Fully support explicit tagging of .Sh and .Ss.

[mandoc.git] / man.c

diff --git a/man.c b/man.c
index cebe1b73f6fa54fc0b7a161c23543ed7f7a86a55..f0e4002b2a2c99cc0039ef780913cdf72af724b0 100644 (file)
--- a/man.c
+++ b/man.c
@@ -1,7 +1,7 @@
-/*     $Id: man.c,v 1.159 2015/04/19 14:00:19 schwarze Exp $ */
+/*     $Id: man.c,v 1.187 2019/01/05 00:36:50 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) 2013, 2014, 2015 Ingo Schwarze <schwarze@openbsd.org>
+ * Copyright (c) 2013-2015, 2017-2019 Ingo Schwarze <schwarze@openbsd.org>

  * Copyright (c) 2011 Joerg Sonnenberger <joerg@netbsd.org>
  *
  * Permission to use, copy, modify, and distribute this software for any
  * Copyright (c) 2011 Joerg Sonnenberger <joerg@netbsd.org>
  *
  * Permission to use, copy, modify, and distribute this software for any


@@ -35,34 +35,11 @@
 #include "roff_int.h"
 #include "libman.h"
 
 #include "roff_int.h"
 #include "libman.h"
 

-const  char *const __man_macronames[MAN_MAX] = {
-       "br",           "TH",           "SH",           "SS",
-       "TP",           "LP",           "PP",           "P",
-       "IP",           "HP",           "SM",           "SB",
-       "BI",           "IB",           "BR",           "RB",
-       "R",            "B",            "I",            "IR",
-       "RI",           "sp",           "nf",
-       "fi",           "RE",           "RS",           "DT",
-       "UC",           "PD",           "AT",           "in",
-       "ft",           "OP",           "EX",           "EE",
-       "UR",           "UE",           "ll"
-       };
-
-const  char * const *man_macronames = __man_macronames;
-
-static void             man_breakscope(struct roff_man *, int);
-static void             man_descope(struct roff_man *, int, int);
+static char            *man_hasc(char *);

 static int              man_ptext(struct roff_man *, int, char *, int);
 static int              man_pmacro(struct roff_man *, int, char *, int);
 
 
 static int              man_ptext(struct roff_man *, int, char *, int);
 static int              man_pmacro(struct roff_man *, int, char *, int);
 
 

-void
-man_endparse(struct roff_man *man)
-{
-
-       man_macroend(man);
-}
-

 int
 man_parseln(struct roff_man *man, int ln, char *buf, int offs)
 {
 int
 man_parseln(struct roff_man *man, int ln, char *buf, int offs)
 {


@@ -70,135 +47,94 @@ man_parseln(struct roff_man *man, int ln, char *buf, int offs)
        if (man->last->type != ROFFT_EQN || ln > man->last->line)
                man->flags |= MAN_NEWLINE;
 
        if (man->last->type != ROFFT_EQN || ln > man->last->line)
                man->flags |= MAN_NEWLINE;
 

-       return (roff_getcontrol(man->roff, buf, &offs) ?
+       return roff_getcontrol(man->roff, buf, &offs) ?

            man_pmacro(man, ln, buf, offs) :
            man_pmacro(man, ln, buf, offs) :

-           man_ptext(man, ln, buf, offs));
-}
-
-void
-man_elem_alloc(struct roff_man *man, int line, int pos, int tok)
-{
-       struct roff_node *p;
-
-       p = roff_node_alloc(man, line, pos, ROFFT_ELEM, tok);
-       roff_node_append(man, p);
-       man->next = ROFF_NEXT_CHILD;
-}
-
-void
-man_block_alloc(struct roff_man *man, int line, int pos, int tok)
-{
-       struct roff_node *p;
-
-       p = roff_node_alloc(man, line, pos, ROFFT_BLOCK, tok);
-       roff_node_append(man, p);
-       man->next = ROFF_NEXT_CHILD;
-}
-
-void
-man_word_alloc(struct roff_man *man, int line, int pos, const char *word)
-{
-       struct roff_node *n;
-
-       n = roff_node_alloc(man, line, pos, ROFFT_TEXT, TOKEN_NONE);
-       n->string = roff_strdup(man->roff, word);
-       roff_node_append(man, n);
-       man_valid_post(man);
-       man->next = ROFF_NEXT_SIBLING;
+           man_ptext(man, ln, buf, offs);

 }
 
 }
 

-void
-man_word_append(struct roff_man *man, const char *word)
-{
-       struct roff_node *n;
-       char            *addstr, *newstr;
-
-       n = man->last;
-       addstr = roff_strdup(man->roff, word);
-       mandoc_asprintf(&newstr, "%s %s", n->string, addstr);
-       free(addstr);
-       free(n->string);
-       n->string = newstr;
-       man->next = ROFF_NEXT_SIBLING;
-}
-
-void
-man_addeqn(struct roff_man *man, const struct eqn *ep)
+/*
+ * If the string ends with \c, return a pointer to the backslash.
+ * Otherwise, return NULL.
+ */
+static char *
+man_hasc(char *start)

 {
 {

-       struct roff_node *n;
+       char    *cp, *ep;

 
 

-       n = roff_node_alloc(man, ep->ln, ep->pos, ROFFT_EQN, TOKEN_NONE);
-       n->eqn = ep;
-       if (ep->ln > man->last->line)
-               n->flags |= MAN_LINE;
-       roff_node_append(man, n);
-       man->next = ROFF_NEXT_SIBLING;
-       man_descope(man, ep->ln, ep->pos);
+       ep = strchr(start, '\0') - 2;
+       if (ep < start || ep[0] != '\\' || ep[1] != 'c')
+               return NULL;
+       for (cp = ep; cp > start; cp--)
+               if (cp[-1] != '\\')
+                       break;
+       return (ep - cp) % 2 ? NULL : ep;

 }
 
 void
 }
 
 void

-man_addspan(struct roff_man *man, const struct tbl_span *sp)
+man_descope(struct roff_man *man, int line, int offs, char *start)

 {
 {

-       struct roff_node *n;
+       /* Trailing \c keeps next-line scope open. */

 
 

-       man_breakscope(man, TOKEN_NONE);
-       n = roff_node_alloc(man, sp->line, 0, ROFFT_TBL, TOKEN_NONE);
-       n->span = sp;
-       roff_node_append(man, n);
-       man_valid_post(man);
-       man->next = ROFF_NEXT_SIBLING;
-       man_descope(man, sp->line, 0);
-}
+       if (start != NULL && man_hasc(start) != NULL)
+               return;

 
 

-static void
-man_descope(struct roff_man *man, int line, int offs)
-{

        /*
         * Co-ordinate what happens with having a next-line scope open:
        /*
         * Co-ordinate what happens with having a next-line scope open:

-        * first close out the element scope (if applicable), then close
-        * out the block scope (also if applicable).
+        * first close out the element scopes (if applicable),
+        * then close out the block scope (also if applicable).

         */
 
        if (man->flags & MAN_ELINE) {
         */
 
        if (man->flags & MAN_ELINE) {

+               while (man->last->parent->type != ROFFT_ROOT &&
+                   man_macro(man->last->parent->tok)->flags & MAN_ESCOPED)
+                       man_unscope(man, man->last->parent);

                man->flags &= ~MAN_ELINE;
                man->flags &= ~MAN_ELINE;

-               man_unscope(man, man->last->parent);

        }
        if ( ! (man->flags & MAN_BLINE))
                return;
        }
        if ( ! (man->flags & MAN_BLINE))
                return;

-       man->flags &= ~MAN_BLINE;

        man_unscope(man, man->last->parent);
        roff_body_alloc(man, line, offs, man->last->tok);
        man_unscope(man, man->last->parent);
        roff_body_alloc(man, line, offs, man->last->tok);

+       man->flags &= ~(MAN_BLINE | ROFF_NONOFILL);

 }
 
 static int
 man_ptext(struct roff_man *man, int line, char *buf, int offs)
 {
        int              i;
 }
 
 static int
 man_ptext(struct roff_man *man, int line, char *buf, int offs)
 {
        int              i;

+       char            *ep;

 
 

-       /* Literal free-form text whitespace is preserved. */
+       /* In no-fill mode, whitespace is preserved on text lines. */

 
 

-       if (man->flags & MAN_LITERAL) {
-               man_word_alloc(man, line, offs, buf + offs);
-               man_descope(man, line, offs);
-               return(1);
+       if (man->flags & ROFF_NOFILL) {
+               roff_word_alloc(man, line, offs, buf + offs);
+               man_descope(man, line, offs, buf + offs);
+               return 1;

        }
 
        for (i = offs; buf[i] == ' '; i++)
                /* Skip leading whitespace. */ ;
 
        /*
        }
 
        for (i = offs; buf[i] == ' '; i++)
                /* Skip leading whitespace. */ ;
 
        /*

-        * Blank lines are ignored right after headings
+        * Blank lines are ignored in next line scope
+        * and right after headings and cancel preceding \c,

         * but add a single vertical space elsewhere.
         */
 
        if (buf[i] == '\0') {
         * but add a single vertical space elsewhere.
         */
 
        if (buf[i] == '\0') {

-               /* Allocate a blank entry. */
-               if (man->last->tok != MAN_SH &&
-                   man->last->tok != MAN_SS) {
-                       man_elem_alloc(man, line, offs, MAN_sp);
-                       man->next = ROFF_NEXT_SIBLING;
+               if (man->flags & (MAN_ELINE | MAN_BLINE)) {
+                       mandoc_msg(MANDOCERR_BLK_BLANK, line, 0, NULL);
+                       return 1;

                }
                }

-               return(1);
+               if (man->last->tok == MAN_SH || man->last->tok == MAN_SS)
+                       return 1;
+               if (man->last->type == ROFFT_TEXT &&
+                   ((ep = man_hasc(man->last->string)) != NULL)) {
+                       *ep = '\0';
+                       return 1;
+               }
+               roff_elem_alloc(man, line, offs, ROFF_sp);
+               man->next = ROFF_NEXT_SIBLING;
+               return 1;

        }
 
        /*
        }
 
        /*


@@ -211,8 +147,7 @@ man_ptext(struct roff_man *man, int line, char *buf, int offs)
 
        if (' ' == buf[i - 1] || '\t' == buf[i - 1]) {
                if (i > 1 && '\\' != buf[i - 2])
 
        if (' ' == buf[i - 1] || '\t' == buf[i - 1]) {
                if (i > 1 && '\\' != buf[i - 2])

-                       mandoc_msg(MANDOCERR_SPACE_EOL, man->parse,
-                           line, i - 1, NULL);
+                       mandoc_msg(MANDOCERR_SPACE_EOL, line, i - 1, NULL);

 
                for (--i; i && ' ' == buf[i]; i--)
                        /* Spin back to non-space. */ ;
 
                for (--i; i && ' ' == buf[i]; i--)
                        /* Spin back to non-space. */ ;


@@ -222,7 +157,7 @@ man_ptext(struct roff_man *man, int line, char *buf, int offs)
 
                buf[i] = '\0';
        }
 
                buf[i] = '\0';
        }

-       man_word_alloc(man, line, offs, buf + offs);
+       roff_word_alloc(man, line, offs, buf + offs);

 
        /*
         * End-of-sentence check.  If the last character is an unescaped
 
        /*
         * End-of-sentence check.  If the last character is an unescaped


@@ -232,10 +167,10 @@ man_ptext(struct roff_man *man, int line, char *buf, int offs)
 
        assert(i);
        if (mandoc_eos(buf, (size_t)i))
 
        assert(i);
        if (mandoc_eos(buf, (size_t)i))

-               man->last->flags |= MAN_EOS;
+               man->last->flags |= NODE_EOS;

 
 

-       man_descope(man, line, offs);
-       return(1);
+       man_descope(man, line, offs, buf + offs);
+       return 1;

 }
 
 static int
 }
 
 static int


@@ -243,30 +178,22 @@ man_pmacro(struct roff_man *man, int ln, char *buf, int offs)
 {
        struct roff_node *n;
        const char      *cp;
 {
        struct roff_node *n;
        const char      *cp;

-       int              tok;
-       int              i, ppos;
+       size_t           sz;
+       enum roff_tok    tok;
+       int              ppos;

        int              bline;
        int              bline;

-       char             mac[5];
-
-       ppos = offs;
-
-       /*
-        * Copy the first word into a nil-terminated buffer.
-        * Stop when a space, tab, escape, or eoln is encountered.
-        */
-
-       i = 0;
-       while (i < 4 && strchr(" \t\\", buf[offs]) == NULL)
-               mac[i++] = buf[offs++];

 
 

-       mac[i] = '\0';
-
-       tok = (i > 0 && i < 4) ? man_hash_find(mac) : TOKEN_NONE;
+       /* Determine the line macro. */

 
 

+       ppos = offs;
+       tok = TOKEN_NONE;
+       for (sz = 0; sz < 4 && strchr(" \t\\", buf[offs]) == NULL; sz++)
+               offs++;
+       if (sz > 0 && sz < 4)
+               tok = roffhash_find(man->manmac, buf + ppos, sz);

        if (tok == TOKEN_NONE) {
        if (tok == TOKEN_NONE) {

-               mandoc_msg(MANDOCERR_MACRO, man->parse,
-                   ln, ppos, buf + ppos - 1);
-               return(1);
+               mandoc_msg(MANDOCERR_MACRO, ln, ppos, "%s", buf + ppos - 1);
+               return 1;

        }
 
        /* Skip a leading escape sequence or tab. */
        }
 
        /* Skip a leading escape sequence or tab. */


@@ -286,7 +213,7 @@ man_pmacro(struct roff_man *man, int ln, char *buf, int offs)
 
        /* Jump to the next non-whitespace word. */
 
 
        /* Jump to the next non-whitespace word. */
 

-       while (buf[offs] && buf[offs] == ' ')
+       while (buf[offs] == ' ')

                offs++;
 
        /*
                offs++;
 
        /*


@@ -295,8 +222,7 @@ man_pmacro(struct roff_man *man, int ln, char *buf, int offs)
         */
 
        if (buf[offs] == '\0' && buf[offs - 1] == ' ')
         */
 
        if (buf[offs] == '\0' && buf[offs - 1] == ' ')

-               mandoc_msg(MANDOCERR_SPACE_EOL, man->parse,
-                   ln, offs - 1, NULL);
+               mandoc_msg(MANDOCERR_SPACE_EOL, ln, offs - 1, NULL);

 
        /*
         * Some macros break next-line scopes; otherwise, remember
 
        /*
         * Some macros break next-line scopes; otherwise, remember


@@ -306,10 +232,20 @@ man_pmacro(struct roff_man *man, int ln, char *buf, int offs)
        man_breakscope(man, tok);
        bline = man->flags & MAN_BLINE;
 
        man_breakscope(man, tok);
        bline = man->flags & MAN_BLINE;
 

+       /*
+        * If the line in next-line scope ends with \c, keep the
+        * next-line scope open for the subsequent input line.
+        * That is not at all portable, only groff >= 1.22.4
+        * does it, but *if* this weird idiom occurs in a manual
+        * page, that's very likely what the author intended.
+        */
+
+       if (bline && man_hasc(buf + offs))
+               bline = 0;
+

        /* Call to handler... */
 
        /* Call to handler... */
 

-       assert(man_macros[tok].fp);
-       (*man_macros[tok].fp)(man, tok, ln, ppos, &offs, buf);
+       (*man_macro(tok)->fp)(man, tok, ln, ppos, &offs, buf);

 
        /* In quick mode (for mandocdb), abort after the NAME section. */
 
 
        /* In quick mode (for mandocdb), abort after the NAME section. */
 


@@ -317,7 +253,7 @@ man_pmacro(struct roff_man *man, int ln, char *buf, int offs)
                n = man->last;
                if (n->type == ROFFT_BODY &&
                    strcmp(n->prev->child->string, "NAME"))
                n = man->last;
                if (n->type == ROFFT_BODY &&
                    strcmp(n->prev->child->string, "NAME"))

-                       return(2);
+                       return 2;

        }
 
        /*
        }
 
        /*


@@ -326,16 +262,16 @@ man_pmacro(struct roff_man *man, int ln, char *buf, int offs)
         * unless the next-line scope is allowed to continue.
         */
 
         * unless the next-line scope is allowed to continue.
         */
 

-       if ( ! bline || man->flags & MAN_ELINE ||
-           man_macros[tok].flags & MAN_NSCOPED)
-               return(1);
-
-       assert(man->flags & MAN_BLINE);
-       man->flags &= ~MAN_BLINE;
+       if (bline == 0 ||
+           (man->flags & MAN_BLINE) == 0 ||
+           man->flags & MAN_ELINE ||
+           man_macro(tok)->flags & MAN_NSCOPED)
+               return 1;

 
        man_unscope(man, man->last->parent);
        roff_body_alloc(man, ln, ppos, man->last->tok);
 
        man_unscope(man, man->last->parent);
        roff_body_alloc(man, ln, ppos, man->last->tok);

-       return(1);
+       man->flags &= ~(MAN_BLINE | ROFF_NONOFILL);
+       return 1;

 }
 
 void
 }
 
 void


@@ -349,101 +285,61 @@ man_breakscope(struct roff_man *man, int tok)
         * Delete the element that is being broken.
         */
 
         * Delete the element that is being broken.
         */
 

-       if (man->flags & MAN_ELINE && (tok == TOKEN_NONE ||
-           ! (man_macros[tok].flags & MAN_NSCOPED))) {
+       if (man->flags & MAN_ELINE && (tok < MAN_TH ||
+           (man_macro(tok)->flags & MAN_NSCOPED) == 0)) {

                n = man->last;
                n = man->last;

-               assert(n->type != ROFFT_TEXT);
-               if (man_macros[n->tok].flags & MAN_NSCOPED)
+               if (n->type == ROFFT_TEXT)
+                       n = n->parent;
+               if (n->tok < MAN_TH ||
+                   (man_macro(n->tok)->flags & (MAN_NSCOPED | MAN_ESCOPED))
+                    == MAN_NSCOPED)

                        n = n->parent;
 
                        n = n->parent;
 

-               mandoc_vmsg(MANDOCERR_BLK_LINE, man->parse,
-                   n->line, n->pos, "%s breaks %s",
-                   tok == TOKEN_NONE ? "TS" : man_macronames[tok],
-                   man_macronames[n->tok]);
+               mandoc_msg(MANDOCERR_BLK_LINE, n->line, n->pos,
+                   "%s breaks %s", roff_name[tok], roff_name[n->tok]);

 
                roff_node_delete(man, n);
                man->flags &= ~MAN_ELINE;
        }
 
 
                roff_node_delete(man, n);
                man->flags &= ~MAN_ELINE;
        }
 

+       /*
+        * Weird special case:
+        * Switching fill mode closes section headers.
+        */
+
+       if (man->flags & MAN_BLINE &&
+           (tok == ROFF_nf || tok == ROFF_fi) &&
+           (man->last->tok == MAN_SH || man->last->tok == MAN_SS)) {
+               n = man->last;
+               man_unscope(man, n);
+               roff_body_alloc(man, n->line, n->pos, n->tok);
+               man->flags &= ~(MAN_BLINE | ROFF_NONOFILL);
+       }
+

        /*
         * A block header next line scope is open,
         * and the new macro is not allowed inside block headers.
         * Delete the block that is being broken.
         */
 
        /*
         * A block header next line scope is open,
         * and the new macro is not allowed inside block headers.
         * Delete the block that is being broken.
         */
 

-       if (man->flags & MAN_BLINE && (tok == TOKEN_NONE ||
-           man_macros[tok].flags & MAN_BSCOPE)) {
+       if (man->flags & MAN_BLINE && tok != ROFF_nf && tok != ROFF_fi &&
+           (tok < MAN_TH || man_macro(tok)->flags & MAN_XSCOPE)) {

                n = man->last;
                if (n->type == ROFFT_TEXT)
                        n = n->parent;
                n = man->last;
                if (n->type == ROFFT_TEXT)
                        n = n->parent;

-               if ( ! (man_macros[n->tok].flags & MAN_BSCOPE))
+               if (n->tok < MAN_TH ||
+                   (man_macro(n->tok)->flags & MAN_XSCOPE) == 0)

                        n = n->parent;
 
                assert(n->type == ROFFT_HEAD);
                n = n->parent;
                assert(n->type == ROFFT_BLOCK);
                        n = n->parent;
 
                assert(n->type == ROFFT_HEAD);
                n = n->parent;
                assert(n->type == ROFFT_BLOCK);

-               assert(man_macros[n->tok].flags & MAN_SCOPED);
+               assert(man_macro(n->tok)->flags & MAN_BSCOPED);

 
 

-               mandoc_vmsg(MANDOCERR_BLK_LINE, man->parse,
-                   n->line, n->pos, "%s breaks %s",
-                   tok == TOKEN_NONE ? "TS" : man_macronames[tok],
-                   man_macronames[n->tok]);
+               mandoc_msg(MANDOCERR_BLK_LINE, n->line, n->pos,
+                   "%s breaks %s", roff_name[tok], roff_name[n->tok]);

 
                roff_node_delete(man, n);
 
                roff_node_delete(man, n);

-               man->flags &= ~MAN_BLINE;
+               man->flags &= ~(MAN_BLINE | ROFF_NONOFILL);

        }
 }
        }
 }

-
-const struct mparse *
-man_mparse(const struct roff_man *man)
-{
-
-       assert(man && man->parse);
-       return(man->parse);
-}
-
-void
-man_deroff(char **dest, const struct roff_node *n)
-{
-       char    *cp;
-       size_t   sz;
-
-       if (n->type != ROFFT_TEXT) {
-               for (n = n->child; n; n = n->next)
-                       man_deroff(dest, n);
-               return;
-       }
-
-       /* Skip leading whitespace and escape sequences. */
-
-       cp = n->string;
-       while ('\0' != *cp) {
-               if ('\\' == *cp) {
-                       cp++;
-                       mandoc_escape((const char **)&cp, NULL, NULL);
-               } else if (isspace((unsigned char)*cp))
-                       cp++;
-               else
-                       break;
-       }
-
-       /* Skip trailing whitespace. */
-
-       for (sz = strlen(cp); sz; sz--)
-               if (0 == isspace((unsigned char)cp[sz-1]))
-                       break;
-
-       /* Skip empty strings. */
-
-       if (0 == sz)
-               return;
-
-       if (NULL == *dest) {
-               *dest = mandoc_strndup(cp, sz);
-               return;
-       }
-
-       mandoc_asprintf(&cp, "%s %*s", *dest, (int)sz, cp);
-       free(*dest);
-       *dest = cp;
-}