WebSVN – wimsdev – Diff – /trunk/wims/src/Misc/modind.c



 
        /* This is an internal program,
         * used to index modules for search engine. */
 
#include "../wims.h"
#include "../Lib/basicstr.c"
 
#define MAX_LANGS       MAX_LANGUAGES
#define MAX_MODULES     65536
char *moduledir=        "public_html/modules";
char *sheetdir=         "public_html/bases/sheet";

{
    char *pp, *p2;
    for(pp=strchr(p,'<'); pp!=NULL; pp=strchr(pp,'<')) {
        p2=find_tag_end(pp);
        if(*p2==0) {*pp=0; return; }
        ovlstrcpy(pp,p2);
    }
}
 
        /* modify a string. Bufferlen must be ast least MAX_LINELEN */
void string_modify(char *start, char *bad_beg, char *bad_end, char *good,...)

    va_start(vp,good);
    vsnprintf(buf,sizeof(buf),good,vp); va_end(vp);
    if(strlen(start)-(bad_end-bad_beg)+strlen(buf)>=MAX_LINELEN)
      return;
    strcat(buf,bad_end);
    ovlstrcpy(bad_beg,buf);
}
 
void _getdef(char buf[], char *name, char value[])
{
    char *p1, *p2, *p3;

    if(langcnt==0) {    /* default languages */
        langcnt=DEFAULT_LANGCNT;
    }
    s=getenv("mlist"); if(s==NULL) exit(1);
    l=strlen(s); if(l<0 || l>100*MAX_LINELEN) exit(1);
    mlist=xmalloc(l+16); ovlstrcpy(mlist,s); old="";
    for(i=0;i<langcnt;i++) {
        snprintf(buf,sizeof(buf),"%s/%s.%s",dicdir,ignoredic,lang[i]);
        f=fopen(buf,"r"); if(f==NULL) continue;
        l=fread(ignore[i],1,MAX_LINELEN,f);fclose(f);
        if(l<0 || l>=MAX_LINELEN) l=0;

    int i,l,thislang;
   
    modcnt=0;
    s=getenv("slist"); if(s==NULL) return;
    l=strlen(s); if(l<0 || l>100*MAX_LINELEN) return;
    mlist=xmalloc(l+16); ovlstrcpy(mlist,s);
    for(p1=find_word_start(mlist); *p1 && modcnt<MAX_MODULES; p1=find_word_start(p2)) {
        p2=find_word_end(p1);
        l=p2-p1; if(*p2) *p2++=0;
        for(i=0;i<langcnt;i++) if(strncasecmp(lang[i],p1,2)==0) break;
        if(i<langcnt) thislang=i; else continue;

    }
    p=find_word_start(indbuf[i_language]);
    if(isalpha(*p) && isalpha(*(p+1))) {
        memmove(module_language,p,2); module_language[2]=0;
    }
    else ovlstrcpy(module_language,"en");
    return 0;
}
 
int sheet_index(int serial)
{

    }
    p2=strstr(p1,"\n:"); if(p2==NULL) p2=p1+strlen(p1);
    else *p2=0;
    p1=find_word_start(p1); strip_trailing_spaces(p1);
    for(p2=p1;*p2;p2++) if(*p2=='\n') *p2=' ';
    ovlstrcpy(sindbuf[s_remark],p1);
    return 0;
}
 
unsigned char categories[16];
char taken[MAX_LINELEN+1];

       takenlen>=MAX_LINELEN-ll-16)
      return;
    if(ll==2 && (!isdigit(word[0]) || !isalpha(word[1]))) return;
    for(p=word;*p;p++) if(!isalnum(*p) && *p!=' ') return;
    taken[takenlen++]=' '; taken[takenlen++]=' ';
    ovlstrcpy(taken+takenlen,word);
    takenlen+=ll; tweight+=weight;
    snprintf(buf,sizeof(buf),"%s:%d?%d\n",word,serial,weight);
    for(i=0;i<catcnt;i++) {
        snprintf(nbuf,sizeof(nbuf),"%s/%c.%s",
                 outdir,categories[i],lang[lind]);

        singlespace(indbuf[trlist[i]]);
        suffix_translate(indbuf[trlist[i]]);
        translate(indbuf[trlist[i]]);
    }
    taken[0]=0; takenlen=tweight=0;
    ovlstrcpy(buf,indbuf[i_title]); towords(buf);
    for(p1=find_word_start(buf);*p1;
        p1=find_word_start(p2)) {
        p2=find_word_end(p1); if(*p2) *p2++=0;
        appenditem(p1,lind,serial,4,module_language);
    }

        appenditem(p1,lind,serial,2,module_language);
    }
    entrycount=gentrycount; dicbuf=gdicbuf;
    memmove(entry,gentry,gentrycount*sizeof(entry[0]));
    unknown_type=unk_delete;
    ovlstrcpy(buf,indbuf[i_title]); translate(buf);
    for(p1=find_word_start(buf); *p1;
        p1=find_word_start(p2)) {
        p2=strchr(p1,',');
        if(p2!=NULL) *p2++=0; else p2=p1+strlen(p1);
        if(strlen(p1)<=0) continue;

        if(p2!=NULL) *p2++=0; else p2=p1+strlen(p1);
        if(strlen(p1)<=0) continue;
        appenditem(p1,lind,serial,2,module_language);
    }
    snprintf(buf,sizeof(buf),"%s",indbuf[i_level]);
    ovlstrcpy(lbuf,"level");
    for(p1=buf; *p1; p1++) if(!isalnum(*p1)) *p1=' ';
    for(p1=find_word_start(buf); *p1;
        p1=find_word_start(p2)) {
        p2=find_word_end(p1);
        if(p2!=NULL) *p2++=0; else p2=p1+strlen(p1);
        if(!isalpha(*p1) ||
           (!isdigit(*(p1+1)) && *(p1+1)!=0) ||
           (*(p1+1)!=0 && *(p1+2)!=0))
          continue;
        *p1=tolower(*p1);
        ovlstrcpy(lbuf+strlen("level"),p1);
        appenditem(lbuf,lind,serial,2,module_language);
    }
    fprintf(weightf,"%d:%d\n",serial,tweight);
}
 

       takenlen>=MAX_LINELEN-ll-16)
      return;
    if(ll==2 && (!isdigit(word[0]) || !isalpha(word[1]))) return;
    for(p=word;*p;p++) if(!isalnum(*p) && *p!=' ') return;
    taken[takenlen++]=' ';taken[takenlen++]=' ';
    ovlstrcpy(taken+takenlen,word);
    takenlen+=ll; tweight+=weight;
    fprintf(indf,"%s:%d?%d\n",word,serial,weight);
}
 
void onesheet(int serial, int lind)

        singlespace(sindbuf[trlist[i]]);
        suffix_translate(sindbuf[trlist[i]]);
        translate(sindbuf[trlist[i]]);
    }
    taken[0]=0; takenlen=tweight=0;
    ovlstrcpy(buf,sindbuf[s_title]); towords(buf);
    for(p1=find_word_start(buf);*p1;
        p1=find_word_start(p2)) {
        p2=find_word_end(p1); if(*p2) *p2++=0;
        sappenditem(p1,lind,serial,4);
    }

        sappenditem(p1,lind,serial,2);
    }
    entrycount=gentrycount; dicbuf=gdicbuf;
    memmove(entry,gentry,gentrycount*sizeof(entry[0]));
    unknown_type=unk_delete;
    ovlstrcpy(buf,sindbuf[s_title]); translate(buf);
    for(p1=find_word_start(buf); *p1;
        p1=find_word_start(p2)) {
        p2=strchr(p1,',');
        if(p2!=NULL) *p2++=0; else p2=p1+strlen(p1);
        if(strlen(p1)<=0) continue;

Rev 3247	Rev 3718
Line 17...	Line 17...
17		17
18	/* This is an internal program,	18	/* This is an internal program,
19	* used to index modules for search engine. */	19	* used to index modules for search engine. */
20		20
21	#include "../wims.h"	21	#include "../wims.h"
-		22	#include "../Lib/basicstr.c"
22		23
23	#define MAX_LANGS MAX_LANGUAGES	24	#define MAX_LANGS MAX_LANGUAGES
24	#define MAX_MODULES 65536	25	#define MAX_MODULES 65536
25	char *moduledir= "public_html/modules";	26	char *moduledir= "public_html/modules";
26	char *sheetdir= "public_html/bases/sheet";	27	char *sheetdir= "public_html/bases/sheet";
Line 190...	Line 191...
190	{	191	{
191	char pp, p2;	192	char pp, p2;
192	for(pp=strchr(p,'<'); pp!=NULL; pp=strchr(pp,'<')) {	193	for(pp=strchr(p,'<'); pp!=NULL; pp=strchr(pp,'<')) {
193	p2=find_tag_end(pp);	194	p2=find_tag_end(pp);
194	if(p2==0) {pp=0; return; }	195	if(p2==0) {pp=0; return; }
195	~~strcpy~~(pp,p2);	196	ovlstrcpy(pp,p2);
196	}	197	}
197	}	198	}
198		199
199	/* modify a string. Bufferlen must be ast least MAX_LINELEN */	200	/* modify a string. Bufferlen must be ast least MAX_LINELEN */
200	void string_modify(char start, char bad_beg, char bad_end, char good,...)	201	void string_modify(char start, char bad_beg, char bad_end, char good,...)
Line 205...	Line 206...
205	va_start(vp,good);	206	va_start(vp,good);
206	vsnprintf(buf,sizeof(buf),good,vp); va_end(vp);	207	vsnprintf(buf,sizeof(buf),good,vp); va_end(vp);
207	if(strlen(start)-(bad_end-bad_beg)+strlen(buf)>=MAX_LINELEN)	208	if(strlen(start)-(bad_end-bad_beg)+strlen(buf)>=MAX_LINELEN)
208	return;	209	return;
209	strcat(buf,bad_end);	210	strcat(buf,bad_end);
210	~~strcpy~~(bad_beg,buf);	211	ovlstrcpy(bad_beg,buf);
211	}	212	}
212		213
213	void _getdef(char buf[], char *name, char value[])	214	void _getdef(char buf[], char *name, char value[])
214	{	215	{
215	char p1, p2, *p3;	216	char p1, p2, *p3;
Line 278...	Line 279...
278	if(langcnt==0) { /* default languages */	279	if(langcnt==0) { /* default languages */
279	langcnt=DEFAULT_LANGCNT;	280	langcnt=DEFAULT_LANGCNT;
280	}	281	}
281	s=getenv("mlist"); if(s==NULL) exit(1);	282	s=getenv("mlist"); if(s==NULL) exit(1);
282	l=strlen(s); if(l<0 \|\| l>100*MAX_LINELEN) exit(1);	283	l=strlen(s); if(l<0 \|\| l>100*MAX_LINELEN) exit(1);
283	mlist=xmalloc(l+16); ~~strcpy~~(mlist,s); old="";	284	mlist=xmalloc(l+16); ovlstrcpy(mlist,s); old="";
284	for(i=0;i<langcnt;i++) {	285	for(i=0;i<langcnt;i++) {
285	snprintf(buf,sizeof(buf),"%s/%s.%s",dicdir,ignoredic,lang[i]);	286	snprintf(buf,sizeof(buf),"%s/%s.%s",dicdir,ignoredic,lang[i]);
286	f=fopen(buf,"r"); if(f==NULL) continue;	287	f=fopen(buf,"r"); if(f==NULL) continue;
287	l=fread(ignore[i],1,MAX_LINELEN,f);fclose(f);	288	l=fread(ignore[i],1,MAX_LINELEN,f);fclose(f);
288	if(l<0 \|\| l>=MAX_LINELEN) l=0;	289	if(l<0 \|\| l>=MAX_LINELEN) l=0;
Line 346...	Line 347...
346	int i,l,thislang;	347	int i,l,thislang;
347		348
348	modcnt=0;	349	modcnt=0;
349	s=getenv("slist"); if(s==NULL) return;	350	s=getenv("slist"); if(s==NULL) return;
350	l=strlen(s); if(l<0 \|\| l>100*MAX_LINELEN) return;	351	l=strlen(s); if(l<0 \|\| l>100*MAX_LINELEN) return;
351	mlist=xmalloc(l+16); ~~strcpy~~(mlist,s);	352	mlist=xmalloc(l+16); ovlstrcpy(mlist,s);
352	for(p1=find_word_start(mlist); *p1 && modcnt<MAX_MODULES; p1=find_word_start(p2)) {	353	for(p1=find_word_start(mlist); *p1 && modcnt<MAX_MODULES; p1=find_word_start(p2)) {
353	p2=find_word_end(p1);	354	p2=find_word_end(p1);
354	l=p2-p1; if(p2) p2++=0;	355	l=p2-p1; if(p2) p2++=0;
355	for(i=0;i<langcnt;i++) if(strncasecmp(lang[i],p1,2)==0) break;	356	for(i=0;i<langcnt;i++) if(strncasecmp(lang[i],p1,2)==0) break;
356	if(i<langcnt) thislang=i; else continue;	357	if(i<langcnt) thislang=i; else continue;
Line 422...	Line 423...
422	}	423	}
423	p=find_word_start(indbuf[i_language]);	424	p=find_word_start(indbuf[i_language]);
424	if(isalpha(p) && isalpha((p+1))) {	425	if(isalpha(p) && isalpha((p+1))) {
425	memmove(module_language,p,2); module_language[2]=0;	426	memmove(module_language,p,2); module_language[2]=0;
426	}	427	}
427	else ~~strcpy~~(module_language,"en");	428	else ovlstrcpy(module_language,"en");
428	return 0;	429	return 0;
429	}	430	}
430		431
431	int sheet_index(int serial)	432	int sheet_index(int serial)
432	{	433	{
Line 449...	Line 450...
449	}	450	}
450	p2=strstr(p1,"\n:"); if(p2==NULL) p2=p1+strlen(p1);	451	p2=strstr(p1,"\n:"); if(p2==NULL) p2=p1+strlen(p1);
451	else *p2=0;	452	else *p2=0;
452	p1=find_word_start(p1); strip_trailing_spaces(p1);	453	p1=find_word_start(p1); strip_trailing_spaces(p1);
453	for(p2=p1;p2;p2++) if(p2=='\n') *p2=' ';	454	for(p2=p1;p2;p2++) if(p2=='\n') *p2=' ';
454	~~strcpy~~(sindbuf[s_remark],p1);	455	ovlstrcpy(sindbuf[s_remark],p1);
455	return 0;	456	return 0;
456	}	457	}
457		458
458	unsigned char categories[16];	459	unsigned char categories[16];
459	char taken[MAX_LINELEN+1];	460	char taken[MAX_LINELEN+1];
Line 472...	Line 473...
472	takenlen>=MAX_LINELEN-ll-16)	473	takenlen>=MAX_LINELEN-ll-16)
473	return;	474	return;
474	if(ll==2 && (!isdigit(word[0]) \|\| !isalpha(word[1]))) return;	475	if(ll==2 && (!isdigit(word[0]) \|\| !isalpha(word[1]))) return;
475	for(p=word;p;p++) if(!isalnum(p) && *p!=' ') return;	476	for(p=word;p;p++) if(!isalnum(p) && *p!=' ') return;
476	taken[takenlen++]=' '; taken[takenlen++]=' ';	477	taken[takenlen++]=' '; taken[takenlen++]=' ';
477	~~strcpy~~(taken+takenlen,word);	478	ovlstrcpy(taken+takenlen,word);
478	takenlen+=ll; tweight+=weight;	479	takenlen+=ll; tweight+=weight;
479	snprintf(buf,sizeof(buf),"%s:%d?%d\n",word,serial,weight);	480	snprintf(buf,sizeof(buf),"%s:%d?%d\n",word,serial,weight);
480	for(i=0;i<catcnt;i++) {	481	for(i=0;i<catcnt;i++) {
481	snprintf(nbuf,sizeof(nbuf),"%s/%c.%s",	482	snprintf(nbuf,sizeof(nbuf),"%s/%c.%s",
482	outdir,categories[i],lang[lind]);	483	outdir,categories[i],lang[lind]);
Line 531...	Line 532...
531	singlespace(indbuf[trlist[i]]);	532	singlespace(indbuf[trlist[i]]);
532	suffix_translate(indbuf[trlist[i]]);	533	suffix_translate(indbuf[trlist[i]]);
533	translate(indbuf[trlist[i]]);	534	translate(indbuf[trlist[i]]);
534	}	535	}
535	taken[0]=0; takenlen=tweight=0;	536	taken[0]=0; takenlen=tweight=0;
536	~~strcpy~~(buf,indbuf[i_title]); towords(buf);	537	ovlstrcpy(buf,indbuf[i_title]); towords(buf);
537	for(p1=find_word_start(buf);*p1;	538	for(p1=find_word_start(buf);*p1;
538	p1=find_word_start(p2)) {	539	p1=find_word_start(p2)) {
539	p2=find_word_end(p1); if(p2) p2++=0;	540	p2=find_word_end(p1); if(p2) p2++=0;
540	appenditem(p1,lind,serial,4,module_language);	541	appenditem(p1,lind,serial,4,module_language);
541	}	542	}
Line 549...	Line 550...
549	appenditem(p1,lind,serial,2,module_language);	550	appenditem(p1,lind,serial,2,module_language);
550	}	551	}
551	entrycount=gentrycount; dicbuf=gdicbuf;	552	entrycount=gentrycount; dicbuf=gdicbuf;
552	memmove(entry,gentry,gentrycount*sizeof(entry[0]));	553	memmove(entry,gentry,gentrycount*sizeof(entry[0]));
553	unknown_type=unk_delete;	554	unknown_type=unk_delete;
554	~~strcpy~~(buf,indbuf[i_title]); translate(buf);	555	ovlstrcpy(buf,indbuf[i_title]); translate(buf);
555	for(p1=find_word_start(buf); *p1;	556	for(p1=find_word_start(buf); *p1;
556	p1=find_word_start(p2)) {	557	p1=find_word_start(p2)) {
557	p2=strchr(p1,',');	558	p2=strchr(p1,',');
558	if(p2!=NULL) *p2++=0; else p2=p1+strlen(p1);	559	if(p2!=NULL) *p2++=0; else p2=p1+strlen(p1);
559	if(strlen(p1)<=0) continue;	560	if(strlen(p1)<=0) continue;
Line 569...	Line 570...
569	if(p2!=NULL) *p2++=0; else p2=p1+strlen(p1);	570	if(p2!=NULL) *p2++=0; else p2=p1+strlen(p1);
570	if(strlen(p1)<=0) continue;	571	if(strlen(p1)<=0) continue;
571	appenditem(p1,lind,serial,2,module_language);	572	appenditem(p1,lind,serial,2,module_language);
572	}	573	}
573	snprintf(buf,sizeof(buf),"%s",indbuf[i_level]);	574	snprintf(buf,sizeof(buf),"%s",indbuf[i_level]);
574	~~strcpy~~(lbuf,"level");	575	ovlstrcpy(lbuf,"level");
575	for(p1=buf; p1; p1++) if(!isalnum(p1)) *p1=' ';	576	for(p1=buf; p1; p1++) if(!isalnum(p1)) *p1=' ';
576	for(p1=find_word_start(buf); *p1;	577	for(p1=find_word_start(buf); *p1;
577	p1=find_word_start(p2)) {	578	p1=find_word_start(p2)) {
578	p2=find_word_end(p1);	579	p2=find_word_end(p1);
579	if(p2!=NULL) *p2++=0; else p2=p1+strlen(p1);	580	if(p2!=NULL) *p2++=0; else p2=p1+strlen(p1);
580	if(!isalpha(*p1) \|\|	581	if(!isalpha(*p1) \|\|
581	(!isdigit((p1+1)) && (p1+1)!=0) \|\|	582	(!isdigit((p1+1)) && (p1+1)!=0) \|\|
582	((p1+1)!=0 && (p1+2)!=0))	583	((p1+1)!=0 && (p1+2)!=0))
583	continue;	584	continue;
584	p1=tolower(p1);	585	p1=tolower(p1);
585	~~strcpy~~(lbuf+strlen("level"),p1);	586	ovlstrcpy(lbuf+strlen("level"),p1);
586	appenditem(lbuf,lind,serial,2,module_language);	587	appenditem(lbuf,lind,serial,2,module_language);
587	}	588	}
588	fprintf(weightf,"%d:%d\n",serial,tweight);	589	fprintf(weightf,"%d:%d\n",serial,tweight);
589	}	590	}
590		591
Line 639...	Line 640...
639	takenlen>=MAX_LINELEN-ll-16)	640	takenlen>=MAX_LINELEN-ll-16)
640	return;	641	return;
641	if(ll==2 && (!isdigit(word[0]) \|\| !isalpha(word[1]))) return;	642	if(ll==2 && (!isdigit(word[0]) \|\| !isalpha(word[1]))) return;
642	for(p=word;p;p++) if(!isalnum(p) && *p!=' ') return;	643	for(p=word;p;p++) if(!isalnum(p) && *p!=' ') return;
643	taken[takenlen++]=' ';taken[takenlen++]=' ';	644	taken[takenlen++]=' ';taken[takenlen++]=' ';
644	~~strcpy~~(taken+takenlen,word);	645	ovlstrcpy(taken+takenlen,word);
645	takenlen+=ll; tweight+=weight;	646	takenlen+=ll; tweight+=weight;
646	fprintf(indf,"%s:%d?%d\n",word,serial,weight);	647	fprintf(indf,"%s:%d?%d\n",word,serial,weight);
647	}	648	}
648		649
649	void onesheet(int serial, int lind)	650	void onesheet(int serial, int lind)
Line 668...	Line 669...
668	singlespace(sindbuf[trlist[i]]);	669	singlespace(sindbuf[trlist[i]]);
669	suffix_translate(sindbuf[trlist[i]]);	670	suffix_translate(sindbuf[trlist[i]]);
670	translate(sindbuf[trlist[i]]);	671	translate(sindbuf[trlist[i]]);
671	}	672	}
672	taken[0]=0; takenlen=tweight=0;	673	taken[0]=0; takenlen=tweight=0;
673	~~strcpy~~(buf,sindbuf[s_title]); towords(buf);	674	ovlstrcpy(buf,sindbuf[s_title]); towords(buf);
674	for(p1=find_word_start(buf);*p1;	675	for(p1=find_word_start(buf);*p1;
675	p1=find_word_start(p2)) {	676	p1=find_word_start(p2)) {
676	p2=find_word_end(p1); if(p2) p2++=0;	677	p2=find_word_end(p1); if(p2) p2++=0;
677	sappenditem(p1,lind,serial,4);	678	sappenditem(p1,lind,serial,4);
678	}	679	}
Line 686...	Line 687...
686	sappenditem(p1,lind,serial,2);	687	sappenditem(p1,lind,serial,2);
687	}	688	}
688	entrycount=gentrycount; dicbuf=gdicbuf;	689	entrycount=gentrycount; dicbuf=gdicbuf;
689	memmove(entry,gentry,gentrycount*sizeof(entry[0]));	690	memmove(entry,gentry,gentrycount*sizeof(entry[0]));
690	unknown_type=unk_delete;	691	unknown_type=unk_delete;
691	~~strcpy~~(buf,sindbuf[s_title]); translate(buf);	692	ovlstrcpy(buf,sindbuf[s_title]); translate(buf);
692	for(p1=find_word_start(buf); *p1;	693	for(p1=find_word_start(buf); *p1;
693	p1=find_word_start(p2)) {	694	p1=find_word_start(p2)) {
694	p2=strchr(p1,',');	695	p2=strchr(p1,',');
695	if(p2!=NULL) *p2++=0; else p2=p1+strlen(p1);	696	if(p2!=NULL) *p2++=0; else p2=p1+strlen(p1);
696	if(strlen(p1)<=0) continue;	697	if(strlen(p1)<=0) continue;

Subversion Repositories wimsdev

(root)/trunk/wims/src/Misc/modind.c – Rev 3247 → 3718