filter.c

   1 /*@ S-nail - a mail user agent derived from Berkeley Mail.
   2  *@ Filter objects.
   3  *
   4  * Copyright (c) 2013 - 2017 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 n_CTAV(n_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 + n_QUOTE_TAB_SPACES) & ~(n_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 && n_QUOTE_IS_A(wc)) {
 216          self->qf_wscnt = 0;
 217          if (self->qf_currq.l >= n_QUOTE_MAX - 3) {
 218             self->qf_currq.s[n_QUOTE_MAX - 3] = '.';
 219             self->qf_currq.s[n_QUOTE_MAX - 2] = '.';
 220             self->qf_currq.s[n_QUOTE_MAX - 1] = '.';
 221             self->qf_currq.l = n_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 < n_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 const *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*: n_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;
 324
 325       /* These magic values ensure we don't bail */
 326       n_idec_ui32_cp(&qmax, cp, 10, &xcp);
 327       if (qmax < self->qf_pfix_len + 6)
 328          qmax = self->qf_pfix_len + 6;
 329       --qmax; /* The newline escape */
 330       if (cp == xcp || *xcp == '\0')
 331          qmin = (qmax >> 1) + (qmax >> 2) + (qmax >> 5);
 332       else {
 333          n_idec_ui32_cp(&qmin, &xcp[1], 10, NULL);
 334          if (qmin < qmax >> 1)
 335             qmin = qmax >> 1;
 336          else if (qmin > qmax - 2)
 337             qmin = qmax - 2;
 338       }
 339       self->qf_qfold_min = qmin;
 340       self->qf_qfold_max = qmax;
 341
 342       /* Add pad for takeover copies, reverse solidus and newline */
 343       self->qf_dat.s = salloc((qmax + 3) * n_mb_cur_max);
 344       self->qf_currq.s = salloc((n_QUOTE_MAX + 1) * n_mb_cur_max);
 345    }
 346 #endif
 347    NYD_LEAVE;
 348 }
 349
 350 FL void
 351 quoteflt_destroy(struct quoteflt *self) /* xxx inline */
 352 {
 353    NYD_ENTER;
 354    n_UNUSED(self);
 355    NYD_LEAVE;
 356 }
 357
 358 FL void
 359 quoteflt_reset(struct quoteflt *self, FILE *f) /* xxx inline */
 360 {
 361    NYD_ENTER;
 362    self->qf_os = f;
 363 #ifdef HAVE_QUOTE_FOLD
 364    self->qf_state = _QF_CLEAN;
 365    self->qf_dat.l =
 366    self->qf_currq.l = 0;
 367    memset(self->qf_mbps, 0, sizeof self->qf_mbps);
 368 #endif
 369    NYD_LEAVE;
 370 }
 371
 372 FL ssize_t
 373 quoteflt_push(struct quoteflt *self, char const *dat, size_t len)
 374 {
 375    /* (xxx Ideally the actual push() [and flush()] would be functions on their
 376     * xxx own, via indirect vtbl call ..) */
 377    ssize_t rv = 0;
 378    NYD_ENTER;
 379
 380    self->qf_nl_last = (len > 0 && dat[len - 1] == '\n'); /* TODO HACK */
 381
 382    if (len == 0)
 383       goto jleave;
 384
 385    /* Bypass? TODO Finally, this filter simply should not be used, then
 386     * (TODO It supercedes prefix_write() or something) */
 387    if (self->qf_pfix_len == 0) {
 388       if (len != fwrite(dat, 1, len, self->qf_os))
 389          goto jerr;
 390       rv = len;
 391    }
 392    /* Normal: place *indentprefix* at every BOL */
 393    else
 394 #ifdef HAVE_QUOTE_FOLD
 395       if (self->qf_qfold_max == 0)
 396 #endif
 397    {
 398       void *vp;
 399       size_t ll;
 400       bool_t pxok = (self->qf_qfold_min != 0);
 401
 402       for (;;) {
 403          if (!pxok) {
 404             ll = self->qf_pfix_len;
 405             if (ll != fwrite(self->qf_pfix, 1, ll, self->qf_os))
 406                goto jerr;
 407             rv += ll;
 408             pxok = TRU1;
 409          }
 410
 411          /* xxx Strictly speaking this is invalid, because only `/' and `.' are
 412           * xxx mandated by POSIX.1-2008 as "invariant across all locales
 413           * xxx supported"; though there is no charset known which uses this
 414           * xxx control char as part of a multibyte character; note that S-nail
 415           * XXX (and the Mail codebase as such) do not support EBCDIC */
 416          if ((vp = memchr(dat, '\n', len)) == NULL)
 417             ll = len;
 418          else {
 419             pxok = FAL0;
 420             ll = PTR2SIZE((char*)vp - dat) + 1;
 421          }
 422
 423          if (ll != fwrite(dat, sizeof *dat, ll, self->qf_os))
 424             goto jerr;
 425          rv += ll;
 426          if ((len -= ll) == 0)
 427             break;
 428          dat += ll;
 429       }
 430
 431       self->qf_qfold_min = pxok;
 432    }
 433    /* Overly complicated, though still only line-per-line: *quote-fold*.
 434     * - If .qf_currq.l is 0, then we are in a clean state.  Reset .qf_mbps;
 435     *   TODO note this means we assume that lines start with reset escape seq,
 436     *   TODO but i don't think this is any worse than what we currently do;
 437     *   TODO in 15.0, with the value carrier, we should carry conversion states
 438     *   TODO all along, only resetting on error (or at words for header =???=);
 439     *   TODO this still is weird for error handling, but we need to act more
 440     *   TODO stream-alike (though in practice i don't think cross-line states
 441     *   TODO can be found, because of compatibility reasons; however, being
 442     *   TODO a problem rather than a solution is not a good thing (tm))
 443     * - Lookout for a newline */
 444 #ifdef HAVE_QUOTE_FOLD
 445    else {
 446       struct qf_vc vc;
 447       ssize_t i;
 448
 449       vc.self = self;
 450       vc.buf = dat;
 451       vc.len = len;
 452       while (vc.len > 0) {
 453          switch (self->qf_state) {
 454          case _QF_CLEAN:
 455          case _QF_PREFIX:
 456             i = _qf_state_prefix(&vc);
 457             break;
 458          default: /* silence cc (`i' unused) */
 459          case _QF_DATA:
 460             i = _qf_state_data(&vc);
 461             break;
 462          }
 463          if (i < 0)
 464             goto jerr;
 465          rv += i;
 466       }
 467    }
 468 #endif /* HAVE_QUOTE_FOLD */
 469
 470 jleave:
 471    NYD_LEAVE;
 472    return rv;
 473 jerr:
 474    rv = -1;
 475    goto jleave;
 476 }
 477
 478 FL ssize_t
 479 quoteflt_flush(struct quoteflt *self)
 480 {
 481    ssize_t rv = 0;
 482    NYD_ENTER;
 483    n_UNUSED(self);
 484
 485 #ifdef HAVE_QUOTE_FOLD
 486    if (self->qf_dat.l > 0) {
 487       rv = _qf_dump_prefix(self);
 488       if (rv >= 0) {
 489          size_t i = self->qf_dat.l;
 490          if (i == fwrite(self->qf_dat.s, 1, i, self->qf_os))
 491             rv += i;
 492          else
 493             rv = -1;
 494          self->qf_dat.l = 0;
 495          self->qf_brk_isws = FAL0;
 496          self->qf_wscnt = self->qf_brkl = self->qf_brkw = 0;
 497          self->qf_datw = self->qf_pfix_len + self->qf_currq.l;
 498       }
 499    }
 500 #endif
 501    NYD_LEAVE;
 502    return rv;
 503 }
 504
 505 /*
 506  * HTML tagsoup filter TODO rewrite wchar_t based (require HAVE_C90AMEND1)
 507  * TODO . Numeric &#NO; entities should also be treated by struct hf_ent
 508  * TODO . Yes, we COULD support CSS based quoting when we'd check type="quote"
 509  * TODO   (nonstandard) and watch out for style="gmail_quote" (or so, VERY
 510  * TODO   nonstandard) and tracking a stack of such elements (to be popped
 511  * TODO   once the closing element is seen).  Then, after writing a newline,
 512  * TODO   place sizeof(stack) ">"s first.  But aren't these HTML mails rude?
 513  * TODO Interlocking and non-well-formed data will break us down
 514  */
 515 #ifdef HAVE_FILTER_HTML_TAGSOUP
 516
 517 enum hf_limits {
 518    _HF_MINLEN  = 10,       /* Minimum line length (can't really be smaller) */
 519    _HF_BRKSUB  = 8         /* Start considering line break MAX - BRKSUB */
 520 };
 521
 522 enum hf_flags {
 523    _HF_UTF8    = 1<<0,     /* Data is in UTF-8 */
 524    _HF_ERROR   = 1<<1,     /* A hard error occurred, bail as soon as possible */
 525    _HF_NOPUT   = 1<<2,     /* (In a tag,) Don't generate output */
 526    _HF_IGN     = 1<<3,     /* Ignore mode on */
 527    _HF_ANY     = 1<<4,     /* Yet seen just any output */
 528    _HF_PRE     = 1<<5,     /* In <pre>formatted mode */
 529    _HF_ENT     = 1<<6,     /* Currently parsing an entity */
 530    _HF_BLANK   = 1<<7,     /* Whitespace last */
 531    _HF_HREF    = 1<<8,     /* External <a href=> was the last href seen */
 532
 533    _HF_NL_1    = 1<<9,     /* One \n seen */
 534    _HF_NL_2    = 2<<9,     /* We have produced an all empty line */
 535    _HF_NL_MASK = _HF_NL_1 | _HF_NL_2
 536 };
 537
 538 enum hf_special_actions {
 539    _HFSA_NEEDSEP  = -1,    /* Need an empty line (paragraph separator) */
 540    _HFSA_NEEDNL   = -2,    /* Need a new line start (table row) */
 541    _HFSA_IGN      = -3,    /* Things like <style>..</style>, <script>.. */
 542    _HFSA_PRE      = -4,    /* <pre>.. */
 543    _HFSA_PRE_END  = -5,
 544    _HFSA_IMG      = -6,    /* <img> */
 545    _HFSA_HREF     = -7,    /* <a>.. */
 546    _HFSA_HREF_END = -8
 547 };
 548
 549 enum hf_entity_flags {
 550    _HFE_HAVE_UNI  = 1<<6,  /* Have a Unicode replacement character */
 551    _HFE_HAVE_CSTR = 1<<7,  /* Have a string replacement */
 552    /* We store the length of the entity name in the flags, too */
 553    _HFE_LENGTH_MASK = (1<<6) - 1
 554 };
 555
 556 struct htmlflt_href {
 557    struct htmlflt_href *hfh_next;
 558    ui32_t      hfh_no;     /* Running sequence */
 559    ui32_t      hfh_len;    /* of .hfh_dat */
 560    char        hfh_dat[n_VFIELD_SIZE(0)];
 561 };
 562
 563 struct htmlflt_tag {
 564    si32_t      hft_act;    /* char or hf_special_actions */
 565    /* Not NUL: character to inject, with high bit set: place a space
 566     * afterwards.  Note: only recognized with _HFSA_NEEDSEP or _HFSA_NEEDNL */
 567    char        hft_injc;
 568    ui8_t       hft_len;    /* Useful bytes in (NUL terminated) .hft_tag */
 569    char const  hft_tag[10]; /* Tag less < and > surroundings (TR, /TR, ..) */
 570 };
 571 n_CTA(n_SIZEOF_FIELD(struct htmlflt_tag, hft_tag) < LINESIZE,
 572    "Structure field too large a size"); /* .hf_ign_tag */
 573
 574 struct hf_ent {
 575    ui8_t       hfe_flags;  /* enum hf_entity_flags plus length of .hfe_ent */
 576    char        hfe_c;      /* Plain replacement character */
 577    ui16_t      hfe_uni;    /* Unicode codepoint if _HFE_HAVE_UNI */
 578    char        hfe_cstr[5]; /* _HFE_HAVE_CSTR (e.g., &hellip; -> ...) */
 579    char const  hfe_ent[7]; /* Entity less & and ; surroundings */
 580 };
 581
 582 /* Tag list; not binary searched :(, so try to take care a bit */
 583 static struct htmlflt_tag const  _hf_tags[] = {
 584 # undef _X
 585 # undef _XC
 586 # define _X(S,A)     {A, '\0', sizeof(S) -1, S "\0"}
 587 # define _XC(S,C,A)  {A, C, sizeof(S) -1, S "\0"}
 588
 589    _X("P", _HFSA_NEEDSEP),       _X("/P", _HFSA_NEEDNL),
 590    _X("DIV", _HFSA_NEEDSEP),     _X("/DIV", _HFSA_NEEDNL),
 591    _X("TR", _HFSA_NEEDNL),
 592                                  _X("/TH", '\t'),
 593                                  _X("/TD", '\t'),
 594    /* Let it stand out; also since we don't support implicit paragraphs after
 595     * block elements, plain running text after a list (seen in Unicode
 596     * announcement via Firefox) */
 597    _X("UL", _HFSA_NEEDSEP),      _X("/UL", _HFSA_NEEDSEP),
 598    _XC("LI", (char)0x80 | '*', _HFSA_NEEDSEP),
 599    _X("DL", _HFSA_NEEDSEP),
 600    _X("DT", _HFSA_NEEDNL),
 601
 602    _X("A", _HFSA_HREF),          _X("/A", _HFSA_HREF_END),
 603    _X("IMG", _HFSA_IMG),
 604    _X("BR", '\n'),
 605    _X("PRE", _HFSA_PRE),         _X("/PRE", _HFSA_PRE_END),
 606    _X("TITLE", _HFSA_NEEDSEP),   /*_X("/TITLE", '\n'),*/
 607    _X("H1", _HFSA_NEEDSEP),      /*_X("/H1", '\n'),*/
 608    _X("H2", _HFSA_NEEDSEP),      /*_X("/H2", '\n'),*/
 609    _X("H3", _HFSA_NEEDSEP),      /*_X("/H3", '\n'),*/
 610    _X("H4", _HFSA_NEEDSEP),      /*_X("/H4", '\n'),*/
 611    _X("H5", _HFSA_NEEDSEP),      /*_X("/H5", '\n'),*/
 612    _X("H6", _HFSA_NEEDSEP),      /*_X("/H6", '\n'),*/
 613
 614    _X("STYLE", _HFSA_IGN),
 615    _X("SCRIPT", _HFSA_IGN),
 616
 617 # undef _X
 618 };
 619
 620 /* Entity list; not binary searched.. */
 621 static struct hf_ent const       _hf_ents[] = {
 622 # undef _X
 623 # undef _XU
 624 # undef _XS
 625 # undef _XUS
 626 # define _X(E,C)     {(sizeof(E) -1), C, 0x0u, "", E "\0"}
 627 # define _XU(E,C,U)  {(sizeof(E) -1) | _HFE_HAVE_UNI, C, U, "", E "\0"}
 628 # define _XS(E,S)    {(sizeof(E) -1) | _HFE_HAVE_CSTR, '\0', 0x0u,S "\0",E "\0"}
 629 # define _XSU(E,S,U) \
 630    {(sizeof(E) -1) | _HFE_HAVE_UNI | _HFE_HAVE_CSTR, '\0', U, S "\0", E "\0"}
 631
 632    _X("quot", '"'),
 633    _X("amp", '&'),
 634    _X("lt", '<'),                _X("gt", '>'),
 635
 636    _XU("nbsp", ' ', 0x0020 /* Note: not 0x00A0 seems to be better for us */),
 637    _XU("middot", '.', 0x00B7),
 638    _XSU("hellip", "...", 0x2026),
 639    _XSU("mdash", "---", 0x2014), _XSU("ndash", "--", 0x2013),
 640    _XSU("laquo", "<<", 0x00AB),  _XSU("raquo", ">>", 0x00BB),
 641    _XSU("lsaquo", "<", 0x2039),  _XSU("rsaquo", ">", 0x203A),
 642    _XSU("lsquo", "'", 0x2018),   _XSU("rsquo", "'", 0x2019),
 643    _XSU("ldquo", "\"", 0x201C),  _XSU("rdquo", "\"", 0x201D),
 644    _XSU("uarr", "^|", 0x2191),   _XSU("darr", "|v", 0x2193),
 645
 646    _XSU("cent", "CENT", 0x00A2),
 647    _XSU("copy", "(C)", 0x00A9),
 648    _XSU("euro", "EUR", 0x20AC),
 649    _XSU("infin", "INFY", 0x221E),
 650    _XSU("pound", "GBP", 0x00A3),
 651    _XSU("reg", "(R)", 0x00AE),
 652    _XSU("sect", "S:", 0x00A7),
 653    _XSU("yen", "JPY", 0x00A5),
 654
 655    /* German umlauts */
 656    _XSU("Auml", "Ae", 0x00C4),   _XSU("auml", "ae", 0x00E4),
 657    _XSU("Ouml", "Oe", 0x00D6),   _XSU("ouml", "oe", 0x00F6),
 658    _XSU("Uuml", "Ue", 0x00DC),   _XSU("uuml", "ue", 0x00FC),
 659    _XSU("szlig", "ss", 0x00DF)
 660
 661 # undef _X
 662 # undef _XU
 663 # undef _XS
 664 # undef _XSU
 665 };
 666
 667 /* Real output */
 668 static struct htmlflt * _hf_dump_hrefs(struct htmlflt *self);
 669 static struct htmlflt * _hf_dump(struct htmlflt *self);
 670 static struct htmlflt * _hf_store(struct htmlflt *self, char c);
 671 # ifdef HAVE_NATCH_CHAR
 672 static struct htmlflt * __hf_sync_mbstuff(struct htmlflt *self);
 673 # endif
 674
 675 /* Virtual output */
 676 static struct htmlflt * _hf_nl(struct htmlflt *self);
 677 static struct htmlflt * _hf_nl_force(struct htmlflt *self);
 678 static struct htmlflt * _hf_putc(struct htmlflt *self, char c);
 679 static struct htmlflt * _hf_putc_premode(struct htmlflt *self, char c);
 680 static struct htmlflt * _hf_puts(struct htmlflt *self, char const *cp);
 681 static struct htmlflt * _hf_putbuf(struct htmlflt *self,
 682                            char const *cp, size_t len);
 683
 684 /* Try to locate a param'eter in >hf_bdat, store it (non-terminated!) or NULL */
 685 static struct htmlflt * _hf_param(struct htmlflt *self, struct str *store,
 686                            char const *param);
 687
 688 /* Expand all entities in the given parameter */
 689 static struct htmlflt * _hf_expand_all_ents(struct htmlflt *self,
 690                            struct str const *param);
 691
 692 /* Completely parsed over a tag / an entity, interpret that */
 693 static struct htmlflt * _hf_check_tag(struct htmlflt *self, char const *s);
 694 static struct htmlflt * _hf_check_ent(struct htmlflt *self, char const *s,
 695                            size_t l);
 696
 697 /* Input handler */
 698 static ssize_t          _hf_add_data(struct htmlflt *self,
 699                            char const *dat, size_t len);
 700
 701 static struct htmlflt *
 702 _hf_dump_hrefs(struct htmlflt *self)
 703 {
 704    struct htmlflt_href *hhp;
 705    NYD2_ENTER;
 706
 707    if (!(self->hf_flags & _HF_NL_2) && putc('\n', self->hf_os) == EOF) {
 708       self->hf_flags |= _HF_ERROR;
 709       goto jleave;
 710    }
 711
 712    /* Reverse the list */
 713    for (hhp = self->hf_hrefs, self->hf_hrefs = NULL; hhp != NULL;) {
 714       struct htmlflt_href *tmp = hhp->hfh_next;
 715       hhp->hfh_next = self->hf_hrefs;
 716       self->hf_hrefs = hhp;
 717       hhp = tmp;
 718    }
 719
 720    /* Then dump it */
 721    while ((hhp = self->hf_hrefs) != NULL) {
 722       self->hf_hrefs = hhp->hfh_next;
 723
 724       if (!(self->hf_flags & _HF_ERROR)) {
 725          int w = fprintf(self->hf_os, "  [%u] %.*s\n",
 726                hhp->hfh_no, (int)hhp->hfh_len, hhp->hfh_dat);
 727          if (w < 0)
 728             self->hf_flags |= _HF_ERROR;
 729       }
 730       free(hhp);
 731    }
 732
 733    self->hf_flags |= (putc('\n', self->hf_os) == EOF)
 734          ?  _HF_ERROR : _HF_NL_1 | _HF_NL_2;
 735    self->hf_href_dist = (ui32_t)n_realscreenheight >> 1;
 736 jleave:
 737    NYD2_LEAVE;
 738    return self;
 739 }
 740
 741 static struct htmlflt *
 742 _hf_dump(struct htmlflt *self)
 743 {
 744    ui32_t f, l;
 745    char c, *cp;
 746    NYD2_ENTER;
 747
 748    f = self->hf_flags & ~_HF_BLANK;
 749    l = self->hf_len;
 750    cp = self->hf_line;
 751    self->hf_mbwidth = self->hf_mboff = self->hf_last_ws = self->hf_len = 0;
 752
 753    for (c = '\0'; l > 0; --l) {
 754       c = *cp++;
 755 jput:
 756       if (putc(c, self->hf_os) == EOF) {
 757          self->hf_flags = (f |= _HF_ERROR);
 758          goto jleave;
 759       }
 760    }
 761
 762    if (c != '\n') {
 763       f |= (f & _HF_NL_1) ? _HF_NL_2 : _HF_NL_1;
 764       l = 1;
 765       c = '\n';
 766       goto jput;
 767    }
 768    self->hf_flags = f;
 769
 770    /* Check whether there are HREFs to dump; there is so much messy tagsoup out
 771     * there that it seems best not to simply dump HREFs in each _dump(), but
 772     * only with some gap, let's say half the real screen height */
 773    if (--self->hf_href_dist < 0 && (f & _HF_NL_2) && self->hf_hrefs != NULL)
 774       self = _hf_dump_hrefs(self);
 775 jleave:
 776    NYD2_LEAVE;
 777    return self;
 778 }
 779
 780 static struct htmlflt *
 781 _hf_store(struct htmlflt *self, char c)
 782 {
 783    ui32_t f, l, i;
 784    NYD2_ENTER;
 785
 786    assert(c != '\n');
 787
 788    f = self->hf_flags;
 789    l = self->hf_len;
 790    self->hf_line[l] = (c == '\t' ? ' ' : c);
 791    self->hf_len = ++l;
 792    if (blankspacechar(c)) {
 793       if (c == '\t') {
 794          i = 8 - ((l - 1) & 7); /* xxx magic tab width of 8 */
 795          if (i > 0) {
 796             do
 797                self = _hf_store(self, ' ');
 798             while (--i > 0);
 799             goto jleave;
 800          }
 801       }
 802       self->hf_last_ws = l;
 803    } else if (/*c == '.' ||*/ c == ',' || c == ';' || c == '-')
 804       self->hf_last_ws = l;
 805
 806    i = l;
 807 # ifdef HAVE_NATCH_CHAR /* XXX This code is really ridiculous! */
 808    if (n_mb_cur_max > 1) { /* XXX should mbrtowc() and THEN store, at least.. */
 809       wchar_t wc;
 810       int w, x = mbtowc(&wc, self->hf_line + self->hf_mboff, l - self->hf_mboff);
 811
 812       if (x > 0) {
 813          if ((w = wcwidth(wc)) == -1 ||
 814                /* Actively filter out L-TO-R and R-TO-R marks TODO ctext */
 815                (wc == 0x200E || wc == 0x200F ||
 816                   (wc >= 0x202A && wc <= 0x202E)) ||
 817                /* And some zero-width messes */
 818                wc == 0x00AD || (wc >= 0x200B && wc <= 0x200D) ||
 819                /* Oh about the ISO C wide character interfaces, baby! */
 820                (wc == 0xFEFF)){
 821             self->hf_len -= x;
 822             goto jleave;
 823          } else if (iswspace(wc))
 824             self->hf_last_ws = l;
 825          self->hf_mboff += x;
 826          i = (self->hf_mbwidth += w);
 827       } else {
 828          if (x < 0) {
 829             (void)mbtowc(&wc, NULL, n_mb_cur_max);
 830             if (UICMP(32, l - self->hf_mboff, >=, n_mb_cur_max)) { /* XXX */
 831                ++self->hf_mboff;
 832                ++self->hf_mbwidth;
 833             }
 834          }
 835          i = self->hf_mbwidth;
 836       }
 837    }
 838 # endif
 839
 840    /* Do we need to break the line? */
 841    if (i >= self->hf_lmax - _HF_BRKSUB) {
 842       ui32_t lim = self->hf_lmax >> 1;
 843
 844       /* Let's hope we saw a sane place to break this line! */
 845       if (self->hf_last_ws >= lim) {
 846 jput:
 847          i = self->hf_len = self->hf_last_ws;
 848          self = _hf_dump(self);
 849          if ((self->hf_len = (l -= i)) > 0) {
 850             self->hf_flags &= ~_HF_NL_MASK;
 851             memmove(self->hf_line, self->hf_line + i, l);
 852 # ifdef HAVE_NATCH_CHAR
 853             __hf_sync_mbstuff(self);
 854 # endif
 855          }
 856          goto jleave;
 857       }
 858
 859       /* Any 7-bit characters? */
 860       for (i = l; i-- >= lim;)
 861          if (asciichar((c = self->hf_line[i]))) {
 862             self->hf_last_ws = ++i;
 863             goto jput;
 864          } else if ((f & _HF_UTF8) && ((ui8_t)c & 0xC0) != 0x80) {
 865             self->hf_last_ws = i;
 866             goto jput;
 867          }
 868
 869       /* Hard break necessary!  xxx really badly done */
 870       if (l >= self->hf_lmax - 1)
 871          self = _hf_dump(self);
 872    }
 873 jleave:
 874    NYD2_LEAVE;
 875    return self;
 876 }
 877
 878 # ifdef HAVE_NATCH_CHAR
 879 static struct htmlflt *
 880 __hf_sync_mbstuff(struct htmlflt *self)
 881 {
 882    wchar_t wc;
 883    char const *b;
 884    ui32_t o, w, l;
 885    NYD2_ENTER;
 886
 887    b = self->hf_line;
 888    o = w = 0;
 889    l = self->hf_len;
 890    goto jumpin;
 891
 892    while (l > 0) {
 893       int x = mbtowc(&wc, b, l);
 894
 895       if (x == 0)
 896          break;
 897
 898       if (x > 0) {
 899          b += x;
 900          l -= x;
 901          o += x;
 902          if ((x = wcwidth(wc)) == -1)
 903             x = 1;
 904          w += x;
 905          continue;
 906       }
 907
 908       /* Bad, skip over a single character.. XXX very bad indeed */
 909       ++b;
 910       ++o;
 911       ++w;
 912       --l;
 913 jumpin:
 914       (void)mbtowc(&wc, NULL, n_mb_cur_max);
 915    }
 916
 917    self->hf_mboff = o;
 918    self->hf_mbwidth = w;
 919
 920    NYD2_LEAVE;
 921    return self;
 922 }
 923 # endif /* HAVE_NATCH_CHAR */
 924
 925 static struct htmlflt *
 926 _hf_nl(struct htmlflt *self)
 927 {
 928    ui32_t f;
 929    NYD2_ENTER;
 930
 931    if (!((f = self->hf_flags) & _HF_ERROR)) {
 932       if (f & _HF_ANY) {
 933          if ((f & _HF_NL_MASK) != _HF_NL_MASK)
 934             self = _hf_dump(self);
 935       } else
 936          self->hf_flags = (f |= _HF_NL_MASK);
 937    }
 938    NYD2_LEAVE;
 939    return self;
 940 }
 941
 942 static struct htmlflt *
 943 _hf_nl_force(struct htmlflt *self)
 944 {
 945    NYD2_ENTER;
 946    if (!(self->hf_flags & _HF_ERROR))
 947       self = _hf_dump(self);
 948    NYD2_LEAVE;
 949    return self;
 950 }
 951
 952 static struct htmlflt *
 953 _hf_putc(struct htmlflt *self, char c)
 954 {
 955    ui32_t f;
 956    NYD2_ENTER;
 957
 958    if ((f = self->hf_flags) & _HF_ERROR)
 959       goto jleave;
 960
 961    if (c == '\n') {
 962       self = _hf_nl(self);
 963       goto jleave;
 964    } else if (c == ' ' || c == '\t') {
 965       if ((f & _HF_BLANK) || self->hf_len == 0)
 966          goto jleave;
 967       f |= _HF_BLANK;
 968    } else
 969       f &= ~_HF_BLANK;
 970    f &= ~_HF_NL_MASK;
 971    self->hf_flags = (f |= _HF_ANY);
 972    self = _hf_store(self, c);
 973 jleave:
 974    NYD2_LEAVE;
 975    return self;
 976 }
 977
 978 static struct htmlflt *
 979 _hf_putc_premode(struct htmlflt *self, char c)
 980 {
 981    ui32_t f;
 982    NYD2_ENTER;
 983
 984    if ((f = self->hf_flags) & _HF_ERROR) {
 985       ;
 986    } else if (c == '\n')
 987       self = _hf_nl_force(self);
 988    else {
 989       f &= ~_HF_NL_MASK;
 990       self->hf_flags = (f |= _HF_ANY);
 991       self = _hf_store(self, c);
 992    }
 993    NYD2_LEAVE;
 994    return self;
 995 }
 996
 997 static struct htmlflt *
 998 _hf_puts(struct htmlflt *self, char const *cp)
 999 {
1000    char c;
1001    NYD2_ENTER;
1002
1003    while ((c = *cp++) != '\0')
1004       self = _hf_putc(self, c);
1005    NYD2_LEAVE;
1006    return self;
1007 }
1008
1009 static struct htmlflt *
1010 _hf_putbuf(struct htmlflt *self, char const *cp, size_t len)
1011 {
1012    NYD2_ENTER;
1013
1014    while (len-- > 0)
1015       self = _hf_putc(self, *cp++);
1016    NYD2_LEAVE;
1017    return self;
1018 }
1019
1020 static struct htmlflt *
1021 _hf_param(struct htmlflt *self, struct str *store, char const *param)
1022 {
1023    char const *cp;
1024    char c, x, quote;
1025    size_t i;
1026    bool_t hot;
1027    NYD2_ENTER;
1028
1029    store->s = NULL;
1030    store->l = 0;
1031    cp = self->hf_bdat;
1032
1033    /* Skip over any non-WS first; be aware of soup, if it slipped through */
1034    for(;;){
1035       if((c = *cp++) == '\0' || c == '>')
1036          goto jleave;
1037       if(whitechar(c))
1038          break;
1039    }
1040
1041    /* Search for the parameter, take care of other quoting along the way */
1042    x = *param++;
1043    x = upperconv(x);
1044    i = strlen(param);
1045
1046    for(hot = TRU1;;){
1047       if((c = *cp++) == '\0' || c == '>')
1048          goto jleave;
1049       if(whitechar(c)){
1050          hot = TRU1;
1051          continue;
1052       }
1053
1054       /* Could it be a parameter? */
1055       if(hot){
1056          hot = FAL0;
1057
1058          /* Is it the desired one? */
1059          if((c = upperconv(c)) == x && !ascncasecmp(param, cp, i)){
1060             char const *cp2 = cp + i;
1061
1062             if((quote = *cp2++) != '='){
1063                if(quote == '\0' || quote == '>')
1064                   goto jleave;
1065                while(whitechar(quote))
1066                   quote = *cp2++;
1067             }
1068             if(quote == '='){
1069                cp = cp2;
1070                break;
1071             }
1072             continue; /* XXX Optimize: i bytes or even cp2 can't be it! */
1073          }
1074       }
1075
1076       /* Not the desired one; but a parameter? */
1077       if(c != '=')
1078          continue;
1079       /* If so, properly skip over the value */
1080       if((c = *cp++) == '"' || c == '\''){
1081          /* TODO i have forgotten whether reverse solidus quoting is allowed
1082           * TODO quoted HTML parameter values?  not supporting that for now.. */
1083          for(quote = c; (c = *cp++) != '\0' && c != quote;)
1084             ;
1085       }else
1086          while(c != '\0' && !whitechar(c) && c != '>')
1087             c = *++cp;
1088       if(c == '\0')
1089          goto jleave;
1090    }
1091
1092    /* Skip further whitespace */
1093    for(;;){
1094       if((c = *cp++) == '\0' || c == '>')
1095          goto jleave;
1096       if(!whitechar(c))
1097          break;
1098    }
1099
1100    if(c == '"' || c == '\''){
1101       /* TODO i have forgotten whether reverse solisud quoting is allowed in
1102        * TODO quoted HTML parameter values?  not supporting that for now.. */
1103       store->s = n_UNCONST(cp);
1104       for(quote = c; (c = *cp) != '\0' && c != quote; ++cp)
1105          ;
1106       /* XXX ... and we simply ignore a missing trailing " :> */
1107    }else{
1108       store->s = n_UNCONST(cp - 1);
1109       if(!whitechar(c))
1110          while((c = *cp) != '\0' && !whitechar(c) && c != '>')
1111             ++cp;
1112    }
1113    i = PTR2SIZE(cp - store->s);
1114
1115    /* Terrible tagsoup out there, e.g., groups.google.com produces href=""
1116     * parameter values prefixed and suffixed by newlines!  Therefore trim the
1117     * value content TODO join into the parse step above! */
1118    for (cp = store->s; i > 0 && spacechar(*cp); ++cp, --i)
1119       ;
1120    store->s = n_UNCONST(cp);
1121    for (cp += i - 1; i > 0 && spacechar(*cp); --cp, --i)
1122       ;
1123    if ((store->l = i) == 0)
1124       store->s = NULL;
1125 jleave:
1126    NYD2_LEAVE;
1127    return self;
1128 }
1129
1130 static struct htmlflt *
1131 _hf_expand_all_ents(struct htmlflt *self, struct str const *param)
1132 {
1133    char const *cp, *maxcp, *ep;
1134    char c;
1135    size_t i;
1136    NYD2_ENTER;
1137
1138    for (cp = param->s, maxcp = cp + param->l; cp < maxcp;)
1139       if ((c = *cp++) != '&')
1140 jputc:
1141          self = _hf_putc(self, c);
1142       else {
1143          for (ep = cp--;;) {
1144             if (ep == maxcp || (c = *ep++) == '\0') {
1145                for (; cp < ep; ++cp)
1146                   self = _hf_putc(self, *cp);
1147                goto jleave;
1148             } else if (c == ';') {
1149                if ((i = PTR2SIZE(ep - cp)) > 1) {
1150                   self = _hf_check_ent(self, cp, i);
1151                   break;
1152                } else {
1153                   c = *cp++;
1154                   goto jputc;
1155                }
1156             }
1157          }
1158          cp = ep;
1159       }
1160 jleave:
1161    NYD2_LEAVE;
1162    return self;
1163 }
1164
1165 static struct htmlflt *
1166 _hf_check_tag(struct htmlflt *self, char const *s)
1167 {
1168    char nobuf[32], c;
1169    struct str param;
1170    size_t i;
1171    struct htmlflt_tag const *hftp;
1172    ui32_t f;
1173    NYD2_ENTER;
1174
1175    /* Extra check only */
1176    assert(s != NULL);
1177    if (*s != '<') {
1178       DBG( n_alert("HTML tagsoup filter _hf_check_tag() called on soup!"); )
1179 jput_as_is:
1180       self = _hf_puts(self, self->hf_bdat);
1181       goto jleave;
1182    }
1183
1184    for (++s, i = 0; (c = s[i]) != '\0' && c != '>' && !whitechar(c); ++i)
1185       /* Special massage for things like <br/>: after the slash only whitespace
1186        * may separate us from the closing right angle! */
1187       if (c == '/') {
1188          size_t j = i + 1;
1189
1190          while ((c = s[j]) != '\0' && c != '>' && whitechar(c))
1191             ++j;
1192          if (c == '>')
1193             break;
1194       }
1195
1196    for (hftp = _hf_tags;;) {
1197       if (i == hftp->hft_len && !ascncasecmp(s, hftp->hft_tag, i)) {
1198          c = s[hftp->hft_len];
1199          if (c == '>' || c == '/' || whitechar(c))
1200             break;
1201       }
1202       if (PTRCMP(++hftp, >=, _hf_tags + n_NELEM(_hf_tags)))
1203          goto jnotknown;
1204    }
1205
1206    f = self->hf_flags;
1207    switch (hftp->hft_act) {
1208    case _HFSA_PRE_END:
1209       f &= ~_HF_PRE;
1210       if (0) {
1211          /* FALLTHRU */
1212    case _HFSA_PRE:
1213          f |= _HF_PRE;
1214       }
1215       self->hf_flags = f;
1216       /* FALLTHRU */
1217
1218    case _HFSA_NEEDSEP:
1219       if (!(self->hf_flags & _HF_NL_2))
1220          self = _hf_nl(self);
1221       /* FALLTHRU */
1222    case _HFSA_NEEDNL:
1223       if (!(f & _HF_NL_1))
1224          self = _hf_nl(self);
1225       if (hftp->hft_injc != '\0') {
1226          self = _hf_putc(self, hftp->hft_injc & 0x7F);
1227          if ((uc_i)hftp->hft_injc & 0x80)
1228             self = _hf_putc(self, ' ');
1229       }
1230       break;
1231
1232    case _HFSA_IGN:
1233       self->hf_ign_tag = hftp;
1234       self->hf_flags = (f |= _HF_IGN | _HF_NOPUT);
1235       break;
1236
1237    case _HFSA_IMG:
1238       self = _hf_param(self, &param, "alt");
1239       self = _hf_putc(self, '[');
1240       if (param.s == NULL) {
1241          param.s = n_UNCONST("IMG");
1242          param.l = 3;
1243          goto jimg_put;
1244       } /* else */ if (memchr(param.s, '&', param.l) != NULL)
1245          self = _hf_expand_all_ents(self, &param);
1246       else
1247 jimg_put:
1248          self = _hf_putbuf(self, param.s, param.l);
1249       self = _hf_putc(self, ']');
1250       break;
1251
1252    case _HFSA_HREF:
1253       self = _hf_param(self, &param, "href");
1254       /* Ignore non-external links */
1255       if (param.s != NULL && *param.s != '#') {
1256          struct htmlflt_href *hhp = smalloc(
1257                n_VSTRUCT_SIZEOF(struct htmlflt_href, hfh_dat) + param.l +1);
1258
1259          hhp->hfh_next = self->hf_hrefs;
1260          hhp->hfh_no = ++self->hf_href_no;
1261          hhp->hfh_len = (ui32_t)param.l;
1262          memcpy(hhp->hfh_dat, param.s, param.l);
1263
1264          snprintf(nobuf, sizeof nobuf, "[%u]", hhp->hfh_no);
1265          self->hf_flags = (f |= _HF_HREF);
1266          self->hf_hrefs = hhp;
1267          self = _hf_puts(self, nobuf);
1268       } else
1269          self->hf_flags = (f &= ~_HF_HREF);
1270       break;
1271    case _HFSA_HREF_END:
1272       if (f & _HF_HREF) {
1273          snprintf(nobuf, sizeof nobuf, "[/%u]", self->hf_href_no);
1274          self = _hf_puts(self, nobuf);
1275       }
1276       break;
1277
1278    default:
1279       c = (char)(hftp->hft_act & 0xFF);
1280       self = _hf_putc(self, c);
1281       break;
1282    case '\0':
1283       break;
1284    }
1285 jleave:
1286    NYD2_LEAVE;
1287    return self;
1288
1289    /* The problem is that even invalid tagsoup is widely used, without real
1290     * searching i have seen e-mail address in <N@H.D> notation, and more.
1291     * To protect us a bit look around and possibly write the content as such */
1292 jnotknown:
1293    switch (*s) {
1294    case '!':
1295    case '?':
1296       /* Ignore <!DOCTYPE, <!-- comments, <? PIs.. */
1297       goto jleave;
1298    case '>':
1299       /* Print out an empty tag as such */
1300       if (s[1] == '\0') {
1301          --s;
1302          goto jput_as_is;
1303       }
1304       break;
1305    case '/':
1306       ++s;
1307       break;
1308    default:
1309       break;
1310    }
1311
1312    /* Also skip over : in order to suppress v:roundrect, w:anchorlock.. */
1313    while ((c = *s++) != '\0' && c != '>' && !whitechar(c) && c != ':')
1314       if (!asciichar(c) || punctchar(c)) {
1315          self = _hf_puts(self, self->hf_bdat);
1316          break;
1317       }
1318    goto jleave;
1319 }
1320
1321 static struct htmlflt *
1322 _hf_check_ent(struct htmlflt *self, char const *s, size_t l)
1323 {
1324    char nobuf[32];
1325    char const *s_save;
1326    size_t l_save;
1327    struct hf_ent const *hfep;
1328    size_t i;
1329    NYD2_ENTER;
1330
1331    s_save = s;
1332    l_save = l;
1333    assert(*s == '&');
1334    assert(l > 0);
1335    /* False entities seen in the wild assert(s[l - 1] == ';'); */
1336    ++s;
1337    l -= 2;
1338
1339    /* Numeric entity, or try named search */
1340    if (*s == '#') {
1341       i = (*++s == 'x' ? 16 : 10);
1342
1343       if ((i != 16 || (++s, --l) > 0) && l < sizeof(nobuf)) {
1344          memcpy(nobuf, s, l);
1345          nobuf[l] = '\0';
1346          n_idec_uiz_cp(&i, nobuf, i, NULL);
1347          if (i <= 0x7F)
1348             self = _hf_putc(self, (char)i);
1349          else if (self->hf_flags & _HF_UTF8) {
1350 jputuni:
1351             l = n_utf32_to_utf8((ui32_t)i, nobuf);
1352             self = _hf_putbuf(self, nobuf, l);
1353          } else
1354             goto jeent;
1355       } else
1356          goto jeent;
1357    } else {
1358       ui32_t f = self->hf_flags, hf;
1359
1360       for (hfep = _hf_ents; PTRCMP(hfep, <, _hf_ents + n_NELEM(_hf_ents));
1361             ++hfep)
1362          if (l == ((hf = hfep->hfe_flags) & _HFE_LENGTH_MASK) &&
1363                !strncmp(s, hfep->hfe_ent, l)) {
1364             if ((hf & _HFE_HAVE_UNI) && (f & _HF_UTF8)) {
1365                i = hfep->hfe_uni;
1366                goto jputuni;
1367             } else if (hf & _HFE_HAVE_CSTR)
1368                self = _hf_puts(self, hfep->hfe_cstr);
1369             else
1370                self = _hf_putc(self, hfep->hfe_c);
1371             goto jleave;
1372          }
1373 jeent:
1374       self = _hf_putbuf(self, s_save, l_save);
1375    }
1376 jleave:
1377    NYD2_LEAVE;
1378    return self;
1379 }
1380
1381 static ssize_t
1382 _hf_add_data(struct htmlflt *self, char const *dat, size_t len)
1383 {
1384    char c, *cp, *cp_max;
1385    bool_t hot;
1386    ssize_t rv = 0;
1387    NYD_ENTER;
1388
1389    /* Final put request? */
1390    if (dat == NULL) {
1391       if (self->hf_len > 0 || self->hf_hrefs != NULL) {
1392          self = _hf_dump(self);
1393          if (self->hf_hrefs != NULL)
1394             self = _hf_dump_hrefs(self);
1395          rv = 1;
1396       }
1397       goto jleave;
1398    }
1399
1400    /* Always ensure some initial buffer */
1401    if ((cp = self->hf_curr) != NULL)
1402       cp_max = self->hf_bmax;
1403    else {
1404       cp = self->hf_curr = self->hf_bdat = smalloc(LINESIZE);
1405       cp_max = self->hf_bmax = cp + LINESIZE -1; /* (Always room for NUL!) */
1406    }
1407    hot = (cp != self->hf_bdat);
1408
1409    for (rv = (ssize_t)len; len > 0; --len) {
1410       ui32_t f = self->hf_flags;
1411
1412       if (f & _HF_ERROR)
1413          break;
1414       c = *dat++;
1415
1416       /* Soup is really weird, and scripts may contain almost anything (and
1417        * newer CSS standards are also cryptic): therefore prefix the _HF_IGN
1418        * test and walk until we see the required end tag */
1419       /* TODO For real safety _HF_IGN soup condome would also need to know
1420        * TODO about quoted strings so that 'var i = "</script>";' couldn't
1421        * TODO fool it!   We really want this mode also for _HF_NOPUT to be
1422        * TODO able to *gracefully* detect the tag-closing '>', but then if
1423        * TODO that is a single mechanism we should have made it! */
1424       if (f & _HF_IGN) {
1425          struct htmlflt_tag const *hftp = self->hf_ign_tag;
1426          size_t i;
1427
1428          if (c == '<') {
1429             hot = TRU1;
1430 jcp_reset:
1431             cp = self->hf_bdat;
1432          } else if (c == '>') {
1433             if (hot) {
1434                if ((i = PTR2SIZE(cp - self->hf_bdat)) > 1 &&
1435                      --i == hftp->hft_len &&
1436                      !ascncasecmp(self->hf_bdat + 1, hftp->hft_tag, i))
1437                   self->hf_flags = (f &= ~(_HF_IGN | _HF_NOPUT));
1438                hot = FAL0;
1439                goto jcp_reset;
1440             }
1441          } else if (hot) {
1442             *cp++ = c;
1443             i = PTR2SIZE(cp - self->hf_bdat);
1444             if ((i == 1 && c != '/') || --i > hftp->hft_len) {
1445                hot = FAL0;
1446                goto jcp_reset;
1447             }
1448          }
1449       } else switch (c) {
1450       case '<':
1451          /* People are using & without &amp;ing it, ditto <; be aware */
1452          if (f & (_HF_NOPUT | _HF_ENT)) {
1453             f &= ~_HF_ENT;
1454             /* Special case "<!--" buffer content to deal with really weird
1455              * things that can be done with "<!--[if gte mso 9]>" syntax */
1456             if (PTR2SIZE(cp - self->hf_bdat) != 4 ||
1457                   memcmp(self->hf_bdat, "<!--", 4)) {
1458                self->hf_flags = f;
1459                *cp = '\0';
1460                self = _hf_puts(self, self->hf_bdat);
1461                f = self->hf_flags;
1462             }
1463          }
1464          cp = self->hf_bdat;
1465          *cp++ = c;
1466          self->hf_flags = (f |= _HF_NOPUT);
1467          break;
1468       case '>':
1469          /* Weird tagsoup around, do we actually parse a tag? */
1470          if (!(f & _HF_NOPUT))
1471             goto jdo_c;
1472          cp[0] = c;
1473          cp[1] = '\0';
1474          f &= ~(_HF_NOPUT | _HF_ENT);
1475          self->hf_flags = f;
1476          self = _hf_check_tag(self, self->hf_bdat);
1477          *(cp = self->hf_bdat) = '\0'; /* xxx extra safety */
1478          /* Quick hack to get rid of redundant newline after <pre> XXX */
1479          if (!(f & _HF_PRE) && (self->hf_flags & _HF_PRE) &&
1480                len > 1 && *dat == '\n')
1481             ++dat, --len;
1482          break;
1483
1484       case '\r': /* TODO CR should be stripped in lower level!! (Only B64!?!) */
1485          break;
1486       case '\n':
1487          /* End of line is not considered unless we are in PRE section.
1488           * However, in _HF_NOPUT mode we must be aware of tagsoup which uses
1489           * newlines for separating parameters */
1490          if (f & _HF_NOPUT)
1491             goto jdo_c;
1492          self = (f & _HF_PRE) ? _hf_nl_force(self) : _hf_putc(self, ' ');
1493          break;
1494
1495       case '\t':
1496          if (!(f & _HF_PRE))
1497             c = ' ';
1498          /* FALLTHRU */
1499       default:
1500 jdo_c:
1501          /* If not currently parsing a tag and bypassing normal output.. */
1502          if (!(f & _HF_NOPUT)) {
1503             if (cntrlchar(c))
1504                break;
1505             if (c == '&') {
1506                cp = self->hf_bdat;
1507                *cp++ = c;
1508                self->hf_flags = (f |= _HF_NOPUT | _HF_ENT);
1509             } else if (f & _HF_PRE) {
1510                self = _hf_putc_premode(self, c);
1511                self->hf_flags &= ~_HF_BLANK;
1512             } else
1513               self = _hf_putc(self, c);
1514          } else if ((f & _HF_ENT) && c == ';') {
1515             cp[0] = c;
1516             cp[1] = '\0';
1517             f &= ~(_HF_NOPUT | _HF_ENT);
1518             self->hf_flags = f;
1519            self = _hf_check_ent(self, self->hf_bdat,
1520                PTR2SIZE(cp + 1 - self->hf_bdat));
1521          } else {
1522             /* We may need to grow the buffer */
1523             if (PTRCMP(cp + 42/2, >=, cp_max)) {
1524                size_t i = PTR2SIZE(cp - self->hf_bdat),
1525                   m = PTR2SIZE(self->hf_bmax - self->hf_bdat) + LINESIZE;
1526
1527                cp = self->hf_bdat = srealloc(self->hf_bdat, m);
1528                self->hf_bmax = cp + m -1;
1529                self->hf_curr = (cp += i);
1530             }
1531             *cp++ = c;
1532          }
1533       }
1534    }
1535    self->hf_curr = cp;
1536 jleave:
1537   NYD_LEAVE;
1538   return (self->hf_flags & _HF_ERROR) ? -1 : rv;
1539 }
1540
1541 /*
1542  * TODO Because we don't support filter chains yet this filter will be run
1543  * TODO in a dedicated subprocess, driven via a special Popen() mode
1544  */
1545 static bool_t __hf_hadpipesig;
1546 static void
1547 __hf_onpipe(int signo)
1548 {
1549    NYD_X; /* Signal handler */
1550    n_UNUSED(signo);
1551    __hf_hadpipesig = TRU1;
1552 }
1553
1554 FL int
1555 htmlflt_process_main(void)
1556 {
1557    char buf[BUFFER_SIZE];
1558    struct htmlflt hf;
1559    size_t i;
1560    int rv;
1561    NYD_ENTER;
1562
1563    __hf_hadpipesig = FAL0;
1564    safe_signal(SIGPIPE, &__hf_onpipe);
1565
1566    htmlflt_init(&hf);
1567    htmlflt_reset(&hf, n_stdout);
1568
1569    for (;;) {
1570       if ((i = fread(buf, sizeof(buf[0]), n_NELEM(buf), n_stdin)) == 0) {
1571          rv = !feof(n_stdin);
1572          break;
1573       }
1574
1575       if ((rv = __hf_hadpipesig))
1576          break;
1577       /* Just use this directly.. */
1578       if (htmlflt_push(&hf, buf, i) < 0) {
1579          rv = 1;
1580          break;
1581       }
1582    }
1583    if (rv == 0 && htmlflt_flush(&hf) < 0)
1584       rv = 1;
1585
1586    htmlflt_destroy(&hf);
1587
1588    rv |= __hf_hadpipesig;
1589    NYD_LEAVE;
1590    return rv;
1591 }
1592
1593 FL void
1594 htmlflt_init(struct htmlflt *self)
1595 {
1596    NYD_ENTER;
1597    /* (Rather redundant though) */
1598    memset(self, 0, sizeof *self);
1599    NYD_LEAVE;
1600 }
1601
1602 FL void
1603 htmlflt_destroy(struct htmlflt *self)
1604 {
1605    NYD_ENTER;
1606    htmlflt_reset(self, NULL);
1607    NYD_LEAVE;
1608 }
1609
1610 FL void
1611 htmlflt_reset(struct htmlflt *self, FILE *f)
1612 {
1613    struct htmlflt_href *hfhp;
1614    NYD_ENTER;
1615
1616    while ((hfhp = self->hf_hrefs) != NULL) {
1617       self->hf_hrefs = hfhp->hfh_next;
1618       free(hfhp);
1619    }
1620
1621    if (self->hf_bdat != NULL)
1622       free(self->hf_bdat);
1623    if (self->hf_line != NULL)
1624       free(self->hf_line);
1625
1626    memset(self, 0, sizeof *self);
1627
1628    if (f != NULL) {
1629       ui32_t sw = n_MAX(_HF_MINLEN, (ui32_t)n_scrnwidth);
1630
1631       self->hf_line = smalloc((size_t)sw * n_mb_cur_max +1);
1632       self->hf_lmax = sw;
1633
1634       if (n_psonce & n_PSO_UNICODE) /* TODO not truly generic */
1635          self->hf_flags = _HF_UTF8;
1636       self->hf_os = f;
1637    }
1638    NYD_LEAVE;
1639 }
1640
1641 FL ssize_t
1642 htmlflt_push(struct htmlflt *self, char const *dat, size_t len)
1643 {
1644    ssize_t rv;
1645    NYD_ENTER;
1646
1647    rv = _hf_add_data(self, dat, len);
1648    NYD_LEAVE;
1649    return rv;
1650 }
1651
1652 FL ssize_t
1653 htmlflt_flush(struct htmlflt *self)
1654 {
1655    ssize_t rv;
1656    NYD_ENTER;
1657
1658    rv = _hf_add_data(self, NULL, 0);
1659    rv |= !fflush(self->hf_os) ? 0 : -1;
1660    NYD_LEAVE;
1661    return rv;
1662 }
1663 #endif /* HAVE_FILTER_HTML_TAGSOUP */
1664
1665 /* s-it-mode */