Subversion Repositories wimsdev

Rev

Rev 6973 | Rev 8100 | Go to most recent revision | Blame | Compare with Previous | Last modification | View Log | RSS feed

  1. /*    Copyright (C) 1998-2003 XIAO, Gang of Universite de Nice - Sophia Antipolis
  2.  *
  3.  *  This program is free software; you can redistribute it and/or modify
  4.  *  it under the terms of the GNU General Public License as published by
  5.  *  the Free Software Foundation; either version 2 of the License, or
  6.  *  (at your option) any later version.
  7.  *
  8.  *  This program is distributed in the hope that it will be useful,
  9.  *  but WITHOUT ANY WARRANTY; without even the implied warranty of
  10.  *  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  11.  *  GNU General Public License for more details.
  12.  *
  13.  *  You should have received a copy of the GNU General Public License
  14.  *  along with this program; if not, write to the Free Software
  15.  *  Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
  16.  */
  17.  
  18. /*  This is an internal program,
  19.  * used to index modules for search engine.
  20.  */
  21.  
  22. #include "../wims.h"
  23. #include "../Lib/basicstr.c"
  24.  
  25. #define MAX_LANGS    MAX_LANGUAGES
  26. #define MAX_MODULES    65536
  27. char *moduledir=    "public_html/modules";
  28. char *sheetdir=     "public_html/bases/sheet";
  29. char *dicdir=       "public_html/bases";
  30. char *outdir=       "public_html/bases/site2";
  31. char *maindic=      "sys/words";
  32. char *groupdic=     "sys/wgrp/wgrp";
  33. char *suffixdic=    "sys/suffix";
  34. char *domaindic=    "sys/domaindic";
  35. char *ignoredic=    "sys/indignore";
  36. char *conffile=     "log/wims.conf";
  37. char *mlistbase=    "list";
  38.  
  39. char lang[MAX_LANGS][4]={
  40.     "en","fr","cn","es","it","nl","si","ca","pt"
  41. };
  42. #define DEFAULT_LANGCNT    6
  43. char allang[MAX_LANGS][4]={
  44.     "en","fr","cn","es","it","nl","de","si","ca","pt"
  45. };
  46. #define allangcnt 8
  47. char ignore[MAX_LANGS][MAX_LINELEN+1];
  48. char mlistfile[MAX_LANGS][256];
  49. int langcnt;
  50. FILE *langf, *titf, *descf, *weightf, *robotf, *indf, *listf, *addrf, *serialf, *authorf, *versionf, *remf;
  51.  
  52. struct cat {
  53.     char *name;
  54.     char typ;
  55. } cat[]={
  56.     {"all_types", 'A'},
  57.     {"exercise",  'X'},
  58.     {"oef",       'O'},
  59.     {"tool",      'T'},
  60.     {"recreation",'R'},
  61.     {"reference", 'Y'},
  62.     {"document",  'D'},
  63.     {"popup",     'P'},
  64.     {"datamodule",'M'}
  65. };
  66. #define catno (sizeof(cat)/sizeof(cat[0]))
  67.  
  68. struct mod {
  69.     char *name;
  70.     unsigned char langs[MAX_LANGS];
  71.     int counts[MAX_LANGS];
  72.     int  langcnt;
  73. } mod[MAX_MODULES];
  74. int modcnt;
  75.  
  76. char *mlist;
  77.  
  78. void *xmalloc(size_t n)
  79. {
  80.     void *p;
  81.     p=malloc(n);
  82.     if(p==NULL) {
  83.     printf("Malloc failure.\n");
  84.     exit(1);
  85.     }
  86.     return p;
  87. }
  88.  
  89. char *acctab="çéèêëúùûüáàâäãóòôöõíìïîñýÇÉÈÊËÚÙÛÜÁÀÂÃÄÓÒÔÖÕÍÌÏÎÑÝ",
  90.      *deatab="ceeeeuuuuaaaaaoooooiiiinyCEEEEUUUUAAAAAOOOOOIIIINY";
  91.  
  92. /*  fold known accented letters to unaccented, other strange characters to space
  93.  *  apostrophe is among the exceptions to be kept (important for multi-word expressions)
  94.  */
  95. void deaccent(char *p)
  96. {
  97.     char *sp;
  98.     char *v;
  99.     for(sp=p;*sp;sp++) {
  100.     if(*sp<0 && (v=strchr(acctab,*sp))!=NULL)
  101.       *sp=*(deatab+(v-acctab));
  102.     if(!isalnum(*sp) && strchr(",.&$+*",*sp)==0) *sp=' ';
  103.     else *sp=tolower(*sp);
  104.     }
  105. }
  106.  
  107. /*  translate everything non-alphanumeric into space */
  108. void towords(char *p)
  109. {
  110.     char *pp;
  111.     for(pp=p;*pp;pp++) if(!isalnum(*pp) && strchr("&$+*",*pp)==0) *pp=' ';
  112. }
  113.  
  114. /*  Points to the end of the word */
  115. char *find_word_end(char *p)
  116. {
  117.     int i;
  118.     for(i=0;!isspace(*p) && *p!=0 && i<MAX_LINELEN; p++,i++);
  119.     return p;
  120. }
  121.  
  122. /*  Strips leading spaces */
  123. char *find_word_start(char *p)
  124. {
  125.     int i;
  126.     for(i=0; isspace(*p) && i<MAX_LINELEN; p++,i++);
  127.     return p;
  128. }
  129.  
  130. /*  Find first occurrence of word */
  131. char *wordchr(char *p, char *w)
  132. {
  133.     char *r;
  134.  
  135.     for(r=strstr(p,w);r!=NULL &&
  136.     ( (r>p && !isspace(*(r-1))) || (!isspace(*(r+strlen(w))) && *(r+strlen(w))!=0) );
  137.     r=strstr(r+1,w));
  138.     return r;
  139. }
  140.  
  141. /*  find a variable in a string (math expression).
  142.  * Returns the pointer or NULL.
  143.  */
  144. char *varchr(char *p, char *v)
  145. {
  146.     char *pp; int n=strlen(v);
  147.     for(pp=strstr(p,v); pp!=NULL; pp=strstr(pp+1,v)) {
  148.     if((pp==p || !isalnum(*(pp-1))) &&
  149.        (!isalnum(*(pp+n)) || *(pp+n)==0)) break;
  150.     }
  151.     return pp;
  152. }
  153.  
  154. /*  strip trailing spaces; return string end. */
  155. char *strip_trailing_spaces(char *p)
  156. {
  157.     char *pp;
  158.     if(*p==0) return p;
  159.     for(pp=p+strlen(p)-1; pp>=p && isspace(*pp); *(pp--)=0);
  160.     return pp;
  161. }
  162.  
  163. char *find_tag_end(char *p)
  164. {
  165.     char *pp;
  166.     pp=p; if(*pp=='<') pp++;
  167.     for(; *pp && *pp!='>'; pp++) {
  168.     if(*pp=='<') {
  169.         pp=find_tag_end(pp)-1; continue;
  170.     }
  171.     if(*pp=='"') {
  172.         pp=strchr(pp+1,'"');
  173.         if(pp==NULL) return p+strlen(p); else continue;
  174.     }
  175.     if(*pp=='\'') {
  176.         pp=strchr(pp+1,'\'');
  177.         if(pp==NULL) return p+strlen(p); else continue;
  178.     }
  179.     }
  180.     if(*pp=='>') pp++; return pp;
  181. }
  182.  
  183. char *find_tag(char *p, char *tag)
  184. {
  185.     char *pp;
  186.     int len;
  187.     len=strlen(tag);
  188.     for(pp=strchr(p,'<'); pp!=NULL && *pp; pp=strchr(pp+1,'<')) {
  189.     if(strncasecmp(pp+1,tag,len)==0 && !isalnum(*(pp+1+len))) return pp;
  190.     }
  191.     return p+strlen(p);
  192. }
  193.  
  194. /*  remove all html tags */
  195. void detag(char *p)
  196. {
  197.     char *pp, *p2;
  198.     for(pp=strchr(p,'<'); pp!=NULL; pp=strchr(pp,'<')) {
  199.     p2=find_tag_end(pp);
  200.     if(*p2==0) {*pp=0; return; }
  201.     ovlstrcpy(pp,p2);
  202.     }
  203. }
  204.  
  205. /*  modify a string. Bufferlen must be at least MAX_LINELEN */
  206. void string_modify(char *start, char *bad_beg, char *bad_end, char *good,...)
  207. {
  208.     char buf[MAX_LINELEN+1];
  209.     va_list vp;
  210.  
  211.     va_start(vp,good);
  212.     vsnprintf(buf,sizeof(buf),good,vp); va_end(vp);
  213.     if(strlen(start)-(bad_end-bad_beg)+strlen(buf)>=MAX_LINELEN)
  214.       return;
  215.     strcat(buf,bad_end);
  216.     ovlstrcpy(bad_beg,buf);
  217. }
  218.  
  219. /* add a space after comma to see end of words */
  220.  
  221. void comma(char *p)
  222. {
  223.     char *pp;
  224.     for(pp=strchr(p,','); pp; pp=strchr(pp+1,','))
  225.       string_modify(p,pp,pp+1,", ");
  226. }
  227.  
  228. void _getdef(char buf[], char *name, char value[])
  229. {
  230.     char *p1, *p2, *p3;
  231.  
  232.     value[0]=0;
  233.     for(p1=strstr(buf,name); p1!=NULL; p1=strstr(p1+1,name)) {
  234.     p2=find_word_start(p1+strlen(name));
  235.     if((p1>buf && !isspace(*(p1-1))) || *p2!='=') continue;
  236.     p3=p1; while(p3>buf && isspace(*(p3-1)) && *(p3-1)!='\n') p3--;
  237.     if(p3>buf && *(p3-1)!='\n') continue;
  238.     p3=strchr(p2,'\n');
  239.     p2=find_word_start(p2+1);
  240.     if(p3 <= p2) continue;
  241.     snprintf(value,MAX_LINELEN,"%s",p2);
  242.     if(p3!=NULL && p3-p2<MAX_LINELEN) value[p3-p2]=0;
  243.     strip_trailing_spaces(value);
  244.     break;
  245.     }
  246. }
  247.  
  248. /*  Get variable definition from a file.
  249.  * Result stored in buffer value of length MAX_LINELEN.
  250.  */
  251. void getdef(char *fname, char *name, char value[])
  252. {
  253.     FILE *f;
  254.     char *buf;
  255.     int l;
  256.  
  257.     value[0]=0;
  258.     f=fopen(fname,"r"); if(f==NULL) return;
  259.     fseek(f,0,SEEK_END); l=ftell(f); fseek(f,0,SEEK_SET);
  260.     buf=xmalloc(l+256); l=fread(buf,1,l,f);
  261.     fclose(f);
  262.     if(l<=0) return; else buf[l]=0;
  263.     _getdef(buf,name,value);
  264.     free(buf);
  265. }
  266.  
  267. #include "translator_.c"
  268.  
  269. char *mdicbuf, *gdicbuf, *ddicbuf;
  270. char gentry[sizeof(entry)], mentry[sizeof(entry)], dentry[sizeof(entry)];
  271. int gentrycount, mentrycount, dentrycount;
  272.  
  273. /*  Preparation of data */
  274. void prep(void)
  275. {
  276.     char buf[MAX_LINELEN+1];
  277.     char *p1,*p2,*s,*old;
  278.     int i,l,thislang,t;
  279.     FILE *f;
  280.  
  281.     s=getenv("modind_outdir"); if(s!=NULL && *s!=0) outdir=s;
  282.     s=getenv("modind_sheetdir"); if(s!=NULL && *s!=0) sheetdir=s;
  283.     snprintf(buf,sizeof(buf),"%s/addr",outdir);
  284.     addrf=fopen(buf,"w");
  285.     snprintf(buf,sizeof(buf),"%s/serial",outdir);
  286.     serialf=fopen(buf,"w");
  287.     modcnt=langcnt=0;
  288. /* take the langs declared in conffile */
  289.     getdef(conffile,"site_languages",buf);
  290.     for(p1=buf;*p1;p1++) if(!isalnum(*p1)) *p1=' ';
  291.     for(p1=find_word_start(buf); *p1 && langcnt<MAX_LANGS; p1=find_word_start(p2)) {
  292.     p2=find_word_end(p1);
  293.     if(p2!=p1+2 || !isalpha(*p1) || !isalpha(*(p1+1))) continue;
  294.     memmove(lang[langcnt],p1,2); lang[langcnt++][2]=0;
  295.     }
  296.     if(langcnt==0) {/*  default languages */
  297.     langcnt=DEFAULT_LANGCNT;
  298.     }
  299.     s=getenv("mlist"); if(s==NULL) exit(1);
  300.     l=strlen(s); if(l<0 || l>100*MAX_LINELEN) exit(1);
  301.     mlist=xmalloc(l+16); ovlstrcpy(mlist,s); old="";
  302.     for(i=0;i<langcnt;i++) {
  303.     snprintf(buf,sizeof(buf),"%s/%s.%s",dicdir,ignoredic,lang[i]);
  304.     f=fopen(buf,"r"); if(f==NULL) continue;
  305.     l=fread(ignore[i],1,MAX_LINELEN,f);fclose(f);
  306.     if(l<0 || l>=MAX_LINELEN) l=0;
  307.     ignore[i][l]=0;
  308.     }
  309.     for(t=0, p1=find_word_start(mlist);
  310.     *p1 && modcnt<MAX_MODULES;
  311.     p1=find_word_start(p2), t++) {
  312.     p2=find_word_end(p1);
  313.     l=p2-p1; if(*p2) *p2++=0;
  314.     fprintf(addrf,"%d:%s\n",t,p1);
  315.     fprintf(serialf,"%s:%d\n",p1,t);
  316.     thislang=-1;
  317. /* language is taken from the address */
  318.     if(l>3 && p1[l-3]=='.') {
  319.         for(i=0;i<langcnt;i++) if(strcasecmp(lang[i],p1+l-2)==0) break;
  320.         if(i<langcnt) {p1[l-3]=0; thislang=i;}
  321.         else {/*  unknown language, not referenced */
  322.         continue;
  323.         }
  324.     }
  325.     if(modcnt>0 && strcmp(old,p1)==0 && thislang>=0) {
  326.         if(mod[modcnt-1].langcnt<langcnt) {
  327.         mod[modcnt-1].langs[mod[modcnt-1].langcnt]=thislang;
  328.         mod[modcnt-1].counts[mod[modcnt-1].langcnt]=t;
  329.         (mod[modcnt-1].langcnt)++;
  330.         }
  331.     }
  332.     else {
  333.         mod[modcnt].name=old=p1;
  334.         if(thislang>=0) {
  335.         mod[modcnt].langs[0]=thislang;
  336.         mod[modcnt].langcnt=1;
  337.         }
  338.         else mod[modcnt].langcnt=0;
  339.         mod[modcnt].counts[0]=t;
  340.         modcnt++;
  341.     }
  342.     }
  343.     snprintf(buf,sizeof(buf),"%s/language",outdir);
  344.     langf=fopen(buf,"w");
  345.     snprintf(buf,sizeof(buf),"%s/title",outdir);
  346.     titf=fopen(buf,"w");
  347.     snprintf(buf,sizeof(buf),"%s/description",outdir);
  348.     descf=fopen(buf,"w");
  349.     snprintf(buf,sizeof(buf),"%s/author",outdir);
  350.     authorf=fopen(buf,"w");
  351.     snprintf(buf,sizeof(buf),"%s/version",outdir);
  352.     versionf=fopen(buf,"w");
  353.     snprintf(buf,sizeof(buf),"%s/lists/robot.phtml",outdir);
  354.     robotf=fopen(buf,"w");
  355.     fclose(addrf); fclose(serialf);
  356.     if(!robotf || !versionf || !authorf || !descf || !titf || !descf) {
  357.     fprintf(stderr,"modind: error creating output files.\n");
  358.     exit(1);
  359.     }
  360. }
  361.  
  362. void sprep(void)
  363. {
  364.     char *p1,*p2,*s;
  365.     int i,l,thislang;
  366.  
  367.     modcnt=0;
  368.     s=getenv("slist"); if(s==NULL) return;
  369.     l=strlen(s); if(l<0 || l>100*MAX_LINELEN) return;
  370.     mlist=xmalloc(l+16); ovlstrcpy(mlist,s);
  371.     for(p1=find_word_start(mlist); *p1 && modcnt<MAX_MODULES; p1=find_word_start(p2)) {
  372.     p2=find_word_end(p1);
  373.     l=p2-p1; if(*p2) *p2++=0;
  374.     for(i=0;i<langcnt;i++) if(strncasecmp(lang[i],p1,2)==0) break;
  375.     if(i<langcnt) thislang=i; else continue;
  376.     mod[modcnt].name=p1;
  377.     mod[modcnt].langs[0]=thislang;
  378.     mod[modcnt].langcnt=1;
  379.     modcnt++;
  380.     }
  381. }
  382.  
  383. void clean(void)
  384. {
  385.     fclose(langf); fclose(titf); fclose(descf); fclose(robotf);
  386.     fclose(authorf); fclose(versionf);
  387. }
  388.  
  389. char *sheetindex[]={
  390.       "title", "description",
  391.       "duration", "severity",
  392.       "level", "domain",
  393.       "keywords", "reserved1", "reserved2", "information"
  394. };
  395. #define SHEETINDEX_NO (sizeof(sheetindex)/sizeof(sheetindex[0]))
  396. char sindbuf[SHEETINDEX_NO][MAX_LINELEN+1];
  397. enum{s_title, s_description,
  398.       s_duration, s_severity,
  399.       s_level, s_domain,
  400.       s_keywords, s_reserved1, s_reserved2,
  401.       s_information
  402. };
  403.  
  404. char *modindex[]={
  405.       "title", "description",
  406.       "author", "address", "copyright",
  407.       "version", "wims_version", "language",
  408.       "category", "level", "domain", "keywords",
  409.       "keywords_ca", "keywords_en", "keywords_fr", "keywords_it", "keywords_nl",
  410.       "title_ca", "title_en", "title_fr", "title_it", "title_nl",
  411.       "require"
  412. };
  413. #define MODINDEX_NO (sizeof(modindex)/sizeof(modindex[0]))
  414. char indbuf[MODINDEX_NO][MAX_LINELEN+1];
  415. enum{i_title, i_description,
  416.       i_author,i_address,i_copyright,
  417.       i_version,i_wims_version,i_language,
  418.       i_category,i_level,i_domain,i_keywords,
  419.       i_keywords_ca,i_keywords_en,i_keywords_fr,i_keywords_it,i_keywords_nl,
  420.       i_title_ca,i_title_en,i_title_fr,i_title_it,i_title_nl,
  421.       i_require
  422. };
  423.  
  424. char *module_special_file[]={
  425.     "intro","help","about"
  426. };
  427. #define MODSPEC_NO (sizeof(module_special_file)/sizeof(module_special_file[0]))
  428. char module_language[4];
  429.  
  430. /*  read and treat module's INDEX file */
  431. int module_index(const char *name)
  432. {
  433.     char *p, fbuf[MAX_LINELEN+1], ibuf[MAX_LINELEN+1];
  434.     FILE *indf;
  435.     int i,l;
  436.  
  437.     snprintf(fbuf,sizeof(fbuf),"%s/%s/INDEX",moduledir,name);
  438.     indf=fopen(fbuf,"r"); if(indf==NULL) return -1;
  439.     l=fread(ibuf,1,MAX_LINELEN,indf); fclose(indf);
  440.     if(l>0 && l<MAX_LINELEN) ibuf[l]=0; else return -1;
  441. /* treate all fields in *modindex */
  442.     for(i=0;i<MODINDEX_NO;i++) {
  443.     _getdef(ibuf,modindex[i],indbuf[i]);
  444. /*  compatibility precaution */
  445.     if(indbuf[i][0]==':') indbuf[i][0]='.';
  446.     }
  447.     p=find_word_start(indbuf[i_language]);
  448.     if(isalpha(*p) && isalpha(*(p+1))) {
  449.     memmove(module_language,p,2); module_language[2]=0;
  450.     }
  451.     else ovlstrcpy(module_language,"en");
  452.     return 0;
  453. }
  454.  
  455. int sheet_index(int serial)
  456. {
  457.     char *p1, *p2, fbuf[MAX_LINELEN+1], ibuf[MAX_LINELEN+1];
  458.     FILE *indf;
  459.     int i,l;
  460.  
  461.     snprintf(fbuf,sizeof(fbuf),"%s/%s.def",sheetdir,mod[serial].name);
  462.     indf=fopen(fbuf,"r"); if(indf==NULL) return -1;
  463.     l=fread(ibuf,1,MAX_LINELEN,indf); fclose(indf);
  464.     if(l>0 && l<MAX_LINELEN) ibuf[l]=0; else return -1;
  465.     for(i=0;i<SHEETINDEX_NO;i++) sindbuf[i][0]=0;
  466.     for(i=0,p1=find_word_start(ibuf);
  467.     i<SHEETINDEX_NO-1 && *p1!=':' && *p1!=0;
  468.     i++,p1=p2) {
  469.     p2=strchr(p1,'\n');
  470.     if(p2!=NULL) *p2++=0; else p2=p1+strlen(p1);
  471.     p1=find_word_start(p1); strip_trailing_spaces(p1);
  472.     snprintf(sindbuf[i],MAX_LINELEN,"%s",p1);
  473.     }
  474.     p2=strstr(p1,"\n:"); if(p2==NULL) p2=p1+strlen(p1);
  475.     else *p2=0;
  476.     p1=find_word_start(p1); strip_trailing_spaces(p1);
  477.     for(p2=p1;*p2;p2++) if(*p2=='\n') *p2=' ';
  478.     ovlstrcpy(sindbuf[s_information],p1);
  479.     return 0;
  480. }
  481.  
  482. unsigned char categories[16];
  483. char taken[MAX_LINELEN+1];
  484. int catcnt, takenlen, tweight;
  485.  
  486. void appenditem(char *word, int lind, int serial, int weight, char *l)
  487. {
  488.     char nbuf[MAX_LINELEN+1], buf[MAX_LINELEN+1];
  489.     int i, ll;
  490.     char *p;
  491.     FILE *f;
  492.  
  493.     if(!isalnum(*word) || (ll=strlen(word))<2 ||
  494.        wordchr(taken,word)!=NULL ||
  495.        wordchr(ignore[lind],word)!=NULL ||
  496.        takenlen>=MAX_LINELEN-ll-16)
  497.       return;
  498.     if(ll==2 && (!isdigit(word[0]) || !isalpha(word[1]))) return;
  499.     for(p=word;*p;p++) if(!isalnum(*p) && *p!=' ') return;
  500.     taken[takenlen++]=' '; taken[takenlen++]=' ';
  501.     ovlstrcpy(taken+takenlen,word);
  502.     takenlen+=ll; tweight+=weight;
  503.     snprintf(buf,sizeof(buf),"%s:%d?%d\n",word,serial,weight);
  504.     for(i=0;i<catcnt;i++) {
  505.     snprintf(nbuf,sizeof(nbuf),"%s/%c.%s",
  506.          outdir,categories[i],lang[lind]);
  507.     f=fopen(nbuf,"a");
  508.     if(f!=NULL) {fputs(buf,f); fclose(f);}
  509.     }
  510. }
  511.  
  512. void appenditem1 (char *buf, int lind, int serial, int weight, char *l )
  513. {
  514.   char *p1, *p2 ;
  515.   for(p1=find_word_start(buf); *p1;
  516.     p1=find_word_start(p2)) {
  517.     p2=strchr(p1,',');
  518.     if(p2!=NULL) *p2++=0; else p2=p1+strlen(p1);
  519.     if(strlen(p1)<=0) continue;
  520.     appenditem(p1,lind,serial,weight,module_language);
  521.   }
  522. }
  523. void appenditem2 (char *buf, int lind, int serial, int weight, char *l )
  524. {
  525.   char *p1, *p2 ;
  526.   for(p1=find_word_start(buf);*p1;
  527.     p1=find_word_start(p2)) {
  528.     p2=find_word_end(p1); if(*p2) *p2++=0;
  529.     appenditem(p1,lind,serial,weight,module_language);
  530.   }
  531. }
  532. void onemodule(const char *name, int serial, int lind)
  533. {
  534.     int i;
  535.     unsigned char trlist[]={
  536.     i_title,i_description,i_category,i_domain,i_keywords,
  537.       i_require,i_author,
  538.       i_keywords_ca,i_keywords_en,i_keywords_fr,i_keywords_it,i_keywords_nl,
  539.       i_title_ca,i_title_en,i_title_fr,i_title_it,i_title_nl
  540.     };
  541.     #define trcnt (sizeof(trlist)/sizeof(trlist[0]))
  542.     char *p1, *p2, *pp, *q, buf[MAX_LINELEN+1], lbuf[16];
  543.     FILE *f;
  544.  
  545.     if(module_index(name)) return;
  546.     towords(indbuf[i_category]);
  547. /*   list the categories (among A=all,X=eXercise,O,D,...) corresponding
  548.  *   to this module
  549.  */
  550.     for(i=catcnt=0;i<catno && catcnt<16;i++) {
  551.     if(wordchr(indbuf[i_category],cat[i].name)!=NULL)
  552.       categories[catcnt++]=cat[i].typ;
  553.     }
  554.     if(catcnt==0) return;
  555.     if(categories[0]!=cat[0].typ)
  556.       categories[catcnt++]=cat[0].typ;
  557. /*  write module's name in the category.language files, for instance lists/X.fr
  558.  * for french exercises
  559.  */
  560.     for(i=0;i<catcnt;i++) {
  561.     snprintf(buf,sizeof(buf),"%s/lists/%c.%s",
  562.          outdir,categories[i],lang[lind]);
  563.     f=fopen(buf,"a");
  564.     if(f!=NULL) {fprintf(f,"%s\n",name); fclose(f);}
  565.     }
  566. /*   add serial number and language (resp.title, ...) to corresponding file  */
  567.     fprintf(langf,"%d:%s\n",serial,module_language);
  568.     fprintf(titf,"%d:%s\n",serial,indbuf[i_title]);
  569.     fprintf(descf,"%d:%s\n",serial,indbuf[i_description]);
  570.     fprintf(authorf,"%d:%s\n",serial,indbuf[i_author]);
  571.     fprintf(versionf,"%d:%s\n",serial,indbuf[i_version]);
  572.  
  573. /*   add module's information in html page for robots  */
  574.     snprintf(buf,sizeof(buf),"%s",indbuf[i_description]);
  575.     for(pp=strchr(buf,','); pp; pp=strchr(pp,','))
  576.       string_modify(buf,pp,pp+1,"&#44;");
  577.     if(strcmp(module_language,lang[lind])==0)
  578.       fprintf(robotf,"%s ,%s,%s,%s,%s\n",name,module_language,name,
  579.           indbuf[i_title], buf);
  580.  
  581. /*   Normalize the information of trlist, using dictionary
  582.  *  -- bases/sys/domain.xx without suffix translation (--> english version)
  583.  */
  584.     entrycount=dentrycount; dicbuf=ddicbuf;
  585.     memmove(entry,dentry,dentrycount*sizeof(entry[0]));
  586.     unknown_type=unk_leave;
  587.     for(i=0;i<trcnt;i++) {
  588.     detag(indbuf[trlist[i]]);
  589.     deaccent(indbuf[trlist[i]]);
  590.     comma(indbuf[trlist[i]]);
  591.     singlespace(indbuf[trlist[i]]);
  592.     translate(indbuf[trlist[i]]);
  593.     }
  594. /*   Normalize the information, using dictionary
  595.  *   bases/sys/words.xx with suffix translation
  596.  */
  597.     entrycount=mentrycount; dicbuf=mdicbuf;
  598.     memmove(entry,mentry,mentrycount*sizeof(entry[0]));
  599.     unknown_type=unk_leave;/*  used in translator_.c */
  600.     for(i=0;i<trcnt;i++) {
  601.     suffix_translate(indbuf[trlist[i]]);
  602.     translate(indbuf[trlist[i]]);
  603.     }
  604.  
  605. /* taken contains all words already seen in the module index */
  606.     taken[0]=0; takenlen=tweight=0;
  607. /*  append words of title  */
  608.     ovlstrcpy(buf,indbuf[i_title]); towords(buf);
  609.     appenditem2(buf,lind,serial,4,module_language);
  610.  
  611. /*  extract words of every other information except level */
  612.     snprintf(buf,sizeof(buf),"%s %s %s %s %s %s %s %s %s %s %s %s %s %s %s",
  613.          indbuf[i_description],indbuf[i_keywords],
  614.          indbuf[i_keywords_ca],indbuf[i_keywords_en],indbuf[i_keywords_fr],
  615.          indbuf[i_keywords_it],indbuf[i_keywords_nl],
  616.          indbuf[i_title_ca],indbuf[i_title_en],indbuf[i_title_fr],
  617.          indbuf[i_title_it],indbuf[i_title_nl],
  618.          indbuf[i_domain],indbuf[i_require],indbuf[i_author]);
  619.     towords(buf);
  620.     appenditem2(buf,lind,serial,2,module_language);
  621.  
  622. /*   this time the dictionary is the group dictionary  sys/wgrp/wgrp
  623.  *   with a g (groupdic), not an m (maindic) . see below main, suffix, group.
  624.  *   and delete unknown ?? and translate
  625.  */
  626.     entrycount=gentrycount; dicbuf=gdicbuf;
  627.     memmove(entry,gentry,gentrycount*sizeof(entry[0]));
  628.  
  629. /*  append words of every title information  */
  630.     ovlstrcpy(buf,indbuf[i_title]);
  631.     unknown_type=unk_delete;
  632.     translate(buf);
  633.     appenditem1(buf,lind,serial,2,module_language);
  634.  
  635. /*  append words of information of description except level  */
  636.     snprintf(buf,sizeof(buf),"%s", indbuf[i_description]);
  637.     unknown_type=unk_delete;
  638.     translate(buf);
  639.     appenditem1(buf,lind,serial,4,module_language);
  640.  
  641. /*  append words (or group of words) of keywords and domain  */
  642.     snprintf(buf,sizeof(buf),"%s, %s, %s, %s, %s, %s, %s",
  643.          indbuf[i_domain],indbuf[i_keywords],
  644.          indbuf[i_keywords_ca], indbuf[i_keywords_en],indbuf[i_keywords_fr],
  645.          indbuf[i_keywords_it], indbuf[i_keywords_nl]);
  646.     unknown_type=unk_leave;
  647.     translate(buf);
  648.     appenditem1(buf,lind,serial,2,module_language);
  649.  
  650. /*   append level information, with weight 2 */
  651.     snprintf(buf,sizeof(buf),"%s",indbuf[i_level]);
  652.     ovlstrcpy(lbuf,"level");
  653.     for(p1=buf; *p1; p1++) if(!isalnum(*p1)) *p1=' ';
  654.     q=buf+strlen(buf);
  655.     for(p1=find_word_start(buf); (*p1) && (p1 < q) ;
  656.     p1=find_word_start(p2)) {
  657.     p2=find_word_end(p1);
  658.     if(p2!=NULL) *p2++=0; else p2=p1+strlen(p1);
  659.     if(!isalpha(*p1) ||
  660.        (!isdigit(*(p1+1)) && *(p1+1)!=0) ||
  661.        (*(p1+1)!=0 && *(p1+2)!=0))
  662.       continue;
  663.     *p1=tolower(*p1);
  664.     ovlstrcpy(lbuf+strlen("level"),p1);
  665.     appenditem(lbuf,lind,serial,2,module_language);
  666.     }
  667. /*   append total weight of module to weight file site2/weight.xx  */
  668.     fprintf(weightf,"%d:%d\n",serial,tweight);
  669. }
  670.  
  671. void modules(void)
  672. {
  673.     int i,j,k,d;
  674.     char namebuf[MAX_LINELEN+1];
  675.     char mdic[MAX_LINELEN+1], sdic[MAX_LINELEN+1], gdic[MAX_LINELEN+1], ddic[MAX_LINELEN+1];
  676.  
  677.     for(j=0;j<langcnt;j++) {
  678.     snprintf(namebuf,sizeof(namebuf),"%s/weight.%s",outdir,lang[j]);
  679.     weightf=fopen(namebuf,"w");
  680.     snprintf(mdic,sizeof(mdic),"%s/%s.%s",dicdir,maindic,lang[j]);
  681.     snprintf(sdic,sizeof(sdic),"%s/%s.%s",dicdir,suffixdic,lang[j]);
  682.     snprintf(gdic,sizeof(gdic),"%s/%s.%s",dicdir,groupdic,lang[j]);
  683.     snprintf(ddic,sizeof(ddic),"%s/%s.%s",dicdir,domaindic,lang[j]);
  684.     suffix_dic(sdic); prepare_dic(gdic);
  685.     gdicbuf=dicbuf; gentrycount=entrycount;
  686.     memmove(gentry,entry,gentrycount*sizeof(entry[0]));
  687.     prepare_dic(mdic);
  688.     mdicbuf=dicbuf; mentrycount=entrycount;
  689.     memmove(mentry,entry,mentrycount*sizeof(entry[0]));
  690.     prepare_dic(ddic);
  691.     ddicbuf=dicbuf; dentrycount=entrycount;
  692.     memmove(dentry,entry,dentrycount*sizeof(entry[0]));
  693.     unknown_type=unk_leave; translate(ignore[j]);
  694.     for(i=0;i<modcnt;i++) {
  695.         if(mod[i].langcnt>0) {
  696.         for(d=k=0;k<mod[i].langcnt;k++)
  697.           if(mod[i].langs[k]<mod[i].langs[d]) d=k;
  698.         for(k=0;k<mod[i].langcnt && mod[i].langs[k]!=j;k++);
  699.         if(k>=mod[i].langcnt) k=d;
  700.         snprintf(namebuf,MAX_LINELEN,"%s.%s",mod[i].name,
  701.              lang[mod[i].langs[k]]);
  702.         onemodule(namebuf,mod[i].counts[k],j);
  703.         }
  704.         else {
  705.         onemodule(mod[i].name,mod[i].counts[0],j);
  706.         }
  707.     }
  708.     if(mentrycount>0) free(mdicbuf);
  709.     if(gentrycount>0) free(gdicbuf);
  710.     if(suffixcnt>0) free(sufbuf);
  711.     if(dentrycount>0) free(ddicbuf);
  712.     if(weightf) fclose(weightf);
  713.     }
  714. }
  715.  
  716. /* FIXME ? differences with appenditem - use fprintf instead of  snprintf */
  717. void sappenditem(char *word, int lind, int serial, int weight)
  718. {
  719.     int ll;
  720.     char *p;
  721.  
  722.     if(!isalnum(*word) || (ll=strlen(word))<2 ||
  723.        wordchr(taken,word)!=NULL ||
  724.        wordchr(ignore[lind],word)!=NULL ||
  725.        takenlen>=MAX_LINELEN-ll-16)
  726.       return;
  727.     if(ll==2 && (!isdigit(word[0]) || !isalpha(word[1]))) return;
  728.     for(p=word;*p;p++) if(!isalnum(*p) && *p!=' ') return;
  729.     taken[takenlen++]=' ';taken[takenlen++]=' ';
  730.     ovlstrcpy(taken+takenlen,word);
  731.     takenlen+=ll; tweight+=weight;
  732.     fprintf(indf,"%s:%d?%d\n",word,serial,weight);
  733. }
  734.  
  735. void onesheet(int serial, int lind)
  736. {
  737.     int i;
  738.     unsigned char trlist[]={
  739.     s_title,s_description,s_domain,s_keywords,s_information
  740.     };
  741.     #define trcnt (sizeof(trlist)/sizeof(trlist[0]))
  742.     char *p1, *p2, buf[MAX_LINELEN+1];
  743.  
  744.     if(sheet_index(serial)) return;
  745.     fprintf(listf,"%s\n",mod[serial].name+3);
  746.     fprintf(titf,"%d:%s\n",serial,sindbuf[s_title]);
  747.     fprintf(descf,"%d:%s\n",serial,sindbuf[s_description]);
  748.     fprintf(remf,"%d:%s\n",serial,sindbuf[s_information]);
  749.  
  750.     entrycount=dentrycount; dicbuf=ddicbuf;
  751.     memmove(entry,dentry,dentrycount*sizeof(entry[0]));
  752.     unknown_type=unk_leave;
  753.     for(i=0;i<trcnt;i++) {
  754.     detag(sindbuf[trlist[i]]);
  755.     deaccent(sindbuf[trlist[i]]);
  756.     comma(sindbuf[trlist[i]]);
  757.     singlespace(sindbuf[trlist[i]]);
  758.     translate(sindbuf[trlist[i]]);
  759.     }
  760.  
  761.     entrycount=mentrycount; dicbuf=mdicbuf;
  762.     memmove(entry,mentry,mentrycount*sizeof(entry[0]));
  763.     unknown_type=unk_leave;
  764.     for(i=0;i<trcnt;i++) {
  765.     suffix_translate(sindbuf[trlist[i]]);
  766.     translate(sindbuf[trlist[i]]);
  767.     }
  768.     taken[0]=0; takenlen=tweight=0;
  769.     ovlstrcpy(buf,sindbuf[s_title]); towords(buf);
  770.     for(p1=find_word_start(buf);*p1;
  771.     p1=find_word_start(p2)) {
  772.     p2=find_word_end(p1); if(*p2) *p2++=0;
  773.     sappenditem(p1,lind,serial,4);
  774.     }
  775.     snprintf(buf,sizeof(buf),"%s %s %s %s",
  776.          sindbuf[s_description],sindbuf[s_keywords],
  777.          sindbuf[s_domain],sindbuf[s_information]);
  778.     towords(buf);
  779.     for(p1=find_word_start(buf);*p1;
  780.     p1=find_word_start(p2)) {
  781.     p2=find_word_end(p1); if(*p2) *p2++=0;
  782.     sappenditem(p1,lind,serial,2);
  783.     }
  784.     entrycount=gentrycount; dicbuf=gdicbuf;
  785.     memmove(entry,gentry,gentrycount*sizeof(entry[0]));
  786.     unknown_type=unk_delete;
  787.     ovlstrcpy(buf,sindbuf[s_title]); translate(buf);
  788.     for(p1=find_word_start(buf); *p1;
  789.     p1=find_word_start(p2)) {
  790.     p2=strchr(p1,',');
  791.     if(p2!=NULL) *p2++=0; else p2=p1+strlen(p1);
  792.     if(strlen(p1)<=0) continue;
  793.     sappenditem(p1,lind,serial,4);
  794.     }
  795.     snprintf(buf,sizeof(buf),"%s, %s, %s, %s",
  796.          sindbuf[s_description],sindbuf[s_keywords],
  797.          sindbuf[s_domain],sindbuf[s_information]);
  798.     translate(buf);
  799.     for(p1=find_word_start(buf); *p1;
  800.     p1=find_word_start(p2)) {
  801.     p2=strchr(p1,',');
  802.     if(p2!=NULL) *p2++=0; else p2=p1+strlen(p1);
  803.     if(strlen(p1)<=0) continue;
  804.     sappenditem(p1,lind,serial,2);
  805.     }
  806.     fprintf(weightf,"%d:%d\n",serial,tweight);
  807. }
  808.  
  809. void sheets(void)
  810. {
  811.     int i,j;
  812.     char mdic[MAX_LINELEN+1], sdic[MAX_LINELEN+1], gdic[MAX_LINELEN+1], ddic[MAX_LINELEN+1];
  813.     char buf[MAX_LINELEN+1];
  814.  
  815.     for(j=0;j<langcnt;j++) {
  816.     snprintf(buf,sizeof(buf),"%s/index/title.%s",sheetdir,lang[j]);
  817.     titf=fopen(buf,"w");
  818.     snprintf(buf,sizeof(buf),"%s/index/description.%s",sheetdir,lang[j]);
  819.     descf=fopen(buf,"w");
  820.     snprintf(buf,sizeof(buf),"%s/index/%s",sheetdir,lang[j]);
  821.     indf=fopen(buf,"w");
  822.     snprintf(buf,sizeof(buf),"%s/index/list.%s",sheetdir,lang[j]);
  823.     listf=fopen(buf,"w");
  824.     snprintf(buf,sizeof(buf),"%s/index/weight.%s",sheetdir,lang[j]);
  825.     weightf=fopen(buf,"w");
  826.     snprintf(buf,sizeof(buf),"%s/index/addr.%s",sheetdir,lang[j]);
  827.     addrf=fopen(buf,"w");
  828.     snprintf(buf,sizeof(buf),"%s/index/information.%s",sheetdir,lang[j]);
  829.     remf=fopen(buf,"w");
  830.     snprintf(buf,sizeof(buf),"%s/index/serial.%s",sheetdir,lang[j]);
  831.     serialf=fopen(buf,"w");
  832.     snprintf(mdic,sizeof(mdic),"%s/%s.%s",dicdir,maindic,lang[j]);
  833.     snprintf(sdic,sizeof(sdic),"%s/%s.%s",dicdir,suffixdic,lang[j]);
  834.     snprintf(gdic,sizeof(gdic),"%s/%s.%s",dicdir,groupdic,lang[j]);
  835.     snprintf(ddic,sizeof(ddic),"%s/%s.%s",dicdir,domaindic,lang[j]);
  836.     suffix_dic(sdic); prepare_dic(gdic);
  837.     gdicbuf=dicbuf; gentrycount=entrycount;
  838.     memmove(gentry,entry,gentrycount*sizeof(entry[0]));
  839.     prepare_dic(mdic);
  840.     mdicbuf=dicbuf; mentrycount=entrycount;
  841.     memmove(mentry,entry,mentrycount*sizeof(entry[0]));
  842.     prepare_dic(ddic);
  843.     ddicbuf=dicbuf; dentrycount=entrycount;
  844.     memmove(dentry,entry,dentrycount*sizeof(entry[0]));
  845.     unknown_type=unk_leave; translate(ignore[j]);
  846.     for(i=0;i<modcnt;i++) {
  847.         if(mod[i].langs[0]!=j) continue;
  848.         fprintf(addrf,"%d:%s\n",i,mod[i].name+3);
  849.         fprintf(serialf,"%s:%d\n",mod[i].name+3,i);
  850.         onesheet(i,j);
  851.     }
  852.     if(mentrycount>0) free(mdicbuf);
  853.     if(gentrycount>0) free(gdicbuf);
  854.     if(suffixcnt>0) free(sufbuf);
  855.     if(dentrycount>0) free(ddicbuf);
  856.     fclose(titf); fclose(descf); fclose(indf); fclose(listf);
  857.     fclose(weightf); fclose(addrf); fclose(serialf);
  858.     }
  859. }
  860.  
  861. int main()
  862. {
  863.     prep();
  864.     if(modcnt>0) modules();
  865.     clean();
  866.     sprep();
  867.     if(modcnt>0) sheets();
  868.     return 0;
  869. }
  870.  
  871.