sys/src/cmd/aux/wikifmt.c

   1 /*
   2  * google code wiki to html converter.
   3  * https://code.google.com/p/support/wiki/WikiSyntax
   4  */
   5 #include <u.h>
   6 #include <libc.h>
   7
   8 enum {
   9         HUNK = 8*1024,
  10 };
  11
  12 char    *buf;
  13 char    *pos;
  14 char    *epos;
  15 char    *opos;
  16
  17 int     inquote = 0;
  18 int     intable = 0;
  19 int     inlist = 0;
  20 int     indent = -1;
  21
  22 void    body(void);
  23
  24 int
  25 match(char *s)
  26 {
  27         int n;
  28
  29         n = strlen(s);
  30         if(pos+n > epos)
  31                 return 0;
  32         return cistrncmp(pos, s, n) == 0;
  33 }
  34
  35 int
  36 got(char *s)
  37 {
  38         if(!match(s))
  39                 return 0;
  40         pos += strlen(s);
  41         return 1;
  42 }
  43
  44 char*
  45 look(char *s, char *e)
  46 {
  47         char *p;
  48         int n;
  49
  50         if(e == nil)
  51                 e = epos;
  52         n = strlen(s);
  53         e -= n;
  54         for(p = pos; p <= e; p++)
  55                 if(cistrncmp(p, s, n) == 0)
  56                         return p;
  57         return nil;
  58 }
  59
  60 void
  61 eatspace(void)
  62 {
  63         while(pos < epos && (*pos == ' ' || *pos == '\t'))
  64                 pos++;
  65 }
  66
  67 char*
  68 trimback(char *s)
  69 {
  70         while(s > pos && strchr("\t ", s[-1]) != nil)
  71                 s--;
  72         return s;
  73 }
  74
  75 void
  76 flush(void)
  77 {
  78         int n;
  79
  80         n = opos - buf;
  81         if(n <= 0)
  82                 return;
  83         if(write(1, buf, n) != n)
  84                 sysfatal("write: %r");
  85         opos = buf;
  86 }
  87
  88 void
  89 output(char *s, int n)
  90 {
  91         int r;
  92
  93         if(n <= 0)
  94                 return;
  95         r = HUNK - (opos - buf);
  96         if(n > r){
  97                 output(s, r);
  98                 output(s+r, n-r);
  99         } else {
 100                 memmove(opos, s, n);
 101                 opos += n;
 102                 if(r == n)
 103                         flush();
 104         }
 105 }
 106
 107 void
 108 string(char *s)
 109 {
 110         output(s, strlen(s));
 111 }
 112
 113 void
 114 escape(char *e)
 115 {
 116         char *p;
 117
 118         for(p = pos; p < e; p++)
 119                 if(*p == '<'){
 120                         output(pos, p - pos);
 121                         pos = p+1;
 122                         string("&lt;");
 123                 } else if(*p == '>'){
 124                         output(pos, p - pos);
 125                         pos = p+1;
 126                         string("&gt;");
 127                 } else if(*p == '&'){
 128                         output(pos, p - pos);
 129                         pos = p+1;
 130                         string("&amp;");
 131                 }
 132         output(pos, p - pos);
 133         pos = p;
 134 }
 135
 136 void
 137 ebody(char *e)
 138 {
 139         char *t;
 140
 141         t = epos;
 142         epos = trimback(e);
 143         body();
 144         pos = e;
 145         epos = t;
 146 }
 147
 148 int
 149 tag(char *term, char *tag)
 150 {
 151         char *e;
 152
 153         if(!got(term))
 154                 return 0;
 155         if(e = look(term, nil)){
 156                 eatspace();
 157                 string("<"); string(tag); string(">");
 158                 ebody(e);
 159                 string("</"); string(tag); string(">");
 160                 pos += strlen(term);
 161         } else
 162                 string(term);
 163         return 1;
 164 }
 165
 166 int
 167 heading(void)
 168 {
 169         char *o, *s, *e;
 170         int n;
 171
 172         for(s = "======"; *s; s++)
 173                 if(got(s))
 174                         break;
 175         if(*s == 0)
 176                 return 0;
 177         n = strlen(s);
 178         e = look("=", look("\n", nil));
 179         if(e == nil)
 180                 e = look("\n", nil);
 181         if(e == nil)
 182                 e = epos;
 183         eatspace();
 184         string("<h");
 185         output("0123456"+n, 1);
 186         string("><a name=\"");
 187         o = pos;
 188         s = trimback(e);
 189         while(pos < s){
 190                 if((*pos >= 'a' && *pos <= 'z')
 191                 || (*pos >= 'A' && *pos <= 'Z')
 192                 || (*pos >= '0' && *pos <= '9')
 193                 || (strchr("!#$%&()_+,-./{|}~:;=?@[\\]^_`", *pos) != 0))
 194                         output(pos, 1);
 195                 else if(*pos == ' ' || *pos == '\t')
 196                         output("_", 1);
 197                 else if(*pos == '<')
 198                         output("&lt;", 4);
 199                 else if(*pos == '>')
 200                         output("&gt;", 4);
 201                 else if(*pos == '&')
 202                         output("&amp;", 5);
 203                 else if(*pos == '"')
 204                         output("&quot;", 6);
 205                 else if(*pos == '\'')
 206                         output("&#39;", 5);
 207                 pos++;
 208         }
 209         string("\"></a>");
 210         pos = o;
 211         ebody(e);
 212         while(got("="))
 213                 ;
 214         string("</h");
 215         output("0123456"+n, 1);
 216         string(">");
 217         return 1;
 218 }
 219
 220 void
 221 link(char *e)
 222 {
 223         char *s, *o;
 224
 225         s = o = pos;
 226         while(s < epos){
 227                 if(e != nil && s >= e)
 228                         break;
 229                 if(*s == 0 || strchr("<>[] \t\r\n", *s) != nil)
 230                         break;
 231                 s++;
 232         }
 233         if(s-4 >= o)
 234         if(cistrncmp(s-4, ".png", 4)
 235         && cistrncmp(s-4, ".jpg", 4)
 236         && cistrncmp(s-4, ".gif", 4)){
 237                 string("<a href=\"");
 238                 escape(s);
 239                 string("\">");
 240                 eatspace();
 241                 if(e != nil && pos < e)
 242                         ebody(e);
 243                 else {
 244                         pos = o;
 245                         escape(s);
 246                 }
 247                 string("</a>");
 248         } else {
 249                 string("<img src=\"");
 250                 escape(s);
 251                 string("\">");
 252         }
 253 }
 254
 255 void
 256 body(void)
 257 {
 258         char *s;
 259         int t;
 260
 261 Next:
 262         if(pos >= epos)
 263                 return;
 264
 265         if(got("\n") || got("\r\n"))
 266                 indent = -1;
 267         if(got("\n") || got("\r\n")){
 268                 string("<br>");
 269                 while(got("\n") || got("\r\n"))
 270                         ;
 271         }
 272
 273         if(indent == -1){
 274                 indent = 0;
 275                 for(;;){
 276                         if(got(" "))
 277                                 indent++;
 278                         else if(got("\t")){
 279                                 indent += 8;
 280                                 indent %= 8;
 281                         }
 282                         else break;
 283                 }
 284
 285                 if(intable && look("||", look("\n", nil)) == nil){
 286                         string("</table>");
 287                         intable = 0;
 288                 }
 289
 290                 string("\n");
 291                 if((indent < inlist) || (indent < inquote))
 292                         return;
 293
 294                 while(indent > 0){
 295                         if(pos >= epos)
 296                                 return;
 297                         if(got("*") || got("#")){
 298                                 s = pos-1;
 299                                 eatspace();
 300                                 if(indent > inlist){
 301                                         if(*s == '*')
 302                                                 string("<ul><li>");
 303                                         else
 304                                                 string("<ol><li>");
 305                                         t = inlist;
 306                                         inlist = indent;
 307                                         body();
 308                                         inlist = t;
 309                                         if(*s == '*')
 310                                                 string("</li></ul>");
 311                                         else
 312                                                 string("</li></ol>");
 313                                 } else {
 314                                         string("</li><li>");
 315                                         break;
 316                                 }
 317                         } else if(indent > inquote){
 318                                 string("<blockquote>");
 319                                 t = inquote;
 320                                 inquote = indent;
 321                                 body();
 322                                 inquote = t;
 323                                 string("</blockquote>");
 324                         } else
 325                                 break;
 326                 }
 327
 328                 if(indent == 0){
 329                         if(got("#")){
 330                                 if((pos = look("\n", nil)) == nil)
 331                                         pos = epos;
 332                                 goto Next;
 333                         }
 334                         if(heading())
 335                                 goto Next;
 336                         if(got("----")){
 337                                 while(got("-"))
 338                                         ;
 339                                 string("<hr>");
 340                                 goto Next;
 341                         }
 342                 }
 343         }
 344
 345         if(got("`")){
 346                 if(s = look("`", nil)){
 347                         escape(s);
 348                         pos = s+1;
 349                 } else
 350                         string("`");
 351         }
 352         else if(got("<")){
 353                 string("<");
 354                 if(s = look(">", nil)){
 355                         s++;
 356                         output(pos, s - pos);
 357                         pos = s;
 358                 }
 359         }
 360         else if(got("[")){
 361                 if(s = look("]", nil)){
 362                         link(s);
 363                         pos = s+1;
 364                 } else
 365                         string("[");
 366         }
 367         else if(tag("*", "b") ||
 368                 tag("_", "i") ||
 369                 tag("^", "sup") ||
 370                 tag(",,", "sub") ||
 371                 tag("~~", "strike")){
 372         }
 373         else if(got("{{{")){
 374                 if(s = look("}}}", nil)){
 375                         if(look("\n", s)){
 376                                 string("<pre>");
 377                                 escape(s);
 378                                 string("</pre>");
 379                         } else {
 380                                 string("<tt>");
 381                                 escape(s);
 382                                 string("</tt>");
 383                         }
 384                         pos = s+3;
 385                 } else
 386                         string("{{{");
 387         }
 388         else if(got("||")){
 389                 if(s = look("||", look("\n", nil))){
 390                         eatspace();
 391                         switch(intable){
 392                         case 0: string("<table>");
 393                                 intable++;
 394                         case 1: string("<tr>");
 395                                 intable++;
 396                         }
 397                         string("<td>");
 398                         ebody(s);
 399                         string("</td>");
 400                 } else if(intable){
 401                         string("</tr>");
 402                         intable = 1;
 403                 }
 404         }
 405         else if(match("http://"))
 406                 link(nil);
 407         else if(match("https://"))
 408                 link(nil);
 409         else if(match("ftp://"))
 410                 link(nil);
 411         else{
 412                 output(pos, 1);
 413                 pos++;
 414         }
 415         goto Next;
 416 }
 417
 418 void
 419 usage(void)
 420 {
 421         fprint(2, "usage: %s [ file ]\n", argv0);
 422         exits("usage");
 423 }
 424
 425 void
 426 main(int argc, char *argv[])
 427 {
 428         int n;
 429
 430         ARGBEGIN{
 431         default:
 432                 usage();
 433         }ARGEND;
 434
 435         if(argc != 0 && argc != 1)
 436                 usage();
 437
 438         if(*argv){
 439                 if((n = open(*argv, OREAD)) < 0)
 440                         sysfatal("open %s: %r", *argv);
 441                 if(dup(n, 0) < 0)
 442                         sysfatal("dup: %r");
 443         }
 444
 445         buf = opos = sbrk(HUNK);
 446         pos = epos = buf + HUNK;
 447         for(;;){
 448                 if(brk(epos + HUNK + 8) < 0)
 449                         sysfatal("brk: %r");
 450                 if((n = read(0, epos, HUNK)) < 0)
 451                         sysfatal("read: %r");
 452                 if(n == 0)
 453                         break;
 454                 epos += n;
 455         }
 456         if(epos > pos && epos[-1] != '\n')
 457                 *epos++ = '\n';
 458
 459         body();
 460         flush();
 461         exits(0);
 462 }