version 1.11, 2009/02/25 13:30:53 |
version 1.49, 2009/03/19 11:49:00 |
|
|
/* $Id$ */ |
/* $Id$ */ |
/* |
/* |
* Copyright (c) 2008 Kristaps Dzonsons <kristaps@kth.se> |
* Copyright (c) 2008, 2009 Kristaps Dzonsons <kristaps@openbsd.org> |
* |
* |
* Permission to use, copy, modify, and distribute this software for any |
* Permission to use, copy, modify, and distribute this software for any |
* purpose with or without fee is hereby granted, provided that the |
* purpose with or without fee is hereby granted, provided that the |
|
|
* TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR |
* TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR |
* PERFORMANCE OF THIS SOFTWARE. |
* PERFORMANCE OF THIS SOFTWARE. |
*/ |
*/ |
|
#include <sys/types.h> |
|
#include <sys/stat.h> |
|
|
#include <assert.h> |
#include <assert.h> |
#include <ctype.h> |
|
#include <err.h> |
#include <err.h> |
|
#include <fcntl.h> |
#include <getopt.h> |
#include <getopt.h> |
#include <stdio.h> |
#include <stdio.h> |
#include <stdlib.h> |
#include <stdlib.h> |
#include <string.h> |
#include <string.h> |
|
#include <unistd.h> |
|
|
#ifndef __OpenBSD__ |
|
#include <time.h> |
|
#endif |
|
|
|
#include "mmain.h" |
|
#include "term.h" |
#include "term.h" |
|
|
#ifdef __NetBSD__ |
#define WARN_WALL 0x03 /* All-warnings mask. */ |
#define xisspace(x) isspace((int)(x)) |
#define WARN_WCOMPAT (1 << 0) /* Compatibility warnings. */ |
#else |
#define WARN_WSYNTAX (1 << 1) /* Syntax warnings. */ |
#define xisspace(x) isspace((x)) |
#define WARN_WERR (1 << 2) /* Warnings->errors. */ |
#endif |
|
|
|
enum termstyle { |
enum termt { |
STYLE_CLEAR, |
TERMT_ASCII, |
STYLE_BOLD, |
TERMT_LINT, |
STYLE_UNDERLINE |
TERMT_TREE |
}; |
}; |
|
|
static void body(struct termp *, |
extern char *__progname; |
|
|
|
__dead static void version(void); |
|
__dead static void usage(void); |
|
#if 0 |
|
__dead static void punt(struct cmdargs *, char **); |
|
#endif |
|
static int foptions(int *, char *); |
|
static int toptions(enum termt *, char *); |
|
static int woptions(int *, char *); |
|
static int merr(void *, int, int, const char *); |
|
static int mwarn(void *, int, int, |
|
enum mdoc_warn, const char *); |
|
static void body(struct termp *, struct termpair *, |
const struct mdoc_meta *, |
const struct mdoc_meta *, |
const struct mdoc_node *); |
const struct mdoc_node *); |
static void header(struct termp *, |
static void header(struct termp *, |
const struct mdoc_meta *); |
const struct mdoc_meta *); |
static void footer(struct termp *, |
static void footer(struct termp *, |
const struct mdoc_meta *); |
const struct mdoc_meta *); |
|
static int file(char **, size_t *, char **, size_t *, |
static void pword(struct termp *, const char *, size_t); |
const char *, struct mdoc *); |
|
static int fdesc(char **, size_t *, char **, size_t *, |
|
const char *, int, struct mdoc *); |
|
static void pword(struct termp *, const char *, int); |
static void pescape(struct termp *, |
static void pescape(struct termp *, |
const char *, size_t *, size_t); |
const char *, int *, int); |
|
static void nescape(struct termp *, |
|
const char *, size_t); |
static void chara(struct termp *, char); |
static void chara(struct termp *, char); |
static void stringa(struct termp *, const char *); |
static void stringa(struct termp *, |
static void style(struct termp *, enum termstyle); |
const char *, size_t); |
|
static void sanity(const struct mdoc_node *); |
|
|
#ifdef __linux__ |
|
extern size_t strlcat(char *, const char *, size_t); |
|
extern size_t strlcpy(char *, const char *, size_t); |
|
#endif |
|
|
|
|
|
int |
int |
main(int argc, char *argv[]) |
main(int argc, char *argv[]) |
{ |
{ |
struct mmain *p; |
|
const struct mdoc *mdoc; |
|
struct termp termp; |
struct termp termp; |
|
int c, fflags, wflags; |
|
struct mdoc_cb cb; |
|
struct mdoc *mdoc; |
|
char *buf, *line; |
|
size_t bufsz, linesz; |
|
enum termt termt; |
|
|
p = mmain_alloc(); |
bzero(&termp, sizeof(struct termp)); |
|
bzero(&cb, sizeof(struct mdoc_cb)); |
|
|
if ( ! mmain_getopt(p, argc, argv, NULL, NULL, NULL, NULL)) |
termt = TERMT_ASCII; |
mmain_exit(p, 1); |
fflags = wflags = 0; |
|
|
if (NULL == (mdoc = mmain_mdoc(p))) |
/* LINTED */ |
mmain_exit(p, 1); |
while (-1 != (c = getopt(argc, argv, "f:VW:T:"))) |
|
switch (c) { |
|
case ('f'): |
|
if ( ! foptions(&fflags, optarg)) |
|
return(0); |
|
break; |
|
case ('T'): |
|
if ( ! toptions(&termt, optarg)) |
|
return(0); |
|
break; |
|
case ('W'): |
|
if ( ! woptions(&wflags, optarg)) |
|
return(0); |
|
break; |
|
case ('V'): |
|
version(); |
|
/* NOTREACHED */ |
|
default: |
|
usage(); |
|
/* NOTREACHED */ |
|
} |
|
|
termp.maxrmargin = 80; /* XXX */ |
argc -= optind; |
termp.rmargin = termp.maxrmargin; |
argv += optind; |
termp.maxcols = 1024; |
|
termp.offset = termp.col = 0; |
|
termp.flags = TERMP_NOSPACE; |
|
|
|
if (NULL == (termp.buf = malloc(termp.maxcols))) |
termp.maxrmargin = 78; /* FIXME */ |
err(1, "malloc"); |
|
|
|
header(&termp, mdoc_meta(mdoc)); |
cb.mdoc_err = merr; |
body(&termp, mdoc_meta(mdoc), mdoc_node(mdoc)); |
cb.mdoc_warn = mwarn; |
footer(&termp, mdoc_meta(mdoc)); |
|
|
|
free(termp.buf); |
/* Line and block buffers persist between parses. */ |
|
|
mmain_exit(p, 0); |
buf = line = NULL; |
|
bufsz = linesz = 0; |
|
|
|
/* Overall mdoc persists between parses. */ |
|
|
|
mdoc = mdoc_alloc(&wflags, fflags, &cb); |
|
|
|
while (*argv) { |
|
if ( ! file(&line, &linesz, &buf, &bufsz, *argv, mdoc)) |
|
break; |
|
|
|
switch (termt) { |
|
case (TERMT_ASCII): |
|
if (NULL == termp.symtab) |
|
termp.symtab = ascii2htab(); |
|
header(&termp, mdoc_meta(mdoc)); |
|
body(&termp, NULL, mdoc_meta(mdoc), |
|
mdoc_node(mdoc)); |
|
footer(&termp, mdoc_meta(mdoc)); |
|
break; |
|
default: |
|
break; |
|
} |
|
|
|
mdoc_reset(mdoc); |
|
argv++; |
|
} |
|
|
|
if (buf) |
|
free(buf); |
|
if (line) |
|
free(line); |
|
if (termp.buf) |
|
free(termp.buf); |
|
if (termp.symtab) |
|
asciifree(termp.symtab); |
|
|
|
mdoc_free(mdoc); |
|
|
|
return(0); |
|
} |
|
|
|
|
|
__dead static void |
|
version(void) |
|
{ |
|
|
|
(void)printf("%s %s\n", __progname, VERSION); |
|
exit(0); |
/* NOTREACHED */ |
/* NOTREACHED */ |
} |
} |
|
|
|
|
|
__dead static void |
|
usage(void) |
|
{ |
|
|
|
(void)fprintf(stderr, "usage: %s\n", __progname); |
|
exit(1); |
|
/* NOTREACHED */ |
|
} |
|
|
|
|
|
static int |
|
file(char **ln, size_t *lnsz, char **buf, size_t *bufsz, |
|
const char *file, struct mdoc *mdoc) |
|
{ |
|
int fd, c; |
|
|
|
if (-1 == (fd = open(file, O_RDONLY, 0))) { |
|
warn("%s", file); |
|
return(0); |
|
} |
|
|
|
c = fdesc(ln, lnsz, buf, bufsz, file, fd, mdoc); |
|
|
|
if (-1 == close(fd)) |
|
warn("%s", file); |
|
|
|
return(c); |
|
} |
|
|
|
|
|
static int |
|
fdesc(char **lnp, size_t *lnsz, char **bufp, size_t *bufsz, |
|
const char *f, int fd, struct mdoc *mdoc) |
|
{ |
|
size_t sz; |
|
ssize_t ssz; |
|
struct stat st; |
|
int j, i, pos, lnn; |
|
char *ln, *buf; |
|
|
|
buf = *bufp; |
|
ln = *lnp; |
|
|
|
/* |
|
* Two buffers: ln and buf. buf is the input buffer, optimised |
|
* for each file's block size. ln is a line buffer. Both |
|
* growable, hence passed in by ptr-ptr. |
|
*/ |
|
|
|
if (-1 == fstat(fd, &st)) { |
|
warnx("%s", f); |
|
sz = BUFSIZ; |
|
} else |
|
sz = (unsigned)BUFSIZ > st.st_blksize ? |
|
(size_t)BUFSIZ : st.st_blksize; |
|
|
|
if (sz > *bufsz) { |
|
if (NULL == (buf = realloc(buf, sz))) |
|
err(1, "realloc"); |
|
*bufp = buf; |
|
*bufsz = sz; |
|
} |
|
|
|
/* |
|
* Fill buf with file blocksize and parse newlines into ln. |
|
*/ |
|
|
|
for (lnn = 1, pos = 0; ; ) { |
|
if (-1 == (ssz = read(fd, buf, sz))) { |
|
warn("%s", f); |
|
return(0); |
|
} else if (0 == ssz) |
|
break; |
|
|
|
for (i = 0; i < (int)ssz; i++) { |
|
if (pos >= (int)*lnsz) { |
|
*lnsz += 256; /* Step-size. */ |
|
ln = realloc(ln, *lnsz); |
|
if (NULL == ln) |
|
err(1, "realloc"); |
|
*lnp = ln; |
|
} |
|
|
|
if ('\n' != buf[i]) { |
|
ln[pos++] = buf[i]; |
|
continue; |
|
} |
|
|
|
/* Check for CPP-escaped newline. */ |
|
|
|
if (pos > 0 && '\\' == ln[pos - 1]) { |
|
for (j = pos - 1; j >= 0; j--) |
|
if ('\\' != ln[j]) |
|
break; |
|
|
|
if ( ! ((pos - j) % 2)) { |
|
pos--; |
|
lnn++; |
|
continue; |
|
} |
|
} |
|
|
|
ln[pos] = 0; |
|
if ( ! mdoc_parseln(mdoc, lnn, ln)) |
|
return(0); |
|
lnn++; |
|
pos = 0; |
|
} |
|
} |
|
|
|
return(mdoc_endparse(mdoc)); |
|
} |
|
|
|
|
|
static int |
|
toptions(enum termt *tflags, char *arg) |
|
{ |
|
|
|
if (0 == strcmp(arg, "ascii")) |
|
*tflags = TERMT_ASCII; |
|
else if (0 == strcmp(arg, "lint")) |
|
*tflags = TERMT_LINT; |
|
else if (0 == strcmp(arg, "tree")) |
|
*tflags = TERMT_TREE; |
|
else { |
|
warnx("bad argument: -T%s", arg); |
|
return(0); |
|
} |
|
|
|
return(1); |
|
} |
|
|
|
|
|
/* |
|
* Parse out the options for [-fopt...] setting compiler options. These |
|
* can be comma-delimited or called again. |
|
*/ |
|
static int |
|
foptions(int *fflags, char *arg) |
|
{ |
|
char *v; |
|
char *toks[] = { "ign-scope", "ign-escape", |
|
"ign-macro", NULL }; |
|
|
|
while (*arg) |
|
switch (getsubopt(&arg, toks, &v)) { |
|
case (0): |
|
*fflags |= MDOC_IGN_SCOPE; |
|
break; |
|
case (1): |
|
*fflags |= MDOC_IGN_ESCAPE; |
|
break; |
|
case (2): |
|
*fflags |= MDOC_IGN_MACRO; |
|
break; |
|
default: |
|
warnx("bad argument: -f%s", arg); |
|
return(0); |
|
} |
|
|
|
return(1); |
|
} |
|
|
|
|
|
/* |
|
* Parse out the options for [-Werr...], which sets warning modes. |
|
* These can be comma-delimited or called again. XXX - should this be |
|
* using -w like troff? |
|
*/ |
|
static int |
|
woptions(int *wflags, char *arg) |
|
{ |
|
char *v; |
|
char *toks[] = { "all", "compat", |
|
"syntax", "error", NULL }; |
|
|
|
while (*arg) |
|
switch (getsubopt(&arg, toks, &v)) { |
|
case (0): |
|
*wflags |= WARN_WALL; |
|
break; |
|
case (1): |
|
*wflags |= WARN_WCOMPAT; |
|
break; |
|
case (2): |
|
*wflags |= WARN_WSYNTAX; |
|
break; |
|
case (3): |
|
*wflags |= WARN_WERR; |
|
break; |
|
default: |
|
warnx("bad argument: -W%s", arg); |
|
return(0); |
|
} |
|
|
|
return(1); |
|
} |
|
|
|
|
|
/* |
|
* 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. |
|
* |
|
* - TERMP_NONOBREAK: don't newline when TERMP_NOBREAK is specified. |
|
* |
|
* 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 |
void |
flushln(struct termp *p) |
flushln(struct termp *p) |
{ |
{ |
size_t i, j, vsz, vis, maxvis; |
int i, j; |
|
size_t vsz, vis, maxvis, mmax, bp; |
|
|
/* |
/* |
* First, establish the maximum columns of "visible" content. |
* First, establish the maximum columns of "visible" content. |
Line 113 flushln(struct termp *p) |
|
Line 427 flushln(struct termp *p) |
|
|
|
assert(p->offset < p->rmargin); |
assert(p->offset < p->rmargin); |
maxvis = p->rmargin - p->offset; |
maxvis = p->rmargin - p->offset; |
|
mmax = p->maxrmargin - p->offset; |
|
bp = TERMP_NOBREAK & p->flags ? mmax : maxvis; |
vis = 0; |
vis = 0; |
|
|
/* |
/* |
Line 123 flushln(struct termp *p) |
|
Line 439 flushln(struct termp *p) |
|
|
|
if ( ! (p->flags & TERMP_NOLPAD)) |
if ( ! (p->flags & TERMP_NOLPAD)) |
/* LINTED */ |
/* LINTED */ |
for (j = 0; j < p->offset; j++) |
for (j = 0; j < (int)p->offset; j++) |
putchar(' '); |
putchar(' '); |
|
|
/* |
for (i = 0; i < (int)p->col; i++) { |
* 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'); |
|
p->col = 0; |
|
return; |
|
} |
|
|
|
for (i = 0; i < p->col; i++) { |
|
/* |
/* |
* Count up visible word characters. Control sequences |
* 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 */ |
/* LINTED */ |
for (j = i, vsz = 0; j < p->col; j++) { |
for (j = i, vsz = 0; j < (int)p->col; j++) { |
if (xisspace(p->buf[j])) |
if (' ' == p->buf[j]) |
break; |
break; |
else if (27 == p->buf[j]) { |
else if (8 == p->buf[j]) |
assert(j + 4 <= p->col); |
j += 1; |
j += 3; |
else |
} else |
|
vsz++; |
vsz++; |
} |
} |
assert(vsz > 0); |
|
|
|
/* |
/* |
* If a word is too long and we're within a line, put it |
* Do line-breaking. If we're greater than our |
* on the next line. Puke if we're being asked to write |
* break-point and already in-line, break to the next |
* something that will exceed the right margin (i.e., |
* line and start writing. If we're at the line start, |
* from a fresh line or when we're not allowed to break |
* then write out the word (TODO: hyphenate) and break |
* the line with TERMP_NOBREAK). |
* in a subsequent loop invocation. |
*/ |
*/ |
|
|
if (vis && vis + vsz >= maxvis) { |
if ( ! (TERMP_NOBREAK & p->flags)) { |
/* FIXME */ |
if (vis && vis + vsz > bp) { |
if (p->flags & TERMP_NOBREAK) |
putchar('\n'); |
errx(1, "word breaks right margin"); |
for (j = 0; j < (int)p->offset; j++) |
putchar('\n'); |
putchar(' '); |
for (j = 0; j < p->offset; j++) |
vis = 0; |
putchar(' '); |
} else if (vis + vsz > bp) |
vis = 0; |
warnx("word breaks right margin"); |
} else if (vis + vsz >= maxvis) |
|
/* FIXME */ |
|
errx(1, "word breaks right margin"); |
|
|
|
|
/* TODO: hyphenate. */ |
|
|
|
} else { |
|
if (vis && vis + vsz > bp) { |
|
putchar('\n'); |
|
for (j = 0; j < (int)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 |
* 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++) { |
for ( ; i < (int)p->col; i++) { |
if (xisspace(p->buf[i])) |
if (' ' == p->buf[i]) |
break; |
break; |
putchar(p->buf[i]); |
putchar(p->buf[i]); |
} |
} |
vis += vsz; |
vis += vsz; |
if (i < p->col) { |
if (i < (int)p->col && vis <= bp) { |
putchar(' '); |
putchar(' '); |
vis++; |
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 < (int)p->rmargin; i++) |
|
putchar(' '); |
|
} |
|
p->col = 0; |
|
return; |
|
} |
|
|
|
/* |
* If we're not to right-marginalise it (newline), then instead |
* If we're not to right-marginalise it (newline), then instead |
* pad to the right margin and stay off. |
* pad to the right margin and stay off. |
*/ |
*/ |
|
|
if (p->flags & TERMP_NOBREAK) { |
if (p->flags & TERMP_NOBREAK) { |
for ( ; vis < maxvis; vis++) |
if ( ! (TERMP_NONOBREAK & p->flags)) |
putchar(' '); |
for ( ; vis < maxvis; vis++) |
|
putchar(' '); |
} else |
} else |
putchar('\n'); |
putchar('\n'); |
|
|
Line 209 flushln(struct termp *p) |
|
Line 540 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 |
void |
newln(struct termp *p) |
newln(struct termp *p) |
{ |
{ |
|
|
/* |
|
* A newline only breaks an existing line; it won't assert |
|
* vertical space. |
|
*/ |
|
p->flags |= TERMP_NOSPACE; |
p->flags |= TERMP_NOSPACE; |
if (0 == p->col) |
if (0 == p->col) { |
|
p->flags &= ~TERMP_NOLPAD; |
return; |
return; |
|
} |
flushln(p); |
flushln(p); |
p->flags &= ~TERMP_NOLPAD; |
p->flags &= ~TERMP_NOLPAD; |
} |
} |
|
|
|
|
|
/* |
|
* 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 |
void |
vspace(struct termp *p) |
vspace(struct termp *p) |
{ |
{ |
|
|
/* |
|
* Asserts a vertical space (a full, empty line-break between |
|
* lines). |
|
*/ |
|
newln(p); |
newln(p); |
putchar('\n'); |
putchar('\n'); |
} |
} |
|
|
|
|
static void |
/* |
stringa(struct termp *p, const char *s) |
* 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 |
/* XXX - speed up if not passing to chara. */ |
* the word and put it verbatim into the output buffer. |
for ( ; *s; s++) |
*/ |
chara(p, *s); |
|
} |
|
|
|
|
|
static void |
|
chara(struct termp *p, char c) |
|
{ |
|
|
|
/* TODO: dynamically expand the buffer. */ |
|
if (p->col + 1 >= p->maxcols) |
|
errx(1, "line overrun"); |
|
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 |
|
pescape(struct termp *p, const char *word, size_t *i, size_t len) |
|
{ |
|
|
|
(*i)++; |
|
assert(*i < len); |
|
|
|
if ('(' == word[*i]) { |
|
/* Two-character escapes. */ |
|
(*i)++; |
|
assert(*i + 1 < len); |
|
|
|
if ('r' == word[*i] && 'B' == word[*i + 1]) |
|
chara(p, ']'); |
|
else if ('l' == word[*i] && 'B' == word[*i + 1]) |
|
chara(p, '['); |
|
else if ('<' == word[*i] && '-' == word[*i + 1]) |
|
stringa(p, "<-"); |
|
else if ('-' == word[*i] && '>' == word[*i + 1]) |
|
stringa(p, "->"); |
|
else if ('l' == word[*i] && 'q' == word[*i + 1]) |
|
chara(p, '\"'); |
|
else if ('r' == word[*i] && 'q' == word[*i + 1]) |
|
chara(p, '\"'); |
|
|
|
(*i)++; |
|
return; |
|
|
|
} else if ('[' != word[*i]) { |
|
/* One-character escapes. */ |
|
switch (word[*i]) { |
|
case ('\\'): |
|
/* FALLTHROUGH */ |
|
case ('\''): |
|
/* FALLTHROUGH */ |
|
case ('`'): |
|
/* FALLTHROUGH */ |
|
case ('-'): |
|
/* FALLTHROUGH */ |
|
case ('.'): |
|
chara(p, word[*i]); |
|
default: |
|
break; |
|
} |
|
return; |
|
} |
|
/* n-character escapes. */ |
|
} |
|
|
|
|
|
static void |
|
pword(struct termp *p, const char *word, size_t len) |
|
{ |
|
size_t i; |
|
|
|
/*assert(len > 0);*/ /* Can be, if literal. */ |
|
|
|
if ( ! (p->flags & TERMP_NOSPACE) && |
|
! (p->flags & TERMP_LITERAL)) |
|
chara(p, ' '); |
|
|
|
p->flags &= ~TERMP_NOSPACE; |
|
|
|
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); |
|
} |
|
|
|
|
|
void |
void |
word(struct termp *p, const char *word) |
word(struct termp *p, const char *word) |
{ |
{ |
size_t i, j, len; |
int i, j, len; |
|
|
if (p->flags & TERMP_LITERAL) { |
if (p->flags & TERMP_LITERAL) { |
pword(p, word, strlen(word)); |
pword(p, word, (int)strlen(word)); |
return; |
return; |
} |
} |
|
|
len = strlen(word); |
if (0 == (len = (int)strlen(word))) |
assert(len > 0); |
errx(1, "blank line not in literal context"); |
|
|
if (mdoc_isdelim(word)) { |
if (mdoc_isdelim(word)) { |
if ( ! (p->flags & TERMP_IGNDELIM)) |
if ( ! (p->flags & TERMP_IGNDELIM)) |
Line 389 word(struct termp *p, const char *word) |
|
Line 601 word(struct termp *p, const char *word) |
|
|
|
/* LINTED */ |
/* LINTED */ |
for (j = i = 0; i < len; i++) { |
for (j = i = 0; i < len; i++) { |
if ( ! xisspace(word[i])) { |
if (' ' != word[i]) { |
j++; |
j++; |
continue; |
continue; |
|
} |
|
|
|
/* Escaped spaces don't delimit... */ |
|
if (i && ' ' == word[i] && '\\' == word[i - 1]) { |
|
j++; |
|
continue; |
} |
} |
|
|
if (0 == j) |
if (0 == j) |
continue; |
continue; |
assert(i >= j); |
assert(i >= j); |
Line 406 word(struct termp *p, const char *word) |
|
Line 625 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 |
static void |
body(struct termp *p, const struct mdoc_meta *meta, |
body(struct termp *p, struct termpair *ppair, |
|
const struct mdoc_meta *meta, |
const struct mdoc_node *node) |
const struct mdoc_node *node) |
{ |
{ |
int dochild; |
int dochild; |
struct termpair pair; |
struct termpair pair; |
|
|
|
/* Some quick sanity-checking. */ |
|
|
|
sanity(node); |
|
|
/* Pre-processing. */ |
/* Pre-processing. */ |
|
|
dochild = 1; |
dochild = 1; |
|
pair.ppair = ppair; |
pair.type = 0; |
pair.type = 0; |
pair.offset = pair.rmargin = 0; |
pair.offset = pair.rmargin = 0; |
pair.flag = 0; |
pair.flag = 0; |
|
pair.count = 0; |
|
|
if (MDOC_TEXT != node->type) { |
if (MDOC_TEXT != node->type) { |
if (termacts[node->tok].pre) |
if (termacts[node->tok].pre) |
if ( ! (*termacts[node->tok].pre)(p, &pair, meta, node)) |
if ( ! (*termacts[node->tok].pre)(p, &pair, meta, node)) |
dochild = 0; |
dochild = 0; |
} else /* MDOC_TEXT == node->type */ |
} else /* MDOC_TEXT == node->type */ |
word(p, node->data.text.string); |
word(p, node->string); |
|
|
/* Children. */ |
/* Children. */ |
|
|
Line 433 body(struct termp *p, const struct mdoc_meta *meta, |
|
Line 665 body(struct termp *p, const struct mdoc_meta *meta, |
|
p->flags |= pair.flag; |
p->flags |= pair.flag; |
|
|
if (dochild && node->child) |
if (dochild && node->child) |
body(p, meta, node->child); |
body(p, &pair, meta, node->child); |
|
|
if (TERMPAIR_FLAG & pair.type) |
if (TERMPAIR_FLAG & pair.type) |
p->flags &= ~pair.flag; |
p->flags &= ~pair.flag; |
Line 447 body(struct termp *p, const struct mdoc_meta *meta, |
|
Line 679 body(struct termp *p, const struct mdoc_meta *meta, |
|
/* Siblings. */ |
/* Siblings. */ |
|
|
if (node->next) |
if (node->next) |
body(p, meta, node->next); |
body(p, ppair, meta, node->next); |
} |
} |
|
|
|
|
Line 456 footer(struct termp *p, const struct mdoc_meta *meta) |
|
Line 688 footer(struct termp *p, const struct mdoc_meta *meta) |
|
{ |
{ |
struct tm *tm; |
struct tm *tm; |
char *buf, *os; |
char *buf, *os; |
size_t sz, osz, ssz, i; |
|
|
|
if (NULL == (buf = malloc(p->rmargin))) |
if (NULL == (buf = malloc(p->rmargin))) |
err(1, "malloc"); |
err(1, "malloc"); |
Line 472 footer(struct termp *p, const struct mdoc_meta *meta) |
|
Line 703 footer(struct termp *p, const struct mdoc_meta *meta) |
|
#endif |
#endif |
err(1, "strftime"); |
err(1, "strftime"); |
|
|
osz = strlcpy(os, meta->os, p->rmargin); |
(void)strlcpy(os, meta->os, p->rmargin); |
|
|
sz = strlen(buf); |
/* |
ssz = sz + osz + 1; |
* This is /slightly/ different from regular groff output |
|
* because we don't have page numbers. Print the following: |
|
* |
|
* OS MDOCDATE |
|
*/ |
|
|
if (ssz > p->rmargin) { |
vspace(p); |
ssz -= p->rmargin; |
|
assert(ssz <= osz); |
|
os[osz - ssz] = 0; |
|
ssz = 1; |
|
} else |
|
ssz = p->rmargin - ssz + 1; |
|
|
|
printf("\n"); |
p->flags |= TERMP_NOSPACE | TERMP_NOBREAK; |
printf("%s", os); |
p->rmargin = p->maxrmargin - strlen(buf); |
for (i = 0; i < ssz; i++) |
p->offset = 0; |
printf(" "); |
|
|
|
printf("%s\n", buf); |
word(p, os); |
fflush(stdout); |
flushln(p); |
|
|
|
p->flags |= TERMP_NOLPAD | TERMP_NOSPACE; |
|
p->offset = p->rmargin; |
|
p->rmargin = p->maxrmargin; |
|
p->flags &= ~TERMP_NOBREAK; |
|
|
|
word(p, buf); |
|
flushln(p); |
|
|
free(buf); |
free(buf); |
free(os); |
free(os); |
} |
} |
|
|
header(struct termp *p, const struct mdoc_meta *meta) |
header(struct termp *p, const struct mdoc_meta *meta) |
{ |
{ |
char *buf, *title; |
char *buf, *title; |
const char *pp, *msec; |
|
size_t ssz, tsz, ttsz, i;; |
|
|
|
|
p->rmargin = p->maxrmargin; |
|
p->offset = 0; |
|
|
if (NULL == (buf = malloc(p->rmargin))) |
if (NULL == (buf = malloc(p->rmargin))) |
err(1, "malloc"); |
err(1, "malloc"); |
if (NULL == (title = malloc(p->rmargin))) |
if (NULL == (title = malloc(p->rmargin))) |
err(1, "malloc"); |
err(1, "malloc"); |
|
|
if (NULL == (pp = mdoc_vol2a(meta->vol))) |
/* |
switch (meta->msec) { |
* The header is strange. It has three components, which are |
case (MSEC_1): |
* 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. |
|
*/ |
|
|
|
assert(meta->vol); |
|
(void)strlcpy(buf, meta->vol, p->rmargin); |
|
|
|
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(%d)", |
|
meta->title, meta->msec); |
|
|
|
p->offset = 0; |
|
p->rmargin = (p->maxrmargin - strlen(buf)) / 2; |
|
p->flags |= TERMP_NOBREAK | TERMP_NOSPACE; |
|
|
|
word(p, title); |
|
flushln(p); |
|
|
|
p->flags |= TERMP_NOLPAD | TERMP_NOSPACE; |
|
p->offset = p->rmargin; |
|
p->rmargin = p->maxrmargin - strlen(title); |
|
|
|
word(p, buf); |
|
flushln(p); |
|
|
|
p->offset = p->rmargin; |
|
p->rmargin = p->maxrmargin; |
|
p->flags &= ~TERMP_NOBREAK; |
|
p->flags |= TERMP_NOLPAD | TERMP_NOSPACE; |
|
|
|
word(p, title); |
|
flushln(p); |
|
|
|
p->rmargin = p->maxrmargin; |
|
p->offset = 0; |
|
p->flags &= ~TERMP_NOSPACE; |
|
|
|
free(title); |
|
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) |
|
{ |
|
const char *rhs; |
|
size_t sz; |
|
|
|
if (NULL == (rhs = a2ascii(p->symtab, word, len, &sz))) |
|
return; |
|
stringa(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 |
|
pescape(struct termp *p, const char *word, int *i, int len) |
|
{ |
|
int j; |
|
|
|
if (++(*i) >= len) |
|
return; |
|
|
|
if ('(' == word[*i]) { |
|
(*i)++; |
|
if (*i + 1 >= len) |
|
return; |
|
|
|
nescape(p, &word[*i], 2); |
|
(*i)++; |
|
return; |
|
|
|
} else if ('*' == word[*i]) { |
|
(*i)++; |
|
if (*i >= len) |
|
return; |
|
|
|
switch (word[*i]) { |
|
case ('('): |
|
(*i)++; |
|
if (*i + 1 >= len) |
|
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]) |
|
return; |
|
|
|
nescape(p, &word[*i - j], (size_t)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, int len) |
|
{ |
|
int i; |
|
|
|
if ( ! (TERMP_NOSPACE & p->flags) && |
|
! (TERMP_LITERAL & p->flags)) |
|
chara(p, ' '); |
|
|
|
if ( ! (p->flags & TERMP_NONOSPACE)) |
|
p->flags &= ~TERMP_NOSPACE; |
|
|
|
/* |
|
* If ANSI (word-length styling), then apply our style now, |
|
* before the word. |
|
*/ |
|
|
|
for (i = 0; i < len; i++) { |
|
if ('\\' == word[i]) { |
|
pescape(p, word, &i, len); |
|
continue; |
|
} |
|
|
|
if (TERMP_STYLE & p->flags) { |
|
if (TERMP_BOLD & p->flags) { |
|
chara(p, word[i]); |
|
chara(p, 8); |
|
} |
|
if (TERMP_UNDER & p->flags) { |
|
chara(p, '_'); |
|
chara(p, 8); |
|
} |
|
} |
|
|
|
chara(p, word[i]); |
|
} |
|
} |
|
|
|
|
|
/* |
|
* 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; |
|
|
|
assert(c); |
|
if (p->col + sz >= p->maxcols) { |
|
if (0 == p->maxcols) |
|
p->maxcols = 256; |
|
s = sz > p->maxcols * 2 ? sz : p->maxcols * 2; |
|
p->buf = realloc(p->buf, s); |
|
if (NULL == p->buf) |
|
err(1, "realloc"); |
|
p->maxcols = s; |
|
} |
|
|
|
(void)memcpy(&p->buf[(int)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) |
|
{ |
|
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; |
|
} |
|
|
|
|
|
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 */ |
/* FALLTHROUGH */ |
case (MSEC_6): |
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 */ |
/* FALLTHROUGH */ |
case (MSEC_7): |
case (MDOC_BODY): |
pp = mdoc_vol2a(VOL_URM); |
/* FALLTHROUGH */ |
|
case (MDOC_HEAD): |
break; |
break; |
case (MSEC_8): |
default: |
pp = mdoc_vol2a(VOL_SMM); |
errx(1, "regular form violated (6)"); |
|
/* NOTREACHED */ |
|
} |
|
if (n->child) switch (n->child->type) { |
|
case (MDOC_TEXT): |
break; |
break; |
case (MSEC_2): |
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 */ |
/* FALLTHROUGH */ |
case (MSEC_3): |
case (MDOC_ELEM): |
/* FALLTHROUGH */ |
/* FALLTHROUGH */ |
case (MSEC_4): |
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 */ |
/* FALLTHROUGH */ |
case (MSEC_5): |
case (MDOC_HEAD): |
pp = mdoc_vol2a(VOL_PRM); |
/* FALLTHROUGH */ |
|
case (MDOC_BODY): |
|
/* FALLTHROUGH */ |
|
case (MDOC_TAIL): |
break; |
break; |
case (MSEC_9): |
|
pp = mdoc_vol2a(VOL_KM); |
|
break; |
|
default: |
default: |
/* FIXME: capitalise. */ |
errx(1, "regular form violated (d)"); |
if (NULL == (pp = mdoc_msec2a(meta->msec))) |
/* NOTREACHED */ |
pp = mdoc_msec2a(MSEC_local); |
} |
|
switch (n->child->type) { |
|
case (MDOC_ROOT): |
|
/* FALLTHROUGH */ |
|
case (MDOC_ELEM): |
|
errx(1, "regular form violated (e)"); |
|
/* NOTREACHED */ |
|
default: |
break; |
break; |
} |
} |
assert(pp); |
break; |
|
case (MDOC_ROOT): |
tsz = strlcpy(buf, pp, p->rmargin); |
if (n->parent) |
assert(tsz < p->rmargin); |
errx(1, "regular form violated (f)"); |
|
if (NULL == n->child) |
if ((pp = mdoc_arch2a(meta->arch))) { |
errx(1, "regular form violated (10)"); |
tsz = strlcat(buf, " (", p->rmargin); |
switch (n->child->type) { |
assert(tsz < p->rmargin); |
case (MDOC_BLOCK): |
tsz = strlcat(buf, pp, p->rmargin); |
break; |
assert(tsz < p->rmargin); |
default: |
tsz = strlcat(buf, ")", p->rmargin); |
errx(1, "regular form violated (11)"); |
assert(tsz < p->rmargin); |
/* NOTREACHED */ |
|
} |
|
break; |
} |
} |
|
} |
|
|
ttsz = strlcpy(title, meta->title, p->rmargin); |
|
|
|
if (NULL == (msec = mdoc_msec2a(meta->msec))) |
static int |
msec = ""; |
merr(void *arg, int line, int col, const char *msg) |
|
{ |
|
|
ssz = (2 * (ttsz + 2 + strlen(msec))) + tsz + 2; |
warnx("error: %s (line %d, column %d)", msg, line, col); |
|
return(0); |
|
} |
|
|
if (ssz > p->rmargin) { |
|
if ((ssz -= p->rmargin) % 2) |
|
ssz++; |
|
ssz /= 2; |
|
|
|
assert(ssz <= ttsz); |
|
title[ttsz - ssz] = 0; |
|
ssz = 1; |
|
} else |
|
ssz = ((p->rmargin - ssz) / 2) + 1; |
|
|
|
printf("%s(%s)", title, msec); |
static int |
|
mwarn(void *arg, int line, int col, |
|
enum mdoc_warn type, const char *msg) |
|
{ |
|
int flags; |
|
char *wtype; |
|
|
for (i = 0; i < ssz; i++) |
flags = *(int *)arg; |
printf(" "); |
wtype = NULL; |
|
|
printf("%s", buf); |
switch (type) { |
|
case (WARN_COMPAT): |
|
wtype = "compat"; |
|
if (flags & WARN_WCOMPAT) |
|
break; |
|
return(1); |
|
case (WARN_SYNTAX): |
|
wtype = "syntax"; |
|
if (flags & WARN_WSYNTAX) |
|
break; |
|
return(1); |
|
} |
|
|
for (i = 0; i < ssz; i++) |
assert(wtype); |
printf(" "); |
warnx("%s warning: %s (line %d, column %d)", |
|
wtype, msg, line, col); |
|
|
printf("%s(%s)\n", title, msec); |
if ( ! (flags & WARN_WERR)) |
fflush(stdout); |
return(1); |
|
|
free(title); |
warnx("%s: considering warnings as errors", |
free(buf); |
__progname); |
|
return(0); |
} |
} |
|
|
|
|