// use this line to compile
// g++ -I. -fPIC -shared -g -o steampowered.com.so steampowered.com.cpp
// regex:
// http.{3,4}cdn\.(\w|\.)*cs\.steampowered\.com\/depot\/[0-9]+\/chunk\/\w{40}
//
string get_filename(string url) {
vector<string> paths, direct;
string file = "";
stringexplode(url, "/", &paths);
int psize = paths.size();
short save = 0;
for(int i=0;i<psize;i++) {
if(save == 1) {
file = paths.at(i);
save = 0;
}
if(save == 2) {
stringexplode(paths.at(i), "?", &direct);
file = file + "_" + direct.at(0);
return file;
}
if(paths.at(i) == "depot")
save = 1;
if(paths.at(i) == "chunk")
save = 2;
}
return "";
}
string get_filename(string url, int *ra, int *rb) {
vector<string> resultado;
string tmp;
if (url.find(".flv?") != string::npos) {
stringexplode(url, "/", &resultado);
url = resultado.at(resultado.size() - 1);
resultado.clear();
if ( (tmp = regex_match("[\\?&]ec_seek=[0-9]+", url)) != "" )
{
tmp.erase(0,9);
*ra = atoi(tmp.c_str());
*rb = -1;
}
stringexplode(url, "?", &resultado);
return resultado.at(0);
} else {
stringexplode(url, "/", &resultado);
tmp = resultado.at(resultado.size()-1);
if( regex_match(".flv$", tmp) == "" )
return "";
else
return tmp;
}
return "";
}
string get_filename(string url) {
vector<string> resultado;
if (url.find("?") != string::npos) {
stringexplode(url, "?", &resultado);
stringexplode(resultado.at(resultado.size()-2), "/", &resultado);
return resultado.at(resultado.size()-1);
} else {
stringexplode(url, "/", &resultado);
return resultado.at(resultado.size()-1);
}
}
string get_filename(string url) {
vector<string> resultado;
stringexplode(url, "/", &resultado);
if( resultado.size() == 4 )
return resultado.at(2) + "-" + resultado.at(3);
else
return "";
}
// use this line to compile
// g++ -I. -fPIC -shared -g -o eset.com.so eset.com.cpp
// http.{1,10}update\.eset\.com\/v[0-9]+(.*\/){2}.*\.nup$
//
string get_filename(string url) {
vector<string> split;
stringexplode(url, "/", &split);
int len = split.size();
if ( len >= 3 ) {
return split.at(len-3) + DELIM + split.at(len-2) + DELIM + split.at(len-1);
} else
return "";
}
// use this line to compile
// g++ -I. -fPIC -shared -g -o xtube.com.so xtube.com.cpp
//~ http.*\.(publicvideo|publicphoto)\.xtube\.com\/(videowall\/)?videos?\/.*(\.flv\?.*|\_Thumb\.flv$)
//~
string get_filename(string url, int *ra, int *rb) {
vector<string> resultado;
stringexplode(url,"/",&resultado);
url = resultado.at(resultado.size() - 1);
if (url.find("?") == string::npos) {
return url;
} else {
resultado.clear();
stringexplode(url, "?", &resultado);
string tmp;
if( (tmp = regex_match("[\\?&]fs=[0-9]+",resultado.at(1))) != "" ) {
tmp.erase(0,4);
*ra = atoi(tmp.c_str());
*rb = -1;
}
return resultado.at(0);
}
}
// use this line to compile
// g++ -I. -fPIC -shared -g -o avira-update.com.so avira-update.com.cpp
// ^http.{3,4}install\.avira-update\.com\/package\/((\w|\_|-)*\/){3}(\w|\_|-)*(\.exe)$
//
string get_filename(string url) {
vector<string> resultado;
stringexplode(url, "/", &resultado);
int size = resultado.size();
string fname = resultado.at(size - 4);
for(int i = 3; i >= 1; i--)
fname = fname + "_-_" + resultado.at(size - i);
return fname;
}
// use this line to compile
// g++ -I. -fPIC -shared -g -o blogspot.com.so blogspot.com.cpp
// Regex
// ^http.{1,10}\.bp\.blogspot\.com\/.*(\.jpg|\.mp4|\.flv|\.swf)$
//
string get_filename(string url) {
vector<string> resultado;
string filename = "";
stringexplode(url, "/", &resultado);
int size = resultado.size();
if ( size < 6 )
return "";
for( int i = 5;i >= 2; i-- )
filename = filename + resultado.at(size - i) + "_";
return filename = filename + resultado.at(size - 1);
}
void
gen_charfield(Datasrc* src, Odometer* odom, int index, Bytebuffer* fieldbuf)
{
int i;
int lastdim = (index == (odom->rank - 1));
size_t declsize = odom->declsize[index];
Constant* con;
ASSERT(declsize != 0);
if(lastdim) {
for(i=0;i<declsize;) {
con = srcnext(src);
if(con == NULL) break;
if(!isstringable(con->nctype)) {
semerror(srcline(src),
"Encountered non-string constant in compound field");
return;
}
i += collectstring(con,declsize,fieldbuf);
}
if(i < declsize) i=fillstring(declsize,i,fieldbuf);
} else { /* ! lastdim*/
int exploded = 0;
size_t slicesize;
/* Compute subslice size */
slicesize = 1;
for(i=index+1;i<odom->rank;i++)
slicesize *= MAX(odom->declsize[i],odom->unlimitedsize[i]);
con = srcpeek(src);
if(con != NULL && !isstringable(con->nctype)) {
semerror(srcline(src),
"Encountered non-string constant in compound field");
return;
}
if(con != NULL && con->value.stringv.len > slicesize) {
/* Constant is larger than just our slice */
/* Explode the constant into subchunks */
exploded = stringexplode(src,slicesize);
}
for(i=0;i<declsize;i++) {
gen_charfield(src,odom,index+1,fieldbuf);
}
if(exploded) srcpop(src);
}
}
void
gen_chararray(Symbol* vsym,
Bytebuffer* databuf,
Datasrc* src,
Odometer* odom,
int index)
{
/* Assume that all dimensions from index+1 to rank-1 are !unlimited */
int i;
int rank = odom->rank;
int lastdim = (index == (rank - 1)); /* last dimension*/
int firstdim = (index == 0);
int isunlimited = (odom->declsize[index] == NC_UNLIMITED);
int exploded = 0;
Constant* con;
if(lastdim) {
gen_chararraysuffix(vsym,databuf,src,odom,index);
return;
}
ASSERT(index >= 0 && index < rank);
odom->index[index] = odom->start[index]; /* reset */
if(isunlimited) {
size_t slicesize;
Constant* con;
if(!firstdim) {
if(!issublist(src)) {
semerror(srcline(src),"Unlimited data must be enclosed in {..}");
return;
}
srcpush(src); /* enter the unlimited data */
}
con=srcpeek(src);
/* Break up the constant if it is too large */
slicesize = odomsubarray(odom,index+1);
if(con != NULL && con->value.stringv.len > slicesize) {
/* Constant is larger than just our slice */
/* Explode the constant into subchunks */
exploded = stringexplode(src,slicesize);
}
while((con=srcpeek(src))!=NULL) {
gen_chararray(vsym,databuf,src,odom,index+1);
odom->index[index]++;
}
odom->unlimitedsize[index] = odom->index[index];
if(exploded) srcpop(src);
if(!firstdim) srcpop(src);
} else { /* !isunlimited*/
size_t slicesize;
con = srcpeek(src);
ASSERT(!lastdim);
/* Break up the constant if it is too large */
slicesize = odomsubarray(odom,index+1);
if(con != NULL && con->value.stringv.len > slicesize) {
/* Constant is larger than just our slice */
/* Explode the constant into subchunks */
exploded = stringexplode(src,slicesize);
}
for(i=0;i<odom->declsize[index];i++) {
gen_chararray(vsym,databuf,src,odom,index+1);
}
if(exploded) srcpop(src);
}
}
static void
gen_chararraysuffix(Symbol* vsym,
Bytebuffer* databuf,
Datasrc* src,
Odometer* odom,
int index)
{
int i;
int rank = odom->rank;
int lastdim = (index == (rank - 1)); /* last dimension*/
int firstdim = (index == 0);
size_t declsize = odom->declsize[index];
int isunlimited = (declsize==NC_UNLIMITED);
Constant* con;
ASSERT(index >= 0 && index < rank);
odom->index[index] = odom->start[index]; /* reset*/
con = srcpeek(src);
if(!isunlimited) {
if(con != NULL && !isstringable(con->nctype)) {
semerror(srcline(src),
"Encountered non-string constant in char data: %s",
vsym->name);
return;
}
if(lastdim) {
/* To mimic ncgen original, it appears we have to hack.
I think firstdim==lastdim may work.
*/
for(i=0;i<declsize;) {
int slen;
con = srcnext(src);
if(con == NULL) break;
slen = collectstring(con,declsize,databuf);
if(!firstdim && slen < declsize) slen=fillstring(declsize,slen,databuf);
i += slen;
}
if(firstdim && i < declsize) i = fillstring(declsize,i,databuf);
odom->index[index] = i;
} else { /* ! lastdim*/
int exploded = 0;
size_t slicesize = odometertotal(odom,index+1);
if(con != NULL && con->nctype == NC_STRING
&& con->value.stringv.len > slicesize) {
/* Constant is larger than just our slice */
/* Explode the constant into subchunks */
exploded = stringexplode(src,slicesize);
}
for(i=0;i<odom->declsize[index];i++) {
gen_chararraysuffix(vsym,databuf,src,odom,index+1);
odom->index[index]++;
}
if(exploded) srcpop(src);
}
} else { /* unlimited => lastdim*/
Constant* con;
if(!firstdim) {
if(!issublist(src)) {
semerror(srcline(src),"Unlimited data must be enclosed in {..}");
return;
}
srcpush(src); /* enter the unlimited data */
}
/* Basically, collect all the strings until we run out */
i = 0;
while((con=srcnext(src))!=NULL) {
i += collectstring(con,0,databuf);
}
odom->index[index] = i;
odom->unlimitedsize[index] = odom->index[index];
if(!firstdim) srcpop(src);
}
}
