=================================================================== RCS file: /cvs/mandoc/Attic/mdocterm.c,v retrieving revision 1.18 retrieving revision 1.38 diff -u -p -r1.18 -r1.38 --- mandoc/Attic/mdocterm.c 2009/02/27 10:55:16 1.18 +++ mandoc/Attic/mdocterm.c 2009/03/10 11:16:43 1.38 @@ -1,4 +1,4 @@ -/* $Id: mdocterm.c,v 1.18 2009/02/27 10:55:16 kristaps Exp $ */ +/* $Id: mdocterm.c,v 1.38 2009/03/10 11:16:43 kristaps Exp $ */ /* * Copyright (c) 2008 Kristaps Dzonsons * @@ -16,7 +16,7 @@ * TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR * PERFORMANCE OF THIS SOFTWARE. */ -#include +#include #include #include @@ -26,51 +26,12 @@ #include #include -#ifndef __OpenBSD__ -#include -#endif - #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 -#define xisspace(x) isspace((x)) -#endif - -enum termstyle { - STYLE_CLEAR, - STYLE_BOLD, - STYLE_UNDERLINE +struct termenc { + const char *enc; + int sym; }; static void body(struct termp *, @@ -83,40 +44,161 @@ static void footer(struct termp *, const struct mdoc_meta *); static void pword(struct termp *, const char *, size_t); -static void pescape(struct termp *, - const char *, size_t *, size_t); -static void nescape(struct termp *, +static void pescape(struct termp *, const char *, + size_t *, size_t); +static void nescape(struct termp *, const char *, size_t); static void chara(struct termp *, char); -static void stringa(struct termp *, const char *); -static void style(struct termp *, enum termstyle); +static void stringa(struct termp *, + const char *, size_t); +static void symbola(struct termp *, enum tsym); +static void sanity(const struct mdoc_node *); +static void stylea(struct termp *, enum tstyle); #ifdef __linux__ extern size_t strlcat(char *, const char *, size_t); extern size_t strlcpy(char *, const char *, size_t); #endif +static struct termenc termenc1[] = { + { "\\", TERMSYM_SLASH }, + { "\'", TERMSYM_RSQUOTE }, + { "`", TERMSYM_LSQUOTE }, + { "-", TERMSYM_HYPHEN }, + { " ", TERMSYM_SPACE }, + { ".", TERMSYM_PERIOD }, + { "&", TERMSYM_BREAK }, + { "e", TERMSYM_SLASH }, + { "q", TERMSYM_DQUOTE }, + { NULL, 0 } +}; +static struct termenc termenc2[] = { + { "rB", TERMSYM_RBRACK }, + { "lB", TERMSYM_LBRACK }, + { "ra", TERMSYM_RANGLE }, + { "la", TERMSYM_LANGLE }, + { "Lq", TERMSYM_LDQUOTE }, + { "lq", TERMSYM_LDQUOTE }, + { "Rq", TERMSYM_RDQUOTE }, + { "rq", TERMSYM_RDQUOTE }, + { "oq", TERMSYM_LSQUOTE }, + { "aq", TERMSYM_RSQUOTE }, + + { "<-", TERMSYM_LARROW }, + { "->", TERMSYM_RARROW }, + { "ua", TERMSYM_UARROW }, + { "da", TERMSYM_DARROW }, + + { "bu", TERMSYM_BULLET }, + { "Ba", TERMSYM_BAR }, + { "ba", TERMSYM_BAR }, + { "co", TERMSYM_COPY }, + { "Am", TERMSYM_AMP }, + + { "Le", TERMSYM_LE }, + { "<=", TERMSYM_LE }, + { "Ge", TERMSYM_GE }, + { ">=", TERMSYM_GE }, + { "==", TERMSYM_EQ }, + { "Ne", TERMSYM_NEQ }, + { "!=", TERMSYM_NEQ }, + { "Pm", TERMSYM_PLUSMINUS }, + { "+-", TERMSYM_PLUSMINUS }, + { "If", TERMSYM_INF2 }, + { "if", TERMSYM_INF }, + { "Na", TERMSYM_NAN }, + { "na", TERMSYM_NAN }, + { "**", TERMSYM_ASTERISK }, + { "Gt", TERMSYM_GT }, + { "Lt", TERMSYM_LT }, + + { "aa", TERMSYM_ACUTE }, + { "ga", TERMSYM_GRAVE }, + + { "en", TERMSYM_EN }, + { "em", TERMSYM_EM }, + + { "Pi", TERMSYM_PI }, + { NULL, 0 } +}; + +static struct termsym termsym_ansi[] = { + { "]", 1 }, /* TERMSYM_RBRACK */ + { "[", 1 }, /* TERMSYM_LBRACK */ + { "<-", 2 }, /* TERMSYM_LARROW */ + { "->", 2 }, /* TERMSYM_RARROW */ + { "^", 1 }, /* TERMSYM_UARROW */ + { "v", 1 }, /* TERMSYM_DARROW */ + { "`", 1 }, /* TERMSYM_LSQUOTE */ + { "\'", 1 }, /* TERMSYM_RSQUOTE */ + { "\'", 1 }, /* TERMSYM_SQUOTE */ + { "``", 2 }, /* TERMSYM_LDQUOTE */ + { "\'\'", 2 }, /* TERMSYM_RDQUOTE */ + { "\"", 1 }, /* TERMSYM_DQUOTE */ + { "<", 1 }, /* TERMSYM_LT */ + { ">", 1 }, /* TERMSYM_GT */ + { "<=", 2 }, /* TERMSYM_LE */ + { ">=", 2 }, /* TERMSYM_GE */ + { "==", 2 }, /* TERMSYM_EQ */ + { "!=", 2 }, /* TERMSYM_NEQ */ + { "\'", 1 }, /* TERMSYM_ACUTE */ + { "`", 1 }, /* TERMSYM_GRAVE */ + { "pi", 2 }, /* TERMSYM_PI */ + { "+=", 2 }, /* TERMSYM_PLUSMINUS */ + { "oo", 2 }, /* TERMSYM_INF */ + { "infinity", 8 }, /* TERMSYM_INF2 */ + { "NaN", 3 }, /* TERMSYM_NAN */ + { "|", 1 }, /* TERMSYM_BAR */ + { "o", 1 }, /* TERMSYM_BULLET */ + { "&", 1 }, /* TERMSYM_AMP */ + { "--", 2 }, /* TERMSYM_EM */ + { "-", 1 }, /* TERMSYM_EN */ + { "(C)", 3 }, /* TERMSYM_COPY */ + { "*", 1 }, /* TERMSYM_ASTERISK */ + { "\\", 1 }, /* TERMSYM_SLASH */ + { "-", 1 }, /* TERMSYM_HYPHEN */ + { " ", 1 }, /* TERMSYM_SPACE */ + { ".", 1 }, /* TERMSYM_PERIOD */ + { "", 0 }, /* TERMSYM_BREAK */ + { "<", 1 }, /* TERMSYM_LANGLE */ + { ">", 1 }, /* TERMSYM_RANGLE */ +}; + +static const char ansi_clear[] = { 27, '[', '0', 'm' }; +static const char ansi_bold[] = { 27, '[', '1', 'm' }; +static const char ansi_under[] = { 27, '[', '4', 'm' }; + +static struct termsym termstyle_ansi[] = { + { ansi_clear, 4 }, + { ansi_bold, 4 }, + { ansi_under, 4 } +}; + + int main(int argc, char *argv[]) { - struct mmain *p; + struct mmain *p; + int c; const struct mdoc *mdoc; - struct termp termp; + struct termp termp; p = mmain_alloc(); - if ( ! mmain_getopt(p, argc, argv, NULL, NULL, NULL, NULL)) - mmain_exit(p, 1); + c = mmain_getopt(p, argc, argv, NULL, NULL, NULL, NULL); + if (1 != c) + mmain_exit(p, -1 == c ? 1 : 0); if (NULL == (mdoc = mmain_mdoc(p))) mmain_exit(p, 1); - termp.maxrmargin = 80; /* XXX */ - termp.rmargin = termp.maxrmargin; + termp.maxrmargin = termp.rmargin = 78; /* XXX */ termp.maxcols = 1024; termp.offset = termp.col = 0; termp.flags = TERMP_NOSPACE; + termp.symtab = termsym_ansi; + termp.styletab = termstyle_ansi; if (NULL == (termp.buf = malloc(termp.maxcols))) err(1, "malloc"); @@ -132,10 +214,42 @@ main(int argc, char *argv[]) } +/* + * 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. + * + * Specifically, a line is whatever's in p->buf of length p->col, which + * is zeroed after this function returns. + * + * The variables TERMP_NOLPAD, TERMP_LITERAL and TERMP_NOBREAK are of + * critical importance here. Their behaviour 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. + * + * - TERMP_NOBREAK: this is the most important and is used when making + * columns. In short: don't print a newline and instead pad to the + * right margin. Used in conjunction with TERMP_NOLPAD. + * + * In-line line breaking: + * + * 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). + */ void flushln(struct termp *p) { - size_t i, j, vsz, vis, maxvis; + size_t i, j, vsz, vis, maxvis, mmax, bp; /* * First, establish the maximum columns of "visible" content. @@ -146,6 +260,8 @@ flushln(struct termp *p) assert(p->offset < p->rmargin); maxvis = p->rmargin - p->offset; + mmax = p->maxrmargin - p->offset; + bp = TERMP_NOBREAK & p->flags ? mmax : maxvis; vis = 0; /* @@ -159,27 +275,19 @@ flushln(struct termp *p) for (j = 0; j < p->offset; j++) putchar(' '); - /* - * If we're literal, print out verbatim. - */ - if (p->flags & TERMP_LITERAL) { - for (i = 0; i < p->col; i++) - putchar(p->buf[i]); - putchar('\n'); - p->col = 0; - return; - } - 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])); + /* FIXME: make non-ANSI friendly. */ + /* LINTED */ for (j = i, vsz = 0; j < p->col; j++) { - if (xisspace(p->buf[j])) + if (isspace((u_char)p->buf[j])) break; else if (27 == p->buf[j]) { assert(j + 4 <= p->col); @@ -187,52 +295,78 @@ flushln(struct termp *p) } else vsz++; } - assert(vsz > 0); /* - * 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). + * Do line-breaking. If we're greater than our + * break-point and already in-line, break to the next + * line and start writing. If we're at the line start, + * then write out the word (TODO: hyphenate) and break + * in a subsequent loop invocation. */ - /* FIXME: allow selective right-margin breaking. */ + if ( ! (TERMP_NOBREAK & p->flags)) { + if (vis && vis + vsz > bp) { + putchar('\n'); + for (j = 0; j < p->offset; j++) + putchar(' '); + vis = 0; + } else if (vis + vsz > bp) + warnx("word breaks right margin"); - if (vis && vis + vsz > maxvis) { - if (p->flags & TERMP_NOBREAK) - errx(1, "word breaks right margin"); - putchar('\n'); - for (j = 0; j < p->offset; j++) - putchar(' '); - vis = 0; - } else if (vis + vsz > maxvis) - errx(1, "word breaks right margin"); + /* TODO: hyphenate. */ + } else { + if (vis && vis + vsz > bp) { + putchar('\n'); + for (j = 0; j < p->rmargin; j++) + putchar(' '); + vis = p->rmargin - p->offset; + } else if (vis + vsz > bp) + warnx("word breaks right margin"); + + /* TODO: hyphenate. */ + } + /* * Write out the word and a trailing space. Omit the - * space if we're the last word in the line. + * space if we're the last word in the line or beyond + * our breakpoint. */ for ( ; i < p->col; i++) { - if (xisspace(p->buf[i])) + if (isspace((u_char)p->buf[i])) break; putchar(p->buf[i]); } vis += vsz; - if (i < p->col) { + if (i < p->col && vis <= bp) { putchar(' '); vis++; } } /* + * If we've overstepped our maximum visible no-break space, then + * cause a newline and offset at the right margin. + */ + + if ((TERMP_NOBREAK & p->flags) && vis >= maxvis) { + if ( ! (TERMP_NONOBREAK & p->flags)) { + 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)) + if ( ! (TERMP_NONOBREAK & p->flags)) for ( ; vis < maxvis; vis++) putchar(' '); } else @@ -242,14 +376,15 @@ flushln(struct termp *p) } +/* + * A newline only breaks an existing line; it won't assert vertical + * space. All data in the output buffer is flushed prior to the newline + * assertion. + */ void newln(struct termp *p) { - /* - * A newline only breaks an existing line; it won't assert - * vertical space. - */ p->flags |= TERMP_NOSPACE; if (0 == p->col) { p->flags &= ~TERMP_NOLPAD; @@ -260,281 +395,32 @@ newln(struct termp *p) } +/* + * Asserts a vertical space (a full, empty line-break between lines). + * Note that if used twice, this will cause two blank spaces and so on. + * All data in the output buffer is flushed prior to the newline + * assertion. + */ void vspace(struct termp *p) { - /* - * Asserts a vertical space (a full, empty line-break between - * lines). - */ newln(p); putchar('\n'); } -static void -stringa(struct termp *p, const char *s) -{ - - /* XXX - speed up if not passing to chara. */ - for ( ; *s; s++) - chara(p, *s); -} - - -static void -chara(struct termp *p, char c) -{ - - /* - * 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; -} - - -static void -style(struct termp *p, enum termstyle esc) -{ - - if (p->col + 4 >= p->maxcols) - errx(1, "line overrun"); - - p->buf[(p->col)++] = 27; - p->buf[(p->col)++] = '['; - switch (esc) { - case (STYLE_CLEAR): - p->buf[(p->col)++] = '0'; - break; - case (STYLE_BOLD): - p->buf[(p->col)++] = '1'; - break; - case (STYLE_UNDERLINE): - p->buf[(p->col)++] = '4'; - break; - default: - abort(); - /* NOTREACHED */ - } - p->buf[(p->col)++] = 'm'; -} - - -static void -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]) - stringa(p, TERMSYM_RBRACK); - else if ('l' == word[0] && 'B' == word[1]) - stringa(p, TERMSYM_LBRACK); - else if ('l' == word[0] && 'q' == word[1]) - stringa(p, TERMSYM_LDQUOTE); - else if ('r' == word[0] && 'q' == word[1]) - 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]) - 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; - } -} - - -static void -pescape(struct termp *p, const char *word, size_t *i, size_t len) -{ - size_t j; - - (*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)++; - assert(*i + 1 < len); - nescape(p, &word[*i], 2); - (*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]) { - case ('\\'): - /* FALLTHROUGH */ - case ('\''): - /* FALLTHROUGH */ - case ('`'): - /* FALLTHROUGH */ - case ('-'): - /* FALLTHROUGH */ - case (' '): - /* FALLTHROUGH */ - case ('.'): - chara(p, word[*i]); - default: - break; - } - return; - } - - (*i)++; - for (j = 0; word[*i] && ']' != word[*i]; (*i)++, j++) - /* Loop... */ ; - - nescape(p, &word[*i - j], j); -} - - -static void -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)) - chara(p, ' '); - - 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) - style(p, STYLE_UNDERLINE); - - for (i = 0; i < len; i++) { - if ('\\' == word[i]) { - pescape(p, word, &i, len); - continue; - } - chara(p, word[i]); - } - - if (p->flags & TERMP_BOLD || - p->flags & TERMP_UNDERLINE) - style(p, STYLE_CLEAR); -} - - +/* + * Break apart a word into "pwords" (partial-words, usually from + * breaking up a phrase into individual words) and, eventually, put them + * into the output buffer. If we're a literal word, then don't break up + * the word and put it verbatim into the output buffer. + */ void 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; @@ -551,10 +437,18 @@ word(struct termp *p, const char *word) /* LINTED */ for (j = i = 0; i < len; i++) { - if ( ! xisspace(word[i])) { + if ( ! isspace((u_char)word[i])) { j++; continue; + } + + /* Escaped spaces don't delimit... */ + if (i > 0 && isspace((u_char)word[i]) && + '\\' == word[i - 1]) { + j++; + continue; } + if (0 == j) continue; assert(i >= j); @@ -568,6 +462,12 @@ word(struct termp *p, const char *word) } +/* + * This is the main function for printing out nodes. It's constituted + * of PRE and POST functions, which correspond to prefix and infix + * processing. The termpair structure allows data to persist between + * prefix and postfix invocations. + */ static void body(struct termp *p, struct termpair *ppair, const struct mdoc_meta *meta, @@ -576,12 +476,10 @@ 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. - */ + /* Some quick sanity-checking. */ + sanity(node); + /* Pre-processing. */ dochild = 1; @@ -596,7 +494,7 @@ body(struct termp *p, struct termpair *ppair, if ( ! (*termacts[node->tok].pre)(p, &pair, meta, node)) dochild = 0; } else /* MDOC_TEXT == node->type */ - word(p, node->data.text.string); + word(p, node->string); /* Children. */ @@ -676,9 +574,7 @@ footer(struct termp *p, const struct mdoc_meta *meta) static void header(struct termp *p, const struct mdoc_meta *meta) { - char *buf, *title, *bufp, *vbuf; - const char *pp; - struct utsname uts; + char *buf, *title, *bufp; p->rmargin = p->maxrmargin; p->offset = 0; @@ -687,49 +583,7 @@ header(struct termp *p, const struct mdoc_meta *meta) 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))) { - switch (meta->msec) { - case (MSEC_1): - /* FALLTHROUGH */ - case (MSEC_6): - /* FALLTHROUGH */ - case (MSEC_7): - pp = mdoc_vol2a(VOL_URM); - break; - case (MSEC_8): - pp = mdoc_vol2a(VOL_SMM); - break; - case (MSEC_2): - /* FALLTHROUGH */ - case (MSEC_3): - /* FALLTHROUGH */ - case (MSEC_4): - /* FALLTHROUGH */ - case (MSEC_5): - pp = mdoc_vol2a(VOL_PRM); - break; - case (MSEC_9): - pp = mdoc_vol2a(VOL_KM); - break; - default: - 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: @@ -743,19 +597,20 @@ header(struct termp *p, const struct mdoc_meta *meta) * switches on the manual section. */ - if (mdoc_arch2a(meta->arch)) - (void)snprintf(buf, p->rmargin, "%s (%s)", - vbuf, mdoc_arch2a(meta->arch)); - else - (void)strlcpy(buf, vbuf, p->rmargin); + assert(meta->vol); + (void)strlcpy(buf, meta->vol, p->rmargin); - pp = mdoc_msec2a(meta->msec); + if (meta->arch) { + (void)strlcat(buf, " (", p->rmargin); + (void)strlcat(buf, meta->arch, p->rmargin); + (void)strlcat(buf, ")", p->rmargin); + } - (void)snprintf(title, p->rmargin, "%s(%s)", - meta->title, pp ? pp : ""); + (void)snprintf(title, p->rmargin, "%s(%d)", + meta->title, meta->msec); for (bufp = title; *bufp; bufp++) - *bufp = toupper(*bufp); + *bufp = toupper((u_char)*bufp); p->offset = 0; p->rmargin = (p->maxrmargin - strlen(buf)) / 2; @@ -784,6 +639,325 @@ header(struct termp *p, const struct mdoc_meta *meta) p->flags &= ~TERMP_NOSPACE; free(title); - free(vbuf); free(buf); } + + +/* + * 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 +nescape(struct termp *p, const char *word, size_t len) +{ + struct termenc *enc; + + switch (len) { + case (1): + enc = termenc1; + break; + case (2): + enc = termenc2; + break; + default: + warnx("unsupported %zu-byte escape sequence", len); + return; + } + + for ( ; enc->enc; enc++) + if (0 == memcmp(enc->enc, word, len)) { + symbola(p, enc->sym); + return; + } + + warnx("unsupported %zu-byte escape sequence", len); +} + + +/* + * 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 +pescape(struct termp *p, const char *word, size_t *i, size_t len) +{ + size_t j; + + if (++(*i) >= len) { + warnx("ignoring bad escape sequence"); + return; + } + + if ('(' == word[*i]) { + (*i)++; + if (*i + 1 >= len) { + warnx("ignoring bad escape sequence"); + return; + } + nescape(p, &word[*i], 2); + (*i)++; + return; + + } else if ('*' == word[*i]) { + (*i)++; + if (*i >= len) { + warnx("ignoring bad escape sequence"); + return; + } + switch (word[*i]) { + case ('('): + (*i)++; + if (*i + 1 >= len) { + warnx("ignoring bad escape sequence"); + return; + } + nescape(p, &word[*i], 2); + (*i)++; + return; + case ('['): + break; + default: + nescape(p, &word[*i], 1); + return; + } + + } else if ('[' != word[*i]) { + nescape(p, &word[*i], 1); + return; + } + + (*i)++; + for (j = 0; word[*i] && ']' != word[*i]; (*i)++, j++) + /* Loop... */ ; + + if (0 == word[*i]) { + warnx("ignoring bad escape sequence"); + return; + } + nescape(p, &word[*i - j], j); +} + + +/* + * 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 +pword(struct termp *p, const char *word, size_t len) +{ + size_t i; + + if ( ! (TERMP_NOSPACE & p->flags) && + ! (TERMP_LITERAL & p->flags)) + chara(p, ' '); + + 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) + stylea(p, TERMSTYLE_BOLD); + if (p->flags & TERMP_UNDERLINE) + stylea(p, TERMSTYLE_UNDER); + + for (i = 0; i < len; i++) { + if ('\\' == word[i]) { + pescape(p, word, &i, len); + continue; + } + chara(p, word[i]); + } + + if (p->flags & TERMP_BOLD || + p->flags & TERMP_UNDERLINE) + stylea(p, TERMSTYLE_CLEAR); +} + + +/* + * Add a symbol to the output line buffer. + */ +static void +symbola(struct termp *p, enum tsym sym) +{ + + assert(p->symtab[sym].sym); + stringa(p, p->symtab[sym].sym, p->symtab[sym].sz); +} + + +/* + * Add a style to the output line buffer. + */ +static void +stylea(struct termp *p, enum tstyle style) +{ + + assert(p->styletab[style].sym); + stringa(p, p->styletab[style].sym, p->styletab[style].sz); +} + + +/* + * Like chara() but for arbitrary-length buffers. Resize the buffer by + * a factor of two (if the buffer is less than that) or the buffer's + * size. + */ +static void +stringa(struct termp *p, const char *c, size_t sz) +{ + size_t s; + + if (0 == sz) + return; + + s = sz > p->maxcols * 2 ? sz : p->maxcols * 2; + + assert(c); + if (p->col + sz >= p->maxcols) { + p->buf = realloc(p->buf, s); + if (NULL == p->buf) + err(1, "realloc"); + p->maxcols = s; + } + + (void)memcpy(&p->buf[p->col], c, sz); + p->col += sz; +} + + +/* + * 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 +chara(struct termp *p, char c) +{ + + 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; +} + + +static void +sanity(const struct mdoc_node *n) +{ + + switch (n->type) { + case (MDOC_TEXT): + if (n->child) + errx(1, "regular form violated (1)"); + if (NULL == n->parent) + errx(1, "regular form violated (2)"); + if (NULL == n->string) + errx(1, "regular form violated (3)"); + switch (n->parent->type) { + case (MDOC_TEXT): + /* FALLTHROUGH */ + case (MDOC_ROOT): + errx(1, "regular form violated (4)"); + /* NOTREACHED */ + default: + break; + } + break; + case (MDOC_ELEM): + if (NULL == n->parent) + errx(1, "regular form violated (5)"); + switch (n->parent->type) { + case (MDOC_TAIL): + /* FALLTHROUGH */ + case (MDOC_BODY): + /* FALLTHROUGH */ + case (MDOC_HEAD): + break; + default: + errx(1, "regular form violated (6)"); + /* NOTREACHED */ + } + if (n->child) switch (n->child->type) { + case (MDOC_TEXT): + break; + default: + errx(1, "regular form violated (7("); + /* NOTREACHED */ + } + break; + case (MDOC_HEAD): + /* FALLTHROUGH */ + case (MDOC_BODY): + /* FALLTHROUGH */ + case (MDOC_TAIL): + if (NULL == n->parent) + errx(1, "regular form violated (8)"); + if (MDOC_BLOCK != n->parent->type) + errx(1, "regular form violated (9)"); + if (n->child) switch (n->child->type) { + case (MDOC_BLOCK): + /* FALLTHROUGH */ + case (MDOC_ELEM): + /* FALLTHROUGH */ + case (MDOC_TEXT): + break; + default: + errx(1, "regular form violated (a)"); + /* NOTREACHED */ + } + break; + case (MDOC_BLOCK): + if (NULL == n->parent) + errx(1, "regular form violated (b)"); + if (NULL == n->child) + errx(1, "regular form violated (c)"); + switch (n->parent->type) { + case (MDOC_ROOT): + /* FALLTHROUGH */ + case (MDOC_HEAD): + /* FALLTHROUGH */ + case (MDOC_BODY): + /* FALLTHROUGH */ + case (MDOC_TAIL): + break; + default: + errx(1, "regular form violated (d)"); + /* NOTREACHED */ + } + switch (n->child->type) { + case (MDOC_ROOT): + /* FALLTHROUGH */ + case (MDOC_ELEM): + errx(1, "regular form violated (e)"); + /* NOTREACHED */ + default: + break; + } + break; + case (MDOC_ROOT): + if (n->parent) + errx(1, "regular form violated (f)"); + if (NULL == n->child) + errx(1, "regular form violated (10)"); + switch (n->child->type) { + case (MDOC_BLOCK): + break; + default: + errx(1, "regular form violated (11)"); + /* NOTREACHED */ + } + break; + } +} +