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 n_CTAV(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    n_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    n_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[n_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 n_CTA(n_SIZEOF_FIELD(struct htmlflt_tag, hft_tag) < LINESIZE,
 570    "Structure field too large a size"); /* .hf_ign_tag */
 571
 572 struct hf_ent {
 573    ui8_t       hfe_flags;  /* enum hf_entity_flags plus length of .hfe_ent */
 574    char        hfe_c;      /* Plain replacement character */
 575    ui16_t      hfe_uni;    /* Unicode codepoint if _HFE_HAVE_UNI */
 576    char        hfe_cstr[5]; /* _HFE_HAVE_CSTR (e.g., &hellip; -> ...) */
 577    char const  hfe_ent[7]; /* Entity less & and ; surroundings */
 578 };
 579
 580 /* Tag list; not binary searched :(, so try to take care a bit */
 581 static struct htmlflt_tag const  _hf_tags[] = {
 582 # undef _X
 583 # undef _XC
 584 # define _X(S,A)     {A, '\0', sizeof(S) -1, S "\0"}
 585 # define _XC(S,C,A)  {A, C, sizeof(S) -1, S "\0"}
 586
 587    _X("P", _HFSA_NEEDSEP),       _X("/P", _HFSA_NEEDNL),
 588    _X("DIV", _HFSA_NEEDSEP),     _X("/DIV", _HFSA_NEEDNL),
 589    _X("TR", _HFSA_NEEDNL),
 590                                  _X("/TH", '\t'),
 591                                  _X("/TD", '\t'),
 592    /* Let it stand out; also since we don't support implicit paragraphs after
 593     * block elements, plain running text after a list (seen in Unicode
 594     * announcement via Firefox) */
 595    _X("UL", _HFSA_NEEDSEP),      _X("/UL", _HFSA_NEEDSEP),
 596    _XC("LI", (char)0x80 | '*', _HFSA_NEEDSEP),
 597    _X("DL", _HFSA_NEEDSEP),
 598    _X("DT", _HFSA_NEEDNL),
 599
 600    _X("A", _HFSA_HREF),          _X("/A", _HFSA_HREF_END),
 601    _X("IMG", _HFSA_IMG),
 602    _X("BR", '\n'),
 603    _X("PRE", _HFSA_PRE),         _X("/PRE", _HFSA_PRE_END),
 604    _X("TITLE", _HFSA_NEEDSEP),   /*_X("/TITLE", '\n'),*/
 605    _X("H1", _HFSA_NEEDSEP),      /*_X("/H1", '\n'),*/
 606    _X("H2", _HFSA_NEEDSEP),      /*_X("/H2", '\n'),*/
 607    _X("H3", _HFSA_NEEDSEP),      /*_X("/H3", '\n'),*/
 608    _X("H4", _HFSA_NEEDSEP),      /*_X("/H4", '\n'),*/
 609    _X("H5", _HFSA_NEEDSEP),      /*_X("/H5", '\n'),*/
 610    _X("H6", _HFSA_NEEDSEP),      /*_X("/H6", '\n'),*/
 611
 612    _X("STYLE", _HFSA_IGN),
 613    _X("SCRIPT", _HFSA_IGN),
 614
 615 # undef _X
 616 };
 617
 618 /* Entity list; not binary searched.. */
 619 static struct hf_ent const       _hf_ents[] = {
 620 # undef _X
 621 # undef _XU
 622 # undef _XS
 623 # undef _XUS
 624 # define _X(E,C)     {(sizeof(E) -1), C, 0x0u, "", E "\0"}
 625 # define _XU(E,C,U)  {(sizeof(E) -1) | _HFE_HAVE_UNI, C, U, "", E "\0"}
 626 # define _XS(E,S)    {(sizeof(E) -1) | _HFE_HAVE_CSTR, '\0', 0x0u,S "\0",E "\0"}
 627 # define _XSU(E,S,U) \
 628    {(sizeof(E) -1) | _HFE_HAVE_UNI | _HFE_HAVE_CSTR, '\0', U, S "\0", E "\0"}
 629
 630    _X("quot", '"'),
 631    _X("amp", '&'),
 632    _X("lt", '<'),                _X("gt", '>'),
 633
 634    _XU("nbsp", ' ', 0x0020 /* Note: not 0x00A0 seems to be better for us */),
 635    _XU("middot", '.', 0x00B7),
 636    _XSU("hellip", "...", 0x2026),
 637    _XSU("mdash", "---", 0x2014), _XSU("ndash", "--", 0x2013),
 638    _XSU("laquo", "<<", 0x00AB),  _XSU("raquo", ">>", 0x00BB),
 639    _XSU("lsaquo", "<", 0x2039),  _XSU("rsaquo", ">", 0x203A),
 640    _XSU("lsquo", "'", 0x2018),   _XSU("rsquo", "'", 0x2019),
 641    _XSU("ldquo", "\"", 0x201C),  _XSU("rdquo", "\"", 0x201D),
 642    _XSU("uarr", "^|", 0x2191),   _XSU("darr", "|v", 0x2193),
 643
 644    _XSU("cent", "CENT", 0x00A2),
 645    _XSU("copy", "(C)", 0x00A9),
 646    _XSU("euro", "EUR", 0x20AC),
 647    _XSU("infin", "INFY", 0x221E),
 648    _XSU("pound", "GBP", 0x00A3),
 649    _XSU("reg", "(R)", 0x00AE),
 650    _XSU("sect", "S:", 0x00A7),
 651    _XSU("yen", "JPY", 0x00A5),
 652
 653    /* German umlauts */
 654    _XSU("Auml", "Ae", 0x00C4),   _XSU("auml", "ae", 0x00E4),
 655    _XSU("Ouml", "Oe", 0x00D6),   _XSU("ouml", "oe", 0x00F6),
 656    _XSU("Uuml", "Ue", 0x00DC),   _XSU("uuml", "ue", 0x00FC),
 657    _XSU("szlig", "ss", 0x00DF)
 658
 659 # undef _X
 660 # undef _XU
 661 # undef _XS
 662 # undef _XSU
 663 };
 664
 665 /* Real output */
 666 static struct htmlflt * _hf_dump_hrefs(struct htmlflt *self);
 667 static struct htmlflt * _hf_dump(struct htmlflt *self);
 668 static struct htmlflt * _hf_store(struct htmlflt *self, char c);
 669 # ifdef HAVE_NATCH_CHAR
 670 static struct htmlflt * __hf_sync_mbstuff(struct htmlflt *self);
 671 # endif
 672
 673 /* Virtual output */
 674 static struct htmlflt * _hf_nl(struct htmlflt *self);
 675 static struct htmlflt * _hf_nl_force(struct htmlflt *self);
 676 static struct htmlflt * _hf_putc(struct htmlflt *self, char c);
 677 static struct htmlflt * _hf_putc_premode(struct htmlflt *self, char c);
 678 static struct htmlflt * _hf_puts(struct htmlflt *self, char const *cp);
 679 static struct htmlflt * _hf_putbuf(struct htmlflt *self,
 680                            char const *cp, size_t len);
 681
 682 /* Try to locate a param'eter in >hf_bdat, store it (non-terminated!) or NULL */
 683 static struct htmlflt * _hf_param(struct htmlflt *self, struct str *store,
 684                            char const *param);
 685
 686 /* Expand all entities in the given parameter */
 687 static struct htmlflt * _hf_expand_all_ents(struct htmlflt *self,
 688                            struct str const *param);
 689
 690 /* Completely parsed over a tag / an entity, interpret that */
 691 static struct htmlflt * _hf_check_tag(struct htmlflt *self, char const *s);
 692 static struct htmlflt * _hf_check_ent(struct htmlflt *self, char const *s,
 693                            size_t l);
 694
 695 /* Input handler */
 696 static ssize_t          _hf_add_data(struct htmlflt *self,
 697                            char const *dat, size_t len);
 698
 699 static struct htmlflt *
 700 _hf_dump_hrefs(struct htmlflt *self)
 701 {
 702    struct htmlflt_href *hhp;
 703    NYD2_ENTER;
 704
 705    if (!(self->hf_flags & _HF_NL_2) && putc('\n', self->hf_os) == EOF) {
 706       self->hf_flags |= _HF_ERROR;
 707       goto jleave;
 708    }
 709
 710    /* Reverse the list */
 711    for (hhp = self->hf_hrefs, self->hf_hrefs = NULL; hhp != NULL;) {
 712       struct htmlflt_href *tmp = hhp->hfh_next;
 713       hhp->hfh_next = self->hf_hrefs;
 714       self->hf_hrefs = hhp;
 715       hhp = tmp;
 716    }
 717
 718    /* Then dump it */
 719    while ((hhp = self->hf_hrefs) != NULL) {
 720       self->hf_hrefs = hhp->hfh_next;
 721
 722       if (!(self->hf_flags & _HF_ERROR)) {
 723          int w = fprintf(self->hf_os, "  [%u] %.*s\n",
 724                hhp->hfh_no, (int)hhp->hfh_len, hhp->hfh_dat);
 725          if (w < 0)
 726             self->hf_flags |= _HF_ERROR;
 727       }
 728       free(hhp);
 729    }
 730
 731    self->hf_flags |= (putc('\n', self->hf_os) == EOF)
 732          ?  _HF_ERROR : _HF_NL_1 | _HF_NL_2;
 733    self->hf_href_dist = (ui32_t)realscreenheight >> 1;
 734 jleave:
 735    NYD2_LEAVE;
 736    return self;
 737 }
 738
 739 static struct htmlflt *
 740 _hf_dump(struct htmlflt *self)
 741 {
 742    ui32_t f, l;
 743    char c, *cp;
 744    NYD2_ENTER;
 745
 746    f = self->hf_flags & ~_HF_BLANK;
 747    l = self->hf_len;
 748    cp = self->hf_line;
 749    self->hf_mbwidth = self->hf_mboff = self->hf_last_ws = self->hf_len = 0;
 750
 751    for (c = '\0'; l > 0; --l) {
 752       c = *cp++;
 753 jput:
 754       if (putc(c, self->hf_os) == EOF) {
 755          self->hf_flags = (f |= _HF_ERROR);
 756          goto jleave;
 757       }
 758    }
 759
 760    if (c != '\n') {
 761       f |= (f & _HF_NL_1) ? _HF_NL_2 : _HF_NL_1;
 762       l = 1;
 763       c = '\n';
 764       goto jput;
 765    }
 766    self->hf_flags = f;
 767
 768    /* Check whether there are HREFs to dump; there is so much messy tagsoup out
 769     * there that it seems best not to simply dump HREFs in each _dump(), but
 770     * only with some gap, let's say half the real screen height */
 771    if (--self->hf_href_dist < 0 && (f & _HF_NL_2) && self->hf_hrefs != NULL)
 772       self = _hf_dump_hrefs(self);
 773 jleave:
 774    NYD2_LEAVE;
 775    return self;
 776 }
 777
 778 static struct htmlflt *
 779 _hf_store(struct htmlflt *self, char c)
 780 {
 781    ui32_t f, l, i;
 782    NYD2_ENTER;
 783
 784    assert(c != '\n');
 785
 786    f = self->hf_flags;
 787    l = self->hf_len;
 788    self->hf_line[l] = (c == '\t' ? ' ' : c);
 789    self->hf_len = ++l;
 790    if (blankspacechar(c)) {
 791       if (c == '\t') {
 792          i = 8 - ((l - 1) & 7); /* xxx magic tab width of 8 */
 793          if (i > 0) {
 794             do
 795                self = _hf_store(self, ' ');
 796             while (--i > 0);
 797             goto jleave;
 798          }
 799       }
 800       self->hf_last_ws = l;
 801    } else if (/*c == '.' ||*/ c == ',' || c == ';' || c == '-')
 802       self->hf_last_ws = l;
 803
 804    i = l;
 805 # ifdef HAVE_NATCH_CHAR /* XXX This code is really ridiculous! */
 806    if (mb_cur_max > 1) { /* XXX should mbrtowc() and THEN store, at least.. */
 807       wchar_t wc;
 808       int w, x = mbtowc(&wc, self->hf_line + self->hf_mboff, l - self->hf_mboff);
 809
 810       if (x > 0) {
 811          if ((w = wcwidth(wc)) == -1 ||
 812                /* Actively filter out L-TO-R and R-TO-R marks TODO ctext */
 813                (wc == 0x200E || wc == 0x200F ||
 814                   (wc >= 0x202A && wc <= 0x202E)) ||
 815                /* And some zero-width messes */
 816                wc == 0x00AD || (wc >= 0x200B && wc <= 0x200D) ||
 817                /* Oh about the ISO C wide character interfaces, baby! */
 818                (wc == 0xFEFF)){
 819             self->hf_len -= x;
 820             goto jleave;
 821          } else if (iswspace(wc))
 822             self->hf_last_ws = l;
 823          self->hf_mboff += x;
 824          i = (self->hf_mbwidth += w);
 825       } else {
 826          if (x < 0) {
 827             mbtowc(&wc, NULL, mb_cur_max);
 828             if (UICMP(32, l - self->hf_mboff, >=, mb_cur_max)) { /* XXX */
 829                ++self->hf_mboff;
 830                ++self->hf_mbwidth;
 831             }
 832          }
 833          i = self->hf_mbwidth;
 834       }
 835    }
 836 # endif
 837
 838    /* Do we need to break the line? */
 839    if (i >= self->hf_lmax - _HF_BRKSUB) {
 840       ui32_t lim = self->hf_lmax >> 1;
 841
 842       /* Let's hope we saw a sane place to break this line! */
 843       if (self->hf_last_ws >= lim) {
 844 jput:
 845          i = self->hf_len = self->hf_last_ws;
 846          self = _hf_dump(self);
 847          if ((self->hf_len = (l -= i)) > 0) {
 848             self->hf_flags &= ~_HF_NL_MASK;
 849             memmove(self->hf_line, self->hf_line + i, l);
 850 # ifdef HAVE_NATCH_CHAR
 851             __hf_sync_mbstuff(self);
 852 # endif
 853          }
 854          goto jleave;
 855       }
 856
 857       /* Any 7-bit characters? */
 858       for (i = l; i-- >= lim;)
 859          if (asciichar((c = self->hf_line[i]))) {
 860             self->hf_last_ws = ++i;
 861             goto jput;
 862          } else if ((f & _HF_UTF8) && ((ui8_t)c & 0xC0) != 0x80) {
 863             self->hf_last_ws = i;
 864             goto jput;
 865          }
 866
 867       /* Hard break necessary!  xxx really badly done */
 868       if (l >= self->hf_lmax - 1)
 869          self = _hf_dump(self);
 870    }
 871 jleave:
 872    NYD2_LEAVE;
 873    return self;
 874 }
 875
 876 # ifdef HAVE_NATCH_CHAR
 877 static struct htmlflt *
 878 __hf_sync_mbstuff(struct htmlflt *self)
 879 {
 880    wchar_t wc;
 881    char const *b;
 882    ui32_t o, w, l;
 883    NYD2_ENTER;
 884
 885    b = self->hf_line;
 886    o = w = 0;
 887    l = self->hf_len;
 888    goto jumpin;
 889
 890    while (l > 0) {
 891       int x = mbtowc(&wc, b, l);
 892
 893       if (x == 0)
 894          break;
 895
 896       if (x > 0) {
 897          b += x;
 898          l -= x;
 899          o += x;
 900          if ((x = wcwidth(wc)) == -1)
 901             x = 1;
 902          w += x;
 903          continue;
 904       }
 905
 906       /* Bad, skip over a single character.. XXX very bad indeed */
 907       ++b;
 908       ++o;
 909       ++w;
 910       --l;
 911 jumpin:
 912       mbtowc(&wc, NULL, mb_cur_max);
 913    }
 914
 915    self->hf_mboff = o;
 916    self->hf_mbwidth = w;
 917
 918    NYD2_LEAVE;
 919    return self;
 920 }
 921 # endif /* HAVE_NATCH_CHAR */
 922
 923 static struct htmlflt *
 924 _hf_nl(struct htmlflt *self)
 925 {
 926    ui32_t f;
 927    NYD2_ENTER;
 928
 929    if (!((f = self->hf_flags) & _HF_ERROR)) {
 930       if (f & _HF_ANY) {
 931          if ((f & _HF_NL_MASK) != _HF_NL_MASK)
 932             self = _hf_dump(self);
 933       } else
 934          self->hf_flags = (f |= _HF_NL_MASK);
 935    }
 936    NYD2_LEAVE;
 937    return self;
 938 }
 939
 940 static struct htmlflt *
 941 _hf_nl_force(struct htmlflt *self)
 942 {
 943    NYD2_ENTER;
 944    if (!(self->hf_flags & _HF_ERROR))
 945       self = _hf_dump(self);
 946    NYD2_LEAVE;
 947    return self;
 948 }
 949
 950 static struct htmlflt *
 951 _hf_putc(struct htmlflt *self, char c)
 952 {
 953    ui32_t f;
 954    NYD2_ENTER;
 955
 956    if ((f = self->hf_flags) & _HF_ERROR)
 957       goto jleave;
 958
 959    if (c == '\n') {
 960       self = _hf_nl(self);
 961       goto jleave;
 962    } else if (c == ' ' || c == '\t') {
 963       if ((f & _HF_BLANK) || self->hf_len == 0)
 964          goto jleave;
 965       f |= _HF_BLANK;
 966    } else
 967       f &= ~_HF_BLANK;
 968    f &= ~_HF_NL_MASK;
 969    self->hf_flags = (f |= _HF_ANY);
 970    self = _hf_store(self, c);
 971 jleave:
 972    NYD2_LEAVE;
 973    return self;
 974 }
 975
 976 static struct htmlflt *
 977 _hf_putc_premode(struct htmlflt *self, char c)
 978 {
 979    ui32_t f;
 980    NYD2_ENTER;
 981
 982    if ((f = self->hf_flags) & _HF_ERROR) {
 983       ;
 984    } else if (c == '\n')
 985       self = _hf_nl_force(self);
 986    else {
 987       f &= ~_HF_NL_MASK;
 988       self->hf_flags = (f |= _HF_ANY);
 989       self = _hf_store(self, c);
 990    }
 991    NYD2_LEAVE;
 992    return self;
 993 }
 994
 995 static struct htmlflt *
 996 _hf_puts(struct htmlflt *self, char const *cp)
 997 {
 998    char c;
 999    NYD2_ENTER;
1000
1001    while ((c = *cp++) != '\0')
1002       self = _hf_putc(self, c);
1003    NYD2_LEAVE;
1004    return self;
1005 }
1006
1007 static struct htmlflt *
1008 _hf_putbuf(struct htmlflt *self, char const *cp, size_t len)
1009 {
1010    NYD2_ENTER;
1011
1012    while (len-- > 0)
1013       self = _hf_putc(self, *cp++);
1014    NYD2_LEAVE;
1015    return self;
1016 }
1017
1018 static struct htmlflt *
1019 _hf_param(struct htmlflt *self, struct str *store, char const *param)
1020 {
1021    char const *cp;
1022    char c, x, quote;
1023    size_t i;
1024    bool_t hot;
1025    NYD2_ENTER;
1026
1027    store->s = NULL;
1028    store->l = 0;
1029    cp = self->hf_bdat;
1030
1031    /* Skip over any non-WS first; be aware of soup, if it slipped through */
1032    for(;;){
1033       if((c = *cp++) == '\0' || c == '>')
1034          goto jleave;
1035       if(whitechar(c))
1036          break;
1037    }
1038
1039    /* Search for the parameter, take care of other quoting along the way */
1040    x = *param++;
1041    x = upperconv(x);
1042    quote = '\0';
1043    i = strlen(param);
1044
1045    for(hot = TRU1;;){
1046       if((c = *cp++) == '\0' || c == '>')
1047          goto jleave;
1048       if(whitechar(c)){
1049          hot = TRU1;
1050          continue;
1051       }
1052
1053       /* Could it be a parameter? */
1054       if(hot){
1055          hot = FAL0;
1056
1057          /* Is it the desired one? */
1058          if((c = upperconv(c)) == x && !ascncasecmp(param, cp, i)){
1059             char const *cp2 = cp + i;
1060
1061             if((quote = *cp2++) != '='){
1062                if(quote == '\0' || quote == '>')
1063                   goto jleave;
1064                while(whitechar(quote))
1065                   quote = *cp2++;
1066             }
1067             if(quote == '='){
1068                cp = cp2;
1069                break;
1070             }
1071             continue; /* XXX Optimize: i bytes or even cp2 can't be it! */
1072          }
1073       }
1074
1075       /* Not the desired one; but a parameter? */
1076       if(c != '=')
1077          continue;
1078       /* If so, properly skip over the value */
1079       if(c == '"' || c == '\''){
1080          /* TODO oops i have forgotten whether backslash quoting is allowed in
1081           * TODO quoted HTML parameter values?  not supporting that for now.. */
1082          for(quote = c; (c = *cp) != '\0' && c != quote; ++cp)
1083             ;
1084       }else
1085          while((c = *cp) != '\0' && !whitechar(c) && c != '>')
1086             ++cp;
1087       if(c == '\0')
1088          goto jleave;
1089    }
1090
1091    /* Skip further whitespace */
1092    for(;;){
1093       if((c = *cp++) == '\0' || c == '>')
1094          goto jleave;
1095       if(!whitechar(c))
1096          break;
1097    }
1098
1099    if(c == '"' || c == '\''){
1100       /* TODO oops i have forgotten whether backslash quoting is allowed in
1101        * TODO quoted HTML parameter values?  not supporting that for now.. */
1102       store->s = n_UNCONST(cp);
1103       for(quote = c; (c = *cp) != '\0' && c != quote; ++cp)
1104          ;
1105       /* XXX ... and we simply ignore a missing trailing " :> */
1106    }else{
1107       store->s = n_UNCONST(cp - 1);
1108       if(!whitechar(c))
1109          while((c = *cp) != '\0' && !whitechar(c) && c != '>')
1110             ++cp;
1111    }
1112    i = PTR2SIZE(cp - store->s);
1113
1114    /* Terrible tagsoup out there, e.g., groups.google.com produces href=""
1115     * parameter values prefixed and suffixed by newlines!  Therefore trim the
1116     * value content TODO join into the parse step above! */
1117    for (cp = store->s; i > 0 && spacechar(*cp); ++cp, --i)
1118       ;
1119    store->s = n_UNCONST(cp);
1120    for (cp += i - 1; i > 0 && spacechar(*cp); --cp, --i)
1121       ;
1122    if ((store->l = i) == 0)
1123       store->s = NULL;
1124 jleave:
1125    NYD2_LEAVE;
1126    return self;
1127 }
1128
1129 static struct htmlflt *
1130 _hf_expand_all_ents(struct htmlflt *self, struct str const *param)
1131 {
1132    char const *cp, *maxcp, *ep;
1133    char c;
1134    size_t i;
1135    NYD2_ENTER;
1136
1137    for (cp = param->s, maxcp = cp + param->l; cp < maxcp;)
1138       if ((c = *cp++) != '&')
1139 jputc:
1140          self = _hf_putc(self, c);
1141       else {
1142          for (ep = cp--;;) {
1143             if (ep == maxcp || (c = *ep++) == '\0') {
1144                for (; cp < ep; ++cp)
1145                   self = _hf_putc(self, *cp);
1146                goto jleave;
1147             } else if (c == ';') {
1148                if ((i = PTR2SIZE(ep - cp)) > 1) {
1149                   self = _hf_check_ent(self, cp, i);
1150                   break;
1151                } else {
1152                   c = *cp++;
1153                   goto jputc;
1154                }
1155             }
1156          }
1157          cp = ep;
1158       }
1159 jleave:
1160    NYD2_LEAVE;
1161    return self;
1162 }
1163
1164 static struct htmlflt *
1165 _hf_check_tag(struct htmlflt *self, char const *s)
1166 {
1167    char nobuf[32], c;
1168    struct str param;
1169    size_t i;
1170    struct htmlflt_tag const *hftp;
1171    ui32_t f;
1172    NYD2_ENTER;
1173
1174    /* Extra check only */
1175    assert(s != NULL);
1176    if (*s != '<') {
1177       DBG( n_alert("HTML tagsoup filter _hf_check_tag() called on soup!"); )
1178 jput_as_is:
1179       self = _hf_puts(self, self->hf_bdat);
1180       goto jleave;
1181    }
1182
1183    for (++s, i = 0; (c = s[i]) != '\0' && c != '>' && !whitechar(c); ++i)
1184       /* Special massage for things like <br/>: after the slash only whitespace
1185        * may separate us from the closing right angle! */
1186       if (c == '/') {
1187          size_t j = i + 1;
1188
1189          while ((c = s[j]) != '\0' && c != '>' && whitechar(c))
1190             ++j;
1191          if (c == '>')
1192             break;
1193       }
1194
1195    for (hftp = _hf_tags;;) {
1196       if (i == hftp->hft_len && !ascncasecmp(s, hftp->hft_tag, i)) {
1197          c = s[hftp->hft_len];
1198          if (c == '>' || c == '/' || whitechar(c))
1199             break;
1200       }
1201       if (PTRCMP(++hftp, >=, _hf_tags + n_NELEM(_hf_tags)))
1202          goto jnotknown;
1203    }
1204
1205    f = self->hf_flags;
1206    switch (hftp->hft_act) {
1207    case _HFSA_PRE_END:
1208       f &= ~_HF_PRE;
1209       if (0) {
1210          /* FALLTHRU */
1211    case _HFSA_PRE:
1212          f |= _HF_PRE;
1213       }
1214       self->hf_flags = f;
1215       /* FALLTHRU */
1216
1217    case _HFSA_NEEDSEP:
1218       if (!(self->hf_flags & _HF_NL_2))
1219          self = _hf_nl(self);
1220       /* FALLTHRU */
1221    case _HFSA_NEEDNL:
1222       if (!(f & _HF_NL_1))
1223          self = _hf_nl(self);
1224       if (hftp->hft_injc != '\0') {
1225          self = _hf_putc(self, hftp->hft_injc & 0x7F);
1226          if ((uc_i)hftp->hft_injc & 0x80)
1227             self = _hf_putc(self, ' ');
1228       }
1229       break;
1230
1231    case _HFSA_IGN:
1232       self->hf_ign_tag = hftp;
1233       self->hf_flags = (f |= _HF_IGN | _HF_NOPUT);
1234       break;
1235
1236    case _HFSA_IMG:
1237       self = _hf_param(self, &param, "alt");
1238       self = _hf_putc(self, '[');
1239       if (param.s == NULL) {
1240          param.s = n_UNCONST("IMG");
1241          param.l = 3;
1242          goto jimg_put;
1243       } /* else */ if (memchr(param.s, '&', param.l) != NULL)
1244          self = _hf_expand_all_ents(self, &param);
1245       else
1246 jimg_put:
1247          self = _hf_putbuf(self, param.s, param.l);
1248       self = _hf_putc(self, ']');
1249       break;
1250
1251    case _HFSA_HREF:
1252       self = _hf_param(self, &param, "href");
1253       /* Ignore non-external links */
1254       if (param.s != NULL && *param.s != '#') {
1255          struct htmlflt_href *hhp = smalloc(
1256                n_VSTRUCT_SIZEOF(struct htmlflt_href, hfh_dat) + param.l +1);
1257
1258          hhp->hfh_next = self->hf_hrefs;
1259          hhp->hfh_no = ++self->hf_href_no;
1260          hhp->hfh_len = (ui32_t)param.l;
1261          memcpy(hhp->hfh_dat, param.s, param.l);
1262
1263          snprintf(nobuf, sizeof nobuf, "[%u]", hhp->hfh_no);
1264          self->hf_flags = (f |= _HF_HREF);
1265          self->hf_hrefs = hhp;
1266          self = _hf_puts(self, nobuf);
1267       } else
1268          self->hf_flags = (f &= ~_HF_HREF);
1269       break;
1270    case _HFSA_HREF_END:
1271       if (f & _HF_HREF) {
1272          snprintf(nobuf, sizeof nobuf, "[/%u]", self->hf_href_no);
1273          self = _hf_puts(self, nobuf);
1274       }
1275       break;
1276
1277    default:
1278       c = (char)(hftp->hft_act & 0xFF);
1279       self = _hf_putc(self, c);
1280       break;
1281    case '\0':
1282       break;
1283    }
1284 jleave:
1285    NYD2_LEAVE;
1286    return self;
1287
1288    /* The problem is that even invalid tagsoup is widely used, without real
1289     * searching i have seen e-mail address in <N@H.D> notation, and more.
1290     * To protect us a bit look around and possibly write the content as such */
1291 jnotknown:
1292    switch (*s) {
1293    case '!':
1294    case '?':
1295       /* Ignore <!DOCTYPE, <!-- comments, <? PIs.. */
1296       goto jleave;
1297    case '>':
1298       /* Print out an empty tag as such */
1299       if (s[1] == '\0') {
1300          --s;
1301          goto jput_as_is;
1302       }
1303       break;
1304    case '/':
1305       ++s;
1306       break;
1307    default:
1308       break;
1309    }
1310
1311    /* Also skip over : in order to suppress v:roundrect, w:anchorlock.. */
1312    while ((c = *s++) != '\0' && c != '>' && !whitechar(c) && c != ':')
1313       if (!asciichar(c) || punctchar(c)) {
1314          self = _hf_puts(self, self->hf_bdat);
1315          break;
1316       }
1317    goto jleave;
1318 }
1319
1320 static struct htmlflt *
1321 _hf_check_ent(struct htmlflt *self, char const *s, size_t l)
1322 {
1323    char nobuf[32];
1324    char const *s_save;
1325    size_t l_save;
1326    struct hf_ent const *hfep;
1327    long i;
1328    NYD2_ENTER;
1329
1330    s_save = s;
1331    l_save = l;
1332    assert(*s == '&');
1333    assert(l > 0);
1334    /* False entities seen in the wild assert(s[l - 1] == ';'); */
1335    ++s;
1336    l -= 2;
1337
1338    /* Numeric entity, or try named search */
1339    if (*s == '#') {
1340       i = (*++s == 'x' ? 16 : 10);
1341
1342       if ((i != 16 || (++s, --l) > 0) && l < sizeof(nobuf)) {
1343          memcpy(nobuf, s, l);
1344          nobuf[l] = '\0';
1345          i = strtol(nobuf, NULL, i);
1346          if (i <= 0x7F)
1347             self = _hf_putc(self, (char)i);
1348          else if (self->hf_flags & _HF_UTF8) {
1349 jputuni:
1350             l = n_utf32_to_utf8((ui32_t)i, nobuf);
1351             self = _hf_putbuf(self, nobuf, l);
1352          } else
1353             goto jeent;
1354       } else
1355          goto jeent;
1356    } else {
1357       ui32_t f = self->hf_flags, hf;
1358
1359       for (hfep = _hf_ents; PTRCMP(hfep, <, _hf_ents + n_NELEM(_hf_ents));
1360             ++hfep)
1361          if (l == ((hf = hfep->hfe_flags) & _HFE_LENGTH_MASK) &&
1362                !strncmp(s, hfep->hfe_ent, l)) {
1363             if ((hf & _HFE_HAVE_UNI) && (f & _HF_UTF8)) {
1364                i = hfep->hfe_uni;
1365                goto jputuni;
1366             } else if (hf & _HFE_HAVE_CSTR)
1367                self = _hf_puts(self, hfep->hfe_cstr);
1368             else
1369                self = _hf_putc(self, hfep->hfe_c);
1370             goto jleave;
1371          }
1372 jeent:
1373       self = _hf_putbuf(self, s_save, l_save);
1374    }
1375 jleave:
1376    NYD2_LEAVE;
1377    return self;
1378 }
1379
1380 static ssize_t
1381 _hf_add_data(struct htmlflt *self, char const *dat, size_t len)
1382 {
1383    char c, *cp, *cp_max;
1384    bool_t hot;
1385    ssize_t rv = 0;
1386    NYD_ENTER;
1387
1388    /* Final put request? */
1389    if (dat == NULL) {
1390       if (self->hf_len > 0 || self->hf_hrefs != NULL) {
1391          self = _hf_dump(self);
1392          if (self->hf_hrefs != NULL)
1393             self = _hf_dump_hrefs(self);
1394          rv = 1;
1395       }
1396       goto jleave;
1397    }
1398
1399    /* Always ensure some initial buffer */
1400    if ((cp = self->hf_curr) != NULL)
1401       cp_max = self->hf_bmax;
1402    else {
1403       cp = self->hf_curr = self->hf_bdat = smalloc(LINESIZE);
1404       cp_max = self->hf_bmax = cp + LINESIZE -1; /* (Always room for NUL!) */
1405    }
1406    hot = (cp != self->hf_bdat);
1407
1408    for (rv = (ssize_t)len; len > 0; --len) {
1409       ui32_t f = self->hf_flags;
1410
1411       if (f & _HF_ERROR)
1412          break;
1413       c = *dat++;
1414
1415       /* Soup is really weird, and scripts may contain almost anything (and
1416        * newer CSS standards are also cryptic): therefore prefix the _HF_IGN
1417        * test and walk until we see the required end tag */
1418       /* TODO For real safety _HF_IGN soup condome would also need to know
1419        * TODO about quoted strings so that 'var i = "</script>";' couldn't
1420        * TODO fool it!   We really want this mode also for _HF_NOPUT to be
1421        * TODO able to *gracefully* detect the tag-closing '>', but then if
1422        * TODO that is a single mechanism we should have made it! */
1423       if (f & _HF_IGN) {
1424          struct htmlflt_tag const *hftp = self->hf_ign_tag;
1425          size_t i;
1426
1427          if (c == '<') {
1428             hot = TRU1;
1429 jcp_reset:
1430             cp = self->hf_bdat;
1431          } else if (c == '>') {
1432             if (hot) {
1433                if ((i = PTR2SIZE(cp - self->hf_bdat)) > 1 &&
1434                      --i == hftp->hft_len &&
1435                      !ascncasecmp(self->hf_bdat + 1, hftp->hft_tag, i))
1436                   self->hf_flags = (f &= ~(_HF_IGN | _HF_NOPUT));
1437                hot = FAL0;
1438                goto jcp_reset;
1439             }
1440          } else if (hot) {
1441             *cp++ = c;
1442             i = PTR2SIZE(cp - self->hf_bdat);
1443             if ((i == 1 && c != '/') || --i > hftp->hft_len) {
1444                hot = FAL0;
1445                goto jcp_reset;
1446             }
1447          }
1448       } else switch (c) {
1449       case '<':
1450          /* People are using & without &amp;ing it, ditto <; be aware */
1451          if (f & (_HF_NOPUT | _HF_ENT)) {
1452             f &= ~_HF_ENT;
1453             /* Special case "<!--" buffer content to deal with really weird
1454              * things that can be done with "<!--[if gte mso 9]>" syntax */
1455             if (PTR2SIZE(cp - self->hf_bdat) != 4 ||
1456                   memcmp(self->hf_bdat, "<!--", 4)) {
1457                self->hf_flags = f;
1458                *cp = '\0';
1459                self = _hf_puts(self, self->hf_bdat);
1460                f = self->hf_flags;
1461             }
1462          }
1463          cp = self->hf_bdat;
1464          *cp++ = c;
1465          self->hf_flags = (f |= _HF_NOPUT);
1466          break;
1467       case '>':
1468          /* Weird tagsoup around, do we actually parse a tag? */
1469          if (!(f & _HF_NOPUT))
1470             goto jdo_c;
1471          cp[0] = c;
1472          cp[1] = '\0';
1473          f &= ~(_HF_NOPUT | _HF_ENT);
1474          self->hf_flags = f;
1475          self = _hf_check_tag(self, self->hf_bdat);
1476          *(cp = self->hf_bdat) = '\0'; /* xxx extra safety */
1477          /* Quick hack to get rid of redundant newline after <pre> XXX */
1478          if (!(f & _HF_PRE) && (self->hf_flags & _HF_PRE) &&
1479                len > 1 && *dat == '\n')
1480             ++dat, --len;
1481          break;
1482
1483       case '\r': /* TODO CR should be stripped in lower level!! (Only B64!?!) */
1484          break;
1485       case '\n':
1486          /* End of line is not considered unless we are in PRE section.
1487           * However, in _HF_NOPUT mode we must be aware of tagsoup which uses
1488           * newlines for separating parameters */
1489          if (f & _HF_NOPUT)
1490             goto jdo_c;
1491          self = (f & _HF_PRE) ? _hf_nl_force(self) : _hf_putc(self, ' ');
1492          break;
1493
1494       case '\t':
1495          if (!(f & _HF_PRE))
1496             c = ' ';
1497          /* FALLTHRU */
1498       default:
1499 jdo_c:
1500          /* If not currently parsing a tag and bypassing normal output.. */
1501          if (!(f & _HF_NOPUT)) {
1502             if (cntrlchar(c))
1503                break;
1504             if (c == '&') {
1505                cp = self->hf_bdat;
1506                *cp++ = c;
1507                self->hf_flags = (f |= _HF_NOPUT | _HF_ENT);
1508             } else if (f & _HF_PRE) {
1509                self = _hf_putc_premode(self, c);
1510                self->hf_flags &= ~_HF_BLANK;
1511             } else
1512               self = _hf_putc(self, c);
1513          } else if ((f & _HF_ENT) && c == ';') {
1514             cp[0] = c;
1515             cp[1] = '\0';
1516             f &= ~(_HF_NOPUT | _HF_ENT);
1517             self->hf_flags = f;
1518            self = _hf_check_ent(self, self->hf_bdat,
1519                PTR2SIZE(cp + 1 - self->hf_bdat));
1520          } else {
1521             /* We may need to grow the buffer */
1522             if (PTRCMP(cp + 42/2, >=, cp_max)) {
1523                size_t i = PTR2SIZE(cp - self->hf_bdat),
1524                   m = PTR2SIZE(self->hf_bmax - self->hf_bdat) + LINESIZE;
1525
1526                cp = self->hf_bdat = srealloc(self->hf_bdat, m);
1527                self->hf_bmax = cp + m -1;
1528                self->hf_curr = (cp += i);
1529             }
1530             *cp++ = c;
1531          }
1532       }
1533    }
1534    self->hf_curr = cp;
1535 jleave:
1536   NYD_LEAVE;
1537   return (self->hf_flags & _HF_ERROR) ? -1 : rv;
1538 }
1539
1540 /*
1541  * TODO Because we don't support filter chains yet this filter will be run
1542  * TODO in a dedicated subprocess, driven via a special Popen() mode
1543  */
1544 static bool_t __hf_hadpipesig;
1545 static void
1546 __hf_onpipe(int signo)
1547 {
1548    NYD_X; /* Signal handler */
1549    n_UNUSED(signo);
1550    __hf_hadpipesig = TRU1;
1551 }
1552
1553 FL int
1554 htmlflt_process_main(void)
1555 {
1556    char buf[BUFFER_SIZE];
1557    struct htmlflt hf;
1558    size_t i;
1559    int rv;
1560    NYD_ENTER;
1561
1562    __hf_hadpipesig = FAL0;
1563    safe_signal(SIGPIPE, &__hf_onpipe);
1564
1565    htmlflt_init(&hf);
1566    htmlflt_reset(&hf, stdout);
1567
1568    for (;;) {
1569       if ((i = fread(buf, sizeof(buf[0]), n_NELEM(buf), stdin)) == 0) {
1570          rv = !feof(stdin);
1571          break;
1572       }
1573
1574       if ((rv = __hf_hadpipesig))
1575          break;
1576       /* Just use this directly.. */
1577       if (htmlflt_push(&hf, buf, i) < 0) {
1578          rv = 1;
1579          break;
1580       }
1581    }
1582    if (rv == 0 && htmlflt_flush(&hf) < 0)
1583       rv = 1;
1584
1585    htmlflt_destroy(&hf);
1586
1587    rv |= __hf_hadpipesig;
1588    NYD_LEAVE;
1589    return rv;
1590 }
1591
1592 FL void
1593 htmlflt_init(struct htmlflt *self)
1594 {
1595    NYD_ENTER;
1596    /* (Rather redundant though) */
1597    memset(self, 0, sizeof *self);
1598    NYD_LEAVE;
1599 }
1600
1601 FL void
1602 htmlflt_destroy(struct htmlflt *self)
1603 {
1604    NYD_ENTER;
1605    htmlflt_reset(self, NULL);
1606    NYD_LEAVE;
1607 }
1608
1609 FL void
1610 htmlflt_reset(struct htmlflt *self, FILE *f)
1611 {
1612    struct htmlflt_href *hfhp;
1613    NYD_ENTER;
1614
1615    while ((hfhp = self->hf_hrefs) != NULL) {
1616       self->hf_hrefs = hfhp->hfh_next;
1617       free(hfhp);
1618    }
1619
1620    if (self->hf_bdat != NULL)
1621       free(self->hf_bdat);
1622    if (self->hf_line != NULL)
1623       free(self->hf_line);
1624
1625    memset(self, 0, sizeof *self);
1626
1627    if (f != NULL) {
1628       ui32_t sw = n_MAX(_HF_MINLEN, (ui32_t)scrnwidth);
1629
1630       self->hf_line = smalloc((size_t)sw * mb_cur_max +1);
1631       self->hf_lmax = sw;
1632
1633       if (options & OPT_UNICODE) /* TODO not truly generic */
1634          self->hf_flags = _HF_UTF8;
1635       self->hf_os = f;
1636    }
1637    NYD_LEAVE;
1638 }
1639
1640 FL ssize_t
1641 htmlflt_push(struct htmlflt *self, char const *dat, size_t len)
1642 {
1643    ssize_t rv;
1644    NYD_ENTER;
1645
1646    rv = _hf_add_data(self, dat, len);
1647    NYD_LEAVE;
1648    return rv;
1649 }
1650
1651 FL ssize_t
1652 htmlflt_flush(struct htmlflt *self)
1653 {
1654    ssize_t rv;
1655    NYD_ENTER;
1656
1657    rv = _hf_add_data(self, NULL, 0);
1658    rv |= !fflush(self->hf_os) ? 0 : -1;
1659    NYD_LEAVE;
1660    return rv;
1661 }
1662 #endif /* HAVE_FILTER_HTML_TAGSOUP */
1663
1664 /* s-it-mode */