size_t begin(0), space;
989
990		do
991		{
#	Line 999 \| Line 999 \| unsigned Ranker::phrase(const string& ph
999
1000		begin = 0;
1001
1002	<	unsigned counter(0);
1002	>	size_t counter(0);
1003
1004		do
1005		{
#	Line 1010 \| Line 1010 \| unsigned Ranker::phrase(const string& ph
1010		return value;
1011		}
1012
1013	<	unsigned Ranker::phrase(const string& phrase, const string& where,
1014	<	map<unsigned, unsigned>& occurrences)
1013	>	size_t Ranker::phrase(const string& phrase, const string& where,
1014	>	map<size_t, size_t>& occurrences)
1015		{
1016	<	unsigned value(0);
1016	>	size_t value(0);
1017		vector<string> words;
1018	<	unsigned begin(0), space;
1018	>	size_t begin(0), space;
1019
1020		do
1021		{
#	Line 1038 \| Line 1038 \| unsigned Ranker::phrase(const string& ph
1038		return value;
1039		}
1040
1041	<	unsigned Ranker::phrase(const vector<string>& words, unsigned word, unsigned&
1041	>	size_t Ranker::phrase(const vector<string>& words, size_t word, size_t&
1042		begin, bool start, const string& where)
1043		{
1044	<	unsigned value(0);
1044	>	size_t value(0);
1045		bool end(!(word + 1 < words.size()));
1046	<	unsigned found(where.find(words[word], begin)), newBegin(found +
1046	>	size_t found(where.find(words[word], begin)), newBegin(found +
1047		words[word].length());
1048
1049		if (found != string::npos)
#	Line 1062 \| Line 1062 \| unsigned Ranker::phrase(const vector<str
1062
1063		if (!start)
1064		{
1065	<	for (unsigned index = begin + 1; index < found - 1; index++)
1065	>	for (size_t index = begin + 1; index < found - 1; index++)
1066		{
1067		if (isalnum(where[index]))
1068		{
#	Line 1102 \| Line 1102 \| unsigned Ranker::phrase(const vector<str
1102		return value;
1103		}
1104
1105	<	unsigned Ranker::phrase(const vector<string>& words, unsigned word, unsigned&
1106	<	begin, bool start, const string& where, map<unsigned, unsigned>&
1105	>	size_t Ranker::phrase(const vector<string>& words, size_t word, size_t&
1106	>	begin, bool start, const string& where, map<size_t, size_t>&
1107		occurrences)
1108		{
1109	<	unsigned value(0);
1109	>	size_t value(0);
1110		bool end(!(word + 1 < words.size()));
1111	<	unsigned found(where.find(words[word], begin)), newBegin(found +
1111	>	size_t found(where.find(words[word], begin)), newBegin(found +
1112		words[word].length());
1113
1114		if (found != string::npos)
#	Line 1127 \| Line 1127 \| unsigned Ranker::phrase(const vector<str
1127
1128		if (!start)
1129		{
1130	<	for (unsigned index = begin + 1; index < found - 1; index++)
1130	>	for (size_t index = begin + 1; index < found - 1; index++)
1131		{
1132		if (isalnum(where[index]))
1133		{
#	Line 1139 \| Line 1139 \| unsigned Ranker::phrase(const vector<str
1139
1140		if (between)
1141		{
1142	<	occurrences.insert(pair<unsigned, unsigned>(found,
1142	>	occurrences.insert(pair<size_t, size_t>(found,
1143		words[word].length()));
1144
1145		if (end)
#	Line 1171 \| Line 1171 \| unsigned Ranker::phrase(const vector<str
1171		return value;
1172		}
1173
1174	<	unsigned Ranker::evaluate(vector<unsigned>& ins)
1174	>	size_t Ranker::evaluate(vector<size_t>& ins)
1175		{
1176	<	unsigned in(0);
1176	>	size_t in(0);
1177
1178	<	for (unsigned index(0); index < ins.size(); index++)
1178	>	for (size_t index(0); index < ins.size(); index++)
1179		{
1180		if (ins[index] > 0)
1181		{
#	Line 1193 \| Line 1193 \| unsigned Ranker::evaluate(vector<unsigne
1193
1194		void Ranker::decrap(string& crap)
1195		{
1196	<	unsigned begin(0), found;
1196	>	size_t begin(0), found;
1197
1198		do
1199		{

Diff Legend

-–
+Removed lines
-+
+Added lines
-<
+Changed lines
->
+Changed lines

Comparing trunk/Search/Ranker.cpp (file contents): Revision 364 by Douglas Thrift, 2004-06-04T04:08:28-07:00 vs. Revision 365 by douglas, 2008-08-11T15:16:35-07:00

Diff Legend

Comparing trunk/Search/Ranker.cpp (file contents):
Revision 364 by Douglas Thrift, 2004-06-04T04:08:28-07:00 vs.
Revision 365 by douglas, 2008-08-11T15:16:35-07:00