-/* $Id: term.c,v 1.91 2009/07/21 13:34:13 kristaps Exp $ */
+/* $Id: term.c,v 1.131 2010/04/08 07:05:38 kristaps Exp $ */
/*
* Copyright (c) 2008, 2009 Kristaps Dzonsons <kristaps@kth.se>
*
* ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
* OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
*/
+#ifdef HAVE_CONFIG_H
+#include "config.h"
+#endif
+
+#include <sys/types.h>
+
#include <assert.h>
-#include <err.h>
+#include <ctype.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
+#include <time.h>
+#include "chars.h"
+#include "out.h"
#include "term.h"
#include "man.h"
#include "mdoc.h"
-
-extern int man_run(struct termp *,
- const struct man *);
-extern int mdoc_run(struct termp *,
- const struct mdoc *);
+#include "main.h"
static struct termp *term_alloc(enum termenc);
static void term_free(struct termp *);
-static void term_pescape(struct termp *, const char **);
-static void term_nescape(struct termp *,
- const char *, size_t);
-static void term_chara(struct termp *, char);
-static void term_encodea(struct termp *, char);
-static int term_isopendelim(const char *);
-static int term_isclosedelim(const char *);
+static void spec(struct termp *, const char *, size_t);
+static void res(struct termp *, const char *, size_t);
+static void buffera(struct termp *, const char *, size_t);
+static void bufferc(struct termp *, char);
+static void adjbuf(struct termp *p, size_t);
+static void encode(struct termp *, const char *, size_t);
void *
}
-int
-terminal_man(void *arg, const struct man *man)
-{
- struct termp *p;
-
- p = (struct termp *)arg;
- if (NULL == p->symtab)
- p->symtab = term_ascii2htab();
-
- return(man_run(p, man));
-}
-
-
-int
-terminal_mdoc(void *arg, const struct mdoc *mdoc)
-{
- struct termp *p;
-
- p = (struct termp *)arg;
- if (NULL == p->symtab)
- p->symtab = term_ascii2htab();
-
- return(mdoc_run(p, mdoc));
-}
-
-
void
terminal_free(void *arg)
{
if (p->buf)
free(p->buf);
- if (TERMENC_ASCII == p->enc && p->symtab)
- term_asciifree(p->symtab);
+ if (p->symtab)
+ chars_free(p->symtab);
free(p);
}
{
struct termp *p;
- if (NULL == (p = malloc(sizeof(struct termp))))
- err(1, "malloc");
- bzero(p, sizeof(struct termp));
- p->maxrmargin = 78;
+ p = calloc(1, sizeof(struct termp));
+ if (NULL == p) {
+ perror(NULL);
+ exit(EXIT_FAILURE);
+ }
p->enc = enc;
return(p);
}
-static int
-term_isclosedelim(const char *p)
-{
-
- if ( ! (*p && 0 == *(p + 1)))
- return(0);
-
- switch (*p) {
- case('.'):
- /* FALLTHROUGH */
- case(','):
- /* FALLTHROUGH */
- case(';'):
- /* FALLTHROUGH */
- case(':'):
- /* FALLTHROUGH */
- case('?'):
- /* FALLTHROUGH */
- case('!'):
- /* FALLTHROUGH */
- case(')'):
- /* FALLTHROUGH */
- case(']'):
- /* FALLTHROUGH */
- case('}'):
- return(1);
- default:
- break;
- }
-
- return(0);
-}
-
-
-static int
-term_isopendelim(const char *p)
-{
-
- if ( ! (*p && 0 == *(p + 1)))
- return(0);
-
- switch (*p) {
- case('('):
- /* FALLTHROUGH */
- case('['):
- /* FALLTHROUGH */
- case('{'):
- return(1);
- default:
- break;
- }
-
- return(0);
-}
-
-
/*
* Flush a line of text. A "line" is loosely defined as being something
* that should be followed by a newline, regardless of whether it's
* broken apart by newlines getting there. A line can also be a
- * fragment of a columnar list.
+ * fragment of a columnar list (`Bl -tag' or `Bl -column'), which does
+ * not have a trailing newline.
*
- * Specifically, a line is whatever's in p->buf of length p->col, which
- * is zeroed after this function returns.
- *
- * The usage of termp:flags is as follows:
+ * The following flags may be specified:
*
* - TERMP_NOLPAD: when beginning to write the line, don't left-pad the
* offset value. This is useful when doing columnar lists where the
* If TERMP_NOBREAK is specified and the line overruns the right
* margin, it will break and pad-right to the right margin after
* writing. If maxrmargin is violated, it will break and continue
- * writing from the right-margin, which will lead to the above
- * scenario upon exit.
- *
- * Otherwise, the line will break at the right margin. Extremely long
- * lines will cause the system to emit a warning (TODO: hyphenate, if
- * possible).
- *
- * FIXME: newline breaks occur (in groff) also occur when a single
- * space follows a NOBREAK (try `Bl -tag')
- *
- * FIXME: there's a newline error where a `Bl -diag' will have a
- * trailing newline if the line is exactly 73 chars long.
+ * writing from the right-margin, which will lead to the above scenario
+ * upon exit. Otherwise, the line will break at the right margin.
*/
void
term_flushln(struct termp *p)
{
- int i, j;
- size_t vbl, vsz, vis, maxvis, mmax, bp;
- static int overstep = 0;
+ int i; /* current input position in p->buf */
+ size_t vis; /* current visual position on output */
+ size_t vbl; /* number of blanks to prepend to output */
+ size_t vsz; /* visual characters to write to output */
+ size_t bp; /* visual right border position */
+ size_t hyph; /* visible position of hyphen */
+ int j; /* temporary loop index */
+ size_t maxvis, mmax;
/*
* 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.
+ * small set of values.
*/
assert(p->offset < p->rmargin);
- maxvis = p->rmargin - p->offset - overstep;
- mmax = p->maxrmargin - p->offset - overstep;
+
+ maxvis = (int)(p->rmargin - p->offset) - p->overstep < 0 ?
+ /* LINTED */
+ 0 : p->rmargin - p->offset - p->overstep;
+ mmax = (int)(p->maxrmargin - p->offset) - p->overstep < 0 ?
+ /* LINTED */
+ 0 : p->maxrmargin - p->offset - p->overstep;
+
bp = TERMP_NOBREAK & p->flags ? mmax : maxvis;
+
+ /*
+ * FIXME: if bp is zero, we still output the first word before
+ * breaking the line.
+ */
+
vis = 0;
- overstep = 0;
/*
* If in the standard case (left-justified), then begin with our
* (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).
+ * Collect the number of printable characters until the
+ * first hyphen, if found. Hyphens aren't included if
+ * they're the first character (so `Fl' doesn't break)
+ * or second consecutive character (`Fl -').
*/
/* LINTED */
- for (j = i, vsz = 0; j < (int)p->col; j++) {
- if (' ' == p->buf[j])
+ for (j = i, vsz = 0, hyph = 0; j < (int)p->col; j++) {
+ if (j && ' ' == p->buf[j])
break;
- else if (8 == p->buf[j])
+ if (8 == p->buf[j])
vsz--;
else
vsz++;
+ if (j > i && '-' == p->buf[j] && 0 == hyph)
+ if ('-' != p->buf[j - 1])
+ hyph = vsz;
}
/*
* beginning of a line, one between words -- but do not
* actually write them yet.
*/
+
vbl = (size_t)(0 == vis ? 0 : 1);
/*
* Find out whether we would exceed the right margin.
- * If so, break to the next line. (TODO: hyphenate)
- * Otherwise, write the chosen number of blanks now.
+ * If so, break to the next line, possibly after
+ * emittign character up to a hyphen. Otherwise, write
+ * the chosen number of blanks.
*/
+
if (vis && vis + vbl + vsz > bp) {
+ /*
+ * Has a hyphen been found before the breakpoint
+ * that we can use?
+ */
+ if (hyph && vis + vbl + hyph <= bp) {
+ /* First prepend blanks. */
+ for (j = 0; j < (int)vbl; j++)
+ putchar(' ');
+
+ /* Emit up to the character. */
+ do {
+ if (31 == p->buf[i])
+ putchar(' ');
+ else
+ putchar(p->buf[i]);
+ if (8 != p->buf[i])
+ vsz--;
+ } while ('-' != p->buf[i++]);
+
+ /* Emit trailing decoration. */
+ if (8 == p->buf[i]) {
+ putchar(p->buf[i]);
+ putchar(p->buf[i + 1]);
+ }
+ }
+
putchar('\n');
if (TERMP_NOBREAK & p->flags) {
for (j = 0; j < (int)p->rmargin; j++)
putchar(' ');
vis = 0;
}
+
+ /* Remove the p->overstep width. */
+
+ bp += (int)/* LINTED */
+ p->overstep;
+ p->overstep = 0;
} else {
for (j = 0; j < (int)vbl; j++)
putchar(' ');
vis += vbl;
}
- /*
- * Finally, write out the word.
- */
- for ( ; i < (int)p->col; i++) {
+ /* Write out the [remaining] word. */
+ for ( ; i < (int)p->col; i++)
if (' ' == p->buf[i])
break;
- putchar(p->buf[i]);
- }
+ else if (31 == p->buf[i])
+ putchar(' ');
+ else
+ putchar(p->buf[i]);
+
vis += vsz;
}
+
p->col = 0;
+ p->overstep = 0;
if ( ! (TERMP_NOBREAK & p->flags)) {
putchar('\n');
}
if (TERMP_HANG & p->flags) {
-
/* We need one blank after the tag. */
- overstep = vis - maxvis + 1;
+ p->overstep = /* LINTED */
+ vis - maxvis + 1;
/*
* Behave exactly the same way as groff:
- * If we have overstepped the margin, temporarily move it
- * to the right and flag the rest of the line to be shorter.
+ * If we have overstepped the margin, temporarily move
+ * it to the right and flag the rest of the line to be
+ * shorter.
* If we landed right at the margin, be happy.
- * If we are one step before the margin, temporarily move it
- * one step LEFT and flag the rest of the line to be longer.
+ * If we are one step before the margin, temporarily
+ * move it one step LEFT and flag the rest of the line
+ * to be longer.
*/
- if (overstep >= -1)
- maxvis += overstep;
- else
- overstep = 0;
+ if (p->overstep >= -1) {
+ assert((int)maxvis + p->overstep >= 0);
+ /* LINTED */
+ maxvis += p->overstep;
+ } else
+ p->overstep = 0;
} else if (TERMP_DANGLE & p->flags)
return;
- if (maxvis > vis + ((TERMP_TWOSPACE & p->flags) ? 1 : 0)) /* pad */
+ /* Right-pad. */
+ if (maxvis > vis + /* LINTED */
+ ((TERMP_TWOSPACE & p->flags) ? 1 : 0))
for ( ; vis < maxvis; vis++)
putchar(' ');
- else { /* break */
+ else { /* ...or newline break. */
putchar('\n');
for (i = 0; i < (int)p->rmargin; i++)
putchar(' ');
}
-/*
- * Determine the symbol indicated by an escape sequences, that is, one
- * starting with a backslash. Once done, we pass this value into the
- * output buffer by way of the symbol table.
- */
static void
-term_nescape(struct termp *p, const char *word, size_t len)
+spec(struct termp *p, const char *word, size_t len)
{
const char *rhs;
size_t sz;
- int i;
- rhs = term_a2ascii(p->symtab, word, len, &sz);
+ rhs = chars_a2ascii(p->symtab, word, len, &sz);
+ if (rhs)
+ encode(p, rhs, sz);
+}
+
+static void
+res(struct termp *p, const char *word, size_t len)
+{
+ const char *rhs;
+ size_t sz;
+
+ rhs = chars_a2res(p->symtab, word, len, &sz);
if (rhs)
- for (i = 0; i < (int)sz; i++)
- term_encodea(p, rhs[i]);
+ encode(p, rhs, sz);
}
-/*
- * Handle an escape sequence: determine its length and pass it to the
- * escape-symbol look table. Note that we assume mdoc(3) has validated
- * the escape sequence (we assert upon badly-formed escape sequences).
- */
-static void
-term_pescape(struct termp *p, const char **word)
+void
+term_fontlast(struct termp *p)
{
- int j;
- const char *wp;
+ enum termfont f;
- wp = *word;
+ f = p->fontl;
+ p->fontl = p->fontq[p->fonti];
+ p->fontq[p->fonti] = f;
+}
- if (0 == *(++wp)) {
- *word = wp;
- return;
- }
- if ('(' == *wp) {
- wp++;
- if (0 == *wp || 0 == *(wp + 1)) {
- *word = 0 == *wp ? wp : wp + 1;
- return;
- }
+void
+term_fontrepl(struct termp *p, enum termfont f)
+{
- term_nescape(p, wp, 2);
- *word = ++wp;
- return;
+ p->fontl = p->fontq[p->fonti];
+ p->fontq[p->fonti] = f;
+}
- } else if ('*' == *wp) {
- if (0 == *(++wp)) {
- *word = wp;
- return;
- }
- switch (*wp) {
- case ('('):
- wp++;
- if (0 == *wp || 0 == *(wp + 1)) {
- *word = 0 == *wp ? wp : wp + 1;
- return;
- }
+void
+term_fontpush(struct termp *p, enum termfont f)
+{
- term_nescape(p, wp, 2);
- *word = ++wp;
- return;
- case ('['):
- break;
- default:
- term_nescape(p, wp, 1);
- *word = wp;
- return;
- }
-
- } else if ('f' == *wp) {
- if (0 == *(++wp)) {
- *word = wp;
- return;
- }
+ assert(p->fonti + 1 < 10);
+ p->fontl = p->fontq[p->fonti];
+ p->fontq[++p->fonti] = f;
+}
- switch (*wp) {
- case ('B'):
- p->flags |= TERMP_BOLD;
- break;
- case ('I'):
- p->flags |= TERMP_UNDER;
- break;
- case ('P'):
- /* FALLTHROUGH */
- case ('R'):
- p->flags &= ~TERMP_STYLE;
- break;
- default:
- break;
- }
- *word = wp;
- return;
+const void *
+term_fontq(struct termp *p)
+{
- } else if ('[' != *wp) {
- term_nescape(p, wp, 1);
- *word = wp;
- return;
- }
+ return(&p->fontq[p->fonti]);
+}
- wp++;
- for (j = 0; *wp && ']' != *wp; wp++, j++)
- /* Loop... */ ;
- if (0 == *wp) {
- *word = wp;
- return;
- }
+enum termfont
+term_fonttop(struct termp *p)
+{
- term_nescape(p, wp - j, (size_t)j);
- *word = wp;
+ return(p->fontq[p->fonti]);
+}
+
+
+void
+term_fontpopq(struct termp *p, const void *key)
+{
+
+ while (p->fonti >= 0 && key != &p->fontq[p->fonti])
+ p->fonti--;
+ assert(p->fonti >= 0);
+}
+
+
+void
+term_fontpop(struct termp *p)
+{
+
+ assert(p->fonti);
+ p->fonti--;
}
void
term_word(struct termp *p, const char *word)
{
- const char *sv;
+ const char *sv, *seq;
+ int sz;
+ size_t ssz;
+ enum roffdeco deco;
- if (term_isclosedelim(word))
- if ( ! (TERMP_IGNDELIM & p->flags))
- p->flags |= TERMP_NOSPACE;
+ sv = word;
+
+ if (word[0] && '\0' == word[1])
+ switch (word[0]) {
+ case('.'):
+ /* FALLTHROUGH */
+ case(','):
+ /* FALLTHROUGH */
+ case(';'):
+ /* FALLTHROUGH */
+ case(':'):
+ /* FALLTHROUGH */
+ case('?'):
+ /* FALLTHROUGH */
+ case('!'):
+ /* FALLTHROUGH */
+ case(')'):
+ /* FALLTHROUGH */
+ case(']'):
+ if ( ! (TERMP_IGNDELIM & p->flags))
+ p->flags |= TERMP_NOSPACE;
+ break;
+ default:
+ break;
+ }
if ( ! (TERMP_NOSPACE & p->flags))
- term_chara(p, ' ');
+ bufferc(p, ' ');
if ( ! (p->flags & TERMP_NONOSPACE))
p->flags &= ~TERMP_NOSPACE;
+ /* FIXME: use strcspn. */
+
+ while (*word) {
+ if ('\\' != *word) {
+ encode(p, word, 1);
+ word++;
+ continue;
+ }
+
+ seq = ++word;
+ sz = a2roffdeco(&deco, &seq, &ssz);
+
+ switch (deco) {
+ case (DECO_RESERVED):
+ res(p, seq, ssz);
+ break;
+ case (DECO_SPECIAL):
+ spec(p, seq, ssz);
+ break;
+ case (DECO_BOLD):
+ term_fontrepl(p, TERMFONT_BOLD);
+ break;
+ case (DECO_ITALIC):
+ term_fontrepl(p, TERMFONT_UNDER);
+ break;
+ case (DECO_ROMAN):
+ term_fontrepl(p, TERMFONT_NONE);
+ break;
+ case (DECO_PREVIOUS):
+ term_fontlast(p);
+ break;
+ default:
+ break;
+ }
+
+ word += sz;
+ if (DECO_NOSPACE == deco && '\0' == *word)
+ p->flags |= TERMP_NOSPACE;
+ }
+
/*
- * If ANSI (word-length styling), then apply our style now,
- * before the word.
+ * Note that we don't process the pipe: the parser sees it as
+ * punctuation, but we don't in terms of typography.
*/
+ if (sv[0] && 0 == sv[1])
+ switch (sv[0]) {
+ case('('):
+ /* FALLTHROUGH */
+ case('['):
+ p->flags |= TERMP_NOSPACE;
+ break;
+ default:
+ break;
+ }
+}
- for (sv = word; *word; word++)
- if ('\\' != *word)
- term_encodea(p, *word);
- else
- term_pescape(p, &word);
- if (term_isopendelim(sv))
- p->flags |= TERMP_NOSPACE;
+static void
+adjbuf(struct termp *p, size_t sz)
+{
+
+ if (0 == p->maxcols)
+ p->maxcols = 1024;
+ while (sz >= p->maxcols)
+ p->maxcols <<= 2;
+
+ p->buf = realloc(p->buf, p->maxcols);
+ if (NULL == p->buf) {
+ perror(NULL);
+ exit(EXIT_FAILURE);
+ }
}
-/*
- * Insert a single character into the line-buffer. If the buffer's
- * space is exceeded, then allocate more space by doubling the buffer
- * size.
- */
static void
-term_chara(struct termp *p, char c)
+buffera(struct termp *p, const char *word, size_t sz)
{
- size_t s;
-
- if (p->col + 1 >= p->maxcols) {
- if (0 == p->maxcols)
- p->maxcols = 256;
- s = p->maxcols * 2;
- p->buf = realloc(p->buf, s);
- if (NULL == p->buf)
- err(1, "realloc");
- p->maxcols = s;
- }
- p->buf[(int)(p->col)++] = c;
+
+ if (p->col + sz >= p->maxcols)
+ adjbuf(p, p->col + sz);
+
+ memcpy(&p->buf[(int)p->col], word, sz);
+ p->col += sz;
}
static void
-term_encodea(struct termp *p, char c)
+bufferc(struct termp *p, char c)
{
-
- if (' ' != c && TERMP_STYLE & p->flags) {
- if (TERMP_BOLD & p->flags) {
- term_chara(p, c);
- term_chara(p, 8);
- }
- if (TERMP_UNDER & p->flags) {
- term_chara(p, '_');
- term_chara(p, 8);
+
+ if (p->col + 1 >= p->maxcols)
+ adjbuf(p, p->col + 1);
+
+ p->buf[(int)p->col++] = c;
+}
+
+
+static void
+encode(struct termp *p, const char *word, size_t sz)
+{
+ enum termfont f;
+ int i;
+
+ /*
+ * Encode and buffer a string of characters. If the current
+ * font mode is unset, buffer directly, else encode then buffer
+ * character by character.
+ */
+
+ if (TERMFONT_NONE == (f = term_fonttop(p))) {
+ buffera(p, word, sz);
+ return;
+ }
+
+ for (i = 0; i < (int)sz; i++) {
+ if ( ! isgraph((u_char)word[i])) {
+ bufferc(p, word[i]);
+ continue;
}
+
+ if (TERMFONT_UNDER == f)
+ bufferc(p, '_');
+ else
+ bufferc(p, word[i]);
+
+ bufferc(p, 8);
+ bufferc(p, word[i]);
+ }
+}
+
+
+size_t
+term_vspan(const struct roffsu *su)
+{
+ double r;
+
+ switch (su->unit) {
+ case (SCALE_CM):
+ r = su->scale * 2;
+ break;
+ case (SCALE_IN):
+ r = su->scale * 6;
+ break;
+ case (SCALE_PC):
+ r = su->scale;
+ break;
+ case (SCALE_PT):
+ r = su->scale / 8;
+ break;
+ case (SCALE_MM):
+ r = su->scale / 1000;
+ break;
+ case (SCALE_VS):
+ r = su->scale;
+ break;
+ default:
+ r = su->scale - 1;
+ break;
}
- term_chara(p, c);
+
+ if (r < 0.0)
+ r = 0.0;
+ return(/* LINTED */(size_t)
+ r);
}
+
+
+size_t
+term_hspan(const struct roffsu *su)
+{
+ double r;
+
+ /* XXX: CM, IN, and PT are approximations. */
+
+ switch (su->unit) {
+ case (SCALE_CM):
+ r = 4 * su->scale;
+ break;
+ case (SCALE_IN):
+ /* XXX: this is an approximation. */
+ r = 10 * su->scale;
+ break;
+ case (SCALE_PC):
+ r = (10 * su->scale) / 6;
+ break;
+ case (SCALE_PT):
+ r = (10 * su->scale) / 72;
+ break;
+ case (SCALE_MM):
+ r = su->scale / 1000; /* FIXME: double-check. */
+ break;
+ case (SCALE_VS):
+ r = su->scale * 2 - 1; /* FIXME: double-check. */
+ break;
+ default:
+ r = su->scale;
+ break;
+ }
+
+ if (r < 0.0)
+ r = 0.0;
+ return((size_t)/* LINTED */
+ r);
+}
+
+
git.ckatri.com / mandoc.git / blobdiff