void LSH::ElCheapoHashing(TQuoteBase *QuoteBase, TInt ShingleLen,
THash<TMd5Sig, TIntSet>& ShingleToQuoteIds) {
fprintf(stderr, "Hashing shingles the el cheapo way...\n");
TIntV QuoteIds;
QuoteBase->GetAllQuoteIds(QuoteIds);
for (int qt = 0; qt < QuoteIds.Len(); qt++) {
if (qt % 1000 == 0) {
fprintf(stderr, "%d out of %d completed\n", qt, QuoteIds.Len());
}
TQuote Q;
QuoteBase->GetQuote(QuoteIds[qt], Q);
// Put x-character (or x-word) shingles into hash table; x is specified by ShingleLen parameter
TStr QContentStr;
Q.GetParsedContentString(QContentStr);
TChA QContentChA = TChA(QContentStr);
for (int i = 0; i < QContentChA.Len() - ShingleLen + 1; i++) {
TChA ShingleChA = TChA();
for (int j = 0; j < ShingleLen; j++) {
ShingleChA.AddCh(QContentChA.GetCh(i + j));
}
TStr Shingle = TStr(ShingleChA);
const TMd5Sig ShingleMd5(Shingle);
TIntSet ShingleQuoteIds;
if (ShingleToQuoteIds.IsKey(ShingleMd5)) {
ShingleQuoteIds = ShingleToQuoteIds.GetDat(ShingleMd5);
}
ShingleQuoteIds.AddKey(QuoteIds[qt]);
ShingleToQuoteIds.AddDat(ShingleMd5, ShingleQuoteIds);
}
}
Err("Done with el cheapo hashing!\n");
}
/// For every quote, add it to corresponding bucket for each hashed x-character shingle of the quote
// (Shingles by characters)
void LSH::HashShingles(TQuoteBase *QuoteBase, TClusterBase *CB, TInt ShingleLen,
THash<TMd5Sig, TShingleIdSet>& ShingleToQuoteIds) {
Err("Hashing shingles...\n");
TIntV QuoteIds;
QuoteBase->GetAllQuoteIds(QuoteIds);
for (int qt = 0; qt < QuoteIds.Len(); qt++) {
if (qt % 1000 == 0) {
fprintf(stderr, "%d out of %d completed\n", qt, QuoteIds.Len());
}
if (CB->IsQuoteInArchivedCluster(QuoteIds[qt]))
continue;
TQuote Q;
QuoteBase->GetQuote(QuoteIds[qt], Q);
// Put x-character (or x-word) shingles into hash table; x is specified by ShingleLen parameter
TStr QContentStr;
Q.GetParsedContentString(QContentStr);
TChA QContentChA = TChA(QContentStr);
int CurWord = 0;
for (int i = 0; i < QContentChA.Len() - ShingleLen + 1; i++) {
TChA ShingleChA = TChA();
for (int j = 0; j < ShingleLen; j++) {
ShingleChA.AddCh(QContentChA.GetCh(i + j));
}
TStr Shingle = TStr(ShingleChA);
const TMd5Sig ShingleMd5(Shingle);
TShingleIdSet ShingleQuoteIds;
if (ShingleToQuoteIds.IsKey(ShingleMd5)) {
ShingleQuoteIds = ShingleToQuoteIds.GetDat(ShingleMd5);
}
for (int j = CurWord; j > CurWord - WordWindow && j >= 0; j--) {
ShingleQuoteIds.AddKey(TShingleId(QuoteIds[qt], j));
}
ShingleToQuoteIds.AddDat(ShingleMd5, ShingleQuoteIds);
// up the current word index if we see a space
if (QContentChA.GetCh(i + ShingleLen - 1) == ' ') {
CurWord++;
}
}
}
Err("Done hashing!\n");
}
int main(int argc, char *argv[]) {
TDocBase *DocBase = new TDocBase;
TChA Url = TChA("http://www.newyorktimes.com/news_story");
TSecTm Date = TSecTm::GetCurTm();
TChA Content = TChA("foo bar foo foo");
TVec<TChA> Links;
Links.Add(TChA("http://www.google.com"));
Links.Add(TChA("http://www.yahoo.com"));
DocBase->AddDoc(Url, Date, Content, Links);
printf("Number of documents: %d\n", DocBase->Len());
TDoc t;
DocBase->GetDoc(0, t);
TStr tUrl;
t.GetUrl(tUrl);
printf("URL: %s\n", tUrl.CStr());
TStrV l;
t.GetLinks(l);
printf("Link1: %s\n", l[0].CStr());
printf("Link2: %s\n", l[1].CStr());
{ TFOut FOut("tmp.bin"); DocBase->Save(FOut); }
printf("Save data successfully\n");
delete DocBase;
TFIn FIn("tmp.bin");
printf("Load data successfully\n");
TDocBase *DocBase2 = new TDocBase;
DocBase2->Load(FIn);
printf("Number of documents: %d\n", DocBase2->Len());
TDoc t2;
DocBase2->GetDoc(0, t2);
TStr t2Url;
t2.GetUrl(t2Url);
printf("URL: %s\n", t2Url.CStr());
t2.GetLinks(l);
printf("Link1: %s\n", l[0].CStr());
printf("Link2: %s\n", l[1].CStr());
delete DocBase2;
return 0;
}
// space seprated sequence of words (includes all non-blank characters, i.e., punctuations)
TChA TStrUtil::GetCleanStr(const TChA& ChA) {
char *b = (char *) ChA.CStr();
while (*b && ! TCh::IsAlNum(*b)) {
b++;
}
if (*b == 0) {
return TChA();
}
TChA OutChA(ChA.Len());
char *e = b;
bool ws=false;
while (*e) {
while (*e && TCh::IsWs(*e)) {
e++;
ws=true;
}
if (! *e) {
break;
}
if (ws) {
OutChA.AddCh(' ');
ws=false;
}
OutChA.AddCh(*e);
e++;
}
//OutChA.ToLc();
return OutChA;
}
// space separated sequence of words, remove all punctuations, etc.
TChA TStrUtil::GetCleanWrdStr(const TChA& ChA) {
char *b = (char *) ChA.CStr();
while (*b && ! TCh::IsAlNum(*b)) {
b++;
}
if (*b == 0) {
return TChA();
}
TChA OutChA(ChA.Len());
char *e = b, tmp;
while (*e) {
b = e;
while (*e && (TCh::IsAlNum(*e) || ((*e=='\'' || *e=='-') && TCh::IsAlNum(*(e+1))))) {
e++;
}
if (b < e) {
tmp = *e;
*e=0;
OutChA += b;
OutChA.AddCh(' ');
*e = tmp;
}
while (*e && ! TCh::IsAlNum(*e)) {
e++;
}
if (! *e) {
break;
}
}
OutChA.DelLastCh();
OutChA.ToLc();
return OutChA;
}
// remove ending /, /index.html, etc. and strip starting www.
bool TStrUtil::GetNormalizedUrl(const TChA& UrlIn, const TChA& BaseUrl, TChA& UrlOut) {
UrlOut = UrlIn;
if (StripEnd(UrlIn, "/", UrlOut)) {}
else if (StripEnd(UrlIn, "/index.html", UrlOut)) {}
else if (StripEnd(UrlIn, "/index.htm", UrlOut)) {}
else if (StripEnd(UrlIn, "/index.php", UrlOut)) {}
if (! (UrlOut.IsPrefix("http://") || UrlOut.IsPrefix("ftp://"))) {
// if UrlIn is relative url, try combine it with BaseUrl
if (UrlIn.Empty() || ! (BaseUrl.IsPrefix("http://") || BaseUrl.IsPrefix("ftp://"))) {
//printf("** Bad URL: base:'%s' url:'%s'\n", BaseUrl.CStr(), UrlIn.CStr());
return false;
}
TChA Out;
if (! GetNormalizedUrl(BaseUrl, TChA(), Out)) {
return false;
}
if (UrlIn[0] != '/') {
Out.AddCh('/');
}
Out += UrlOut;
UrlOut = Out;
}
// http://www. --> http://
if (UrlOut.IsPrefix("http://www.")) {
TStr prefix("http://");
UrlOut = prefix + UrlOut.GetSubStr(11, TInt::Mx);
}
UrlOut.ToLc();
return true;
}
// http://www.ijs.si/fdfd/blah.html --> www.ijs.si
TChA TStrUtil::GetDomNm(const TChA& UrlChA) {
int EndSlash = UrlChA.SearchCh('/', 7)-1; // skip starting http://
if (EndSlash > 0) {
const int BegSlash = UrlChA.SearchChBack('/', EndSlash);
if (BegSlash > 0) { return UrlChA.GetSubStr(BegSlash+1, EndSlash).ToLc(); }
else { return UrlChA.GetSubStr(0, UrlChA.SearchCh('/', 0)-1).ToLc(); }
} else {
if (UrlChA.IsPrefix("http://")) { return UrlChA.GetSubStr(7, UrlChA.Len()-1).ToLc(); }
EndSlash = UrlChA.SearchCh('/', 0);
if (EndSlash > 0) { return UrlChA.GetSubStr(0, EndSlash-1).ToLc(); }
else { return TChA(UrlChA).ToLc(); }
}
}
TEST(TStr, Constructors) {
TStr Default;
TStr CStr("abc");
TStr OneChar('a');
TStr CopyCStr(CStr);
TStr Move(TStr("abc"));
TStr ChA(TChA("abc"));
TStr SStr(TSStr("abc"));
EXPECT_EQ(Default, "");
EXPECT_EQ(CStr, "abc");
EXPECT_EQ(OneChar, "a");
EXPECT_EQ(CopyCStr, "abc");
EXPECT_EQ(Move, "abc");
EXPECT_EQ(ChA, "abc");
EXPECT_EQ(SStr, "abc");
EXPECT_EQ(TStr(nullptr), "");
}
