filter.c

   1 /*@ S-nail - a mail user agent derived from Berkeley Mail.
   2  *@ Filter objects.
   3  *
   4  * Copyright (c) 2013 - 2016 Steffen (Daode) Nurpmeso <steffen@sdaoden.eu>.
   5  *
   6  * Permission to use, copy, modify, and/or distribute this software for any
   7  * purpose with or without fee is hereby granted, provided that the above
   8  * copyright notice and this permission notice appear in all copies.
   9  *
  10  * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
  11  * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
  12  * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
  13  * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
  14  * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
  15  * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
  16  * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
  17  */
  18 #undef n_FILE
  19 #define n_FILE filter
  20
  21 #ifndef HAVE_AMALGAMATION
  22 # include "nail.h"
  23 #endif
  24
  25 /*
  26  * Quotation filter
  27  */
  28
  29 /*
  30  * TODO quotation filter: anticipate in future data: don't break if only WS
  31  * TODO or a LF escaping \ follows on the line (simply reuse the latter).
  32  */
  33
  34 #ifdef HAVE_QUOTE_FOLD
  35 CTA(QUOTE_MAX > 3);
  36
  37 enum qf_state {
  38    _QF_CLEAN,
  39    _QF_PREFIX,
  40    _QF_DATA
  41 };
  42
  43 struct qf_vc {
  44    struct quoteflt   *self;
  45    char const        *buf;
  46    size_t            len;
  47 };
  48
  49 /* Print out prefix and current quote */
  50 static ssize_t _qf_dump_prefix(struct quoteflt *self);
  51
  52 /* Add one data character */
  53 static ssize_t _qf_add_data(struct quoteflt *self, wchar_t wc);
  54
  55 /* State machine handlers */
  56 static ssize_t _qf_state_prefix(struct qf_vc *vc);
  57 static ssize_t _qf_state_data(struct qf_vc *vc);
  58
  59 static ssize_t
  60 _qf_dump_prefix(struct quoteflt *self)
  61 {
  62    ssize_t rv;
  63    size_t i;
  64    NYD_ENTER;
  65
  66    if ((i = self->qf_pfix_len) > 0 && i != fwrite(self->qf_pfix, 1, i,
  67          self->qf_os))
  68       goto jerr;
  69    rv = i;
  70
  71    if ((i = self->qf_currq.l) > 0 && i != fwrite(self->qf_currq.s, 1, i,
  72          self->qf_os))
  73       goto jerr;
  74    rv += i;
  75 jleave:
  76    NYD_LEAVE;
  77    return rv;
  78 jerr:
  79    rv = -1;
  80    goto jleave;
  81 }
  82
  83 static ssize_t
  84 _qf_add_data(struct quoteflt *self, wchar_t wc)
  85 {
  86    char *save_b;
  87    ui32_t save_l, save_w;
  88    ssize_t rv = 0;
  89    int w, l;
  90    NYD_ENTER;
  91
  92    save_l = save_w = 0; /* silence cc */
  93    save_b = NULL;
  94    /* <newline> ends state */
  95    if (wc == L'\n')
  96       goto jflush;
  97    if (wc == L'\r') /* TODO CR should be stripped in lower level!! */
  98       goto jleave;
  99
 100    /* Unroll <tab> to spaces */
 101    if (wc == L'\t') {
 102       save_l = self->qf_datw;
 103       save_w = (save_l + QUOTE_TAB_SPACES) & ~(QUOTE_TAB_SPACES - 1);
 104       save_w -= save_l;
 105       while (save_w-- > 0) {
 106          ssize_t j = _qf_add_data(self, L' ');
 107          if (j < 0) {
 108             rv = j;
 109             break;
 110          }
 111          rv += j;
 112       }
 113       goto jleave;
 114    }
 115
 116    w = wcwidth(wc);
 117    if (w == -1) {
 118 jbad:
 119       ++self->qf_datw;
 120       self->qf_dat.s[self->qf_dat.l++] = '?';
 121    } else {
 122       l = wctomb(self->qf_dat.s + self->qf_dat.l, wc);
 123       if (l < 0)
 124          goto jbad;
 125       self->qf_datw += (ui32_t)w;
 126       self->qf_dat.l += (size_t)l;
 127    }
 128
 129    /* TODO The last visual may excess (adjusted!) *qfold-max* if it's a wide;
 130     * TODO place it on the next line, break before */
 131    if (self->qf_datw >= self->qf_qfold_max) {
 132       /* If we have seen a nice breakpoint during traversal, shuffle data
 133        * around a bit so as to restore the trailing part after flushing */
 134       if (self->qf_brkl > 0) {
 135          save_w = self->qf_datw - self->qf_brkw;
 136          save_l = self->qf_dat.l - self->qf_brkl;
 137          save_b = self->qf_dat.s + self->qf_brkl + 2;
 138          memmove(save_b, save_b - 2, save_l);
 139          self->qf_dat.l = self->qf_brkl;
 140       }
 141
 142       self->qf_dat.s[self->qf_dat.l++] = '\\';
 143 jflush:
 144       self->qf_dat.s[self->qf_dat.l++] = '\n';
 145       rv = quoteflt_flush(self);
 146
 147       /* Restore takeovers, if any */
 148       if (save_b != NULL) {
 149          self->qf_brk_isws = FAL0;
 150          self->qf_datw += save_w;
 151          self->qf_dat.l = save_l;
 152          memmove(self->qf_dat.s, save_b, save_l);
 153       }
 154    } else if (self->qf_datw >= self->qf_qfold_min && !self->qf_brk_isws) {
 155       bool_t isws = (iswspace(wc) != 0);
 156
 157       if (isws || !self->qf_brk_isws || self->qf_brkl == 0) {
 158          self->qf_brkl = self->qf_dat.l;
 159          self->qf_brkw = self->qf_datw;
 160          self->qf_brk_isws = isws;
 161       }
 162    }
 163
 164    /* If state changed to prefix, perform full reset (note this implies that
 165     * quoteflt_flush() performs too much work..) */
 166    if (wc == '\n') {
 167       self->qf_state = _QF_PREFIX;
 168       self->qf_wscnt = self->qf_datw = 0;
 169       self->qf_currq.l = 0;
 170    }
 171 jleave:
 172    NYD_LEAVE;
 173    return rv;
 174 }
 175
 176 static ssize_t
 177 _qf_state_prefix(struct qf_vc *vc)
 178 {
 179    struct quoteflt *self;
 180    ssize_t rv;
 181    char const *buf;
 182    size_t len, i;
 183    wchar_t wc;
 184    NYD_ENTER;
 185
 186    self = vc->self;
 187    rv = 0;
 188
 189    for (buf = vc->buf, len = vc->len; len > 0;) {
 190       /* xxx NULL BYTE! */
 191       i = mbrtowc(&wc, buf, len, self->qf_mbps);
 192       if (i == (size_t)-1) {
 193          /* On hard error, don't modify mbstate_t and step one byte */
 194          self->qf_mbps[0] = self->qf_mbps[1];
 195          ++buf;
 196          --len;
 197          self->qf_wscnt = 0;
 198          continue;
 199       }
 200       self->qf_mbps[1] = self->qf_mbps[0];
 201       if (i == (size_t)-2) {
 202          /* Redundant shift sequence, out of buffer */
 203          len = 0;
 204          break;
 205       }
 206       buf += i;
 207       len -= i;
 208
 209       if (wc == L'\n')
 210          goto jfin;
 211       if (iswspace(wc)) {
 212          ++self->qf_wscnt;
 213          continue;
 214       }
 215       if (i == 1 && ISQUOTE(wc)) {
 216          self->qf_wscnt = 0;
 217          if (self->qf_currq.l >= QUOTE_MAX - 3) {
 218             self->qf_currq.s[QUOTE_MAX - 3] = '.';
 219             self->qf_currq.s[QUOTE_MAX - 2] = '.';
 220             self->qf_currq.s[QUOTE_MAX - 1] = '.';
 221             self->qf_currq.l = QUOTE_MAX;
 222          } else
 223             self->qf_currq.s[self->qf_currq.l++] = buf[-1];
 224          continue;
 225       }
 226
 227       /* The quote is parsed and compressed; dump it */
 228 jfin:
 229       self->qf_state = _QF_DATA;
 230       /* Overtake WS to the current quote in order to preserve it for eventual
 231        * necessary follow lines, too */
 232       /* TODO we de-facto "normalize" to ASCII SP here which MESSES tabs!! */
 233       while (self->qf_wscnt-- > 0 && self->qf_currq.l < QUOTE_MAX)
 234          self->qf_currq.s[self->qf_currq.l++] = ' ';
 235       self->qf_datw = self->qf_pfix_len + self->qf_currq.l;
 236       self->qf_wscnt = 0;
 237       rv = _qf_add_data(self, wc);
 238       break;
 239    }
 240
 241    vc->buf = buf;
 242    vc->len = len;
 243    NYD_LEAVE;
 244    return rv;
 245 }
 246
 247 static ssize_t
 248 _qf_state_data(struct qf_vc *vc)
 249 {
 250    struct quoteflt *self;
 251    ssize_t rv;
 252    char const *buf;
 253    size_t len, i;
 254    wchar_t wc;
 255    NYD_ENTER;
 256
 257    self = vc->self;
 258    rv = 0;
 259
 260    for (buf = vc->buf, len = vc->len; len > 0;) {
 261       /* xxx NULL BYTE! */
 262       i = mbrtowc(&wc, buf, len, self->qf_mbps);
 263       if (i == (size_t)-1) {
 264          /* On hard error, don't modify mbstate_t and step one byte */
 265          self->qf_mbps[0] = self->qf_mbps[1];
 266          ++buf;
 267          --len;
 268          continue;
 269       }
 270       self->qf_mbps[1] = self->qf_mbps[0];
 271       if (i == (size_t)-2) {
 272          /* Redundant shift sequence, out of buffer */
 273          len = 0;
 274          break;
 275       }
 276       buf += i;
 277       len -= i;
 278
 279       {  ssize_t j = _qf_add_data(self, wc);
 280          if (j < 0) {
 281             rv = j;
 282             break;
 283          }
 284          rv += j;
 285       }
 286
 287       if (self->qf_state != _QF_DATA)
 288          break;
 289    }
 290
 291    vc->buf = buf;
 292    vc->len = len;
 293    NYD_LEAVE;
 294    return rv;
 295 }
 296 #endif /* HAVE_QUOTE_FOLD */
 297
 298 FL struct quoteflt *
 299 quoteflt_dummy(void) /* TODO LEGACY (until filters are plugged when needed) */
 300 {
 301    static struct quoteflt qf_i;
 302
 303    return &qf_i;
 304 }
 305
 306 FL void
 307 quoteflt_init(struct quoteflt *self, char const *prefix)
 308 {
 309 #ifdef HAVE_QUOTE_FOLD
 310    char *xcp, *cp;
 311 #endif
 312    NYD_ENTER;
 313
 314    memset(self, 0, sizeof *self);
 315
 316    if ((self->qf_pfix = prefix) != NULL)
 317       self->qf_pfix_len = (ui32_t)strlen(prefix);
 318
 319    /* Check whether the user wants the more fancy quoting algorithm */
 320    /* TODO *quote-fold*: QUOTE_MAX may excess it! */
 321 #ifdef HAVE_QUOTE_FOLD
 322    if (self->qf_pfix_len > 0 && (cp = ok_vlook(quote_fold)) != NULL) {
 323       ui32_t qmin, qmax = (ui32_t)strtol(cp, &xcp, 10);
 324       /* These magic values ensure we don't bail :) */
 325       if (qmax < self->qf_pfix_len + 6)
 326          qmax = self->qf_pfix_len + 6;
 327       --qmax; /* The newline escape */
 328       if (cp == xcp || *xcp == '\0')
 329          qmin = (qmax >> 1) + (qmax >> 2) + (qmax >> 5);
 330       else {
 331          qmin = (ui32_t)strtol(xcp + 1, NULL, 10);
 332          if (qmin < qmax >> 1)
 333             qmin = qmax >> 1;
 334          else if (qmin > qmax - 2)
 335             qmin = qmax - 2;
 336       }
 337       self->qf_qfold_min = qmin;
 338       self->qf_qfold_max = qmax;
 339
 340       /* Add pad for takeover copies, backslash and newline */
 341       self->qf_dat.s = salloc((qmax + 3) * mb_cur_max);
 342       self->qf_currq.s = salloc((QUOTE_MAX + 1) * mb_cur_max);
 343    }
 344 #endif
 345    NYD_LEAVE;
 346 }
 347
 348 FL void
 349 quoteflt_destroy(struct quoteflt *self) /* xxx inline */
 350 {
 351    NYD_ENTER;
 352    UNUSED(self);
 353    NYD_LEAVE;
 354 }
 355
 356 FL void
 357 quoteflt_reset(struct quoteflt *self, FILE *f) /* xxx inline */
 358 {
 359    NYD_ENTER;
 360    self->qf_os = f;
 361 #ifdef HAVE_QUOTE_FOLD
 362    self->qf_state = _QF_CLEAN;
 363    self->qf_dat.l =
 364    self->qf_currq.l = 0;
 365    memset(self->qf_mbps, 0, sizeof self->qf_mbps);
 366 #endif
 367    NYD_LEAVE;
 368 }
 369
 370 FL ssize_t
 371 quoteflt_push(struct quoteflt *self, char const *dat, size_t len)
 372 {
 373    /* (xxx Ideally the actual push() [and flush()] would be functions on their
 374     * xxx own, via indirect vtbl call ..) */
 375    ssize_t rv = 0;
 376    NYD_ENTER;
 377
 378    self->qf_nl_last = (len > 0 && dat[len - 1] == '\n'); /* TODO HACK */
 379
 380    if (len == 0)
 381       goto jleave;
 382
 383    /* Bypass? TODO Finally, this filter simply should not be used, then
 384     * (TODO It supercedes prefix_write() or something) */
 385    if (self->qf_pfix_len == 0) {
 386       if (len != fwrite(dat, 1, len, self->qf_os))
 387          goto jerr;
 388       rv = len;
 389    }
 390    /* Normal: place *indentprefix* at every BOL */
 391    else
 392 #ifdef HAVE_QUOTE_FOLD
 393       if (self->qf_qfold_max == 0)
 394 #endif
 395    {
 396       void *vp;
 397       size_t ll;
 398       bool_t pxok = (self->qf_qfold_min != 0);
 399
 400       for (;;) {
 401          if (!pxok) {
 402             ll = self->qf_pfix_len;
 403             if (ll != fwrite(self->qf_pfix, 1, ll, self->qf_os))
 404                goto jerr;
 405             rv += ll;
 406             pxok = TRU1;
 407          }
 408
 409          /* xxx Strictly speaking this is invalid, because only `/' and `.' are
 410           * xxx mandated by POSIX.1-2008 as "invariant across all locales
 411           * xxx supported"; though there is no charset known which uses this
 412           * xxx control char as part of a multibyte character; note that S-nail
 413           * XXX (and the Mail codebase as such) do not support EBCDIC */
 414          if ((vp = memchr(dat, '\n', len)) == NULL)
 415             ll = len;
 416          else {
 417             pxok = FAL0;
 418             ll = PTR2SIZE((char*)vp - dat) + 1;
 419          }
 420
 421          if (ll != fwrite(dat, sizeof *dat, ll, self->qf_os))
 422             goto jerr;
 423          rv += ll;
 424          if ((len -= ll) == 0)
 425             break;
 426          dat += ll;
 427       }
 428
 429       self->qf_qfold_min = pxok;
 430    }
 431    /* Overly complicated, though still only line-per-line: *quote-fold*.
 432     * - If .qf_currq.l is 0, then we are in a clean state.  Reset .qf_mbps;
 433     *   TODO note this means we assume that lines start with reset escape seq,
 434     *   TODO but i don't think this is any worse than what we currently do;
 435     *   TODO in 15.0, with the value carrier, we should carry conversion states
 436     *   TODO all along, only resetting on error (or at words for header =???=);
 437     *   TODO this still is weird for error handling, but we need to act more
 438     *   TODO stream-alike (though in practice i don't think cross-line states
 439     *   TODO can be found, because of compatibility reasons; however, being
 440     *   TODO a problem rather than a solution is not a good thing (tm))
 441     * - Lookout for a newline */
 442 #ifdef HAVE_QUOTE_FOLD
 443    else {
 444       struct qf_vc vc;
 445       ssize_t i;
 446
 447       vc.self = self;
 448       vc.buf = dat;
 449       vc.len = len;
 450       while (vc.len > 0) {
 451          switch (self->qf_state) {
 452          case _QF_CLEAN:
 453          case _QF_PREFIX:
 454             i = _qf_state_prefix(&vc);
 455             break;
 456          default: /* silence cc (`i' unused) */
 457          case _QF_DATA:
 458             i = _qf_state_data(&vc);
 459             break;
 460          }
 461          if (i < 0)
 462             goto jerr;
 463          rv += i;
 464       }
 465    }
 466 #endif /* HAVE_QUOTE_FOLD */
 467
 468 jleave:
 469    NYD_LEAVE;
 470    return rv;
 471 jerr:
 472    rv = -1;
 473    goto jleave;
 474 }
 475
 476 FL ssize_t
 477 quoteflt_flush(struct quoteflt *self)
 478 {
 479    ssize_t rv = 0;
 480    NYD_ENTER;
 481    UNUSED(self);
 482
 483 #ifdef HAVE_QUOTE_FOLD
 484    if (self->qf_dat.l > 0) {
 485       rv = _qf_dump_prefix(self);
 486       if (rv >= 0) {
 487          size_t i = self->qf_dat.l;
 488          if (i == fwrite(self->qf_dat.s, 1, i, self->qf_os))
 489             rv += i;
 490          else
 491             rv = -1;
 492          self->qf_dat.l = 0;
 493          self->qf_brk_isws = FAL0;
 494          self->qf_wscnt = self->qf_brkl = self->qf_brkw = 0;
 495          self->qf_datw = self->qf_pfix_len + self->qf_currq.l;
 496       }
 497    }
 498 #endif
 499    NYD_LEAVE;
 500    return rv;
 501 }
 502
 503 /*
 504  * HTML tagsoup filter TODO rewrite wchar_t based (require HAVE_C90AMEND1)
 505  * TODO . Numeric &#NO; entities should also be treated by struct hf_ent
 506  * TODO . Yes, we COULD support CSS based quoting when we'd check type="quote"
 507  * TODO   (nonstandard) and watch out for style="gmail_quote" (or so, VERY
 508  * TODO   nonstandard) and tracking a stack of such elements (to be popped
 509  * TODO   once the closing element is seen).  Then, after writing a newline,
 510  * TODO   place sizeof(stack) ">"s first.  But aren't these HTML mails rude?
 511  * TODO Interlocking and non-well-formed data will break us down
 512  */
 513 #ifdef HAVE_FILTER_HTML_TAGSOUP
 514
 515 enum hf_limits {
 516    _HF_MINLEN  = 10,       /* Minimum line length (can't really be smaller) */
 517    _HF_BRKSUB  = 8         /* Start considering line break MAX - BRKSUB */
 518 };
 519
 520 enum hf_flags {
 521    _HF_UTF8    = 1<<0,     /* Data is in UTF-8 */
 522    _HF_ERROR   = 1<<1,     /* A hard error occurred, bail as soon as possible */
 523    _HF_NOPUT   = 1<<2,     /* (In a tag,) Don't generate output */
 524    _HF_IGN     = 1<<3,     /* Ignore mode on */
 525    _HF_ANY     = 1<<4,     /* Yet seen just any output */
 526    _HF_PRE     = 1<<5,     /* In <pre>formatted mode */
 527    _HF_ENT     = 1<<6,     /* Currently parsing an entity */
 528    _HF_BLANK   = 1<<7,     /* Whitespace last */
 529    _HF_HREF    = 1<<8,     /* External <a href=> was the last href seen */
 530
 531    _HF_NL_1    = 1<<9,     /* One \n seen */
 532    _HF_NL_2    = 2<<9,     /* We have produced an all empty line */
 533    _HF_NL_MASK = _HF_NL_1 | _HF_NL_2
 534 };
 535
 536 enum hf_special_actions {
 537    _HFSA_NEEDSEP  = -1,    /* Need an empty line (paragraph separator) */
 538    _HFSA_NEEDNL   = -2,    /* Need a new line start (table row) */
 539    _HFSA_IGN      = -3,    /* Things like <style>..</style>, <script>.. */
 540    _HFSA_PRE      = -4,    /* <pre>.. */
 541    _HFSA_PRE_END  = -5,
 542    _HFSA_IMG      = -6,    /* <img> */
 543    _HFSA_HREF     = -7,    /* <a>.. */
 544    _HFSA_HREF_END = -8
 545 };
 546
 547 enum hf_entity_flags {
 548    _HFE_HAVE_UNI  = 1<<6,  /* Have a Unicode replacement character */
 549    _HFE_HAVE_CSTR = 1<<7,  /* Have a string replacement */
 550    /* We store the length of the entity name in the flags, too */
 551    _HFE_LENGTH_MASK = (1<<6) - 1
 552 };
 553
 554 struct htmlflt_href {
 555    struct htmlflt_href *hfh_next;
 556    ui32_t      hfh_no;     /* Running sequence */
 557    ui32_t      hfh_len;    /* of .hfh_dat */
 558    char        hfh_dat[VFIELD_SIZE(0)];
 559 };
 560
 561 struct htmlflt_tag {
 562    si32_t      hft_act;    /* char or hf_special_actions */
 563    /* Not NUL: character to inject, with high bit set: place a space
 564     * afterwards.  Note: only recognized with _HFSA_NEEDSEP or _HFSA_NEEDNL */
 565    char        hft_injc;
 566    ui8_t       hft_len;    /* Useful bytes in (NUL terminated) .hft_tag */
 567    char const  hft_tag[10]; /* Tag less < and > surroundings (TR, /TR, ..) */
 568 };
 569 CTA(SIZEOF_FIELD(struct htmlflt_tag, hft_tag) < LINESIZE); /* .hf_ign_tag */
 570
 571 struct hf_ent {
 572    ui8_t       hfe_flags;  /* enum hf_entity_flags plus length of .hfe_ent */
 573    char        hfe_c;      /* Plain replacement character */
 574    ui16_t      hfe_uni;    /* Unicode codepoint if _HFE_HAVE_UNI */
 575    char        hfe_cstr[5]; /* _HFE_HAVE_CSTR (e.g., &hellip; -> ...) */
 576    char const  hfe_ent[7]; /* Entity less & and ; surroundings */
 577 };
 578
 579 /* Tag list; not binary searched :(, so try to take care a bit */
 580 static struct htmlflt_tag const  _hf_tags[] = {
 581 # undef _X
 582 # undef _XC
 583 # define _X(S,A)     {A, '\0', sizeof(S) -1, S "\0"}
 584 # define _XC(S,C,A)  {A, C, sizeof(S) -1, S "\0"}
 585
 586    _X("P", _HFSA_NEEDSEP),       _X("/P", _HFSA_NEEDNL),
 587    _X("DIV", _HFSA_NEEDSEP),     _X("/DIV", _HFSA_NEEDNL),
 588    _X("TR", _HFSA_NEEDNL),
 589                                  _X("/TH", '\t'),
 590                                  _X("/TD", '\t'),
 591    /* Let it stand out; also since we don't support implicit paragraphs after
 592     * block elements, plain running text after a list (seen in Unicode
 593     * announcement via Firefox) */
 594    _X("UL", _HFSA_NEEDSEP),      _X("/UL", _HFSA_NEEDSEP),
 595    _XC("LI", (char)0x80 | '*', _HFSA_NEEDSEP),
 596    _X("DL", _HFSA_NEEDSEP),
 597    _X("DT", _HFSA_NEEDNL),
 598
 599    _X("A", _HFSA_HREF),          _X("/A", _HFSA_HREF_END),
 600    _X("IMG", _HFSA_IMG),
 601    _X("BR", '\n'),
 602    _X("PRE", _HFSA_PRE),         _X("/PRE", _HFSA_PRE_END),
 603    _X("TITLE", _HFSA_NEEDSEP),   /*_X("/TITLE", '\n'),*/
 604    _X("H1", _HFSA_NEEDSEP),      /*_X("/H1", '\n'),*/
 605    _X("H2", _HFSA_NEEDSEP),      /*_X("/H2", '\n'),*/
 606    _X("H3", _HFSA_NEEDSEP),      /*_X("/H3", '\n'),*/
 607    _X("H4", _HFSA_NEEDSEP),      /*_X("/H4", '\n'),*/
 608    _X("H5", _HFSA_NEEDSEP),      /*_X("/H5", '\n'),*/
 609    _X("H6", _HFSA_NEEDSEP),      /*_X("/H6", '\n'),*/
 610
 611    _X("STYLE", _HFSA_IGN),
 612    _X("SCRIPT", _HFSA_IGN),
 613
 614 # undef _X
 615 };
 616
 617 /* Entity list; not binary searched.. */
 618 static struct hf_ent const       _hf_ents[] = {
 619 # undef _X
 620 # undef _XU
 621 # undef _XS
 622 # undef _XUS
 623 # define _X(E,C)     {(sizeof(E) -1), C, 0x0u, "", E "\0"}
 624 # define _XU(E,C,U)  {(sizeof(E) -1) | _HFE_HAVE_UNI, C, U, "", E "\0"}
 625 # define _XS(E,S)    {(sizeof(E) -1) | _HFE_HAVE_CSTR, '\0', 0x0u,S "\0",E "\0"}
 626 # define _XSU(E,S,U) \
 627    {(sizeof(E) -1) | _HFE_HAVE_UNI | _HFE_HAVE_CSTR, '\0', U, S "\0", E "\0"}
 628
 629    _X("quot", '"'),
 630    _X("amp", '&'),
 631    _X("lt", '<'),                _X("gt", '>'),
 632
 633    _XU("nbsp", ' ', 0x0020 /* Note: not 0x00A0 seems to be better for us */),
 634    _XU("middot", '.', 0x00B7),
 635    _XSU("hellip", "...", 0x2026),
 636    _XSU("mdash", "---", 0x2014), _XSU("ndash", "--", 0x2013),
 637    _XSU("laquo", "<<", 0x00AB),  _XSU("raquo", ">>", 0x00BB),
 638    _XSU("lsaquo", "<", 0x2039),  _XSU("rsaquo", ">", 0x203A),
 639    _XSU("lsquo", "'", 0x2018),   _XSU("rsquo", "'", 0x2019),
 640    _XSU("ldquo", "\"", 0x201C),  _XSU("rdquo", "\"", 0x201D),
 641    _XSU("uarr", "^|", 0x2191),   _XSU("darr", "|v", 0x2193),
 642
 643    _XSU("cent", "CENT", 0x00A2),
 644    _XSU("copy", "(C)", 0x00A9),
 645    _XSU("euro", "EUR", 0x20AC),
 646    _XSU("infin", "INFY", 0x221E),
 647    _XSU("pound", "GBP", 0x00A3),
 648    _XSU("reg", "(R)", 0x00AE),
 649    _XSU("sect", "S:", 0x00A7),
 650    _XSU("yen", "JPY", 0x00A5),
 651
 652    /* German umlauts */
 653    _XSU("Auml", "Ae", 0x00C4),   _XSU("auml", "ae", 0x00E4),
 654    _XSU("Ouml", "Oe", 0x00D6),   _XSU("ouml", "oe", 0x00F6),
 655    _XSU("Uuml", "Ue", 0x00DC),   _XSU("uuml", "ue", 0x00FC),
 656    _XSU("szlig", "ss", 0x00DF)
 657
 658 # undef _X
 659 # undef _XU
 660 # undef _XS
 661 # undef _XSU
 662 };
 663
 664 /* Real output */
 665 static struct htmlflt * _hf_dump_hrefs(struct htmlflt *self);
 666 static struct htmlflt * _hf_dump(struct htmlflt *self);
 667 static struct htmlflt * _hf_store(struct htmlflt *self, char c);
 668 # ifdef HAVE_NATCH_CHAR
 669 static struct htmlflt * __hf_sync_mbstuff(struct htmlflt *self);
 670 # endif
 671
 672 /* Virtual output */
 673 static struct htmlflt * _hf_nl(struct htmlflt *self);
 674 static struct htmlflt * _hf_nl_force(struct htmlflt *self);
 675 static struct htmlflt * _hf_putc(struct htmlflt *self, char c);
 676 static struct htmlflt * _hf_putc_premode(struct htmlflt *self, char c);
 677 static struct htmlflt * _hf_puts(struct htmlflt *self, char const *cp);
 678 static struct htmlflt * _hf_putbuf(struct htmlflt *self,
 679                            char const *cp, size_t len);
 680
 681 /* Try to locate a param'eter in >hf_bdat, store it (non-terminated!) or NULL */
 682 static struct htmlflt * _hf_param(struct htmlflt *self, struct str *store,
 683                            char const *param);
 684
 685 /* Expand all entities in the given parameter */
 686 static struct htmlflt * _hf_expand_all_ents(struct htmlflt *self,
 687                            struct str const *param);
 688
 689 /* Completely parsed over a tag / an entity, interpret that */
 690 static struct htmlflt * _hf_check_tag(struct htmlflt *self, char const *s);
 691 static struct htmlflt * _hf_check_ent(struct htmlflt *self, char const *s,
 692                            size_t l);
 693
 694 /* Input handler */
 695 static ssize_t          _hf_add_data(struct htmlflt *self,
 696                            char const *dat, size_t len);
 697
 698 static struct htmlflt *
 699 _hf_dump_hrefs(struct htmlflt *self)
 700 {
 701    struct htmlflt_href *hhp;
 702    NYD2_ENTER;
 703
 704    if (!(self->hf_flags & _HF_NL_2) && putc('\n', self->hf_os) == EOF) {
 705       self->hf_flags |= _HF_ERROR;
 706       goto jleave;
 707    }
 708
 709    /* Reverse the list */
 710    for (hhp = self->hf_hrefs, self->hf_hrefs = NULL; hhp != NULL;) {
 711       struct htmlflt_href *tmp = hhp->hfh_next;
 712       hhp->hfh_next = self->hf_hrefs;
 713       self->hf_hrefs = hhp;
 714       hhp = tmp;
 715    }
 716
 717    /* Then dump it */
 718    while ((hhp = self->hf_hrefs) != NULL) {
 719       self->hf_hrefs = hhp->hfh_next;
 720
 721       if (!(self->hf_flags & _HF_ERROR)) {
 722          int w = fprintf(self->hf_os, "  [%u] %.*s\n",
 723                hhp->hfh_no, (int)hhp->hfh_len, hhp->hfh_dat);
 724          if (w < 0)
 725             self->hf_flags |= _HF_ERROR;
 726       }
 727       free(hhp);
 728    }
 729
 730    self->hf_flags |= (putc('\n', self->hf_os) == EOF)
 731          ?  _HF_ERROR : _HF_NL_1 | _HF_NL_2;
 732    self->hf_href_dist = (ui32_t)realscreenheight >> 1;
 733 jleave:
 734    NYD2_LEAVE;
 735    return self;
 736 }
 737
 738 static struct htmlflt *
 739 _hf_dump(struct htmlflt *self)
 740 {
 741    ui32_t f, l;
 742    char c, *cp;
 743    NYD2_ENTER;
 744
 745    f = self->hf_flags & ~_HF_BLANK;
 746    l = self->hf_len;
 747    cp = self->hf_line;
 748    self->hf_mbwidth = self->hf_mboff = self->hf_last_ws = self->hf_len = 0;
 749
 750    for (c = '\0'; l > 0; --l) {
 751       c = *cp++;
 752 jput:
 753       if (putc(c, self->hf_os) == EOF) {
 754          self->hf_flags = (f |= _HF_ERROR);
 755          goto jleave;
 756       }
 757    }
 758
 759    if (c != '\n') {
 760       f |= (f & _HF_NL_1) ? _HF_NL_2 : _HF_NL_1;
 761       l = 1;
 762       c = '\n';
 763       goto jput;
 764    }
 765    self->hf_flags = f;
 766
 767    /* Check whether there are HREFs to dump; there is so much messy tagsoup out
 768     * there that it seems best not to simply dump HREFs in each _dump(), but
 769     * only with some gap, let's say half the real screen height */
 770    if (--self->hf_href_dist < 0 && (f & _HF_NL_2) && self->hf_hrefs != NULL)
 771       self = _hf_dump_hrefs(self);
 772 jleave:
 773    NYD2_LEAVE;
 774    return self;
 775 }
 776
 777 static struct htmlflt *
 778 _hf_store(struct htmlflt *self, char c)
 779 {
 780    ui32_t f, l, i;
 781    NYD2_ENTER;
 782
 783    assert(c != '\n');
 784
 785    f = self->hf_flags;
 786    l = self->hf_len;
 787    self->hf_line[l] = (c == '\t' ? ' ' : c);
 788    self->hf_len = ++l;
 789    if (blankspacechar(c)) {
 790       if (c == '\t') {
 791          i = 8 - ((l - 1) & 7); /* xxx magic tab width of 8 */
 792          if (i > 0) {
 793             do
 794                self = _hf_store(self, ' ');
 795             while (--i > 0);
 796             goto jleave;
 797          }
 798       }
 799       self->hf_last_ws = l;
 800    } else if (/*c == '.' ||*/ c == ',' || c == ';' || c == '-')
 801       self->hf_last_ws = l;
 802
 803    i = l;
 804 # ifdef HAVE_NATCH_CHAR /* XXX This code is really ridiculous! */
 805    if (mb_cur_max > 1) { /* XXX should mbrtowc() and THEN store, at least.. */
 806       wchar_t wc;
 807       int w, x = mbtowc(&wc, self->hf_line + self->hf_mboff, l - self->hf_mboff);
 808
 809       if (x > 0) {
 810          if ((w = wcwidth(wc)) == -1 ||
 811                /* Actively filter out L-TO-R and R-TO-R marks TODO ctext */
 812                (wc == 0x200E || wc == 0x200F ||
 813                   (wc >= 0x202A && wc <= 0x202E)) ||
 814                /* And some zero-width messes */
 815                wc == 0x00AD || (wc >= 0x200B && wc <= 0x200D) ||
 816                /* Oh about the ISO C wide character interfaces, baby! */
 817                (wc == 0xFEFF)){
 818             self->hf_len -= x;
 819             goto jleave;
 820          } else if (iswspace(wc))
 821             self->hf_last_ws = l;
 822          self->hf_mboff += x;
 823          i = (self->hf_mbwidth += w);
 824       } else {
 825          if (x < 0) {
 826             mbtowc(&wc, NULL, mb_cur_max);
 827             if (UICMP(32, l - self->hf_mboff, >=, mb_cur_max)) { /* XXX */
 828                ++self->hf_mboff;
 829                ++self->hf_mbwidth;
 830             }
 831          }
 832          i = self->hf_mbwidth;
 833       }
 834    }
 835 # endif
 836
 837    /* Do we need to break the line? */
 838    if (i >= self->hf_lmax - _HF_BRKSUB) {
 839       ui32_t lim = self->hf_lmax >> 1;
 840
 841       /* Let's hope we saw a sane place to break this line! */
 842       if (self->hf_last_ws >= lim) {
 843 jput:
 844          i = self->hf_len = self->hf_last_ws;
 845          self = _hf_dump(self);
 846          if ((self->hf_len = (l -= i)) > 0) {
 847             self->hf_flags &= ~_HF_NL_MASK;
 848             memmove(self->hf_line, self->hf_line + i, l);
 849 # ifdef HAVE_NATCH_CHAR
 850             __hf_sync_mbstuff(self);
 851 # endif
 852          }
 853          goto jleave;
 854       }
 855
 856       /* Any 7-bit characters? */
 857       for (i = l; i-- >= lim;)
 858          if (asciichar((c = self->hf_line[i]))) {
 859             self->hf_last_ws = ++i;
 860             goto jput;
 861          } else if ((f & _HF_UTF8) && ((ui8_t)c & 0xC0) != 0x80) {
 862             self->hf_last_ws = i;
 863             goto jput;
 864          }
 865
 866       /* Hard break necessary!  xxx really badly done */
 867       if (l >= self->hf_lmax - 1)
 868          self = _hf_dump(self);
 869    }
 870 jleave:
 871    NYD2_LEAVE;
 872    return self;
 873 }
 874
 875 # ifdef HAVE_NATCH_CHAR
 876 static struct htmlflt *
 877 __hf_sync_mbstuff(struct htmlflt *self)
 878 {
 879    wchar_t wc;
 880    char const *b;
 881    ui32_t o, w, l;
 882    NYD2_ENTER;
 883
 884    b = self->hf_line;
 885    o = w = 0;
 886    l = self->hf_len;
 887    goto jumpin;
 888
 889    while (l > 0) {
 890       int x = mbtowc(&wc, b, l);
 891
 892       if (x == 0)
 893          break;
 894
 895       if (x > 0) {
 896          b += x;
 897          l -= x;
 898          o += x;
 899          if ((x = wcwidth(wc)) == -1)
 900             x = 1;
 901          w += x;
 902          continue;
 903       }
 904
 905       /* Bad, skip over a single character.. XXX very bad indeed */
 906       ++b;
 907       ++o;
 908       ++w;
 909       --l;
 910 jumpin:
 911       mbtowc(&wc, NULL, mb_cur_max);
 912    }
 913
 914    self->hf_mboff = o;
 915    self->hf_mbwidth = w;
 916
 917    NYD2_LEAVE;
 918    return self;
 919 }
 920 # endif /* HAVE_NATCH_CHAR */
 921
 922 static struct htmlflt *
 923 _hf_nl(struct htmlflt *self)
 924 {
 925    ui32_t f;
 926    NYD2_ENTER;
 927
 928    if (!((f = self->hf_flags) & _HF_ERROR)) {
 929       if (f & _HF_ANY) {
 930          if ((f & _HF_NL_MASK) != _HF_NL_MASK)
 931             self = _hf_dump(self);
 932       } else
 933          self->hf_flags = (f |= _HF_NL_MASK);
 934    }
 935    NYD2_LEAVE;
 936    return self;
 937 }
 938
 939 static struct htmlflt *
 940 _hf_nl_force(struct htmlflt *self)
 941 {
 942    NYD2_ENTER;
 943    if (!(self->hf_flags & _HF_ERROR))
 944       self = _hf_dump(self);
 945    NYD2_LEAVE;
 946    return self;
 947 }
 948
 949 static struct htmlflt *
 950 _hf_putc(struct htmlflt *self, char c)
 951 {
 952    ui32_t f;
 953    NYD2_ENTER;
 954
 955    if ((f = self->hf_flags) & _HF_ERROR)
 956       goto jleave;
 957
 958    if (c == '\n') {
 959       self = _hf_nl(self);
 960       goto jleave;
 961    } else if (c == ' ' || c == '\t') {
 962       if ((f & _HF_BLANK) || self->hf_len == 0)
 963          goto jleave;
 964       f |= _HF_BLANK;
 965    } else
 966       f &= ~_HF_BLANK;
 967    f &= ~_HF_NL_MASK;
 968    self->hf_flags = (f |= _HF_ANY);
 969    self = _hf_store(self, c);
 970 jleave:
 971    NYD2_LEAVE;
 972    return self;
 973 }
 974
 975 static struct htmlflt *
 976 _hf_putc_premode(struct htmlflt *self, char c)
 977 {
 978    ui32_t f;
 979    NYD2_ENTER;
 980
 981    if ((f = self->hf_flags) & _HF_ERROR) {
 982       ;
 983    } else if (c == '\n')
 984       self = _hf_nl_force(self);
 985    else {
 986       f &= ~_HF_NL_MASK;
 987       self->hf_flags = (f |= _HF_ANY);
 988       self = _hf_store(self, c);
 989    }
 990    NYD2_LEAVE;
 991    return self;
 992 }
 993
 994 static struct htmlflt *
 995 _hf_puts(struct htmlflt *self, char const *cp)
 996 {
 997    char c;
 998    NYD2_ENTER;
 999
1000    while ((c = *cp++) != '\0')
1001       self = _hf_putc(self, c);
1002    NYD2_LEAVE;
1003    return self;
1004 }
1005
1006 static struct htmlflt *
1007 _hf_putbuf(struct htmlflt *self, char const *cp, size_t len)
1008 {
1009    NYD2_ENTER;
1010
1011    while (len-- > 0)
1012       self = _hf_putc(self, *cp++);
1013    NYD2_LEAVE;
1014    return self;
1015 }
1016
1017 static struct htmlflt *
1018 _hf_param(struct htmlflt *self, struct str *store, char const *param)
1019 {
1020    char const *cp;
1021    char c, x, quote;
1022    size_t i;
1023    bool_t hot;
1024    NYD2_ENTER;
1025
1026    store->s = NULL;
1027    store->l = 0;
1028    cp = self->hf_bdat;
1029
1030    /* Skip over any non-WS first; be aware of soup, if it slipped through */
1031    for(;;){
1032       if((c = *cp++) == '\0' || c == '>')
1033          goto jleave;
1034       if(whitechar(c))
1035          break;
1036    }
1037
1038    /* Search for the parameter, take care of other quoting along the way */
1039    x = *param++;
1040    x = upperconv(x);
1041    quote = '\0';
1042    i = strlen(param);
1043
1044    for(hot = TRU1;;){
1045       if((c = *cp++) == '\0' || c == '>')
1046          goto jleave;
1047       if(whitechar(c)){
1048          hot = TRU1;
1049          continue;
1050       }
1051
1052       /* Could it be a parameter? */
1053       if(hot){
1054          hot = FAL0;
1055
1056          /* Is it the desired one? */
1057          if((c = upperconv(c)) == x && !ascncasecmp(param, cp, i)){
1058             char const *cp2 = cp + i;
1059
1060             if((quote = *cp2++) != '='){
1061                if(quote == '\0' || quote == '>')
1062                   goto jleave;
1063                while(whitechar(quote))
1064                   quote = *cp2++;
1065             }
1066             if(quote == '='){
1067                cp = cp2;
1068                break;
1069             }
1070             continue; /* XXX Optimize: i bytes or even cp2 can't be it! */
1071          }
1072       }
1073
1074       /* Not the desired one; but a parameter? */
1075       if(c != '=')
1076          continue;
1077       /* If so, properly skip over the value */
1078       if(c == '"' || c == '\''){
1079          /* TODO oops i have forgotten whether backslash quoting is allowed in
1080           * TODO quoted HTML parameter values?  not supporting that for now.. */
1081          for(quote = c; (c = *cp) != '\0' && c != quote; ++cp)
1082             ;
1083       }else
1084          while((c = *cp) != '\0' && !whitechar(c) && c != '>')
1085             ++cp;
1086       if(c == '\0')
1087          goto jleave;
1088    }
1089
1090    /* Skip further whitespace */
1091    for(;;){
1092       if((c = *cp++) == '\0' || c == '>')
1093          goto jleave;
1094       if(!whitechar(c))
1095          break;
1096    }
1097
1098    if(c == '"' || c == '\''){
1099       /* TODO oops i have forgotten whether backslash quoting is allowed in
1100        * TODO quoted HTML parameter values?  not supporting that for now.. */
1101       store->s = UNCONST(cp);
1102       for(quote = c; (c = *cp) != '\0' && c != quote; ++cp)
1103          ;
1104       /* XXX ... and we simply ignore a missing trailing " :> */
1105    }else{
1106       store->s = UNCONST(cp - 1);
1107       if(!whitechar(c))
1108          while((c = *cp) != '\0' && !whitechar(c) && c != '>')
1109             ++cp;
1110    }
1111    i = PTR2SIZE(cp - store->s);
1112
1113    /* Terrible tagsoup out there, e.g., groups.google.com produces href=""
1114     * parameter values prefixed and suffixed by newlines!  Therefore trim the
1115     * value content TODO join into the parse step above! */
1116    for (cp = store->s; i > 0 && spacechar(*cp); ++cp, --i)
1117       ;
1118    store->s = UNCONST(cp);
1119    for (cp += i - 1; i > 0 && spacechar(*cp); --cp, --i)
1120       ;
1121    if ((store->l = i) == 0)
1122       store->s = NULL;
1123 jleave:
1124    NYD2_LEAVE;
1125    return self;
1126 }
1127
1128 static struct htmlflt *
1129 _hf_expand_all_ents(struct htmlflt *self, struct str const *param)
1130 {
1131    char const *cp, *maxcp, *ep;
1132    char c;
1133    size_t i;
1134    NYD2_ENTER;
1135
1136    for (cp = param->s, maxcp = cp + param->l; cp < maxcp;)
1137       if ((c = *cp++) != '&')
1138 jputc:
1139          self = _hf_putc(self, c);
1140       else {
1141          for (ep = cp--;;) {
1142             if (ep == maxcp || (c = *ep++) == '\0') {
1143                for (; cp < ep; ++cp)
1144                   self = _hf_putc(self, *cp);
1145                goto jleave;
1146             } else if (c == ';') {
1147                if ((i = PTR2SIZE(ep - cp)) > 1) {
1148                   self = _hf_check_ent(self, cp, i);
1149                   break;
1150                } else {
1151                   c = *cp++;
1152                   goto jputc;
1153                }
1154             }
1155          }
1156          cp = ep;
1157       }
1158 jleave:
1159    NYD2_LEAVE;
1160    return self;
1161 }
1162
1163 static struct htmlflt *
1164 _hf_check_tag(struct htmlflt *self, char const *s)
1165 {
1166    char nobuf[32], c;
1167    struct str param;
1168    size_t i;
1169    struct htmlflt_tag const *hftp;
1170    ui32_t f;
1171    NYD2_ENTER;
1172
1173    /* Extra check only */
1174    assert(s != NULL);
1175    if (*s != '<') {
1176       DBG( n_alert("HTML tagsoup filter _hf_check_tag() called on soup!"); )
1177 jput_as_is:
1178       self = _hf_puts(self, self->hf_bdat);
1179       goto jleave;
1180    }
1181
1182    for (++s, i = 0; (c = s[i]) != '\0' && c != '>' && !whitechar(c); ++i)
1183       /* Special massage for things like <br/>: after the slash only whitespace
1184        * may separate us from the closing right angle! */
1185       if (c == '/') {
1186          size_t j = i + 1;
1187
1188          while ((c = s[j]) != '\0' && c != '>' && whitechar(c))
1189             ++j;
1190          if (c == '>')
1191             break;
1192       }
1193
1194    for (hftp = _hf_tags;;) {
1195       if (i == hftp->hft_len && !ascncasecmp(s, hftp->hft_tag, i)) {
1196          c = s[hftp->hft_len];
1197          if (c == '>' || c == '/' || whitechar(c))
1198             break;
1199       }
1200       if (PTRCMP(++hftp, >=, _hf_tags + NELEM(_hf_tags)))
1201          goto jnotknown;
1202    }
1203
1204    f = self->hf_flags;
1205    switch (hftp->hft_act) {
1206    case _HFSA_PRE_END:
1207       f &= ~_HF_PRE;
1208       if (0) {
1209          /* FALLTHRU */
1210    case _HFSA_PRE:
1211          f |= _HF_PRE;
1212       }
1213       self->hf_flags = f;
1214       /* FALLTHRU */
1215
1216    case _HFSA_NEEDSEP:
1217       if (!(self->hf_flags & _HF_NL_2))
1218          self = _hf_nl(self);
1219       /* FALLTHRU */
1220    case _HFSA_NEEDNL:
1221       if (!(f & _HF_NL_1))
1222          self = _hf_nl(self);
1223       if (hftp->hft_injc != '\0') {
1224          self = _hf_putc(self, hftp->hft_injc & 0x7F);
1225          if ((uc_i)hftp->hft_injc & 0x80)
1226             self = _hf_putc(self, ' ');
1227       }
1228       break;
1229
1230    case _HFSA_IGN:
1231       self->hf_ign_tag = hftp;
1232       self->hf_flags = (f |= _HF_IGN | _HF_NOPUT);
1233       break;
1234
1235    case _HFSA_IMG:
1236       self = _hf_param(self, &param, "alt");
1237       self = _hf_putc(self, '[');
1238       if (param.s == NULL) {
1239          param.s = UNCONST("IMG");
1240          param.l = 3;
1241          goto jimg_put;
1242       } /* else */ if (memchr(param.s, '&', param.l) != NULL)
1243          self = _hf_expand_all_ents(self, &param);
1244       else
1245 jimg_put:
1246          self = _hf_putbuf(self, param.s, param.l);
1247       self = _hf_putc(self, ']');
1248       break;
1249
1250    case _HFSA_HREF:
1251       self = _hf_param(self, &param, "href");
1252       /* Ignore non-external links */
1253       if (param.s != NULL && *param.s != '#') {
1254          struct htmlflt_href *hhp = smalloc(sizeof(*hhp) -
1255                VFIELD_SIZEOF(struct htmlflt_href, hfh_dat) + param.l +1);
1256
1257          hhp->hfh_next = self->hf_hrefs;
1258          hhp->hfh_no = ++self->hf_href_no;
1259          hhp->hfh_len = (ui32_t)param.l;
1260          memcpy(hhp->hfh_dat, param.s, param.l);
1261
1262          snprintf(nobuf, sizeof nobuf, "[%u]", hhp->hfh_no);
1263          self->hf_flags = (f |= _HF_HREF);
1264          self->hf_hrefs = hhp;
1265          self = _hf_puts(self, nobuf);
1266       } else
1267          self->hf_flags = (f &= ~_HF_HREF);
1268       break;
1269    case _HFSA_HREF_END:
1270       if (f & _HF_HREF) {
1271          snprintf(nobuf, sizeof nobuf, "[/%u]", self->hf_href_no);
1272          self = _hf_puts(self, nobuf);
1273       }
1274       break;
1275
1276    default:
1277       c = (char)(hftp->hft_act & 0xFF);
1278       self = _hf_putc(self, c);
1279       break;
1280    case '\0':
1281       break;
1282    }
1283 jleave:
1284    NYD2_LEAVE;
1285    return self;
1286
1287    /* The problem is that even invalid tagsoup is widely used, without real
1288     * searching i have seen e-mail address in <N@H.D> notation, and more.
1289     * To protect us a bit look around and possibly write the content as such */
1290 jnotknown:
1291    switch (*s) {
1292    case '!':
1293    case '?':
1294       /* Ignore <!DOCTYPE, <!-- comments, <? PIs.. */
1295       goto jleave;
1296    case '>':
1297       /* Print out an empty tag as such */
1298       if (s[1] == '\0') {
1299          --s;
1300          goto jput_as_is;
1301       }
1302       break;
1303    case '/':
1304       ++s;
1305       break;
1306    default:
1307       break;
1308    }
1309
1310    /* Also skip over : in order to suppress v:roundrect, w:anchorlock.. */
1311    while ((c = *s++) != '\0' && c != '>' && !whitechar(c) && c != ':')
1312       if (!asciichar(c) || punctchar(c)) {
1313          self = _hf_puts(self, self->hf_bdat);
1314          break;
1315       }
1316    goto jleave;
1317 }
1318
1319 static struct htmlflt *
1320 _hf_check_ent(struct htmlflt *self, char const *s, size_t l)
1321 {
1322    char nobuf[32];
1323    char const *s_save;
1324    size_t l_save;
1325    struct hf_ent const *hfep;
1326    long i;
1327    NYD2_ENTER;
1328
1329    s_save = s;
1330    l_save = l;
1331    assert(*s == '&');
1332    assert(l > 0);
1333    /* False entities seen in the wild assert(s[l - 1] == ';'); */
1334    ++s;
1335    l -= 2;
1336
1337    /* Numeric entity, or try named search */
1338    if (*s == '#') {
1339       i = (*++s == 'x' ? 16 : 10);
1340
1341       if ((i != 16 || (++s, --l) > 0) && l < sizeof(nobuf)) {
1342          memcpy(nobuf, s, l);
1343          nobuf[l] = '\0';
1344          i = strtol(nobuf, NULL, i);
1345          if (i <= 0x7F)
1346             self = _hf_putc(self, (char)i);
1347          else if (self->hf_flags & _HF_UTF8) {
1348 jputuni:
1349             l = n_utf32_to_utf8((ui32_t)i, nobuf);
1350             self = _hf_putbuf(self, nobuf, l);
1351          } else
1352             goto jeent;
1353       } else
1354          goto jeent;
1355    } else {
1356       ui32_t f = self->hf_flags, hf;
1357
1358       for (hfep = _hf_ents; PTRCMP(hfep, <, _hf_ents + NELEM(_hf_ents)); ++hfep)
1359          if (l == ((hf = hfep->hfe_flags) & _HFE_LENGTH_MASK) &&
1360                !strncmp(s, hfep->hfe_ent, l)) {
1361             if ((hf & _HFE_HAVE_UNI) && (f & _HF_UTF8)) {
1362                i = hfep->hfe_uni;
1363                goto jputuni;
1364             } else if (hf & _HFE_HAVE_CSTR)
1365                self = _hf_puts(self, hfep->hfe_cstr);
1366             else
1367                self = _hf_putc(self, hfep->hfe_c);
1368             goto jleave;
1369          }
1370 jeent:
1371       self = _hf_putbuf(self, s_save, l_save);
1372    }
1373 jleave:
1374    NYD2_LEAVE;
1375    return self;
1376 }
1377
1378 static ssize_t
1379 _hf_add_data(struct htmlflt *self, char const *dat, size_t len)
1380 {
1381    char c, *cp, *cp_max;
1382    bool_t hot;
1383    ssize_t rv = 0;
1384    NYD_ENTER;
1385
1386    /* Final put request? */
1387    if (dat == NULL) {
1388       if (self->hf_len > 0 || self->hf_hrefs != NULL) {
1389          self = _hf_dump(self);
1390          if (self->hf_hrefs != NULL)
1391             self = _hf_dump_hrefs(self);
1392          rv = 1;
1393       }
1394       goto jleave;
1395    }
1396
1397    /* Always ensure some initial buffer */
1398    if ((cp = self->hf_curr) != NULL)
1399       cp_max = self->hf_bmax;
1400    else {
1401       cp = self->hf_curr = self->hf_bdat = smalloc(LINESIZE);
1402       cp_max = self->hf_bmax = cp + LINESIZE -1; /* (Always room for NUL!) */
1403    }
1404    hot = (cp != self->hf_bdat);
1405
1406    for (rv = (ssize_t)len; len > 0; --len) {
1407       ui32_t f = self->hf_flags;
1408
1409       if (f & _HF_ERROR)
1410          break;
1411       c = *dat++;
1412
1413       /* Soup is really weird, and scripts may contain almost anything (and
1414        * newer CSS standards are also cryptic): therefore prefix the _HF_IGN
1415        * test and walk until we see the required end tag */
1416       /* TODO For real safety _HF_IGN soup condome would also need to know
1417        * TODO about quoted strings so that 'var i = "</script>";' couldn't
1418        * TODO fool it!   We really want this mode also for _HF_NOPUT to be
1419        * TODO able to *gracefully* detect the tag-closing '>', but then if
1420        * TODO that is a single mechanism we should have made it! */
1421       if (f & _HF_IGN) {
1422          struct htmlflt_tag const *hftp = self->hf_ign_tag;
1423          size_t i;
1424
1425          if (c == '<') {
1426             hot = TRU1;
1427 jcp_reset:
1428             cp = self->hf_bdat;
1429          } else if (c == '>') {
1430             if (hot) {
1431                if ((i = PTR2SIZE(cp - self->hf_bdat)) > 1 &&
1432                      --i == hftp->hft_len &&
1433                      !ascncasecmp(self->hf_bdat + 1, hftp->hft_tag, i))
1434                   self->hf_flags = (f &= ~(_HF_IGN | _HF_NOPUT));
1435                hot = FAL0;
1436                goto jcp_reset;
1437             }
1438          } else if (hot) {
1439             *cp++ = c;
1440             i = PTR2SIZE(cp - self->hf_bdat);
1441             if ((i == 1 && c != '/') || --i > hftp->hft_len) {
1442                hot = FAL0;
1443                goto jcp_reset;
1444             }
1445          }
1446       } else switch (c) {
1447       case '<':
1448          /* People are using & without &amp;ing it, ditto <; be aware */
1449          if (f & (_HF_NOPUT | _HF_ENT)) {
1450             f &= ~_HF_ENT;
1451             /* Special case "<!--" buffer content to deal with really weird
1452              * things that can be done with "<!--[if gte mso 9]>" syntax */
1453             if (PTR2SIZE(cp - self->hf_bdat) != 4 ||
1454                   memcmp(self->hf_bdat, "<!--", 4)) {
1455                self->hf_flags = f;
1456                *cp = '\0';
1457                self = _hf_puts(self, self->hf_bdat);
1458                f = self->hf_flags;
1459             }
1460          }
1461          cp = self->hf_bdat;
1462          *cp++ = c;
1463          self->hf_flags = (f |= _HF_NOPUT);
1464          break;
1465       case '>':
1466          /* Weird tagsoup around, do we actually parse a tag? */
1467          if (!(f & _HF_NOPUT))
1468             goto jdo_c;
1469          cp[0] = c;
1470          cp[1] = '\0';
1471          f &= ~(_HF_NOPUT | _HF_ENT);
1472          self->hf_flags = f;
1473          self = _hf_check_tag(self, self->hf_bdat);
1474          *(cp = self->hf_bdat) = '\0'; /* xxx extra safety */
1475          /* Quick hack to get rid of redundant newline after <pre> XXX */
1476          if (!(f & _HF_PRE) && (self->hf_flags & _HF_PRE) &&
1477                len > 1 && *dat == '\n')
1478             ++dat, --len;
1479          break;
1480
1481       case '\r': /* TODO CR should be stripped in lower level!! (Only B64!?!) */
1482          break;
1483       case '\n':
1484          /* End of line is not considered unless we are in PRE section.
1485           * However, in _HF_NOPUT mode we must be aware of tagsoup which uses
1486           * newlines for separating parameters */
1487          if (f & _HF_NOPUT)
1488             goto jdo_c;
1489          self = (f & _HF_PRE) ? _hf_nl_force(self) : _hf_putc(self, ' ');
1490          break;
1491
1492       case '\t':
1493          if (!(f & _HF_PRE))
1494             c = ' ';
1495          /* FALLTHRU */
1496       default:
1497 jdo_c:
1498          /* If not currently parsing a tag and bypassing normal output.. */
1499          if (!(f & _HF_NOPUT)) {
1500             if (cntrlchar(c))
1501                break;
1502             if (c == '&') {
1503                cp = self->hf_bdat;
1504                *cp++ = c;
1505                self->hf_flags = (f |= _HF_NOPUT | _HF_ENT);
1506             } else if (f & _HF_PRE) {
1507                self = _hf_putc_premode(self, c);
1508                self->hf_flags &= ~_HF_BLANK;
1509             } else
1510               self = _hf_putc(self, c);
1511          } else if ((f & _HF_ENT) && c == ';') {
1512             cp[0] = c;
1513             cp[1] = '\0';
1514             f &= ~(_HF_NOPUT | _HF_ENT);
1515             self->hf_flags = f;
1516            self = _hf_check_ent(self, self->hf_bdat,
1517                PTR2SIZE(cp + 1 - self->hf_bdat));
1518          } else {
1519             /* We may need to grow the buffer */
1520             if (PTRCMP(cp + 42/2, >=, cp_max)) {
1521                size_t i = PTR2SIZE(cp - self->hf_bdat),
1522                   m = PTR2SIZE(self->hf_bmax - self->hf_bdat) + LINESIZE;
1523
1524                cp = self->hf_bdat = srealloc(self->hf_bdat, m);
1525                self->hf_bmax = cp + m -1;
1526                self->hf_curr = (cp += i);
1527             }
1528             *cp++ = c;
1529          }
1530       }
1531    }
1532    self->hf_curr = cp;
1533 jleave:
1534   NYD_LEAVE;
1535   return (self->hf_flags & _HF_ERROR) ? -1 : rv;
1536 }
1537
1538 /*
1539  * TODO Because we don't support filter chains yet this filter will be run
1540  * TODO in a dedicated subprocess, driven via a special Popen() mode
1541  */
1542 static bool_t __hf_hadpipesig;
1543 static void
1544 __hf_onpipe(int signo)
1545 {
1546    NYD_X; /* Signal handler */
1547    UNUSED(signo);
1548    __hf_hadpipesig = TRU1;
1549 }
1550
1551 FL int
1552 htmlflt_process_main(void)
1553 {
1554    char buf[BUFFER_SIZE];
1555    struct htmlflt hf;
1556    size_t i;
1557    int rv;
1558    NYD_ENTER;
1559
1560    __hf_hadpipesig = FAL0;
1561    safe_signal(SIGPIPE, &__hf_onpipe);
1562
1563    htmlflt_init(&hf);
1564    htmlflt_reset(&hf, stdout);
1565
1566    for (;;) {
1567       if ((i = fread(buf, sizeof(buf[0]), NELEM(buf), stdin)) == 0) {
1568          rv = !feof(stdin);
1569          break;
1570       }
1571
1572       if ((rv = __hf_hadpipesig))
1573          break;
1574       /* Just use this directly.. */
1575       if (htmlflt_push(&hf, buf, i) < 0) {
1576          rv = 1;
1577          break;
1578       }
1579    }
1580    if (rv == 0 && htmlflt_flush(&hf) < 0)
1581       rv = 1;
1582
1583    htmlflt_destroy(&hf);
1584
1585    rv |= __hf_hadpipesig;
1586    NYD_LEAVE;
1587    return rv;
1588 }
1589
1590 FL void
1591 htmlflt_init(struct htmlflt *self)
1592 {
1593    NYD_ENTER;
1594    /* (Rather redundant though) */
1595    memset(self, 0, sizeof *self);
1596    NYD_LEAVE;
1597 }
1598
1599 FL void
1600 htmlflt_destroy(struct htmlflt *self)
1601 {
1602    NYD_ENTER;
1603    htmlflt_reset(self, NULL);
1604    NYD_LEAVE;
1605 }
1606
1607 FL void
1608 htmlflt_reset(struct htmlflt *self, FILE *f)
1609 {
1610    struct htmlflt_href *hfhp;
1611    NYD_ENTER;
1612
1613    while ((hfhp = self->hf_hrefs) != NULL) {
1614       self->hf_hrefs = hfhp->hfh_next;
1615       free(hfhp);
1616    }
1617
1618    if (self->hf_bdat != NULL)
1619       free(self->hf_bdat);
1620    if (self->hf_line != NULL)
1621       free(self->hf_line);
1622
1623    memset(self, 0, sizeof *self);
1624
1625    if (f != NULL) {
1626       ui32_t sw = MAX(_HF_MINLEN, (ui32_t)scrnwidth);
1627
1628       self->hf_line = smalloc((size_t)sw * mb_cur_max +1);
1629       self->hf_lmax = sw;
1630
1631       if (options & OPT_UNICODE) /* TODO not truly generic */
1632          self->hf_flags = _HF_UTF8;
1633       self->hf_os = f;
1634    }
1635    NYD_LEAVE;
1636 }
1637
1638 FL ssize_t
1639 htmlflt_push(struct htmlflt *self, char const *dat, size_t len)
1640 {
1641    ssize_t rv;
1642    NYD_ENTER;
1643
1644    rv = _hf_add_data(self, dat, len);
1645    NYD_LEAVE;
1646    return rv;
1647 }
1648
1649 FL ssize_t
1650 htmlflt_flush(struct htmlflt *self)
1651 {
1652    ssize_t rv;
1653    NYD_ENTER;
1654
1655    rv = _hf_add_data(self, NULL, 0);
1656    rv |= !fflush(self->hf_os) ? 0 : -1;
1657    NYD_LEAVE;
1658    return rv;
1659 }
1660 #endif /* HAVE_FILTER_HTML_TAGSOUP */
1661
1662 /* s-it-mode */