=================================================================== RCS file: /cvs/mandoc/term.c,v retrieving revision 1.261 retrieving revision 1.267 diff -u -p -r1.261 -r1.267 --- mandoc/term.c 2017/06/01 19:05:37 1.261 +++ mandoc/term.c 2017/06/07 20:01:19 1.267 @@ -1,4 +1,4 @@ -/* $Id: term.c,v 1.261 2017/06/01 19:05:37 schwarze Exp $ */ +/* $Id: term.c,v 1.267 2017/06/07 20:01:19 schwarze Exp $ */ /* * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons * Copyright (c) 2010-2017 Ingo Schwarze @@ -32,17 +32,19 @@ #include "main.h" static size_t cond_width(const struct termp *, int, int *); -static void adjbuf(struct termp *p, size_t); +static void adjbuf(struct termp_col *, size_t); static void bufferc(struct termp *, char); static void encode(struct termp *, const char *, size_t); static void encode1(struct termp *, int); +static void endline(struct termp *); void term_free(struct termp *p) { - - free(p->buf); + for (p->tcol = p->tcols; p->tcol < p->tcols + p->maxtcol; p->tcol++) + free(p->tcol->buf); + free(p->tcols); free(p->fontq); free(p); } @@ -84,69 +86,53 @@ term_end(struct termp *p) * to be broken, start the next line at the right margin instead * of at the offset. Used together with TERMP_NOBREAK for the tags * in various kinds of tagged lists. - * - TERMP_DANGLE: Do not break the output line at the right margin, + * - TERMP_HANG: Do not break the output line at the right margin, * append the next chunk after it even if this one is too long. * To be used together with TERMP_NOBREAK. - * - TERMP_HANG: Like TERMP_DANGLE, and also suppress padding before - * the next chunk if this column is not full. + * - TERMP_NOPAD: Start writing at the current position, + * do not pad with blank characters up to the offset. */ void term_flushln(struct termp *p) { - size_t i; /* current input position in p->buf */ - int ntab; /* number of tabs to prepend */ size_t vis; /* current visual position on output */ size_t vbl; /* number of blanks to prepend to output */ size_t vend; /* end of word visual position on output */ size_t bp; /* visual right border position */ size_t dv; /* temporary for visual pos calculations */ - size_t j; /* temporary loop index for p->buf */ + size_t j; /* temporary loop index for p->tcol->buf */ size_t jhy; /* last hyph before overflow w/r/t j */ size_t maxvis; /* output position of visible boundary */ + int ntab; /* number of tabs to prepend */ - /* - * First, establish the maximum columns of "visible" content. - * This is usually the difference between the right-margin and - * an indentation, but can be, for tagged lists or columns, a - * small set of values. - * - * The following unsigned-signed subtractions look strange, - * but they are actually correct. If the int p->overstep - * is negative, it gets sign extended. Subtracting that - * very large size_t effectively adds a small number to dv. - */ - dv = p->rmargin > p->offset ? p->rmargin - p->offset : 0; - maxvis = (int)dv > p->overstep ? dv - (size_t)p->overstep : 0; - - if (p->flags & TERMP_NOBREAK) { - dv = p->maxrmargin > p->offset ? - p->maxrmargin - p->offset : 0; - bp = (int)dv > p->overstep ? - dv - (size_t)p->overstep : 0; - } else - bp = maxvis; - - /* - * Calculate the required amount of padding. - */ - vbl = p->offset + p->overstep > p->viscol ? - p->offset + p->overstep - p->viscol : 0; - + vbl = (p->flags & TERMP_NOPAD) || p->tcol->offset < p->viscol ? + 0 : p->tcol->offset - p->viscol; + if (p->minbl && vbl < p->minbl) + vbl = p->minbl; + maxvis = p->tcol->rmargin > p->viscol + vbl ? + p->tcol->rmargin - p->viscol - vbl : 0; + bp = !(p->flags & TERMP_NOBREAK) ? maxvis : + p->maxrmargin > p->viscol + vbl ? + p->maxrmargin - p->viscol - vbl : 0; vis = vend = 0; - i = 0; - while (i < p->col) { + if (p->lasttcol == 0) + p->tcol->col = 0; + while (p->tcol->col < p->lastcol) { + /* * Handle literal tab characters: collapse all * subsequent tabs into a single huge set of spaces. */ + ntab = 0; - while (i < p->col && p->buf[i] == '\t') { + while (p->tcol->col < p->lastcol && + p->tcol->buf[p->tcol->col] == '\t') { vend = term_tab_next(vis); vbl += vend - vis; vis = vend; ntab++; - i++; + p->tcol->col++; } /* @@ -156,42 +142,47 @@ term_flushln(struct termp *p) * space is printed according to regular spacing rules). */ - for (j = i, jhy = 0; j < p->col; j++) { - if (' ' == p->buf[j] || '\t' == p->buf[j]) + jhy = 0; + for (j = p->tcol->col; j < p->lastcol; j++) { + if (p->tcol->buf[j] == ' ' || p->tcol->buf[j] == '\t') break; /* Back over the last printed character. */ - if (8 == p->buf[j]) { + if (p->tcol->buf[j] == '\b') { assert(j); - vend -= (*p->width)(p, p->buf[j - 1]); + vend -= (*p->width)(p, p->tcol->buf[j - 1]); continue; } /* Regular word. */ /* Break at the hyphen point if we overrun. */ if (vend > vis && vend < bp && - (ASCII_HYPH == p->buf[j] || - ASCII_BREAK == p->buf[j])) + (p->tcol->buf[j] == ASCII_HYPH|| + p->tcol->buf[j] == ASCII_BREAK)) jhy = j; /* * Hyphenation now decided, put back a real * hyphen such that we get the correct width. */ - if (ASCII_HYPH == p->buf[j]) - p->buf[j] = '-'; + if (p->tcol->buf[j] == ASCII_HYPH) + p->tcol->buf[j] = '-'; - vend += (*p->width)(p, p->buf[j]); + vend += (*p->width)(p, p->tcol->buf[j]); } /* * Find out whether we would exceed the right margin. * If so, break to the next line. */ - if (vend > bp && 0 == jhy && vis > 0) { + + if (vend > bp && jhy == 0 && vis > 0 && + (p->flags & TERMP_BRNEVER) == 0) { + if (p->lasttcol) + return; + + endline(p); vend -= vis; - (*p->endline)(p); - p->viscol = 0; /* Use pending tabs on the new line. */ @@ -201,42 +192,40 @@ term_flushln(struct termp *p) /* Re-establish indentation. */ - if (p->flags & TERMP_BRIND) { - vbl += p->rmargin; - vend += p->rmargin - p->offset; - } else - vbl += p->offset; - - /* - * Remove the p->overstep width. - * Again, if p->overstep is negative, - * sign extension does the right thing. - */ - - bp += (size_t)p->overstep; - p->overstep = 0; + if (p->flags & TERMP_BRIND) + vbl += p->tcol->rmargin; + else + vbl += p->tcol->offset; + maxvis = p->tcol->rmargin > vbl ? + p->tcol->rmargin - vbl : 0; + bp = !(p->flags & TERMP_NOBREAK) ? maxvis : + p->maxrmargin > vbl ? p->maxrmargin - vbl : 0; } - /* Write out the [remaining] word. */ - for ( ; i < p->col; i++) { - if (vend > bp && jhy > 0 && i > jhy) + /* + * Write out the rest of the word. + */ + + for ( ; p->tcol->col < p->lastcol; p->tcol->col++) { + if (vend > bp && jhy > 0 && p->tcol->col > jhy) break; - if ('\t' == p->buf[i]) + if (p->tcol->buf[p->tcol->col] == '\t') break; - if (' ' == p->buf[i]) { - j = i; - while (i < p->col && ' ' == p->buf[i]) - i++; - dv = (i - j) * (*p->width)(p, ' '); + if (p->tcol->buf[p->tcol->col] == ' ') { + j = p->tcol->col; + while (p->tcol->col < p->lastcol && + p->tcol->buf[p->tcol->col] == ' ') + p->tcol->col++; + dv = (p->tcol->col - j) * (*p->width)(p, ' '); vbl += dv; vend += dv; break; } - if (ASCII_NBRSP == p->buf[i]) { + if (p->tcol->buf[p->tcol->col] == ASCII_NBRSP) { vbl += (*p->width)(p, ' '); continue; } - if (ASCII_BREAK == p->buf[i]) + if (p->tcol->buf[p->tcol->col] == ASCII_BREAK) continue; /* @@ -250,11 +239,13 @@ term_flushln(struct termp *p) vbl = 0; } - (*p->letter)(p, p->buf[i]); - if (8 == p->buf[i]) - p->viscol -= (*p->width)(p, p->buf[i-1]); + (*p->letter)(p, p->tcol->buf[p->tcol->col]); + if (p->tcol->buf[p->tcol->col] == '\b') + p->viscol -= (*p->width)(p, + p->tcol->buf[p->tcol->col - 1]); else - p->viscol += (*p->width)(p, p->buf[i]); + p->viscol += (*p->width)(p, + p->tcol->buf[p->tcol->col]); } vis = vend; } @@ -263,48 +254,45 @@ term_flushln(struct termp *p) * If there was trailing white space, it was not printed; * so reset the cursor position accordingly. */ + if (vis > vbl) vis -= vbl; else vis = 0; - p->col = 0; - p->overstep = 0; - p->flags &= ~(TERMP_BACKAFTER | TERMP_BACKBEFORE); + p->col = p->lastcol = 0; + p->minbl = p->trailspace; + p->flags &= ~(TERMP_BACKAFTER | TERMP_BACKBEFORE | TERMP_NOPAD); - if ( ! (TERMP_NOBREAK & p->flags)) { - p->viscol = 0; - (*p->endline)(p); - return; - } - - if (TERMP_HANG & p->flags) { - p->overstep += (int)(p->offset + vis - p->rmargin + - p->trailspace * (*p->width)(p, ' ')); - - /* - * If we have overstepped the margin, temporarily move - * it to the right and flag the rest of the line to be - * shorter. - * If there is a request to keep the columns together, - * allow negative overstep when the column is not full. - */ - if (p->trailspace && p->overstep < 0) - p->overstep = 0; - return; - - } else if (TERMP_DANGLE & p->flags) - return; - /* Trailing whitespace is significant in some columns. */ + if (vis && vbl && (TERMP_BRTRSP & p->flags)) vis += vbl; /* If the column was overrun, break the line. */ - if (maxvis < vis + p->trailspace * (*p->width)(p, ' ')) { - (*p->endline)(p); - p->viscol = 0; + if ((p->flags & TERMP_NOBREAK) == 0 || + ((p->flags & TERMP_HANG) == 0 && + vis + p->trailspace * (*p->width)(p, ' ') > maxvis)) + endline(p); +} + +static void +endline(struct termp *p) +{ + if ((p->flags & (TERMP_NEWMC | TERMP_ENDMC)) == TERMP_ENDMC) { + p->mc = NULL; + p->flags &= ~TERMP_ENDMC; } + if (p->mc != NULL) { + if (p->viscol && p->maxrmargin >= p->viscol) + (*p->advance)(p, p->maxrmargin - p->viscol + 1); + p->flags |= TERMP_NOBUF | TERMP_NOSPACE; + term_word(p, p->mc); + p->flags &= ~(TERMP_NOBUF | TERMP_NEWMC); + } + p->viscol = 0; + p->minbl = 0; + (*p->endline)(p); } /* @@ -317,7 +305,7 @@ term_newln(struct termp *p) { p->flags |= TERMP_NOSPACE; - if (p->col || p->viscol) + if (p->lastcol || p->viscol) term_flushln(p); } @@ -333,6 +321,7 @@ term_vspace(struct termp *p) term_newln(p); p->viscol = 0; + p->minbl = 0; if (0 < p->skipvsp) p->skipvsp--; else @@ -404,28 +393,28 @@ term_word(struct termp *p, const char *word) const char nbrsp[2] = { ASCII_NBRSP, 0 }; const char *seq, *cp; int sz, uc; - size_t ssz; + size_t csz, lsz, ssz; enum mandoc_esc esc; - if ( ! (TERMP_NOSPACE & p->flags)) { - if ( ! (TERMP_KEEP & p->flags)) { - bufferc(p, ' '); - if (TERMP_SENTENCE & p->flags) + if ((p->flags & TERMP_NOBUF) == 0) { + if ((p->flags & TERMP_NOSPACE) == 0) { + if ((p->flags & TERMP_KEEP) == 0) { bufferc(p, ' '); - } else - bufferc(p, ASCII_NBRSP); + if (p->flags & TERMP_SENTENCE) + bufferc(p, ' '); + } else + bufferc(p, ASCII_NBRSP); + } + if (p->flags & TERMP_PREKEEP) + p->flags |= TERMP_KEEP; + if (p->flags & TERMP_NONOSPACE) + p->flags |= TERMP_NOSPACE; + else + p->flags &= ~TERMP_NOSPACE; + p->flags &= ~(TERMP_SENTENCE | TERMP_NONEWLINE); + p->skipvsp = 0; } - if (TERMP_PREKEEP & p->flags) - p->flags |= TERMP_KEEP; - if ( ! (p->flags & TERMP_NONOSPACE)) - p->flags &= ~TERMP_NOSPACE; - else - p->flags |= TERMP_NOSPACE; - - p->flags &= ~(TERMP_SENTENCE | TERMP_NONEWLINE); - p->skipvsp = 0; - while ('\0' != *word) { if ('\\' != *word) { if (TERMP_NBRWORD & p->flags) { @@ -501,15 +490,71 @@ term_word(struct termp *p, const char *word) else { uc += p->col; p->col = 0; - if (p->offset > (size_t)(-uc)) { + if (p->tcol->offset > (size_t)(-uc)) { p->ti += uc; - p->offset += uc; + p->tcol->offset += uc; } else { - p->ti -= p->offset; - p->offset = 0; + p->ti -= p->tcol->offset; + p->tcol->offset = 0; } } continue; + case ESCAPE_HLINE: + if (a2roffsu(seq, &su, SCALE_EM) == 0) + continue; + uc = term_hspan(p, &su) / 24; + if (uc <= 0) { + if (p->tcol->rmargin <= p->tcol->offset) + continue; + lsz = p->tcol->rmargin - p->tcol->offset; + } else + lsz = uc; + while (sz && + strchr(" %&()*+-./0123456789:<=>", *seq)) { + seq++; + sz--; + } + if (sz && strchr("cifMmnPpuv", *seq)) { + seq++; + sz--; + } + if (sz == 0) + uc = -1; + else if (*seq == '\\') { + seq++; + esc = mandoc_escape(&seq, &cp, &sz); + switch (esc) { + case ESCAPE_UNICODE: + uc = mchars_num2uc(cp + 1, sz - 1); + break; + case ESCAPE_NUMBERED: + uc = mchars_num2char(cp, sz); + break; + case ESCAPE_SPECIAL: + uc = mchars_spec2cp(cp, sz); + break; + default: + uc = -1; + break; + } + } else + uc = *seq; + if (uc < 0x20 || (uc > 0x7E && uc < 0xA0)) + uc = '_'; + if (p->enc == TERMENC_ASCII) { + cp = ascii_uc2str(uc); + csz = term_strlen(p, cp); + ssz = strlen(cp); + } else + csz = (*p->width)(p, uc); + while (lsz >= csz) { + if (p->enc == TERMENC_ASCII) + encode(p, cp, ssz); + else + encode1(p, uc); + lsz -= csz; + } + continue; case ESCAPE_SKIPCHAR: p->flags |= TERMP_BACKAFTER; continue; @@ -529,10 +574,12 @@ term_word(struct termp *p, const char *word) } } /* Trim trailing backspace/blank pair. */ - if (p->col > 2 && - (p->buf[p->col - 1] == ' ' || - p->buf[p->col - 1] == '\t')) - p->col -= 2; + if (p->lastcol > 2 && + (p->tcol->buf[p->lastcol - 1] == ' ' || + p->tcol->buf[p->lastcol - 1] == '\t')) + p->lastcol -= 2; + if (p->col > p->lastcol) + p->col = p->lastcol; continue; default: continue; @@ -557,25 +604,28 @@ term_word(struct termp *p, const char *word) } static void -adjbuf(struct termp *p, size_t sz) +adjbuf(struct termp_col *c, size_t sz) { - - if (0 == p->maxcols) - p->maxcols = 1024; - while (sz >= p->maxcols) - p->maxcols <<= 2; - - p->buf = mandoc_reallocarray(p->buf, p->maxcols, sizeof(int)); + if (c->maxcols == 0) + c->maxcols = 1024; + while (c->maxcols <= sz) + c->maxcols <<= 2; + c->buf = mandoc_reallocarray(c->buf, c->maxcols, sizeof(*c->buf)); } static void bufferc(struct termp *p, char c) { - - if (p->col + 1 >= p->maxcols) - adjbuf(p, p->col + 1); - - p->buf[p->col++] = c; + if (p->flags & TERMP_NOBUF) { + (*p->letter)(p, c); + return; + } + if (p->col + 1 >= p->tcol->maxcols) + adjbuf(p->tcol, p->col + 1); + if (p->lastcol <= p->col || (c != ' ' && c != ASCII_NBRSP)) + p->tcol->buf[p->col] = c; + if (p->lastcol < ++p->col) + p->lastcol = p->col; } /* @@ -588,31 +638,40 @@ encode1(struct termp *p, int c) { enum termfont f; - if (p->col + 7 >= p->maxcols) - adjbuf(p, p->col + 7); + if (p->flags & TERMP_NOBUF) { + (*p->letter)(p, c); + return; + } + if (p->col + 7 >= p->tcol->maxcols) + adjbuf(p->tcol, p->col + 7); + f = (c == ASCII_HYPH || c > 127 || isgraph(c)) ? p->fontq[p->fonti] : TERMFONT_NONE; if (p->flags & TERMP_BACKBEFORE) { - if (p->buf[p->col - 1] == ' ' || p->buf[p->col - 1] == '\t') + if (p->tcol->buf[p->col - 1] == ' ' || + p->tcol->buf[p->col - 1] == '\t') p->col--; else - p->buf[p->col++] = 8; + p->tcol->buf[p->col++] = '\b'; p->flags &= ~TERMP_BACKBEFORE; } - if (TERMFONT_UNDER == f || TERMFONT_BI == f) { - p->buf[p->col++] = '_'; - p->buf[p->col++] = 8; + if (f == TERMFONT_UNDER || f == TERMFONT_BI) { + p->tcol->buf[p->col++] = '_'; + p->tcol->buf[p->col++] = '\b'; } - if (TERMFONT_BOLD == f || TERMFONT_BI == f) { - if (ASCII_HYPH == c) - p->buf[p->col++] = '-'; + if (f == TERMFONT_BOLD || f == TERMFONT_BI) { + if (c == ASCII_HYPH) + p->tcol->buf[p->col++] = '-'; else - p->buf[p->col++] = c; - p->buf[p->col++] = 8; + p->tcol->buf[p->col++] = c; + p->tcol->buf[p->col++] = '\b'; } - p->buf[p->col++] = c; + if (p->lastcol <= p->col || (c != ' ' && c != ASCII_NBRSP)) + p->tcol->buf[p->col] = c; + if (p->lastcol < ++p->col) + p->lastcol = p->col; if (p->flags & TERMP_BACKAFTER) { p->flags |= TERMP_BACKBEFORE; p->flags &= ~TERMP_BACKAFTER; @@ -624,15 +683,24 @@ encode(struct termp *p, const char *word, size_t sz) { size_t i; - if (p->col + 2 + (sz * 5) >= p->maxcols) - adjbuf(p, p->col + 2 + (sz * 5)); + if (p->flags & TERMP_NOBUF) { + for (i = 0; i < sz; i++) + (*p->letter)(p, word[i]); + return; + } + if (p->col + 2 + (sz * 5) >= p->tcol->maxcols) + adjbuf(p->tcol, p->col + 2 + (sz * 5)); + for (i = 0; i < sz; i++) { if (ASCII_HYPH == word[i] || isgraph((unsigned char)word[i])) encode1(p, word[i]); else { - p->buf[p->col++] = word[i]; + if (p->lastcol <= p->col || + (word[i] != ' ' && word[i] != ASCII_NBRSP)) + p->tcol->buf[p->col] = word[i]; + p->col++; /* * Postpone the effect of \z while handling @@ -646,6 +714,8 @@ encode(struct termp *p, const char *word, size_t sz) } } } + if (p->lastcol < p->col) + p->lastcol = p->col; } void