tagging issues from weerd@ regarding hyphens

[mandoc.git] / man_macro.c

diff --git a/man_macro.c b/man_macro.c
index 4bb02356f98fa2d703dfba012b1325297baf08e1..b3c3a3cb488b273bca0d81bc65dca51bfc1c6bbe 100644 (file)
--- a/man_macro.c
+++ b/man_macro.c
@@ -1,7 +1,7 @@
-/*     $Id: man_macro.c,v 1.125 2018/08/15 15:37:25 schwarze Exp $ */
+/* $Id: man_macro.c,v 1.145 2020/09/09 17:01:10 schwarze Exp $ */

 /*
  * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
 /*
  * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>

- * Copyright (c) 2012-2015, 2017, 2018 Ingo Schwarze <schwarze@openbsd.org>
+ * Copyright (c) 2012-2015, 2017-2020 Ingo Schwarze <schwarze@openbsd.org>

  * Copyright (c) 2013 Franco Fichtner <franco@lastsummer.de>
  *
  * Permission to use, copy, modify, and distribute this software for any
  * Copyright (c) 2013 Franco Fichtner <franco@lastsummer.de>
  *
  * Permission to use, copy, modify, and distribute this software for any


@@ -22,6 +22,7 @@
 
 #include <assert.h>
 #include <ctype.h>
 
 #include <assert.h>
 #include <ctype.h>

+#include <stdio.h>

 #include <stdlib.h>
 #include <string.h>
 
 #include <stdlib.h>
 #include <string.h>
 


@@ -40,47 +41,54 @@ static      int              man_args(struct roff_man *, int,
                                int *, char *, char **);
 static void             rew_scope(struct roff_man *, enum roff_tok);
 
                                int *, char *, char **);
 static void             rew_scope(struct roff_man *, enum roff_tok);
 

-const  struct man_macro __man_macros[MAN_MAX - MAN_TH] = {
-       { in_line_eoln, MAN_BSCOPE }, /* TH */
-       { blk_imp, MAN_BSCOPE | MAN_SCOPED }, /* SH */
-       { blk_imp, MAN_BSCOPE | MAN_SCOPED }, /* SS */
-       { blk_imp, MAN_BSCOPE | MAN_SCOPED }, /* TP */
-       { blk_imp, MAN_BSCOPE }, /* LP */
-       { blk_imp, MAN_BSCOPE }, /* PP */
-       { blk_imp, MAN_BSCOPE }, /* P */
-       { blk_imp, MAN_BSCOPE }, /* IP */
-       { blk_imp, MAN_BSCOPE }, /* HP */
-       { in_line_eoln, MAN_SCOPED | MAN_JOIN }, /* SM */
-       { in_line_eoln, MAN_SCOPED | MAN_JOIN }, /* SB */
+static const struct man_macro man_macros[MAN_MAX - MAN_TH] = {
+       { in_line_eoln, MAN_XSCOPE }, /* TH */
+       { blk_imp, MAN_XSCOPE | MAN_BSCOPED }, /* SH */
+       { blk_imp, MAN_XSCOPE | MAN_BSCOPED }, /* SS */
+       { blk_imp, MAN_XSCOPE | MAN_BSCOPED }, /* TP */
+       { blk_imp, MAN_XSCOPE | MAN_BSCOPED }, /* TQ */
+       { blk_imp, MAN_XSCOPE }, /* LP */
+       { blk_imp, MAN_XSCOPE }, /* PP */
+       { blk_imp, MAN_XSCOPE }, /* P */
+       { blk_imp, MAN_XSCOPE }, /* IP */
+       { blk_imp, MAN_XSCOPE }, /* HP */
+       { in_line_eoln, MAN_NSCOPED | MAN_ESCOPED | MAN_JOIN }, /* SM */
+       { in_line_eoln, MAN_NSCOPED | MAN_ESCOPED | MAN_JOIN }, /* SB */

        { in_line_eoln, 0 }, /* BI */
        { in_line_eoln, 0 }, /* IB */
        { in_line_eoln, 0 }, /* BR */
        { in_line_eoln, 0 }, /* RB */
        { in_line_eoln, 0 }, /* BI */
        { in_line_eoln, 0 }, /* IB */
        { in_line_eoln, 0 }, /* BR */
        { in_line_eoln, 0 }, /* RB */

-       { in_line_eoln, MAN_SCOPED | MAN_JOIN }, /* R */
-       { in_line_eoln, MAN_SCOPED | MAN_JOIN }, /* B */
-       { in_line_eoln, MAN_SCOPED | MAN_JOIN }, /* I */
+       { in_line_eoln, MAN_NSCOPED | MAN_ESCOPED | MAN_JOIN }, /* R */
+       { in_line_eoln, MAN_NSCOPED | MAN_ESCOPED | MAN_JOIN }, /* B */
+       { in_line_eoln, MAN_NSCOPED | MAN_ESCOPED | MAN_JOIN }, /* I */

        { in_line_eoln, 0 }, /* IR */
        { in_line_eoln, 0 }, /* RI */
        { in_line_eoln, 0 }, /* IR */
        { in_line_eoln, 0 }, /* RI */

-       { in_line_eoln, MAN_NSCOPED }, /* nf */
-       { in_line_eoln, MAN_NSCOPED }, /* fi */
-       { blk_close, MAN_BSCOPE }, /* RE */
-       { blk_exp, MAN_BSCOPE }, /* RS */
+       { blk_close, MAN_XSCOPE }, /* RE */
+       { blk_exp, MAN_XSCOPE }, /* RS */

        { in_line_eoln, 0 }, /* DT */
        { in_line_eoln, 0 }, /* UC */
        { in_line_eoln, MAN_NSCOPED }, /* PD */
        { in_line_eoln, 0 }, /* AT */
        { in_line_eoln, MAN_NSCOPED }, /* in */
        { in_line_eoln, 0 }, /* DT */
        { in_line_eoln, 0 }, /* UC */
        { in_line_eoln, MAN_NSCOPED }, /* PD */
        { in_line_eoln, 0 }, /* AT */
        { in_line_eoln, MAN_NSCOPED }, /* in */

+       { blk_imp, MAN_XSCOPE }, /* SY */
+       { blk_close, MAN_XSCOPE }, /* YS */

        { in_line_eoln, 0 }, /* OP */
        { in_line_eoln, 0 }, /* OP */

-       { in_line_eoln, MAN_BSCOPE }, /* EX */
-       { in_line_eoln, MAN_BSCOPE }, /* EE */
-       { blk_exp, MAN_BSCOPE }, /* UR */
-       { blk_close, MAN_BSCOPE }, /* UE */
-       { blk_exp, MAN_BSCOPE }, /* MT */
-       { blk_close, MAN_BSCOPE }, /* ME */
+       { in_line_eoln, MAN_XSCOPE }, /* EX */
+       { in_line_eoln, MAN_XSCOPE }, /* EE */
+       { blk_exp, MAN_XSCOPE }, /* UR */
+       { blk_close, MAN_XSCOPE }, /* UE */
+       { blk_exp, MAN_XSCOPE }, /* MT */
+       { blk_close, MAN_XSCOPE }, /* ME */

 };
 };

-const  struct man_macro *const man_macros = __man_macros - MAN_TH;

 
 
 
 

+const struct man_macro *
+man_macro(enum roff_tok tok)
+{
+       assert(tok >= MAN_TH && tok <= MAN_MAX);
+       return man_macros + (tok - MAN_TH);
+}
+

 void
 man_unscope(struct roff_man *man, const struct roff_node *to)
 {
 void
 man_unscope(struct roff_man *man, const struct roff_node *to)
 {


@@ -94,13 +102,16 @@ man_unscope(struct roff_man *man, const struct roff_node *to)
 
                if (to == NULL && ! (n->flags & NODE_VALID)) {
                        if (man->flags & (MAN_BLINE | MAN_ELINE) &&
 
                if (to == NULL && ! (n->flags & NODE_VALID)) {
                        if (man->flags & (MAN_BLINE | MAN_ELINE) &&

-                           man_macros[n->tok].flags & MAN_SCOPED) {
-                               mandoc_vmsg(MANDOCERR_BLK_LINE,
-                                   man->parse, n->line, n->pos,
+                           man_macro(n->tok)->flags &
+                            (MAN_BSCOPED | MAN_NSCOPED)) {
+                               mandoc_msg(MANDOCERR_BLK_LINE,
+                                   n->line, n->pos,

                                    "EOF breaks %s", roff_name[n->tok]);
                                    "EOF breaks %s", roff_name[n->tok]);

-                               if (man->flags & MAN_ELINE)
-                                       man->flags &= ~MAN_ELINE;
-                               else {
+                               if (man->flags & MAN_ELINE) {
+                                       if ((man_macro(n->parent->tok)->flags &
+                                           MAN_ESCOPED) == 0)
+                                               man->flags &= ~MAN_ELINE;
+                               } else {

                                        assert(n->type == ROFFT_HEAD);
                                        n = n->parent;
                                        man->flags &= ~MAN_BLINE;
                                        assert(n->type == ROFFT_HEAD);
                                        n = n->parent;
                                        man->flags &= ~MAN_BLINE;


@@ -111,9 +122,9 @@ man_unscope(struct roff_man *man, const struct roff_node *to)
                                continue;
                        }
                        if (n->type == ROFFT_BLOCK &&
                                continue;
                        }
                        if (n->type == ROFFT_BLOCK &&

-                           man_macros[n->tok].fp == blk_exp)
+                           man_macro(n->tok)->fp == blk_exp)

                                mandoc_msg(MANDOCERR_BLK_NOEND,
                                mandoc_msg(MANDOCERR_BLK_NOEND,

-                                   man->parse, n->line, n->pos,
+                                   n->line, n->pos, "%s",

                                    roff_name[n->tok]);
                }
 
                                    roff_name[n->tok]);
                }
 


@@ -175,7 +186,7 @@ rew_scope(struct roff_man *man, enum roff_tok tok)
                }
                if (tok != MAN_SH && (n->tok == MAN_SH ||
                    (tok != MAN_SS && (n->tok == MAN_SS ||
                }
                if (tok != MAN_SH && (n->tok == MAN_SH ||
                    (tok != MAN_SS && (n->tok == MAN_SS ||

-                    man_macros[n->tok].fp == blk_exp))))
+                    man_macro(n->tok)->fp == blk_exp))))

                        return;
                man_unscope(man, n);
                n = man->last;
                        return;
                man_unscope(man, n);
                n = man->last;


@@ -191,31 +202,37 @@ blk_close(MACRO_PROT_ARGS)
 {
        enum roff_tok            ctok, ntok;
        const struct roff_node  *nn;
 {
        enum roff_tok            ctok, ntok;
        const struct roff_node  *nn;

-       char                    *p;
-       int                      cline, cpos, nrew, target;
+       char                    *p, *ep;
+       int                      cline, cpos, la, nrew, target;

 
        nrew = 1;
        switch (tok) {
        case MAN_RE:
                ntok = MAN_RS;
 
        nrew = 1;
        switch (tok) {
        case MAN_RE:
                ntok = MAN_RS;

+               la = *pos;

                if ( ! man_args(man, line, pos, buf, &p))
                        break;
                for (nn = man->last->parent; nn; nn = nn->parent)
                        if (nn->tok == ntok && nn->type == ROFFT_BLOCK)
                                nrew++;
                if ( ! man_args(man, line, pos, buf, &p))
                        break;
                for (nn = man->last->parent; nn; nn = nn->parent)
                        if (nn->tok == ntok && nn->type == ROFFT_BLOCK)
                                nrew++;

-               target = strtol(p, &p, 10);
-               if (*p != '\0')
-                       mandoc_vmsg(MANDOCERR_ARG_EXCESS, man->parse,
-                           line, p - buf, "RE ... %s", p);
+               target = strtol(p, &ep, 10);
+               if (*ep != '\0')
+                       mandoc_msg(MANDOCERR_ARG_EXCESS, line,
+                           la + (buf[la] == '"') + (int)(ep - p),
+                           "RE ... %s", ep);
+               free(p);

                if (target == 0)
                        target = 1;
                nrew -= target;
                if (nrew < 1) {
                if (target == 0)
                        target = 1;
                nrew -= target;
                if (nrew < 1) {

-                       mandoc_vmsg(MANDOCERR_RE_NOTOPEN, man->parse,
+                       mandoc_msg(MANDOCERR_RE_NOTOPEN,

                            line, ppos, "RE %d", target);
                        return;
                }
                break;
                            line, ppos, "RE %d", target);
                        return;
                }
                break;

+       case MAN_YS:
+               ntok = MAN_SY;
+               break;

        case MAN_UE:
                ntok = MAN_UR;
                break;
        case MAN_UE:
                ntok = MAN_UR;
                break;


@@ -231,9 +248,15 @@ blk_close(MACRO_PROT_ARGS)
                        break;
 
        if (nn == NULL) {
                        break;
 
        if (nn == NULL) {

-               mandoc_msg(MANDOCERR_BLK_NOTOPEN, man->parse,
-                   line, ppos, roff_name[tok]);
+               mandoc_msg(MANDOCERR_BLK_NOTOPEN,
+                   line, ppos, "%s", roff_name[tok]);

                rew_scope(man, MAN_PP);
                rew_scope(man, MAN_PP);

+               if (tok == MAN_RE) {
+                       roff_elem_alloc(man, line, ppos, ROFF_br);
+                       man->last->flags |= NODE_LINE |
+                           NODE_VALID | NODE_ENDED;
+                       man->next = ROFF_NEXT_SIBLING;
+               }

                return;
        }
 
                return;
        }
 


@@ -250,6 +273,8 @@ blk_close(MACRO_PROT_ARGS)
        if (buf[*pos] != '\0') {
                roff_word_alloc(man, line, ppos, buf + *pos);
                man->last->flags |= NODE_DELIMC;
        if (buf[*pos] != '\0') {
                roff_word_alloc(man, line, ppos, buf + *pos);
                man->last->flags |= NODE_DELIMC;

+               if (mandoc_eos(man->last->string, strlen(man->last->string)))
+                       man->last->flags |= NODE_EOS;

        }
 
        /* Move a trailing paragraph behind the block. */
        }
 
        /* Move a trailing paragraph behind the block. */


@@ -258,6 +283,13 @@ blk_close(MACRO_PROT_ARGS)
                *pos = strlen(buf);
                blk_imp(man, ctok, cline, cpos, pos, buf);
        }
                *pos = strlen(buf);
                blk_imp(man, ctok, cline, cpos, pos, buf);
        }

+
+       /* Synopsis blocks need an explicit end marker for spacing. */
+
+       if (tok == MAN_YS && man->last == nn) {
+               roff_elem_alloc(man, line, ppos, tok);
+               man_unscope(man, man->last);
+       }

 }
 
 void
 }
 
 void


@@ -267,7 +299,10 @@ blk_exp(MACRO_PROT_ARGS)
        char            *p;
        int              la;
 
        char            *p;
        int              la;
 

-       rew_scope(man, tok);
+       if (tok == MAN_RS) {
+               rew_scope(man, tok);
+               man->flags |= ROFF_NONOFILL;
+       }

        roff_block_alloc(man, line, ppos, tok);
        head = roff_head_alloc(man, line, ppos, tok);
 
        roff_block_alloc(man, line, ppos, tok);
        head = roff_head_alloc(man, line, ppos, tok);
 


@@ -282,14 +317,16 @@ blk_exp(MACRO_PROT_ARGS)
                                roff_setreg(man->roff, "an-margin",
                                    head->aux, '+');
                }
                                roff_setreg(man->roff, "an-margin",
                                    head->aux, '+');
                }

+               free(p);

        }
 
        if (buf[*pos] != '\0')
        }
 
        if (buf[*pos] != '\0')

-               mandoc_vmsg(MANDOCERR_ARG_EXCESS, man->parse, line,
-                   *pos, "%s ... %s", roff_name[tok], buf + *pos);
+               mandoc_msg(MANDOCERR_ARG_EXCESS, line, *pos,
+                   "%s ... %s", roff_name[tok], buf + *pos);

 
        man_unscope(man, head);
        roff_body_alloc(man, line, ppos, tok);
 
        man_unscope(man, head);
        roff_body_alloc(man, line, ppos, tok);

+       man->flags &= ~ROFF_NONOFILL;

 }
 
 /*
 }
 
 /*


@@ -306,9 +343,10 @@ blk_imp(MACRO_PROT_ARGS)
        struct roff_node *n;
 
        rew_scope(man, tok);
        struct roff_node *n;
 
        rew_scope(man, tok);

-       n = roff_block_alloc(man, line, ppos, tok);
-       if (n->tok == MAN_SH || n->tok == MAN_SS)
-               man->flags &= ~MAN_LITERAL;
+       man->flags |= ROFF_NONOFILL;
+       if (tok == MAN_SH || tok == MAN_SS)
+               man->flags &= ~ROFF_NOFILL;
+       roff_block_alloc(man, line, ppos, tok);

        n = roff_head_alloc(man, line, ppos, tok);
 
        /* Add line arguments. */
        n = roff_head_alloc(man, line, ppos, tok);
 
        /* Add line arguments. */


@@ -318,16 +356,17 @@ blk_imp(MACRO_PROT_ARGS)
                if ( ! man_args(man, line, pos, buf, &p))
                        break;
                roff_word_alloc(man, line, la, p);
                if ( ! man_args(man, line, pos, buf, &p))
                        break;
                roff_word_alloc(man, line, la, p);

+               free(p);

        }
 
        /*
         * For macros having optional next-line scope,
         * keep the head open if there were no arguments.
        }
 
        /*
         * For macros having optional next-line scope,
         * keep the head open if there were no arguments.

-        * For `TP', always keep the head open.
+        * For `TP' and `TQ', always keep the head open.

         */
 
         */
 

-       if (man_macros[tok].flags & MAN_SCOPED &&
-           (tok == MAN_TP || n == man->last)) {
+       if (man_macro(tok)->flags & MAN_BSCOPED &&
+           (tok == MAN_TP || tok == MAN_TQ || n == man->last)) {

                man->flags |= MAN_BLINE;
                return;
        }
                man->flags |= MAN_BLINE;
                return;
        }


@@ -336,6 +375,7 @@ blk_imp(MACRO_PROT_ARGS)
 
        man_unscope(man, n);
        roff_body_alloc(man, line, ppos, tok);
 
        man_unscope(man, n);
        roff_body_alloc(man, line, ppos, tok);

+       man->flags &= ~ROFF_NONOFILL;

 }
 
 void
 }
 
 void


@@ -348,27 +388,26 @@ in_line_eoln(MACRO_PROT_ARGS)
        roff_elem_alloc(man, line, ppos, tok);
        n = man->last;
 
        roff_elem_alloc(man, line, ppos, tok);
        n = man->last;
 

+       if (tok == MAN_EX)
+               man->flags |= ROFF_NOFILL;
+       else if (tok == MAN_EE)
+               man->flags &= ~ROFF_NOFILL;
+

        for (;;) {
        for (;;) {

-               if (buf[*pos] != '\0' && (tok == MAN_fi || tok == MAN_nf)) {
-                       mandoc_vmsg(MANDOCERR_ARG_SKIP,
-                           man->parse, line, *pos, "%s %s",
-                           roff_name[tok], buf + *pos);
-                       break;
-               }

                if (buf[*pos] != '\0' && man->last != n && tok == MAN_PD) {
                if (buf[*pos] != '\0' && man->last != n && tok == MAN_PD) {

-                       mandoc_vmsg(MANDOCERR_ARG_EXCESS,
-                           man->parse, line, *pos, "%s ... %s",
-                           roff_name[tok], buf + *pos);
+                       mandoc_msg(MANDOCERR_ARG_EXCESS, line, *pos,
+                           "%s ... %s", roff_name[tok], buf + *pos);

                        break;
                }
                la = *pos;
                if ( ! man_args(man, line, pos, buf, &p))
                        break;
                        break;
                }
                la = *pos;
                if ( ! man_args(man, line, pos, buf, &p))
                        break;

-               if (man_macros[tok].flags & MAN_JOIN &&
+               if (man_macro(tok)->flags & MAN_JOIN &&

                    man->last->type == ROFFT_TEXT)
                        roff_word_append(man, p);
                else
                        roff_word_alloc(man, line, la, p);
                    man->last->type == ROFFT_TEXT)
                        roff_word_append(man, p);
                else
                        roff_word_alloc(man, line, la, p);

+               free(p);

        }
 
        /*
        }
 
        /*


@@ -381,13 +420,12 @@ in_line_eoln(MACRO_PROT_ARGS)
                man->last->flags |= NODE_EOS;
 
        /*
                man->last->flags |= NODE_EOS;
 
        /*

-        * If no arguments are specified and this is MAN_SCOPED (i.e.,
+        * If no arguments are specified and this is MAN_ESCOPED (i.e.,

         * next-line scoped), then set our mode to indicate that we're
         * waiting for terms to load into our context.
         */
 
         * next-line scoped), then set our mode to indicate that we're
         * waiting for terms to load into our context.
         */
 

-       if (n == man->last && man_macros[tok].flags & MAN_SCOPED) {
-               assert( ! (man_macros[tok].flags & MAN_NSCOPED));
+       if (n == man->last && man_macro(tok)->flags & MAN_ESCOPED) {

                man->flags |= MAN_ELINE;
                return;
        }
                man->flags |= MAN_ELINE;
                return;
        }


@@ -398,18 +436,21 @@ in_line_eoln(MACRO_PROT_ARGS)
        /* Rewind our element scope. */
 
        for ( ; man->last; man->last = man->last->parent) {
        /* Rewind our element scope. */
 
        for ( ; man->last; man->last = man->last->parent) {

-               man_state(man, man->last);
+               man->last->flags |= NODE_VALID;

                if (man->last == n)
                        break;
        }
                if (man->last == n)
                        break;
        }

+
+       /* Rewind next-line scoped ancestors, if any. */
+
+       if (man_macro(tok)->flags & MAN_ESCOPED)
+               man_descope(man, line, ppos, NULL);

 }
 
 void
 man_endparse(struct roff_man *man)
 {
 }
 
 void
 man_endparse(struct roff_man *man)
 {

-
-       man_unscope(man, man->first);
-       man->flags &= ~MAN_LITERAL;
+       man_unscope(man, man->meta.first);

 }
 
 static int
 }
 
 static int


@@ -424,6 +465,6 @@ man_args(struct roff_man *man, int line, int *pos, char *buf, char **v)
        if ('\0' == *start)
                return 0;
 
        if ('\0' == *start)
                return 0;
 

-       *v = mandoc_getarg(man->parse, v, line, pos);
+       *v = roff_getarg(man->roff, v, line, pos);

        return 1;
 }
        return 1;
 }