=================================================================== RCS file: /cvs/mandoc/term.c,v retrieving revision 1.94 retrieving revision 1.137 diff -u -p -r1.94 -r1.137 --- mandoc/term.c 2009/07/27 12:02:49 1.94 +++ mandoc/term.c 2010/05/17 22:11:42 1.137 @@ -1,4 +1,4 @@ -/* $Id: term.c,v 1.94 2009/07/27 12:02:49 kristaps Exp $ */ +/* $Id: term.c,v 1.137 2010/05/17 22:11:42 kristaps Exp $ */ /* * Copyright (c) 2008, 2009 Kristaps Dzonsons * @@ -14,68 +14,45 @@ * 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 + #include -#include +#include #include #include #include +#include +#include "mandoc.h" +#include "chars.h" +#include "out.h" #include "term.h" #include "man.h" #include "mdoc.h" +#include "main.h" -extern int man_run(struct termp *, - const struct man *); -extern int mdoc_run(struct termp *, - const struct mdoc *); - -static struct termp *term_alloc(enum termenc); +static struct termp *term_alloc(enum termenc, size_t); 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_sescape(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 * -ascii_alloc(void) +ascii_alloc(size_t width) { - return(term_alloc(TERMENC_ASCII)); + return(term_alloc(TERMENC_ASCII, width)); } -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) { @@ -90,94 +67,41 @@ term_free(struct termp *p) 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); } static struct termp * -term_alloc(enum termenc enc) +term_alloc(enum termenc enc, size_t width) { 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; + /* Enforce some lower boundary. */ + if (width < 60) + width = 60; + p->defrmargin = width - 2; 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 following flags may be specified: * - * The usage of termp:flags is as follows: - * * - TERMP_NOLPAD: when beginning to write the line, don't left-pad the * offset value. This is useful when doing columnar lists where the * prior column has right-padded. @@ -201,57 +125,51 @@ term_isopendelim(const char *p) * 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 vend; /* end of word visual position on output */ + size_t bp; /* visual right border position */ + 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); - assert((int)(p->rmargin - p->offset) - overstep > 0); - maxvis = /* LINTED */ - p->rmargin - p->offset - overstep; - mmax = /* LINTED */ - 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; - vis = 0; - overstep = 0; /* - * If in the standard case (left-justified), then begin with our - * indentation, otherwise (columns, etc.) just start spitting - * out text. + * Indent the first line of a paragraph. */ + vbl = p->flags & TERMP_NOLPAD ? 0 : p->offset; - if ( ! (p->flags & TERMP_NOLPAD)) - /* LINTED */ - for (j = 0; j < (int)p->offset; j++) - putchar(' '); + /* + * FIXME: if bp is zero, we still output the first word before + * breaking the line. + */ - for (i = 0; i < (int)p->col; i++) { + vis = vend = i = 0; + while (i < (int)p->col) { + /* * Count up visible word characters. Control sequences * (starting with the CSI) aren't counted. A space @@ -260,55 +178,69 @@ term_flushln(struct termp *p) */ /* LINTED */ - for (j = i, vsz = 0; j < (int)p->col; j++) { + for (j = i; j < (int)p->col; j++) { if (j && ' ' == p->buf[j]) break; - else if (8 == p->buf[j]) - vsz--; + if (8 == p->buf[j]) + vend--; else - vsz++; + vend++; } /* - * Choose the number of blanks to prepend: no blank at the - * 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. */ - if (vis && vis + vbl + vsz > bp) { + if (vend > bp && vis > 0) { + vend -= vis; putchar('\n'); if (TERMP_NOBREAK & p->flags) { for (j = 0; j < (int)p->rmargin; j++) putchar(' '); - vis = p->rmargin - p->offset; + vend += p->rmargin - p->offset; } else { - for (j = 0; j < (int)p->offset; j++) - putchar(' '); - vis = 0; + vbl = p->offset; } - } else { - for (j = 0; j < (int)vbl; j++) - putchar(' '); - vis += vbl; + + /* Remove the p->overstep width. */ + + bp += (int)/* LINTED */ + p->overstep; + p->overstep = 0; } - /* - * Finally, write out the word. - */ + /* Write out the [remaining] word. */ for ( ; i < (int)p->col; i++) { - if (' ' == p->buf[i]) + if (' ' == p->buf[i]) { + while (' ' == p->buf[i]) { + vbl++; + i++; + } break; + } + if (ASCII_NBRSP == p->buf[i]) { + vbl++; + continue; + } + + /* + * Now we definitely know there will be + * printable characters to output, + * so write preceding white space now. + */ + if (vbl) { + for (j = 0; j < (int)vbl; j++) + putchar(' '); + vbl = 0; + } putchar(p->buf[i]); } - vis += vsz; + vend += vbl; + vis = vend; } + p->col = 0; + p->overstep = 0; if ( ! (TERMP_NOBREAK & p->flags)) { putchar('\n'); @@ -317,7 +249,7 @@ term_flushln(struct termp *p) if (TERMP_HANG & p->flags) { /* We need one blank after the tag. */ - overstep = /* LINTED */ + p->overstep = /* LINTED */ vis - maxvis + 1; /* @@ -330,12 +262,12 @@ term_flushln(struct termp *p) * move it one step LEFT and flag the rest of the line * to be longer. */ - if (overstep >= -1) { - assert((int)maxvis + overstep >= 0); + if (p->overstep >= -1) { + assert((int)maxvis + p->overstep >= 0); /* LINTED */ - maxvis += overstep; + maxvis += p->overstep; } else - overstep = 0; + p->overstep = 0; } else if (TERMP_DANGLE & p->flags) return; @@ -387,214 +319,348 @@ term_vspace(struct termp *p) } -/* - * 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); - - if (NULL == rhs) - return; - for (i = 0; i < (int)sz; i++) - term_encodea(p, rhs[i]); + rhs = chars_a2ascii(p->symtab, word, len, &sz); + if (rhs) + encode(p, rhs, sz); } static void -term_sescape(struct termp *p, const char *word, size_t len) +res(struct termp *p, const char *word, size_t len) { const char *rhs; size_t sz; - int i; - rhs = term_a2res(p->symtab, word, len, &sz); + rhs = chars_a2res(p->symtab, word, len, &sz); + if (rhs) + encode(p, rhs, sz); +} - if (NULL == rhs) - return; - for (i = 0; i < (int)sz; i++) - term_encodea(p, rhs[i]); + +void +term_fontlast(struct termp *p) +{ + enum termfont f; + + f = p->fontl; + p->fontl = p->fontq[p->fonti]; + p->fontq[p->fonti] = f; } +void +term_fontrepl(struct termp *p, enum termfont f) +{ + + p->fontl = p->fontq[p->fonti]; + p->fontq[p->fonti] = f; +} + + +void +term_fontpush(struct termp *p, enum termfont f) +{ + + assert(p->fonti + 1 < 10); + p->fontl = p->fontq[p->fonti]; + p->fontq[++p->fonti] = f; +} + + +const void * +term_fontq(struct termp *p) +{ + + return(&p->fontq[p->fonti]); +} + + +enum termfont +term_fonttop(struct termp *p) +{ + + 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--; +} + + /* - * 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). + * Handle pwords, partial words, which may be either a single word or a + * phrase that cannot be broken down (such as a literal string). This + * handles word styling. */ -static void -term_pescape(struct termp *p, const char **word) +void +term_word(struct termp *p, const char *word) { - int j; - const char *wp; + const char *sv, *seq; + int sz; + size_t ssz; + enum roffdeco deco; - wp = *word; + sv = word; - if (0 == *(++wp)) { - *word = wp; - return; + 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)) { + bufferc(p, ' '); + if (TERMP_SENTENCE & p->flags) + bufferc(p, ' '); } - if ('(' == *wp) { - wp++; - if (0 == *wp || 0 == *(wp + 1)) { - *word = 0 == *wp ? wp : wp + 1; - return; - } + if ( ! (p->flags & TERMP_NONOSPACE)) + p->flags &= ~TERMP_NOSPACE; - term_nescape(p, wp, 2); - *word = ++wp; - return; + p->flags &= ~TERMP_SENTENCE; - } else if ('*' == *wp) { - if (0 == *(++wp)) { - *word = wp; - return; + /* FIXME: use strcspn. */ + + while (*word) { + if ('\\' != *word) { + encode(p, word, 1); + word++; + continue; } - switch (*wp) { - case ('('): - wp++; - if (0 == *wp || 0 == *(wp + 1)) { - *word = 0 == *wp ? wp : wp + 1; - return; - } + seq = ++word; + sz = a2roffdeco(&deco, &seq, &ssz); - term_sescape(p, wp, 2); - *word = ++wp; - return; - case ('['): + 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: - term_sescape(p, wp, 1); - *word = wp; - return; + break; } - - } else if ('f' == *wp) { - if (0 == *(++wp)) { - *word = wp; - return; - } - switch (*wp) { - case ('B'): - p->flags |= TERMP_BOLD; - break; - case ('I'): - p->flags |= TERMP_UNDER; - break; - case ('P'): + word += sz; + if (DECO_NOSPACE == deco && '\0' == *word) + p->flags |= TERMP_NOSPACE; + } + + /* + * 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 ('R'): - p->flags &= ~TERMP_STYLE; + case('['): + p->flags |= TERMP_NOSPACE; break; default: break; } +} - *word = wp; - return; - } else if ('[' != *wp) { - term_nescape(p, wp, 1); - *word = wp; - return; - } +static void +adjbuf(struct termp *p, size_t sz) +{ - wp++; - for (j = 0; *wp && ']' != *wp; wp++, j++) - /* Loop... */ ; + if (0 == p->maxcols) + p->maxcols = 1024; + while (sz >= p->maxcols) + p->maxcols <<= 2; - if (0 == *wp) { - *word = wp; - return; + p->buf = realloc(p->buf, p->maxcols); + if (NULL == p->buf) { + perror(NULL); + exit(EXIT_FAILURE); } +} - term_nescape(p, wp - j, (size_t)j); - *word = wp; + +static void +buffera(struct termp *p, const char *word, size_t sz) +{ + + if (p->col + sz >= p->maxcols) + adjbuf(p, p->col + sz); + + memcpy(&p->buf[(int)p->col], word, sz); + p->col += sz; } -/* - * Handle pwords, partial words, which may be either a single word or a - * phrase that cannot be broken down (such as a literal string). This - * handles word styling. - */ -void -term_word(struct termp *p, const char *word) +static void +bufferc(struct termp *p, char c) { - const char *sv; - if (term_isclosedelim(word)) - if ( ! (TERMP_IGNDELIM & p->flags)) - p->flags |= TERMP_NOSPACE; + if (p->col + 1 >= p->maxcols) + adjbuf(p, p->col + 1); - if ( ! (TERMP_NOSPACE & p->flags)) - term_chara(p, ' '); + p->buf[(int)p->col++] = c; +} - if ( ! (p->flags & TERMP_NONOSPACE)) - p->flags &= ~TERMP_NOSPACE; - /* - * If ANSI (word-length styling), then apply our style now, - * before the word. +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. */ - for (sv = word; *word; word++) - if ('\\' != *word) - term_encodea(p, *word); + 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 - term_pescape(p, &word); + bufferc(p, word[i]); - if (term_isopendelim(sv)) - p->flags |= TERMP_NOSPACE; + bufferc(p, 8); + bufferc(p, word[i]); + } } -/* - * 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) +size_t +term_vspan(const struct roffsu *su) { - size_t s; + double r; - 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; + 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; } - p->buf[(int)(p->col)++] = c; + + if (r < 0.0) + r = 0.0; + return(/* LINTED */(size_t) + r); } -static void -term_encodea(struct termp *p, char c) +size_t +term_hspan(const struct roffsu *su) { - - 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); - } + 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; } - term_chara(p, c); + + if (r < 0.0) + r = 0.0; + return((size_t)/* LINTED */ + r); } + +