-/* $Id: term.c,v 1.269 2017/06/12 19:05:47 schwarze Exp $ */
+/* $Id: term.c,v 1.285 2022/01/10 18:01:35 schwarze Exp $ */
/*
+ * Copyright (c) 2010-2022 Ingo Schwarze <schwarze@openbsd.org>
* Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
- * Copyright (c) 2010-2017 Ingo Schwarze <schwarze@openbsd.org>
*
* Permission to use, copy, modify, and distribute this software for any
* purpose with or without fee is hereby granted, provided that the above
#include <assert.h>
#include <ctype.h>
+#include <stdint.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
static void encode(struct termp *, const char *, size_t);
static void encode1(struct termp *, int);
static void endline(struct termp *);
+static void term_field(struct termp *, size_t, size_t);
+static void term_fill(struct termp *, size_t *, size_t *,
+ size_t);
void
void
term_free(struct termp *p)
{
+ term_tab_free();
for (p->tcol = p->tcols; p->tcol < p->tcols + p->maxtcol; p->tcol++)
free(p->tcol->buf);
free(p->tcols);
* Flush a chunk of text. By default, break the output line each time
* the right margin is reached, and continue output on the next line
* at the same offset as the chunk itself. By default, also break the
- * output line at the end of the chunk.
- * The following flags may be specified:
- *
- * - TERMP_NOBREAK: Do not break the output line at the right margin,
- * but only at the max right margin. Also, do not break the output
- * line at the end of the chunk, such that the next call can pad to
- * the next column. However, if less than p->trailspace blanks,
- * which can be 0, 1, or 2, remain to the right margin, the line
- * will be broken.
- * - TERMP_BRTRSP: Consider trailing whitespace significant
- * when deciding whether the chunk fits or not.
- * - TERMP_BRIND: If the chunk does not fit and the output line has
- * 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_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_NOPAD: Start writing at the current position,
- * do not pad with blank characters up to the offset.
+ * output line at the end of the chunk. There are many flags modifying
+ * this behaviour, see the comments in the body of the function.
*/
void
term_flushln(struct termp *p)
{
- 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->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 */
+ size_t vbl; /* Number of blanks to prepend to the output. */
+ size_t vbr; /* Actual visual position of the end of field. */
+ size_t vfield; /* Desired visual field width. */
+ size_t vtarget; /* Desired visual position of the right margin. */
+ size_t ic; /* Character position in the input buffer. */
+ size_t nbr; /* Number of characters to print in this field. */
+
+ /*
+ * Normally, start writing at the left margin, but with the
+ * NOPAD flag, start writing at the current position instead.
+ */
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;
-
- if ((p->flags && TERMP_MULTICOL) == 0)
+
+ if ((p->flags & TERMP_MULTICOL) == 0)
p->tcol->col = 0;
- while (p->tcol->col < p->tcol->lastcol) {
+
+ /* Loop over output lines. */
+
+ for (;;) {
+ vfield = p->tcol->rmargin > p->viscol + vbl ?
+ p->tcol->rmargin - p->viscol - vbl : 0;
/*
- * Handle literal tab characters: collapse all
- * subsequent tabs into a single huge set of spaces.
+ * Normally, break the line at the the right margin
+ * of the field, but with the NOBREAK flag, only
+ * break it at the max right margin of the screen,
+ * and with the BRNEVER flag, never break it at all.
*/
- ntab = 0;
- while (p->tcol->col < p->tcol->lastcol &&
- p->tcol->buf[p->tcol->col] == '\t') {
- vend = term_tab_next(vis);
- vbl += vend - vis;
- vis = vend;
- ntab++;
- p->tcol->col++;
- }
+ vtarget = (p->flags & TERMP_NOBREAK) == 0 ? vfield :
+ p->maxrmargin > p->viscol + vbl ?
+ p->maxrmargin - p->viscol - vbl : 0;
/*
- * Count up visible word characters. Control sequences
- * (starting with the CSI) aren't counted. A space
- * generates a non-printing word, which is valid (the
- * space is printed according to regular spacing rules).
+ * Figure out how much text will fit in the field.
+ * If there is whitespace only, print nothing.
*/
- jhy = 0;
- for (j = p->tcol->col; j < p->tcol->lastcol; j++) {
- if (p->tcol->buf[j] == ' ' || p->tcol->buf[j] == '\t')
- break;
+ term_fill(p, &nbr, &vbr,
+ p->flags & TERMP_BRNEVER ? SIZE_MAX : vtarget);
+ if (nbr == 0)
+ break;
- /* Back over the last printed character. */
- if (p->tcol->buf[j] == '\b') {
- assert(j);
- vend -= (*p->width)(p, p->tcol->buf[j - 1]);
- continue;
- }
+ /*
+ * With the CENTER or RIGHT flag, increase the indentation
+ * to center the text between the left and right margins
+ * or to adjust it to the right margin, respectively.
+ */
- /* Regular word. */
- /* Break at the hyphen point if we overrun. */
- if (vend > vis && vend < bp &&
- (p->tcol->buf[j] == ASCII_HYPH||
- p->tcol->buf[j] == ASCII_BREAK))
- jhy = j;
+ if (vbr < vtarget) {
+ if (p->flags & TERMP_CENTER)
+ vbl += (vtarget - vbr) / 2;
+ else if (p->flags & TERMP_RIGHT)
+ vbl += vtarget - vbr;
+ }
- /*
- * Hyphenation now decided, put back a real
- * hyphen such that we get the correct width.
- */
- if (p->tcol->buf[j] == ASCII_HYPH)
- p->tcol->buf[j] = '-';
+ /* Finally, print the field content. */
- vend += (*p->width)(p, p->tcol->buf[j]);
- }
+ term_field(p, vbl, nbr);
/*
- * Find out whether we would exceed the right margin.
- * If so, break to the next line.
+ * If there is no text left in the field, exit the loop.
+ * If the BRTRSP flag is set, consider trailing
+ * whitespace significant when deciding whether
+ * the field fits or not.
*/
- if (vend > bp && jhy == 0 && vis > 0 &&
- (p->flags & TERMP_BRNEVER) == 0) {
- if (p->flags & TERMP_MULTICOL)
- return;
+ for (ic = p->tcol->col; ic < p->tcol->lastcol; ic++) {
+ switch (p->tcol->buf[ic]) {
+ case '\t':
+ if (p->flags & TERMP_BRTRSP)
+ vbr = term_tab_next(vbr);
+ continue;
+ case ' ':
+ if (p->flags & TERMP_BRTRSP)
+ vbr += (*p->width)(p, ' ');
+ continue;
+ case '\n':
+ case ASCII_BREAK:
+ continue;
+ default:
+ break;
+ }
+ break;
+ }
+ if (ic == p->tcol->lastcol)
+ break;
- endline(p);
- vend -= vis;
+ /*
+ * At the location of an automtic line break, input
+ * space characters are consumed by the line break.
+ */
- /* Use pending tabs on the new line. */
+ while (p->tcol->col < p->tcol->lastcol &&
+ p->tcol->buf[p->tcol->col] == ' ')
+ p->tcol->col++;
- vbl = 0;
- while (ntab--)
- vbl = term_tab_next(vbl);
+ /*
+ * In multi-column mode, leave the rest of the text
+ * in the buffer to be handled by a subsequent
+ * invocation, such that the other columns of the
+ * table can be handled first.
+ * In single-column mode, simply break the line.
+ */
- /* Re-establish indentation. */
+ if (p->flags & TERMP_MULTICOL)
+ return;
- 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;
- }
+ endline(p);
+ p->viscol = 0;
/*
- * Write out the rest of the word.
+ * Normally, start the next line at the same indentation
+ * as this one, but with the BRIND flag, start it at the
+ * right margin instead. This is used together with
+ * NOBREAK for the tags in various kinds of tagged lists.
*/
- for ( ; p->tcol->col < p->tcol->lastcol; p->tcol->col++) {
- if (vend > bp && jhy > 0 && p->tcol->col > jhy)
+ vbl = p->flags & TERMP_BRIND ?
+ p->tcol->rmargin : p->tcol->offset;
+ }
+
+ /* Reset output state in preparation for the next field. */
+
+ p->col = p->tcol->col = p->tcol->lastcol = 0;
+ p->minbl = p->trailspace;
+ p->flags &= ~(TERMP_BACKAFTER | TERMP_BACKBEFORE | TERMP_NOPAD);
+
+ if (p->flags & TERMP_MULTICOL)
+ return;
+
+ /*
+ * The HANG flag means that the next field
+ * always follows on the same line.
+ * The NOBREAK flag means that the next field
+ * follows on the same line unless the field was overrun.
+ * Normally, break the line at the end of each field.
+ */
+
+ if ((p->flags & TERMP_HANG) == 0 &&
+ ((p->flags & TERMP_NOBREAK) == 0 ||
+ vbr + term_len(p, p->trailspace) > vfield))
+ endline(p);
+}
+
+/*
+ * Store the number of input characters to print in this field in *nbr
+ * and their total visual width to print in *vbr.
+ * If there is only whitespace in the field, both remain zero.
+ * The desired visual width of the field is provided by vtarget.
+ * If the first word is longer, the field will be overrun.
+ */
+static void
+term_fill(struct termp *p, size_t *nbr, size_t *vbr, size_t vtarget)
+{
+ size_t ic; /* Character position in the input buffer. */
+ size_t vis; /* Visual position of the current character. */
+ size_t vn; /* Visual position of the next character. */
+ int breakline; /* Break at the end of this word. */
+ int graph; /* Last character was non-blank. */
+
+ *nbr = *vbr = vis = 0;
+ breakline = graph = 0;
+ for (ic = p->tcol->col; ic < p->tcol->lastcol; ic++) {
+ switch (p->tcol->buf[ic]) {
+ case '\b': /* Escape \o (overstrike) or backspace markup. */
+ assert(ic > 0);
+ vis -= (*p->width)(p, p->tcol->buf[ic - 1]);
+ continue;
+
+ case '\t': /* Normal ASCII whitespace. */
+ case ' ':
+ case ASCII_BREAK: /* Escape \: (breakpoint). */
+ switch (p->tcol->buf[ic]) {
+ case '\t':
+ vn = term_tab_next(vis);
break;
- if (p->tcol->buf[p->tcol->col] == '\t')
+ case ' ':
+ vn = vis + (*p->width)(p, ' ');
break;
- if (p->tcol->buf[p->tcol->col] == ' ') {
- j = p->tcol->col;
- while (p->tcol->col < p->tcol->lastcol &&
- p->tcol->buf[p->tcol->col] == ' ')
- p->tcol->col++;
- dv = (p->tcol->col - j) * (*p->width)(p, ' ');
- vbl += dv;
- vend += dv;
+ case ASCII_BREAK:
+ vn = vis;
break;
+ default:
+ abort();
}
- if (p->tcol->buf[p->tcol->col] == ASCII_NBRSP) {
- vbl += (*p->width)(p, ' ');
- continue;
+ /* Can break at the end of a word. */
+ if (breakline || vn > vtarget)
+ break;
+ if (graph) {
+ *nbr = ic;
+ *vbr = vis;
+ graph = 0;
}
- if (p->tcol->buf[p->tcol->col] == ASCII_BREAK)
- continue;
+ vis = vn;
+ continue;
+
+ case '\n': /* Escape \p (break at the end of the word). */
+ breakline = 1;
+ continue;
+ case ASCII_HYPH: /* Breakable hyphen. */
+ graph = 1;
/*
- * Now we definitely know there will be
- * printable characters to output,
- * so write preceding white space now.
+ * We are about to decide whether to break the
+ * line or not, so we no longer need this hyphen
+ * to be marked as breakable. Put back a real
+ * hyphen such that we get the correct width.
*/
- if (vbl) {
- (*p->advance)(p, vbl);
- p->viscol += vbl;
- vbl = 0;
+ p->tcol->buf[ic] = '-';
+ vis += (*p->width)(p, '-');
+ if (vis > vtarget) {
+ ic++;
+ break;
}
+ *nbr = ic + 1;
+ *vbr = vis;
+ continue;
- (*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->tcol->buf[p->tcol->col]);
+ case ASCII_NBRSP: /* Non-breakable space. */
+ p->tcol->buf[ic] = ' ';
+ /* FALLTHROUGH */
+ default: /* Printable character. */
+ graph = 1;
+ vis += (*p->width)(p, p->tcol->buf[ic]);
+ if (vis > vtarget && *nbr > 0)
+ return;
+ continue;
}
- vis = vend;
+ break;
}
/*
- * If there was trailing white space, it was not printed;
- * so reset the cursor position accordingly.
+ * If the last word extends to the end of the field without any
+ * trailing whitespace, the loop could not check yet whether it
+ * can remain on this line. So do the check now.
*/
- if (vis > vbl)
- vis -= vbl;
- else
- vis = 0;
+ if (graph && (vis <= vtarget || *nbr == 0)) {
+ *nbr = ic;
+ *vbr = vis;
+ }
+}
- p->col = p->tcol->col = p->tcol->lastcol = 0;
- p->minbl = p->trailspace;
- p->flags &= ~(TERMP_BACKAFTER | TERMP_BACKBEFORE | TERMP_NOPAD);
+/*
+ * Print the contents of one field
+ * with an indentation of vbl visual columns,
+ * and an input string length of nbr characters.
+ */
+static void
+term_field(struct termp *p, size_t vbl, size_t nbr)
+{
+ size_t ic; /* Character position in the input buffer. */
+ size_t vis; /* Visual position of the current character. */
+ size_t dv; /* Visual width of the current character. */
+ size_t vn; /* Visual position of the next character. */
- if (p->flags & TERMP_MULTICOL)
- return;
+ vis = 0;
+ for (ic = p->tcol->col; ic < nbr; ic++) {
- /* Trailing whitespace is significant in some columns. */
+ /*
+ * To avoid the printing of trailing whitespace,
+ * do not print whitespace right away, only count it.
+ */
- if (vis && vbl && (TERMP_BRTRSP & p->flags))
- vis += vbl;
+ switch (p->tcol->buf[ic]) {
+ case '\n':
+ case ASCII_BREAK:
+ continue;
+ case '\t':
+ vn = term_tab_next(vis);
+ vbl += vn - vis;
+ vis = vn;
+ continue;
+ case ' ':
+ case ASCII_NBRSP:
+ dv = (*p->width)(p, ' ');
+ vbl += dv;
+ vis += dv;
+ continue;
+ default:
+ break;
+ }
- /* If the column was overrun, break the line. */
- if ((p->flags & TERMP_NOBREAK) == 0 ||
- ((p->flags & TERMP_HANG) == 0 &&
- vis + p->trailspace * (*p->width)(p, ' ') > maxvis))
- endline(p);
+ /*
+ * We found a non-blank character to print,
+ * so write preceding white space now.
+ */
+
+ if (vbl > 0) {
+ (*p->advance)(p, vbl);
+ p->viscol += vbl;
+ vbl = 0;
+ }
+
+ /* Print the character and adjust the visual position. */
+
+ (*p->letter)(p, p->tcol->buf[ic]);
+ if (p->tcol->buf[ic] == '\b') {
+ dv = (*p->width)(p, p->tcol->buf[ic - 1]);
+ p->viscol -= dv;
+ vis -= dv;
+ } else {
+ dv = (*p->width)(p, p->tcol->buf[ic]);
+ p->viscol += dv;
+ vis += dv;
+ }
+ }
+ p->tcol->col = nbr;
}
static void
word++;
esc = mandoc_escape(&word, &seq, &sz);
- if (ESCAPE_ERROR == esc)
- continue;
-
switch (esc) {
case ESCAPE_UNICODE:
uc = mchars_num2uc(seq + 1, sz - 1);
encode1(p, uc);
}
continue;
+ case ESCAPE_UNDEF:
+ uc = *seq;
+ break;
case ESCAPE_FONTBOLD:
+ case ESCAPE_FONTCB:
term_fontrepl(p, TERMFONT_BOLD);
continue;
case ESCAPE_FONTITALIC:
+ case ESCAPE_FONTCI:
term_fontrepl(p, TERMFONT_UNDER);
continue;
case ESCAPE_FONTBI:
term_fontrepl(p, TERMFONT_BI);
continue;
case ESCAPE_FONT:
+ case ESCAPE_FONTCR:
case ESCAPE_FONTROMAN:
term_fontrepl(p, TERMFONT_NONE);
continue;
case ESCAPE_FONTPREV:
term_fontlast(p);
continue;
+ case ESCAPE_BREAK:
+ bufferc(p, '\n');
+ continue;
case ESCAPE_NOSPACE:
if (p->flags & TERMP_BACKAFTER)
p->flags &= ~TERMP_BACKAFTER;
else if (*word == '\0')
p->flags |= (TERMP_NOSPACE | TERMP_NONEWLINE);
continue;
+ case ESCAPE_DEVICE:
+ if (p->type == TERMTYPE_PDF)
+ encode(p, "pdf", 3);
+ else if (p->type == TERMTYPE_PS)
+ encode(p, "ps", 2);
+ else if (p->enc == TERMENC_ASCII)
+ encode(p, "ascii", 5);
+ else
+ encode(p, "utf8", 4);
+ continue;
case ESCAPE_HORIZ:
+ if (*seq == '|') {
+ seq++;
+ uc = -p->col;
+ } else
+ uc = 0;
if (a2roffsu(seq, &su, SCALE_EM) == NULL)
continue;
- uc = term_hspan(p, &su) / 24;
- if (uc > 0)
- while (uc-- > 0)
+ uc += term_hen(p, &su);
+ if (uc > 0) {
+ while (uc > 0) {
bufferc(p, ASCII_NBRSP);
- else if (p->col > (size_t)(-uc))
+ uc -= term_len(p, 1);
+ }
+ } else if (p->col > (size_t)(-uc)) {
p->col += uc;
- else {
+ } else {
uc += p->col;
p->col = 0;
if (p->tcol->offset > (size_t)(-uc)) {
}
continue;
case ESCAPE_HLINE:
- if ((seq = a2roffsu(seq, &su, SCALE_EM)) == NULL)
+ if ((cp = a2roffsu(seq, &su, SCALE_EM)) == NULL)
continue;
- uc = term_hspan(p, &su) / 24;
+ uc = term_hen(p, &su);
if (uc <= 0) {
if (p->tcol->rmargin <= p->tcol->offset)
continue;
lsz = p->tcol->rmargin - p->tcol->offset;
} else
lsz = uc;
- if (*seq == '\0')
+ if (*cp == seq[-1])
uc = -1;
- else if (*seq == '\\') {
- seq++;
+ else if (*cp == '\\') {
+ seq = cp + 1;
esc = mandoc_escape(&seq, &cp, &sz);
switch (esc) {
case ESCAPE_UNICODE:
case ESCAPE_SPECIAL:
uc = mchars_spec2cp(cp, sz);
break;
+ case ESCAPE_UNDEF:
+ uc = *seq;
+ break;
default:
uc = -1;
break;
}
} else
- uc = *seq;
+ uc = *cp;
if (uc < 0x20 || (uc > 0x7E && uc < 0xA0))
uc = '_';
if (p->enc == TERMENC_ASCII) {
switch (*cp) {
case '\\':
cp++;
- esc = mandoc_escape(&cp, &seq, &ssz);
- if (ESCAPE_ERROR == esc)
- continue;
-
rhs = NULL;
-
+ esc = mandoc_escape(&cp, &seq, &ssz);
switch (esc) {
case ESCAPE_UNICODE:
uc = mchars_num2uc(seq + 1, ssz - 1);
sz += cond_width(p, uc, &skip);
}
continue;
+ case ESCAPE_UNDEF:
+ uc = *seq;
+ break;
+ case ESCAPE_DEVICE:
+ if (p->type == TERMTYPE_PDF) {
+ rhs = "pdf";
+ rsz = 3;
+ } else if (p->type == TERMTYPE_PS) {
+ rhs = "ps";
+ rsz = 2;
+ } else if (p->enc == TERMENC_ASCII) {
+ rhs = "ascii";
+ rsz = 5;
+ } else {
+ rhs = "utf8";
+ rsz = 4;
+ }
+ break;
case ESCAPE_SKIPCHAR:
skip = 1;
continue;
}
/*
- * Convert a scaling width to basic units, rounding down.
+ * Convert a scaling width to basic units, rounding towards 0.
*/
int
term_hspan(const struct termp *p, const struct roffsu *su)
return (*p->hspan)(p, su);
}
+
+/*
+ * Convert a scaling width to basic units, rounding to closest.
+ */
+int
+term_hen(const struct termp *p, const struct roffsu *su)
+{
+ int bu;
+
+ if ((bu = (*p->hspan)(p, su)) >= 0)
+ return (bu + 11) / 24;
+ else
+ return -((-bu + 11) / 24);
+}
git.ckatri.com / mandoc.git / blobdiff