void hash_init(hash_struct * hash, int normalized) {
hash->sav = inthash_new(0);
hash->adrfil = inthash_new(0);
hash->former_adrfil = inthash_new(0);
hash->normalized = normalized;
/* Case-insensitive comparison ; keys are direct char* filenames */
inthash_value_set_key_handler(hash->sav,
key_duphandler,
key_freehandler,
key_sav_hashes,
key_sav_equals,
hash);
/* URL-style comparison ; keys are lien_url structure pointers casted
to char* */
inthash_value_set_key_handler(hash->adrfil,
key_duphandler,
key_freehandler,
key_adrfil_hashes,
key_adrfil_equals,
hash);
inthash_value_set_key_handler(hash->former_adrfil,
key_duphandler,
key_freehandler,
key_former_adrfil_hashes,
key_former_adrfil_equals,
hash);
}
HTSEXT_API char* hts_getcategories(char* path, int type) {
String categ = STRING_EMPTY;
String profiles = STRING_EMPTY;
char* rpath = path;
find_handle h;
inthash hashCateg = NULL;
if (rpath[0]) {
if (rpath[strlen(rpath)-1]=='/') {
rpath[strlen(rpath)-1]='\0'; /* note: patching stored (inhash) value */
}
}
h = hts_findfirst(rpath);
if (h) {
String iname = STRING_EMPTY;
if (type == 1) {
hashCateg = inthash_new(127);
StringCat(categ, "Test category 1");
StringCat(categ, "\r\nTest category 2");
}
do {
if (hts_findisdir(h)) {
char BIGSTK line2[1024];
StringCopy(iname,rpath);
StringCat(iname,"/");
StringCat(iname,hts_findgetname(h));
StringCat(iname,"/hts-cache/winprofile.ini");
if (fexist(StringBuff(iname))) {
if (type == 1) {
FILE* fp = fopen(StringBuff(iname), "rb");
if (fp != NULL) {
int done=0;
while(!feof(fp) && !done) {
int n = linput(fp, line2, sizeof(line2) - 2);
if (n > 0) {
if (strfield(line2, "category=")) {
if (*(line2+9)) {
if (!inthash_read(hashCateg, line2+9, NULL)) {
inthash_write(hashCateg, line2+9, 0);
if (StringLength(categ) > 0) {
StringCat(categ, "\r\n");
}
unescapehttp(line2+9, &categ);
}
}
done=1;
}
}
}
line2[0] = '\0';
fclose(fp);
}
} else {
if (StringLength(profiles) > 0) {
StringCat(profiles, "\r\n");
}
StringCat(profiles, hts_findgetname(h));
}
}
}
} while(hts_findnext(h));
hts_findclose(h);
StringFree(iname);
}
if (hashCateg) {
inthash_delete(&hashCateg);
hashCateg = NULL;
}
if (type == 1)
return StringBuffRW(categ);
else
return StringBuffRW(profiles);
}
/*
Indexing system
A little bit dirty, (quick'n dirty, in fact)
But should be okay on most cases
Tags and javascript handled (ignored)
*/
int index_keyword(const char* html_data,LLint size,const char* mime,const char* filename,const char* indexpath) {
#if HTS_MAKE_KEYWORD_INDEX
char catbuff[CATBUFF_SIZE];
int intag=0,inscript=0,incomment=0;
char keyword[KEYW_LEN+32];
int i=0;
//
int WordIndexSize=1024;
inthash WordIndexHash=NULL;
FILE *tmpfp=NULL;
//
// Check parameters
if (!html_data)
return 0;
if (!size)
return 0;
if (!mime)
return 0;
if (!filename)
return 0;
// Init ?
if (hts_index_init) {
remove(concat(catbuff,indexpath,"index.txt"));
remove(concat(catbuff,indexpath,"sindex.html"));
hts_index_init=0;
}
// Check MIME type
if (is_html_mime_type(mime)) {
inscript=0;
}
// FIXME - temporary fix for image/svg+xml (svg)
// "IN XML" (html like, in fact :) )
else if (
(strfield2(mime,"image/svg+xml"))
||
(strfield2(mime,"image/svg-xml"))
#if HTS_USEMMS
||
strfield2(mime,"video/x-ms-asf")
#endif
) {
inscript=0;
}
else if (
(strfield2(mime,"application/x-javascript"))
|| (strfield2(mime,"text/css"))
) {
inscript=1;
//} else if (strfield2(mime, "text/vnd.wap.wml")) { // humm won't work in many cases
// inscript=0;
} else
return 0;
// Temporary file
tmpfp = tmpfile();
if (!tmpfp)
return 0;
// Create hash structure
// Hash tables rulez da world!
WordIndexHash=inthash_new(WordIndexSize);
if (!WordIndexHash)
return 0;
// Start indexing this page
keyword[0]='\0';
while(i<size) {
if (strfield(html_data + i , "<script")) {
inscript=1;
}
else if (strfield(html_data + i , "<!--")) {
incomment=1;
}
else if (strfield(html_data + i , "</script")) {
if (!incomment)
inscript=0;
}
else if (strfield(html_data + i , "-->")) {
incomment=0;
}
else if (html_data[i]=='<') {
if (!inscript)
intag=1;
}
else if (html_data[i]=='>') {
intag=0;
}
else {
// Okay, parse keywords
if ( (!inscript) && (!incomment) && (!intag) ) {
char cchar=html_data[i];
int pos;
int len = (int) strlen(keyword);
// Replace (ignore case, and so on..)
if ((pos=strcpos(KEYW_TRANSCODE_FROM,cchar))>=0)
cchar=KEYW_TRANSCODE_TO[pos];
if (strchr(KEYW_ACCEPT,cchar)) {
/* Ignore some characters at begining */
if ((len>0) || (!strchr(KEYW_IGNORE_BEG,cchar))) {
keyword[len++]=cchar;
keyword[len]='\0';
}
} else if ( (strchr(KEYW_SPACE,cchar)) || (!cchar) ) {
/* Avoid these words */
if (len>0) {
if (strchr(KEYW_NOT_BEG,keyword[0])) {
keyword[(len=0)]='\0';
}
}
/* Strip ending . and so */
{
int ok=0;
while((len = (int) strlen(keyword)) && (!ok)) {
if (strchr(KEYW_STRIP_END,keyword[len-1])) { /* strip it */
keyword[len-1]='\0';
} else
ok=1;
}
}
/* Store it ? */
if (len >= KEYW_MIN_LEN ) {
hts_primindex_words++;
if (inthash_inc(WordIndexHash,keyword)) { /* added new */
fprintf(tmpfp,"%s\n",keyword);
}
}
keyword[(len=0)]='\0';
} else /* Invalid */
keyword[(len=0)]='\0';
if (len>KEYW_LEN) {
keyword[(len=0)]='\0';
}
}
}
i++;
}
// Reset temp file
fseek(tmpfp,0,SEEK_SET);
// Process indexing for this page
{
//FILE* fp=NULL;
//fp=fopen(concat(indexpath,"index.txt"),"ab");
if (fp_tmpproject) {
while(!feof(tmpfp)) {
char line[KEYW_LEN + 32];
linput(tmpfp,line,KEYW_LEN + 2);
if (strnotempty(line)) {
intptr_t e=0;
if (inthash_read(WordIndexHash,line,&e)) {
//if (e) {
char BIGSTK savelst[HTS_URLMAXSIZE*2];
e++; /* 0 means "once" */
if (strncmp((const char*)fslash(catbuff,(char*)indexpath),filename,strlen(indexpath))==0) // couper
strcpybuff(savelst,filename+strlen(indexpath));
else
strcpybuff(savelst,filename);
// Add entry for this file and word
fprintf(fp_tmpproject,"%s %d %s\n",line,(int) (KEYW_SORT_MAXCOUNT - e),savelst);
hts_primindex_size++;
//}
}
}
}
//fclose(fp);
}
}
// Delete temp file
fclose(tmpfp);
tmpfp=NULL;
// Clear hash table
inthash_delete(&WordIndexHash);
#endif
return 1;
}