]> git.lizzy.rs Git - plan9front.git/blob - sys/src/cmd/mothra/rdhtml.c
merge
[plan9front.git] / sys / src / cmd / mothra / rdhtml.c
1 #include <u.h>
2 #include <libc.h>
3 #include <draw.h>
4 #include <event.h>
5 #include <panel.h>
6 #include "mothra.h"
7 #include "html.h"
8 #include "rtext.h"
9
10 typedef struct Fontdata Fontdata;
11 struct Fontdata{
12         char *name;
13         Font *font;
14         int space;
15 }fontlist[4][4]={
16         "lucidasans/unicode.7", 0, 0,
17         "lucidasans/unicode.8", 0, 0,
18         "lucidasans/unicode.10", 0, 0,
19         "lucidasans/unicode.13", 0, 0,
20
21         "lucidasans/italicunicode.7", 0, 0,
22         "lucidasans/italicunicode.8", 0, 0,
23         "lucidasans/italicunicode.10", 0, 0,
24         "lucidasans/italicunicode.13", 0, 0,
25
26         "lucidasans/boldunicode.7", 0, 0,
27         "lucidasans/boldunicode.8", 0, 0,
28         "lucidasans/boldunicode.10", 0, 0,
29         "lucidasans/boldunicode.13", 0, 0,
30
31         "lucidasans/typeunicode.7", 0, 0,
32         "pelm/unicode.8", 0, 0,
33         "lucidasans/typeunicode.12", 0, 0,
34         "lucidasans/typeunicode.16", 0, 0,
35 };
36 Fontdata *pl_whichfont(int f, int s){
37         char name[NNAME];
38
39         assert(f >= 0 && f < 4);
40         assert(s >= 0 && s < 4);
41
42         if(fontlist[f][s].font==0){
43                 snprint(name, sizeof(name), "/lib/font/bit/%s.font", fontlist[f][s].name);
44                 fontlist[f][s].font=openfont(display, name);
45                 if(fontlist[f][s].font==0) fontlist[f][s].font=font;
46                 fontlist[f][s].space=stringwidth(fontlist[f][s].font, "0");
47         }
48         return &fontlist[f][s];
49         
50 }
51 void getfonts(void){
52         int f, s;
53         for(f=0;f!=4;f++)
54                 for(s=0;s!=4;s++)
55                         pl_whichfont(f, s);
56 }
57 void pl_pushstate(Hglob *g, int t){
58         ++g->state;
59         if(g->state==&g->stack[NSTACK]){
60                 htmlerror(g->name, g->lineno, "stack overflow at <%s>", tag[t].name);
61                 --g->state;
62         }
63         g->state[0]=g->state[-1];
64         g->state->tag=t;
65 }
66 void pl_linespace(Hglob *g){
67         plrtbitmap(&g->dst->text, 1000000, 0, linespace, 0, 0);
68         g->para=0;
69         g->linebrk=0;
70 }
71 enum{
72         HORIZ,
73         VERT,
74 };
75 int strtolength(Hglob *g, int dir, char *str)
76 {
77         double f;
78
79         f = atof(str);
80         if(cistrstr(str, "px"))
81                 return floor(f);
82         if(cistrstr(str, "%"))
83                 return floor(f*((dir==HORIZ) ? Dx(g->dst->text->r) : Dy(g->dst->text->r))/100);
84         if(cistrstr(str, "em")){
85                 Point z;
86                 z = stringsize(g->dst->text->font, "M");
87                 return floor(f*((dir==HORIZ) ? z.x : z.y));
88         }
89         return floor(f);
90 }
91
92 void pl_htmloutput(Hglob *g, int nsp, char *s, Field *field){
93         Fontdata *f;
94         int space, indent;
95         Action *ap;
96         if(g->state->tag==Tag_title
97 /*      || g->state->tag==Tag_textarea */
98         || g->state->tag==Tag_select){
99                 if(s){
100                         if(g->tp!=g->text && g->tp!=g->etext && g->tp[-1]!=' ')
101                                 *g->tp++=' ';
102                         while(g->tp!=g->etext && *s) *g->tp++=*s++;
103                         if(g->state->tag==Tag_title) g->dst->changed=1;
104                         *g->tp='\0';
105                 }
106                 return;
107         }
108         f=pl_whichfont(g->state->font, g->state->size);
109         space=f->space;
110         indent=g->state->margin;
111         if(g->para){
112                 space=1000000;
113                 indent+=g->state->indent;
114         }
115         else if(g->linebrk)
116                 space=1000000;
117         else if(nsp<=0)
118                 space=0;
119         if(g->state->image[0]==0 && g->state->link[0]==0 && g->state->name[0]==0 && field==0)
120                 ap=0;
121         else{
122                 ap=mallocz(sizeof(Action), 1);
123                 if(ap!=0){
124                         if(g->state->image[0])
125                                 ap->image = strdup(g->state->image);
126                         if(g->state->link[0])
127                                 ap->link = strdup(g->state->link);
128                         if(g->state->name[0])
129                                 ap->name = strdup(g->state->name);
130                         ap->ismap=g->state->ismap;
131                         ap->width=g->state->width;
132                         ap->height=g->state->height;
133                         ap->field=field;
134                 }
135         }
136         if(space<0) space=0;
137         if(indent<0) indent=0;
138         if(g->state->pre && s[0]=='\t'){
139                 space=0;
140                 while(s[0]=='\t'){
141                         space++;
142                         s++;
143                 }
144                 space=PL_TAB|space;
145                 if(g->linebrk){
146                         indent=space;
147                         space=1000000;
148                 }
149         }
150         plrtstr(&g->dst->text, space, indent, f->font, strdup(s),
151                 g->state->link[0] || g->state->image[0], ap);
152         g->para=0;
153         g->linebrk=0;
154         g->dst->changed=1;
155 }
156
157 /*
158  * Buffered read, no translation
159  * Save in cache.
160  */
161 int pl_bread(Hglob *g){
162         int n, c;
163         char err[1024];
164         if(g->hbufp==g->ehbuf){
165                 n=read(g->hfd, g->hbuf, NHBUF);
166                 if(n<=0){
167                         if(n<0){
168                                 snprint(err, sizeof(err), "%r reading %s", g->name);
169                                 pl_htmloutput(g, 1, err, 0);
170                         }
171                         g->heof=1;
172                         return EOF;
173                 }
174                 g->hbufp=g->hbuf;
175                 g->ehbuf=g->hbuf+n;
176         }
177         c=*g->hbufp++&255;
178         if(c=='\n') g->lineno++;
179         return c;
180 }
181 /*
182  * Read a character, translating \r\n, \n\r, \r and \n into \n
183  */
184 int pl_readc(Hglob *g){
185         int c;
186         static int peek=-1;
187         if(peek!=-1){
188                 c=peek;
189                 peek=-1;
190         }
191         else
192                 c=pl_bread(g);
193         if(c=='\r'){
194                 c=pl_bread(g);
195                 if(c!='\n') peek=c;
196                 return '\n';
197         }
198         if(c=='\n'){
199                 c=pl_bread(g);
200                 if(c!='\r') peek=c;
201                 return '\n';
202         }
203         return c;
204 }
205 void pl_putback(Hglob *g, int c){
206         if(g->npeekc==NPEEKC) htmlerror(g->name, g->lineno, "too much putback!");
207         else if(c!=EOF) g->peekc[g->npeekc++]=c;
208 }
209 int pl_nextc(Hglob *g){
210         int c;
211         int n;
212         Rune r;
213         char crune[4];
214         if(g->heof) return EOF;
215         if(g->npeekc!=0) return g->peekc[--g->npeekc];
216         c=pl_readc(g);
217         if(c=='<'){
218                 c=pl_readc(g);
219                 if(c=='/'){
220                         c=pl_readc(g);
221                         pl_putback(g, c);
222                         pl_putback(g, '/');
223                         if('a'<=c && c<='z' || 'A'<=c && c<='Z') return STAG;
224                         return '<';
225                 }
226                 pl_putback(g, c);
227                 if(c=='!' || 'a'<=c && c<='z' || 'A'<=c && c<='Z' || c=='?') return STAG;
228                 return '<';
229         }
230         if(c=='>') return ETAG;
231         if(c==EOF) return c;
232         n=0;
233         for (;;){
234                 crune[n++]=c;
235                 if(fullrune(crune, n)){
236                         chartorune(&r, crune);
237                         return r;
238                 }
239                 c=pl_readc(g);
240                 if(c==EOF)
241                         return EOF;
242         }
243         return c;
244 }
245 int entchar(int c){
246         return c=='#' || 'a'<=c && c<='z' || 'A'<=c && c<='Z' || '0'<=c && c<='9';
247 }
248 /*
249  * remove entity references, in place.
250  * Potential bug:
251  *      This doesn't work if removing an entity reference can lengthen the string!
252  *      Fortunately, this doesn't happen.
253  */
254 void pl_rmentities(Hglob *g, char *s){
255         char *t, *u, c, svc;
256         Entity *ep;
257         Rune r;
258         t=s;
259         do{
260                 c=*s++;
261                 if(c=='&'
262                 && ((*s=='#' && strchr("0123456789Xx", s[1]))
263                   || 'a'<=*s && *s<='z'
264                   || 'A'<=*s && *s<='Z')){
265                         u=s;
266                         while(entchar(*s)) s++;
267                         svc=*s;
268                         *s = 0;
269                         if(svc==';') s++;
270                         if(strcmp(u, "lt") == 0)
271                                 *t++='<';
272                         else if(strcmp(u, "gt") == 0)
273                                 *t++='>';
274                         else if(strcmp(u, "quot") == 0)
275                                 *t++='"';
276                         else if(strcmp(u, "apos") == 0)
277                                 *t++='\'';
278                         else if(strcmp(u, "amp") == 0)
279                                 *t++='&';
280                         else {
281                                 if(svc==';') s--;
282                                 *s=svc;
283                                 *t++='&';
284                                 while(u<s)
285                                         *t++=*u++;
286                         }
287                 }       
288                 else *t++=c;
289         }while(c);
290 }
291 /*
292  * Skip over white space
293  */
294 char *pl_white(char *s){
295         while(*s==' ' || *s=='\t' || *s=='\n' || *s=='\r') s++;
296         return s;
297 }
298 /*
299  * Skip over HTML word
300  */
301 char *pl_word(char *s){
302         if ('a'<=*s && *s<='z' || 'A'<=*s && *s<='Z') {
303                 s++;
304                 while('a'<=*s && *s<='z' || 'A'<=*s && *s<='Z' || '0'<=*s && *s<='9' || *s=='-' || *s=='.') s++;
305         }
306         return s;
307 }
308 /*
309  * Skip to matching quote
310  */
311 char *pl_quote(char *s){
312         char q;
313         q=*s++;
314         while(*s!=q && *s!='\0') s++;
315         return s;
316 }
317 void pl_dnl(char *s){
318         char *t;
319         for(t=s;*s;s++) if(*s!='\r' && *s!='\n') *t++=*s;
320         *t='\0';
321 }
322 void pl_tagparse(Hglob *g, char *str){
323         char *s, *t, *name, c;
324         Pair *ap;
325         Tag *tagp;
326         g->tag=Tag_end;
327         ap=g->attr;
328         if(str[0]=='!'){        /* test should be strncmp(str, "!--", 3)==0 */
329                 g->tag=Tag_comment;
330                 ap->name=0;
331                 return;
332         }
333         if(str[0]=='/') str++;
334         name=str;
335         s=pl_word(str);
336         if(*s!=' ' && *s!='\n' && *s!='\t' && *s!='\0'){
337                 htmlerror(g->name, g->lineno, "bad tag name in %s", str);
338                 ap->name=0;
339                 return;
340         }
341         if(*s!='\0') *s++='\0';
342         for(t=name;t!=s;t++) if('A'<=*t && *t<='Z') *t+='a'-'A';
343         /*
344          * Binary search would be faster here
345          */
346         for(tagp=tag;tagp->name;tagp++) if(strcmp(name, tagp->name)==0) break;
347         g->tag=tagp-tag;
348         if(g->tag==Tag_end) htmlerror(g->name, g->lineno, "no tag %s", name);
349         for(;;){
350                 s=pl_white(s);
351                 if(*s=='\0'){
352                         ap->name=0;
353                         return;
354                 }
355                 ap->name=s;
356                 s=pl_word(s);
357                 t=pl_white(s);
358                 c=*t;
359                 *s='\0';
360                 for(s=ap->name;*s;s++) if('A'<=*s && *s<='Z') *s+='a'-'A';
361                 if(c=='='){
362                         s=pl_white(t+1);
363                         if(*s=='\'' || *s=='"'){
364                                 ap->value=s+1;
365                                 s=pl_quote(s);
366                                 if(*s=='\0'){
367                                         htmlerror(g->name, g->lineno,
368                                                 "No terminating quote in rhs of attribute %s",
369                                                 ap->name);
370                                         ap->name=0;
371                                         return;
372                                 }
373                                 *s++='\0';
374                                 pl_dnl(ap->value);
375                         }
376                         else{
377                                 /* read up to white space or > */
378                                 ap->value=s;
379                                 while(*s!=' ' && *s!='\t' && *s!='\n' && *s!='\0') s++;
380                                 if(*s!='\0') *s++='\0';
381                         }
382                         pl_rmentities(g, ap->value);
383                 }
384                 else{
385                         if(c!='\0') s++;
386                         ap->value="";
387                 }
388                 if(ap==&g->attr[NATTR-1])
389                         htmlerror(g->name, g->lineno, "too many attributes!");
390                 else ap++;
391         }
392 }
393 int pl_getcomment(Hglob *g){
394         int c;
395         if((c=pl_nextc(g))=='-' && (c=pl_nextc(g))=='-'){
396                 /* <!-- eats everything until --> or EOF */
397                 for(;;){
398                         while((c=pl_nextc(g))!='-' && c!=EOF)
399                                 ;
400                         if(c==EOF)
401                                 break;
402                         if((c=pl_nextc(g))=='-'){
403                                 while((c=pl_nextc(g))=='-')
404                                         ;
405                                 if(c==ETAG || c==EOF)
406                                         break;
407                         }
408                 }
409         } else {
410                 /* <! eats everything until > or EOF */
411                 while(c!=ETAG && c!=EOF)
412                         c=pl_nextc(g);
413         }
414         if(c==EOF)
415                 htmlerror(g->name, g->lineno, "EOF in comment");
416         g->tag=Tag_comment;
417         g->attr->name=0;
418         g->token[0]='\0';
419         return TAG;
420 }
421 int lrunetochar(char *p, int v)
422 {
423         Rune r;
424
425         r=v;
426         return runetochar(p, &r);
427 }
428
429 /*
430  * Read a start or end tag -- the caller has read the initial <
431  */
432 int pl_gettag(Hglob *g){
433         char *tokp;
434         int c;
435         tokp=g->token;
436         if((c=pl_nextc(g))=='!' || c=='?')
437                 return pl_getcomment(g);
438         pl_putback(g, c);
439         while((c=pl_nextc(g))!=ETAG && c!=EOF)
440                 if(tokp!=&g->token[NTOKEN-3]) tokp += lrunetochar(tokp, c);
441         *tokp='\0';
442         if(c==EOF) htmlerror(g->name, g->lineno, "EOF in tag");
443         pl_tagparse(g, g->token);
444         if(g->token[0]!='/') return TAG;
445         if(g->attr[0].name!=0)
446                 htmlerror(g->name, g->lineno, "end tag should not have attributes");
447         return ENDTAG;
448 }
449 /*
450  * The next token is a tag, an end tag or a sequence of
451  * non-white characters.
452  * If inside <pre>, newlines are converted to <br> and spaces are preserved.
453  * Otherwise, spaces and newlines are noted and discarded.
454  */
455 int pl_gettoken(Hglob *g){
456         char *tokp;
457         int c;
458         if(g->state->pre) switch(c=pl_nextc(g)){
459         case STAG: return pl_gettag(g);
460         case EOF: return EOF;
461         case '\n':
462                 pl_tagparse(g, "br");
463                 return TAG;
464         default:
465                 tokp=g->token;
466                 while(c=='\t'){
467                         if(tokp!=&g->token[NTOKEN-3]) tokp += lrunetochar(tokp, c);
468                         c=pl_nextc(g);
469                 }
470                 while(c!='\t' && c!='\n' && c!=STAG && c!=EOF){
471                         if(c==ETAG) c='>';
472                         if(tokp!=&g->token[NTOKEN-3]) tokp += lrunetochar(tokp, c);
473                         c=pl_nextc(g);
474                 }
475                 *tokp='\0';
476                 pl_rmentities(g, g->token);
477                 pl_putback(g, c);
478                 g->nsp=0;
479                 g->spacc=0;
480                 return TEXT;
481         }
482         while((c=pl_nextc(g))==' ' || c=='\t' || c=='\n')
483                 if(g->spacc!=-1)
484                         g->spacc++;
485         switch(c){
486         case STAG: return pl_gettag(g);
487         case EOF: return EOF;
488         default:
489                 tokp=g->token;
490                 do{
491                         if(c==ETAG) c='>';
492                         if(tokp!=&g->token[NTOKEN-3]) tokp += lrunetochar(tokp, c);
493                         c=pl_nextc(g);
494                 }while(c!=' ' && c!='\t' && c!='\n' && c!=STAG && c!=EOF);
495                 *tokp='\0';
496                 pl_rmentities(g, g->token);
497                 pl_putback(g, c);
498                 g->nsp=g->spacc;
499                 g->spacc=0;
500                 return TEXT;
501         }
502 }
503 char *pl_getattr(Pair *attr, char *name){
504         for(;attr->name;attr++)
505                 if(strcmp(attr->name, name)==0)
506                         return attr->value;
507         return 0;
508 }
509 int pl_hasattr(Pair *attr, char *name){
510         for(;attr->name;attr++)
511                 if(strcmp(attr->name, name)==0)
512                         return 1;
513         return 0;
514 }
515 void plaintext(Hglob *g){
516         char line[NLINE];
517         char *lp, *elp;
518         int c;
519         g->state->font=CWIDTH;
520         g->state->size=NORMAL;
521         elp=&line[NLINE+1];
522         lp=line;
523         for(;;){
524                 c=pl_readc(g);
525                 if(c==EOF) break;
526                 if(c=='\n' || lp==elp){
527                         *lp='\0';
528                         g->linebrk=1;
529                         pl_htmloutput(g, 0, line, 0);
530                         lp=line;
531                 }
532                 if(c=='\t'){
533                         do *lp++=' '; while(lp!=elp && utfnlen(line, lp-line)%8!=0);
534                 }
535                 else if(c!='\n')
536                         lp += lrunetochar(lp, c);
537         }
538         if(lp!=line){
539                 *lp='\0';
540                 g->linebrk=1;
541                 pl_htmloutput(g, 0, line, 0);
542         }
543 }
544 void plrdplain(char *name, int fd, Www *dst){
545         Hglob g;
546         g.state=g.stack;
547         g.state->tag=Tag_html;
548         g.state->font=CWIDTH;
549         g.state->size=NORMAL;
550         g.state->pre=0;
551         g.state->image[0]=0;
552         g.state->link[0]=0;
553         g.state->name[0]=0;
554         g.state->margin=0;
555         g.state->indent=20;
556         g.state->ismap=0;
557         g.state->table=0;
558         g.dst=dst;
559         g.hfd=fd;
560         g.name=name;
561         g.ehbuf=g.hbufp=g.hbuf;
562         g.npeekc=0;
563         g.heof=0;
564         g.lineno=1;
565         g.linebrk=1;
566         g.para=0;
567         g.text=dst->title;
568         g.tp=g.text;
569         g.etext=g.text+NTITLE-1;
570         g.spacc=0;
571         g.form=0;
572         strncpy(g.text, name, NTITLE);
573         plaintext(&g);
574         dst->finished=1;
575 }
576 void plrdhtml(char *name, int fd, Www *dst){
577         Stack *sp;
578         char buf[20];
579         char *str;
580         Hglob g;
581         int t;
582         int tagerr;
583         g.state=g.stack;
584         g.state->tag=Tag_html;
585         g.state->font=ROMAN;
586         g.state->size=NORMAL;
587         g.state->pre=0;
588         g.state->image[0]=0;
589         g.state->link[0]=0;
590         g.state->name[0]=0;
591         g.state->margin=0;
592         g.state->indent=25;
593         g.state->ismap=0;
594         g.state->width=0;
595         g.state->height=0;
596         g.state->table=0;
597         g.dst=dst;
598         g.hfd=fd;
599         g.name=name;
600         g.ehbuf=g.hbufp=g.hbuf;
601         g.npeekc=0;
602         g.heof=0;
603         g.lineno=1;
604         g.linebrk=1;
605         g.para=0;
606         g.text=dst->title;
607         g.tp=g.text;
608         g.etext=g.text+NTITLE-1;
609         dst->title[0]='\0';
610         g.spacc=0;
611         g.form=0;
612
613         for(;;) switch(pl_gettoken(&g)){
614         case TAG:
615                 switch(tag[g.tag].action){
616                 case OPTEND:
617                         for(sp=g.state;sp!=g.stack && sp->tag!=g.tag;--sp);
618                         if(sp->tag!=g.tag)
619                                 pl_pushstate(&g, g.tag);
620                         else
621                                 for(;g.state!=sp;--g.state)
622                                         if(tag[g.state->tag].action!=OPTEND)
623                                                 htmlerror(g.name, g.lineno,
624                                                         "end tag </%s> missing",
625                                                         tag[g.state->tag].name);
626                         break;
627                 case END:
628                         pl_pushstate(&g, g.tag);
629                         break;
630                 }
631                 switch(g.tag){
632                 default:
633                         htmlerror(g.name, g.lineno,
634                                 "unimplemented tag <%s>", tag[g.tag].name);
635                         break;
636                 case Tag_end:   /* unrecognized start tag */
637                         break;
638                 case Tag_meta:
639                         break;
640                 case Tag_img:
641                         if(str=pl_getattr(g.attr, "src"))
642                                 strncpy(g.state->image, str, sizeof(g.state->image));
643                         g.state->ismap=pl_hasattr(g.attr, "ismap");
644                         if(str=pl_getattr(g.attr, "width"))
645                                 g.state->width = strtolength(&g, HORIZ, str);
646                         if(str=pl_getattr(g.attr, "height"))
647                                 g.state->height = strtolength(&g, VERT, str);
648                         str=pl_getattr(g.attr, "alt");
649                         if(str==0){
650                                 if(g.state->image[0])
651                                         str=g.state->image;
652                                 else
653                                         str="[[image]]";
654                         }
655                         pl_htmloutput(&g, 0, str, 0);
656                         g.state->image[0]=0;
657                         g.state->ismap=0;
658                         g.state->width=0;
659                         g.state->height=0;
660                         break;
661                 case Tag_plaintext:
662                         g.spacc=0;
663                         plaintext(&g);
664                         break;
665                 case Tag_comment:
666                 case Tag_html:
667                 case Tag_link:
668                 case Tag_nextid:
669                         break;
670                 case Tag_table:
671                         g.state->table++;
672                         break;
673                 case Tag_tr:
674                         if(g.state->table==1){
675                                 g.spacc=0;
676                                 g.linebrk=1;
677                         } else
678                                 g.spacc++;
679                         break;
680                 case Tag_td:
681                         g.spacc++;
682                         break;
683                 case Tag_a:
684                         if(str=pl_getattr(g.attr, "href"))
685                                 strncpy(g.state->link, str, sizeof(g.state->link));
686                         if(str=pl_getattr(g.attr, "name")){
687                                 strncpy(g.state->name, str, sizeof(g.state->name));
688                                 pl_htmloutput(&g, 0, "", 0);
689                         }
690                         break;
691                 case Tag_frame:
692                         pl_htmloutput(&g, 0, "FRAME: ", 0);
693                         if(str=pl_getattr(g.attr, "src"))
694                                 strncpy(g.state->link, str, sizeof(g.state->link));
695                         if(str=pl_getattr(g.attr, "name"))
696                                 strncpy(g.state->name, str, sizeof(g.state->name));
697                         else
698                                 str = g.state->link;
699                         pl_htmloutput(&g, 0, str, 0);
700                         g.state->link[0]=0;
701                         g.state->name[0] =0;
702                         g.spacc=0;
703                         g.linebrk=1;
704                         break;
705                 case Tag_address:
706                         g.spacc=0;
707                         g.linebrk=1;
708                         g.state->font=ROMAN;
709                         g.state->size=NORMAL;
710                         g.state->margin=300;
711                         g.state->indent=50;
712                         break;
713                 case Tag_b:
714                 case Tag_strong:
715                         g.state->font=BOLD;
716                         break;
717                 case Tag_blockquot:
718                         g.spacc=0;
719                         g.linebrk=1;
720                         g.state->margin+=50;
721                         g.state->indent=20;
722                         break;
723                 case Tag_body:
724                         break;
725                 case Tag_head:
726                         g.state->font=ROMAN;
727                         g.state->size=NORMAL;
728                         g.state->margin=0;
729                         g.state->indent=20;
730                         g.spacc=0;
731                         break;
732                 case Tag_br:
733                         g.spacc=0;
734                         g.linebrk=1;
735                         break;
736                 case Tag_center:
737                         /* more to come */
738                         break;
739                 case Tag_cite:
740                         g.state->font=ITALIC;
741                         g.state->size=NORMAL;
742                         break;
743                 case Tag_code:
744                         g.state->font=CWIDTH;
745                         g.state->size=NORMAL;
746                         break;
747                 case Tag_dd:
748                         g.linebrk=1;
749                         g.state->indent=0;
750                         g.state->font=ROMAN;
751                         g.spacc=0;
752                         break;
753                 case Tag_dfn:
754                         htmlerror(g.name, g.lineno, "<dfn> deprecated");
755                         g.state->font=BOLD;
756                         g.state->size=NORMAL;
757                         break;
758                 case Tag_dl:
759                         g.state->font=BOLD;
760                         g.state->size=NORMAL;
761                         g.state->margin+=40;
762                         g.spacc=0;
763                         break;
764                 case Tag_dt:
765                         g.para=1;
766                         g.state->indent=-40;
767                         g.state->font=BOLD;
768                         g.spacc=0;
769                         break;
770                 case Tag_font:
771                         /* more to come */
772                         break;
773                 case Tag_u:
774                         htmlerror(g.name, g.lineno, "<u> deprecated");
775                 case Tag_em:
776                 case Tag_i:
777                 case Tag_var:
778                         g.state->font=ITALIC;
779                         break;
780                 case Tag_h1:
781                         g.linebrk=1;
782                         g.state->font=BOLD;
783                         g.state->size=ENORMOUS;
784                         g.state->margin+=100;
785                         g.spacc=0;
786                         break;
787                 case Tag_h2:
788                         pl_linespace(&g);
789                         g.state->font=BOLD;
790                         g.state->size=ENORMOUS;
791                         g.spacc=0;
792                         break;
793                 case Tag_h3:
794                         g.linebrk=1;
795                         pl_linespace(&g);
796                         g.state->font=ITALIC;
797                         g.state->size=ENORMOUS;
798                         g.state->margin+=20;
799                         g.spacc=0;
800                         break;
801                 case Tag_h4:
802                         pl_linespace(&g);
803                         g.state->font=BOLD;
804                         g.state->size=LARGE;
805                         g.state->margin+=10;
806                         g.spacc=0;
807                         break;
808                 case Tag_h5:
809                         pl_linespace(&g);
810                         g.state->font=ITALIC;
811                         g.state->size=LARGE;
812                         g.state->margin+=10;
813                         g.spacc=0;
814                         break;
815                 case Tag_h6:
816                         pl_linespace(&g);
817                         g.state->font=BOLD;
818                         g.state->size=LARGE;
819                         g.spacc=0;
820                         break;
821                 case Tag_hr:
822                         g.spacc=0;
823                         plrtbitmap(&g.dst->text, 1000000, g.state->margin, hrule, 0, 0);
824                         break;
825                 case Tag_key:
826                         htmlerror(g.name, g.lineno, "<key> deprecated");
827                 case Tag_kbd:
828                         g.state->font=CWIDTH;
829                         break;
830                 case Tag_dir:
831                 case Tag_menu:
832                 case Tag_ol:
833                 case Tag_ul:
834                         g.state->number=0;
835                         g.linebrk=1;
836                         g.state->margin+=25;
837                         g.state->indent=-25;
838                         g.spacc=0;
839                         break;
840                 case Tag_li:
841                         g.spacc=0;
842                         switch(g.state->tag){
843                         default:
844                                 htmlerror(g.name, g.lineno, "can't have <li> in <%s>",
845                                         tag[g.state->tag].name);
846                         case Tag_dir:   /* supposed to be multi-columns, can't do! */
847                         case Tag_menu:
848                                 g.linebrk=1;
849                                 break;
850                         case Tag_ol:
851                                 g.para=1;
852                                 snprint(buf, sizeof(buf), "%2d  ", ++g.state->number);
853                                 pl_htmloutput(&g, 0, buf, 0);
854                                 break;
855                         case Tag_ul:
856                                 g.para=0;
857                                 g.linebrk=0;
858                                 g.spacc=-1;
859                                 plrtbitmap(&g.dst->text, 100000,
860                                         g.state->margin+g.state->indent, bullet, 0, 0);
861                                 break;
862                         }
863                         break;
864                 case Tag_p:
865                         pl_linespace(&g);
866                         g.linebrk=1;
867                         g.spacc=0;
868                         break;
869                 case Tag_listing:
870                 case Tag_xmp:
871                         htmlerror(g.name, g.lineno, "<%s> deprecated", tag[g.tag].name);
872                 case Tag_pre:
873                 case Tag_samp:
874                         g.state->indent=0;
875                         g.state->pre=1;
876                         g.state->font=CWIDTH;
877                         g.state->size=NORMAL;
878                         pl_linespace(&g);
879                         break;
880                 case Tag_tt:
881                         g.state->font=CWIDTH;
882                         g.state->size=NORMAL;
883                         break;
884                 case Tag_title:
885                         g.text=dst->title+strlen(dst->title);
886                         g.tp=g.text;
887                         g.etext=dst->title+NTITLE-1;
888                         break;
889                 case Tag_form:
890                 case Tag_input:
891                 case Tag_button:
892                 case Tag_select:
893                 case Tag_option:
894                 case Tag_textarea:
895                 case Tag_isindex:
896                         rdform(&g);
897                         break;
898                 case Tag_script:
899                 case Tag_style:
900                         /*
901                          * ignore the content of these tags, eat tokens until we
902                          * reach a matching endtag.
903                          */
904                         t = g.tag;
905                         for(;;){
906                                 switch(pl_gettoken(&g)){
907                                 default:
908                                         continue;
909                                 case ENDTAG:
910                                         if(g.tag != t)
911                                                 continue;
912                                 case EOF:
913                                         break;
914                                 }
915                                 break;
916                         }
917                         break;
918                 }
919                 break;
920
921         case ENDTAG:
922                 /*
923                  * If the end tag doesn't match the top, we try to uncover a match
924                  * on the stack.
925                  */
926                 if(g.state->tag!=g.tag){
927                         tagerr=0;
928                         for(sp=g.state;sp!=g.stack;--sp){
929                                 if(sp->tag==g.tag)
930                                         break;
931                                 if(tag[g.state->tag].action!=OPTEND) tagerr++;
932                         }
933                         if(sp==g.stack){
934                                 if(tagerr)
935                                         htmlerror(g.name, g.lineno,
936                                                 "end tag mismatch <%s>...</%s>, ignored",
937                                                 tag[g.state->tag].name, tag[g.tag].name);
938                         }
939                         else{
940                                 if(tagerr)
941                                         htmlerror(g.name, g.lineno,
942                                                 "end tag mismatch <%s>...</%s>, "
943                                                 "intervening tags popped",
944                                                 tag[g.state->tag].name, tag[g.tag].name);
945                                 g.state=sp-1;
946                         }
947                 }
948                 else if(g.state==g.stack)
949                         htmlerror(g.name, g.lineno, "end tag </%s> at stack bottom",
950                                 tag[g.tag].name);
951                 else
952                         --g.state;
953                 switch(g.tag){
954                 case Tag_select:
955                 case Tag_form:
956                 case Tag_textarea:
957                         endform(&g);
958                         break;
959                 case Tag_h1:
960                 case Tag_h2:
961                 case Tag_h3:
962                 case Tag_h4:
963                         pl_linespace(&g);
964                         break;
965                 case Tag_address:
966                 case Tag_blockquot:
967                 case Tag_body:
968                 case Tag_dir:
969                 case Tag_dl:
970                 case Tag_dt:
971                 case Tag_h5:
972                 case Tag_h6:
973                 case Tag_listing:
974                 case Tag_menu:
975                 case Tag_ol:
976                 case Tag_samp:
977                 case Tag_title:
978                 case Tag_ul:
979                 case Tag_xmp:
980                         g.linebrk=1;
981                         break;
982                 case Tag_table:
983                         if(g.state->table==0)
984                                 g.linebrk=1;
985                         break;
986                 case Tag_pre:
987                         pl_linespace(&g);
988                         break;
989                 }
990                 break;
991         case TEXT:
992                 pl_htmloutput(&g, g.nsp, g.token, 0);
993                 break;
994         case EOF:
995                 for(;g.state!=g.stack;--g.state)
996                         if(tag[g.state->tag].action!=OPTEND)
997                                 htmlerror(g.name, g.lineno,
998                                         "missing </%s> at EOF", tag[g.state->tag].name);
999                 *g.tp='\0';
1000                 dst->changed=1;
1001                 getpix(dst->text, dst);
1002                 dst->finished=1;
1003                 return;
1004         }
1005 }