=================================================================== RCS file: /cvs/mandoc/Attic/mdocterm.c,v retrieving revision 1.15 retrieving revision 1.24 diff -u -p -r1.15 -r1.24 --- mandoc/Attic/mdocterm.c 2009/02/26 17:11:38 1.15 +++ mandoc/Attic/mdocterm.c 2009/03/02 12:09:32 1.24 @@ -1,4 +1,4 @@ -/* $Id: mdocterm.c,v 1.15 2009/02/26 17:11:38 kristaps Exp $ */ +/* $Id: mdocterm.c,v 1.24 2009/03/02 12:09:32 kristaps Exp $ */ /* * Copyright (c) 2008 Kristaps Dzonsons * @@ -16,6 +16,8 @@ * TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR * PERFORMANCE OF THIS SOFTWARE. */ +#include + #include #include #include @@ -31,6 +33,34 @@ #include "mmain.h" #include "term.h" +#define TERMSYM_RBRACK "]" +#define TERMSYM_LBRACK "[" +#define TERMSYM_LARROW "<-" +#define TERMSYM_RARROW "->" +#define TERMSYM_UARROW "^" +#define TERMSYM_DARROW "v" +#define TERMSYM_LSQUOTE "`" +#define TERMSYM_RSQUOTE "\'" +#define TERMSYM_SQUOTE "\'" +#define TERMSYM_LDQUOTE "``" +#define TERMSYM_RDQUOTE "\'\'" +#define TERMSYM_DQUOTE "\"" +#define TERMSYM_LT "<" +#define TERMSYM_GT ">" +#define TERMSYM_LE "<=" +#define TERMSYM_GE ">=" +#define TERMSYM_EQ "==" +#define TERMSYM_NEQ "!=" +#define TERMSYM_ACUTE "\'" +#define TERMSYM_GRAVE "`" +#define TERMSYM_PI "pi" +#define TERMSYM_PLUSMINUS "+=" +#define TERMSYM_INF "oo" +#define TERMSYM_INF2 "infinity" +#define TERMSYM_NAN "NaN" +#define TERMSYM_BAR "|" +#define TERMSYM_BULLET "o" + #ifdef __NetBSD__ #define xisspace(x) isspace((int)(x)) #else @@ -55,6 +85,7 @@ static void footer(struct termp *, static void pword(struct termp *, const char *, size_t); static void pescape(struct termp *, const char *, size_t *, size_t); +static void pgraph(struct termp *, char); static void nescape(struct termp *, const char *, size_t); static void chara(struct termp *, char); @@ -82,7 +113,7 @@ main(int argc, char *argv[]) if (NULL == (mdoc = mmain_mdoc(p))) mmain_exit(p, 1); - termp.maxrmargin = 80; /* XXX */ + termp.maxrmargin = 78; /* XXX */ termp.rmargin = termp.maxrmargin; termp.maxcols = 1024; termp.offset = termp.col = 0; @@ -133,7 +164,6 @@ flushln(struct termp *p) * If we're literal, print out verbatim. */ if (p->flags & TERMP_LITERAL) { - /* FIXME: count non-printing chars. */ for (i = 0; i < p->col; i++) putchar(p->buf[i]); putchar('\n'); @@ -144,9 +174,10 @@ flushln(struct termp *p) for (i = 0; i < p->col; i++) { /* * Count up visible word characters. Control sequences - * (starting with the CSI) aren't counted. + * (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). */ - assert( ! xisspace(p->buf[i])); /* LINTED */ for (j = i, vsz = 0; j < p->col; j++) { @@ -158,27 +189,35 @@ flushln(struct termp *p) } else vsz++; } - assert(vsz > 0); /* + * If we're breaking normally... + * * If a word is too long and we're within a line, put it * on the next line. Puke if we're being asked to write * something that will exceed the right margin (i.e., - * from a fresh line or when we're not allowed to break - * the line with TERMP_NOBREAK). + * from a fresh line). + * + * If we're not breaking... + * + * Don't let the visible size exceed the full right + * margin. */ - if (vis && vis + vsz > maxvis) { - /* FIXME */ - if (p->flags & TERMP_NOBREAK) + if ( ! (TERMP_NOBREAK & p->flags)) { + if (vis && vis + vsz > maxvis) { + putchar('\n'); + for (j = 0; j < p->offset; j++) + putchar(' '); + vis = 0; + } else if (vis + vsz > maxvis) errx(1, "word breaks right margin"); + } else if (vis + vsz > p->maxrmargin - p->offset) { putchar('\n'); - for (j = 0; j < p->offset; j++) + for (j = 0; j < p->rmargin; j++) putchar(' '); - vis = 0; - } else if (vis + vsz > maxvis) - /* FIXME */ - errx(1, "word breaks right margin"); + vis = p->rmargin; + } /* * Write out the word and a trailing space. Omit the @@ -197,15 +236,22 @@ flushln(struct termp *p) } } + if ((TERMP_NOBREAK & p->flags) && vis >= maxvis) { + putchar('\n'); + for (i = 0; i < p->rmargin; i++) + putchar(' '); + p->col = 0; + return; + } + /* * If we're not to right-marginalise it (newline), then instead * pad to the right margin and stay off. */ if (p->flags & TERMP_NOBREAK) { - if ( ! (p->flags & TERMP_NORPAD)) - for ( ; vis < maxvis; vis++) - putchar(' '); + for ( ; vis < maxvis; vis++) + putchar(' '); } else putchar('\n'); @@ -258,9 +304,16 @@ static void chara(struct termp *p, char c) { - /* TODO: dynamically expand the buffer. */ - if (p->col + 1 >= p->maxcols) - errx(1, "line overrun"); + /* + * Insert a single character into the line-buffer. If the + * buffer's space is exceeded, then allocate more space. + */ + if (p->col + 1 >= p->maxcols) { + p->buf = realloc(p->buf, p->maxcols * 2); + if (NULL == p->buf) + err(1, "malloc"); + p->maxcols *= 2; + } p->buf[(p->col)++] = c; } @@ -297,21 +350,79 @@ nescape(struct termp *p, const char *word, size_t len) { switch (len) { + case (1): + if ('q' == word[0]) + stringa(p, TERMSYM_DQUOTE); + break; case (2): if ('r' == word[0] && 'B' == word[1]) - chara(p, ']'); + stringa(p, TERMSYM_RBRACK); else if ('l' == word[0] && 'B' == word[1]) - chara(p, '['); - else if ('<' == word[0] && '-' == word[1]) - stringa(p, "<-"); - else if ('-' == word[0] && '>' == word[1]) - stringa(p, "->"); + stringa(p, TERMSYM_LBRACK); else if ('l' == word[0] && 'q' == word[1]) - chara(p, '\"'); + stringa(p, TERMSYM_LDQUOTE); else if ('r' == word[0] && 'q' == word[1]) - chara(p, '\"'); + stringa(p, TERMSYM_RDQUOTE); + else if ('o' == word[0] && 'q' == word[1]) + stringa(p, TERMSYM_LSQUOTE); + else if ('a' == word[0] && 'q' == word[1]) + stringa(p, TERMSYM_RSQUOTE); + else if ('<' == word[0] && '-' == word[1]) + stringa(p, TERMSYM_LARROW); + else if ('-' == word[0] && '>' == word[1]) + stringa(p, TERMSYM_RARROW); else if ('b' == word[0] && 'u' == word[1]) - chara(p, 'o'); + stringa(p, TERMSYM_BULLET); + else if ('<' == word[0] && '=' == word[1]) + stringa(p, TERMSYM_LE); + else if ('>' == word[0] && '=' == word[1]) + stringa(p, TERMSYM_GE); + else if ('=' == word[0] && '=' == word[1]) + stringa(p, TERMSYM_EQ); + else if ('+' == word[0] && '-' == word[1]) + stringa(p, TERMSYM_PLUSMINUS); + else if ('u' == word[0] && 'a' == word[1]) + stringa(p, TERMSYM_UARROW); + else if ('d' == word[0] && 'a' == word[1]) + stringa(p, TERMSYM_DARROW); + else if ('a' == word[0] && 'a' == word[1]) + stringa(p, TERMSYM_ACUTE); + else if ('g' == word[0] && 'a' == word[1]) + stringa(p, TERMSYM_GRAVE); + else if ('!' == word[0] && '=' == word[1]) + stringa(p, TERMSYM_NEQ); + else if ('i' == word[0] && 'f' == word[1]) + stringa(p, TERMSYM_INF); + else if ('n' == word[0] && 'a' == word[1]) + stringa(p, TERMSYM_NAN); + else if ('b' == word[0] && 'a' == word[1]) + stringa(p, TERMSYM_BAR); + + /* Deprecated forms. */ + else if ('B' == word[0] && 'a' == word[1]) + stringa(p, TERMSYM_BAR); + else if ('I' == word[0] && 'f' == word[1]) + stringa(p, TERMSYM_INF2); + else if ('G' == word[0] && 'e' == word[1]) + stringa(p, TERMSYM_GE); + else if ('G' == word[0] && 't' == word[1]) + stringa(p, TERMSYM_GT); + else if ('L' == word[0] && 'e' == word[1]) + stringa(p, TERMSYM_LE); + else if ('L' == word[0] && 'q' == word[1]) + stringa(p, TERMSYM_LDQUOTE); + else if ('L' == word[0] && 't' == word[1]) + stringa(p, TERMSYM_LT); + else if ('N' == word[0] && 'a' == word[1]) + stringa(p, TERMSYM_NAN); + else if ('N' == word[0] && 'e' == word[1]) + stringa(p, TERMSYM_NEQ); + else if ('P' == word[0] && 'i' == word[1]) + stringa(p, TERMSYM_PI); + else if ('P' == word[0] && 'm' == word[1]) + stringa(p, TERMSYM_PLUSMINUS); + else if ('R' == word[0] && 'q' == word[1]) + stringa(p, TERMSYM_RDQUOTE); break; default: break; @@ -320,6 +431,26 @@ nescape(struct termp *p, const char *word, size_t len) static void +pgraph(struct termp *p, char byte) +{ + int i; + + switch (byte) { + case (' '): + chara(p, ' '); + break; + case ('\t'): + for (i = 0; i < INDENT; i++) + chara(p, ' '); + break; + default: + warnx("unknown non-graphing character"); + break; + } +} + + +static void pescape(struct termp *p, const char *word, size_t *i, size_t len) { size_t j; @@ -327,6 +458,11 @@ pescape(struct termp *p, const char *word, size_t *i, (*i)++; assert(*i < len); + /* + * Handle an escape sequence. This must manage both groff-style + * escapes and mdoc-style escapes. + */ + if ('(' == word[*i]) { /* Two-character escapes. */ (*i)++; @@ -335,6 +471,22 @@ pescape(struct termp *p, const char *word, size_t *i, (*i)++; return; + } else if ('*' == word[*i]) { + (*i)++; + assert(*i < len); + switch (word[*i]) { + case ('('): + (*i)++; + assert(*i + 1 < len); + nescape(p, &word[*i], 2); + (*i)++; + return; + default: + break; + } + nescape(p, &word[*i], 1); + return; + } else if ('[' != word[*i]) { /* One-character escapes. */ switch (word[*i]) { @@ -350,6 +502,10 @@ pescape(struct termp *p, const char *word, size_t *i, /* FALLTHROUGH */ case ('.'): chara(p, word[*i]); + break; + case ('e'): + chara(p, '\\'); + break; default: break; } @@ -369,7 +525,11 @@ pword(struct termp *p, const char *word, size_t len) { size_t i; - /*assert(len > 0);*/ /* Can be, if literal. */ + /* + * 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. + */ if ( ! (p->flags & TERMP_NOSPACE) && ! (p->flags & TERMP_LITERAL)) @@ -378,6 +538,11 @@ pword(struct termp *p, const char *word, size_t len) if ( ! (p->flags & TERMP_NONOSPACE)) p->flags &= ~TERMP_NOSPACE; + /* + * XXX - if literal and underlining, this will underline the + * spaces between literal words. + */ + if (p->flags & TERMP_BOLD) style(p, STYLE_BOLD); if (p->flags & TERMP_UNDERLINE) @@ -388,6 +553,10 @@ pword(struct termp *p, const char *word, size_t len) pescape(p, word, &i, len); continue; } + if ( ! isgraph((int)word[i])) { + pgraph(p, word[i]); + continue; + } chara(p, word[i]); } @@ -402,6 +571,13 @@ word(struct termp *p, const char *word) { size_t i, j, len; + /* + * Break apart a word into tokens. If we're a literal word, + * then don't. This doesn't handle zero-length words (there + * should be none) and makes sure that pword doesn't get spaces + * or nil words unless literal. + */ + if (p->flags & TERMP_LITERAL) { pword(p, word, strlen(word)); return; @@ -421,7 +597,14 @@ word(struct termp *p, const char *word) if ( ! xisspace(word[i])) { j++; continue; + } + + /* Escaped spaces don't delimit... */ + if (i > 0 && xisspace(word[i]) && '\\' == word[i - 1]) { + j++; + continue; } + if (0 == j) continue; assert(i >= j); @@ -443,6 +626,12 @@ body(struct termp *p, struct termpair *ppair, int dochild; struct termpair pair; + /* + * This is the main function for printing out nodes. It's + * constituted of PRE and POST functions, which correspond to + * prefix and infix processing. + */ + /* Pre-processing. */ dochild = 1; @@ -505,6 +694,13 @@ footer(struct termp *p, const struct mdoc_meta *meta) (void)strlcpy(os, meta->os, p->rmargin); + /* + * This is /slightly/ different from regular groff output + * because we don't have page numbers. Print the following: + * + * OS MDOCDATE + */ + vspace(p); p->flags |= TERMP_NOSPACE | TERMP_NOBREAK; @@ -530,15 +726,21 @@ footer(struct termp *p, const struct mdoc_meta *meta) static void header(struct termp *p, const struct mdoc_meta *meta) { - char *buf, *title; + char *buf, *title, *bufp, *vbuf; const char *pp; + struct utsname uts; + p->rmargin = p->maxrmargin; + p->offset = 0; + if (NULL == (buf = malloc(p->rmargin))) err(1, "malloc"); if (NULL == (title = malloc(p->rmargin))) err(1, "malloc"); + if (NULL == (vbuf = malloc(p->rmargin))) + err(1, "malloc"); - if (NULL == (pp = mdoc_vol2a(meta->vol))) + if (NULL == (pp = mdoc_vol2a(meta->vol))) { switch (meta->msec) { case (MSEC_1): /* FALLTHROUGH */ @@ -563,23 +765,48 @@ header(struct termp *p, const struct mdoc_meta *meta) pp = mdoc_vol2a(VOL_KM); break; default: - /* FIXME: capitalise. */ - if (NULL == (pp = mdoc_msec2a(meta->msec))) - pp = mdoc_msec2a(MSEC_local); break; } + } + vbuf[0] = 0; + if (pp) { + if (-1 == uname(&uts)) + err(1, "uname"); + (void)strlcat(vbuf, uts.sysname, p->rmargin); + (void)strlcat(vbuf, " ", p->rmargin); + } else if (NULL == (pp = mdoc_msec2a(meta->msec))) + pp = mdoc_msec2a(MSEC_local); + + (void)strlcat(vbuf, pp, p->rmargin); + + /* + * The header is strange. It has three components, which are + * really two with the first duplicated. It goes like this: + * + * IDENTIFIER TITLE IDENTIFIER + * + * The IDENTIFIER is NAME(SECTION), which is the command-name + * (if given, or "unknown" if not) followed by the manual page + * section. These are given in `Dt'. The TITLE is a free-form + * string depending on the manual volume. If not specified, it + * switches on the manual section. + */ + if (mdoc_arch2a(meta->arch)) - (void)snprintf(buf, p->rmargin, "%s(%s)", - pp, mdoc_arch2a(meta->arch)); + (void)snprintf(buf, p->rmargin, "%s (%s)", + vbuf, mdoc_arch2a(meta->arch)); else - (void)strlcpy(buf, pp, p->rmargin); + (void)strlcpy(buf, vbuf, p->rmargin); pp = mdoc_msec2a(meta->msec); (void)snprintf(title, p->rmargin, "%s(%s)", meta->title, pp ? pp : ""); + for (bufp = title; *bufp; bufp++) + *bufp = toupper(*bufp); + p->offset = 0; p->rmargin = (p->maxrmargin - strlen(buf)) / 2; p->flags |= TERMP_NOBREAK | TERMP_NOSPACE; @@ -607,5 +834,6 @@ header(struct termp *p, const struct mdoc_meta *meta) p->flags &= ~TERMP_NOSPACE; free(title); + free(vbuf); free(buf); }