-/* $Id: roff.c,v 1.193 2014/02/14 23:05:20 schwarze Exp $ */
+/* $Id: roff.c,v 1.202 2014/03/23 12:11:18 schwarze Exp $ */
/*
* Copyright (c) 2010, 2011, 2012 Kristaps Dzonsons <kristaps@bsd.lv>
* Copyright (c) 2010-2014 Ingo Schwarze <schwarze@openbsd.org>
#include <string.h>
#include "mandoc.h"
+#include "mandoc_aux.h"
#include "libroff.h"
#include "libmandoc.h"
ROFF_am1,
ROFF_as,
ROFF_cc,
+ ROFF_ce,
ROFF_de,
ROFF_dei,
ROFF_de1,
ROFF_EQ,
ROFF_EN,
ROFF_cblock,
- ROFF_ccond,
ROFF_USERDEF,
ROFF_MAX
};
-enum roffrule {
- ROFFRULE_DENY,
- ROFFRULE_ALLOW
-};
-
/*
* An incredibly-simple string buffer.
*/
};
struct roff {
- enum mparset parsetype; /* requested parse type */
struct mparse *parse; /* parse point */
- int quick; /* skip standard macro deletion */
+ int options; /* parse options */
struct roffnode *last; /* leaf of stack */
- enum roffrule rstack[RSTACK_MAX]; /* stack of !`ie' rules */
+ int rstack[RSTACK_MAX]; /* stack of !`ie' rules */
char control; /* control character */
int rstackpos; /* position in rstack */
struct roffreg *regtab; /* number registers */
char *name; /* node name, e.g. macro name */
char *end; /* end-rules: custom token */
int endspan; /* end-rules: next-line or infty */
- enum roffrule rule; /* current evaluation rule */
+ int rule; /* current evaluation rule */
};
#define ROFF_ARGS struct roff *r, /* parse ctx */ \
static enum rofferr roff_block_sub(ROFF_ARGS);
static enum rofferr roff_cblock(ROFF_ARGS);
static enum rofferr roff_cc(ROFF_ARGS);
-static enum rofferr roff_ccond(ROFF_ARGS);
+static void roff_ccond(struct roff *, int, int);
static enum rofferr roff_cond(ROFF_ARGS);
static enum rofferr roff_cond_text(ROFF_ARGS);
static enum rofferr roff_cond_sub(ROFF_ARGS);
static enum rofferr roff_ds(ROFF_ARGS);
-static enum roffrule roff_evalcond(const char *, int *);
+static int roff_evalcond(const char *, int *);
+static int roff_evalstrcond(const char *, int *);
static void roff_free1(struct roff *);
static void roff_freereg(struct roffreg *);
static void roff_freestr(struct roffkv *);
{ "am1", roff_block, roff_block_text, roff_block_sub, 0, NULL },
{ "as", roff_ds, NULL, NULL, 0, NULL },
{ "cc", roff_cc, NULL, NULL, 0, NULL },
+ { "ce", roff_line_ignore, NULL, NULL, 0, NULL },
{ "de", roff_block, roff_block_text, roff_block_sub, 0, NULL },
{ "dei", roff_block, roff_block_text, roff_block_sub, 0, NULL },
{ "de1", roff_block, roff_block_text, roff_block_sub, 0, NULL },
{ "EQ", roff_EQ, NULL, NULL, 0, NULL },
{ "EN", roff_EN, NULL, NULL, 0, NULL },
{ ".", roff_cblock, NULL, NULL, 0, NULL },
- { "\\}", roff_ccond, NULL, NULL, 0, NULL },
{ NULL, roff_userdef, NULL, NULL, 0, NULL },
};
+/* not currently implemented: Ds em Eq LP Me PP pp Or Rd Sf SH */
const char *const __mdoc_reserved[] = {
"Ac", "Ad", "An", "Ao", "Ap", "Aq", "Ar", "At",
"Bc", "Bd", "Bf", "Bk", "Bl", "Bo", "Bq",
"Brc", "Bro", "Brq", "Bsx", "Bt", "Bx",
"Cd", "Cm", "Db", "Dc", "Dd", "Dl", "Do", "Dq",
- "Ds", "Dt", "Dv", "Dx", "D1",
- "Ec", "Ed", "Ef", "Ek", "El", "Em", "em",
- "En", "Eo", "Eq", "Er", "Es", "Ev", "Ex",
+ "Dt", "Dv", "Dx", "D1",
+ "Ec", "Ed", "Ef", "Ek", "El", "Em",
+ "En", "Eo", "Er", "Es", "Ev", "Ex",
"Fa", "Fc", "Fd", "Fl", "Fn", "Fo", "Fr", "Ft", "Fx",
- "Hf", "Ic", "In", "It", "Lb", "Li", "Lk", "Lp", "LP",
- "Me", "Ms", "Mt", "Nd", "Nm", "No", "Ns", "Nx",
+ "Hf", "Ic", "In", "It", "Lb", "Li", "Lk", "Lp",
+ "Ms", "Mt", "Nd", "Nm", "No", "Ns", "Nx",
"Oc", "Oo", "Op", "Os", "Ot", "Ox",
- "Pa", "Pc", "Pf", "Po", "Pp", "PP", "pp", "Pq",
- "Qc", "Ql", "Qo", "Qq", "Or", "Rd", "Re", "Rs", "Rv",
- "Sc", "Sf", "Sh", "SH", "Sm", "So", "Sq",
+ "Pa", "Pc", "Pf", "Po", "Pp", "Pq",
+ "Qc", "Ql", "Qo", "Qq", "Re", "Rs", "Rv",
+ "Sc", "Sh", "Sm", "So", "Sq",
"Ss", "St", "Sx", "Sy",
"Ta", "Tn", "Ud", "Ux", "Va", "Vt", "Xc", "Xo", "Xr",
- "%A", "%B", "%D", "%I", "%J", "%N", "%O",
+ "%A", "%B", "%C", "%D", "%I", "%J", "%N", "%O",
"%P", "%Q", "%R", "%T", "%U", "%V",
NULL
};
+/* not currently implemented: BT DE DS ME MT PT SY TQ YS */
const char *const __man_reserved[] = {
- "AT", "B", "BI", "BR", "BT", "DE", "DS", "DT",
- "EE", "EN", "EQ", "EX", "HF", "HP", "I", "IB", "IP", "IR",
- "LP", "ME", "MT", "OP", "P", "PD", "PP", "PT",
- "R", "RB", "RE", "RI", "RS", "SB", "SH", "SM", "SS", "SY",
- "TE", "TH", "TP", "TQ", "TS", "T&", "UC", "UE", "UR", "YS",
+ "AT", "B", "BI", "BR", "DT",
+ "EE", "EN", "EQ", "EX", "HP", "I", "IB", "IP", "IR",
+ "LP", "OP", "P", "PD", "PP",
+ "R", "RB", "RE", "RI", "RS", "SB", "SH", "SM", "SS",
+ "TE", "TH", "TP", "TS", "T&", "UC", "UE", "UR",
NULL
};
p->parent = r->last;
p->line = line;
p->col = col;
- p->rule = p->parent ? p->parent->rule : ROFFRULE_DENY;
+ p->rule = p->parent ? p->parent->rule : 0;
r->last = p;
}
struct roff *
-roff_alloc(enum mparset type, struct mparse *parse, int quick)
+roff_alloc(struct mparse *parse, int options)
{
struct roff *r;
r = mandoc_calloc(1, sizeof(struct roff));
- r->parsetype = type;
r->parse = parse;
- r->quick = quick;
+ r->options = options;
r->rstackpos = -1;
roffhash_init();
/* Spring the input line trap. */
if (1 == roffit_lines) {
- isz = asprintf(&p, "%s\n.%s", *bufp, roffit_macro);
- if (-1 == isz) {
- perror(NULL);
- exit((int)MANDOCLEVEL_SYSERR);
- }
+ isz = mandoc_asprintf(&p, "%s\n.%s", *bufp, roffit_macro);
free(*bufp);
*bufp = p;
*szp = isz + 1;
'\t' == buf[*pos] || ' ' == buf[*pos])
return(ROFF_MAX);
- /*
- * We stop the macro parse at an escape, tab, space, or nil.
- * However, `\}' is also a valid macro, so make sure we don't
- * clobber it by seeing the `\' as the end of token.
- */
+ /* We stop the macro parse at an escape, tab, space, or nil. */
mac = buf + *pos;
- maclen = strcspn(mac + 1, " \\\t\0") + 1;
+ maclen = strcspn(mac, " \\\t\0");
t = (r->current_string = roff_getstrn(r, mac, maclen))
? ROFF_USERDEF : roffhash_find(mac, maclen);
}
-/* ARGSUSED */
-static enum rofferr
-roff_ccond(ROFF_ARGS)
+static void
+roff_ccond(struct roff *r, int ln, int ppos)
{
if (NULL == r->last) {
mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
- return(ROFF_IGN);
+ return;
}
switch (r->last->tok) {
break;
default:
mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
- return(ROFF_IGN);
+ return;
}
if (r->last->endspan > -1) {
mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
- return(ROFF_IGN);
+ return;
}
- if ((*bufp)[pos])
- mandoc_msg(MANDOCERR_ARGSLOST, r->parse, ln, pos, NULL);
-
roffnode_pop(r);
roffnode_cleanscope(r);
- return(ROFF_IGN);
+ return;
}
roff_cond_sub(ROFF_ARGS)
{
enum rofft t;
- enum roffrule rr;
char *ep;
+ int rr;
rr = r->last->rule;
roffnode_cleanscope(r);
*/
if ((ROFF_MAX != t) &&
- (ROFF_ccond == t || ROFFRULE_ALLOW == rr ||
- ROFFMAC_STRUCT & roffs[t].flags)) {
+ (rr || ROFFMAC_STRUCT & roffs[t].flags)) {
assert(roffs[t].proc);
return((*roffs[t].proc)(r, t, bufp, szp,
ln, ppos, pos, offs));
}
- /* Always check for the closing delimiter `\}'. */
+ /*
+ * If `\}' occurs on a macro line without a preceding macro,
+ * drop the line completely.
+ */
- ep = &(*bufp)[pos];
- while (NULL != (ep = strchr(ep, '\\'))) {
- if ('}' != *(++ep))
- continue;
+ ep = *bufp + pos;
+ if ('\\' == ep[0] && '}' == ep[1])
+ rr = 0;
- /*
- * If we're at the end of line, then just chop
- * off the \} and resize the buffer.
- * If we aren't, then convert it to spaces.
- */
-
- if ('\0' == *(ep + 1)) {
- *--ep = '\0';
- *szp -= 2;
- } else
- *(ep - 1) = *ep = ' ';
+ /* Always check for the closing delimiter `\}'. */
- roff_ccond(r, ROFF_ccond, bufp, szp,
- ln, pos, pos + 2, offs);
- break;
+ while (NULL != (ep = strchr(ep, '\\'))) {
+ if ('}' == *(++ep)) {
+ *ep = '&';
+ roff_ccond(r, ln, ep - *bufp - 1);
+ }
+ ++ep;
}
- return(ROFFRULE_DENY == rr ? ROFF_IGN : ROFF_CONT);
+ return(rr ? ROFF_CONT : ROFF_IGN);
}
/* ARGSUSED */
roff_cond_text(ROFF_ARGS)
{
char *ep;
- enum roffrule rr;
+ int rr;
rr = r->last->rule;
roffnode_cleanscope(r);
- ep = &(*bufp)[pos];
- for ( ; NULL != (ep = strchr(ep, '\\')); ep++) {
- ep++;
- if ('}' != *ep)
- continue;
- *ep = '&';
- roff_ccond(r, ROFF_ccond, bufp, szp,
- ln, pos, pos + 2, offs);
+ ep = *bufp + pos;
+ while (NULL != (ep = strchr(ep, '\\'))) {
+ if ('}' == *(++ep)) {
+ *ep = '&';
+ roff_ccond(r, ln, ep - *bufp - 1);
+ }
+ ++ep;
}
- return(ROFFRULE_DENY == rr ? ROFF_IGN : ROFF_CONT);
+ return(rr ? ROFF_CONT : ROFF_IGN);
}
static int
return(*res);
}
-static enum roffrule
+/*
+ * Evaluate a string comparison condition.
+ * The first character is the delimiter.
+ * Succeed if the string up to its second occurrence
+ * matches the string up to its third occurence.
+ * Advance the cursor after the third occurrence
+ * or lacking that, to the end of the line.
+ */
+static int
+roff_evalstrcond(const char *v, int *pos)
+{
+ const char *s1, *s2, *s3;
+ int match;
+
+ match = 0;
+ s1 = v + *pos; /* initial delimiter */
+ s2 = s1 + 1; /* for scanning the first string */
+ s3 = strchr(s2, *s1); /* for scanning the second string */
+
+ if (NULL == s3) /* found no middle delimiter */
+ goto out;
+
+ while ('\0' != *++s3) {
+ if (*s2 != *s3) { /* mismatch */
+ s3 = strchr(s3, *s1);
+ break;
+ }
+ if (*s3 == *s1) { /* found the final delimiter */
+ match = 1;
+ break;
+ }
+ s2++;
+ }
+
+out:
+ if (NULL == s3)
+ s3 = strchr(s2, '\0');
+ else
+ s3++;
+ *pos = s3 - v;
+ return(match);
+}
+
+static int
roff_evalcond(const char *v, int *pos)
{
- int not, lh, rh;
+ int wanttrue, lh, rh;
char op;
+ if ('!' == v[*pos]) {
+ wanttrue = 0;
+ (*pos)++;
+ } else
+ wanttrue = 1;
+
switch (v[*pos]) {
case ('n'):
+ /* FALLTHROUGH */
+ case ('o'):
(*pos)++;
- return(ROFFRULE_ALLOW);
+ return(wanttrue);
+ case ('c'):
+ /* FALLTHROUGH */
+ case ('d'):
+ /* FALLTHROUGH */
case ('e'):
/* FALLTHROUGH */
- case ('o'):
+ case ('r'):
/* FALLTHROUGH */
case ('t'):
(*pos)++;
- return(ROFFRULE_DENY);
- case ('!'):
- (*pos)++;
- not = 1;
- break;
+ return(!wanttrue);
default:
- not = 0;
break;
}
if (!roff_getnum(v, pos, &lh))
- return ROFFRULE_DENY;
- if (!roff_getop(v, pos, &op)) {
- if (lh < 0)
- lh = 0;
- goto out;
- }
+ return(roff_evalstrcond(v, pos) == wanttrue);
+ if (!roff_getop(v, pos, &op))
+ return((lh > 0) == wanttrue);
if (!roff_getnum(v, pos, &rh))
- return ROFFRULE_DENY;
+ return(0);
+
switch (op) {
case 'g':
- lh = lh >= rh;
- break;
+ return((lh >= rh) == wanttrue);
case 'l':
- lh = lh <= rh;
- break;
+ return((lh <= rh) == wanttrue);
case '=':
- lh = lh == rh;
- break;
+ return((lh == rh) == wanttrue);
case '>':
- lh = lh > rh;
- break;
+ return((lh > rh) == wanttrue);
case '<':
- lh = lh < rh;
- break;
+ return((lh < rh) == wanttrue);
default:
- return ROFFRULE_DENY;
+ return(0);
}
-out:
- if (not)
- lh = !lh;
- return lh ? ROFFRULE_ALLOW : ROFFRULE_DENY;
}
/* ARGSUSED */
*/
r->last->rule = ROFF_el == tok ?
- (r->rstackpos < 0 ?
- ROFFRULE_DENY : r->rstack[r->rstackpos--]) :
+ (r->rstackpos < 0 ? 0 : r->rstack[r->rstackpos--]) :
roff_evalcond(*bufp, &pos);
/*
r->parse, ln, ppos, NULL);
return(ROFF_ERR);
}
- r->rstack[++r->rstackpos] =
- ROFFRULE_DENY == r->last->rule ?
- ROFFRULE_ALLOW : ROFFRULE_DENY;
+ r->rstack[++r->rstackpos] = !r->last->rule;
}
/* If the parent has false as its rule, then so do we. */
- if (r->last->parent && ROFFRULE_DENY == r->last->parent->rule)
- r->last->rule = ROFFRULE_DENY;
+ if (r->last->parent && !r->last->parent->rule)
+ r->last->rule = 0;
/*
* Determine scope.
{
const char *const *cp;
- if (0 == r->quick && MPARSE_MDOC != r->parsetype)
+ if (0 == ((MPARSE_MDOC | MPARSE_QUICK) & r->options))
for (cp = __mdoc_reserved; *cp; cp++)
roff_setstr(r, *cp, NULL, 0);
{
const char *const *cp;
- if (0 == r->quick && MPARSE_MDOC != r->parsetype)
+ if (0 == (MPARSE_QUICK & r->options))
for (cp = __man_reserved; *cp; cp++)
roff_setstr(r, *cp, NULL, 0);