Removed superfluous XXX.

[mandoc.git] / man.c

diff --git a/man.c b/man.c
index 14004c550fcb6d29a9f1871b99a5cffca4a94bcc..5763c9ea948aebbe5dee427e8cab5164b6a2dadc 100644 (file)
--- a/man.c
+++ b/man.c
@@ -1,4 +1,4 @@
-/*     $Id: man.c,v 1.21 2009/06/16 19:55:28 kristaps Exp $ */
+/*     $Id: man.c,v 1.36 2009/08/21 13:18:32 kristaps Exp $ */

 /*
  * Copyright (c) 2008, 2009 Kristaps Dzonsons <kristaps@kth.se>
  *
 /*
  * Copyright (c) 2008, 2009 Kristaps Dzonsons <kristaps@kth.se>
  *


@@ -23,13 +23,38 @@
 
 #include "libman.h"
 
 
 #include "libman.h"
 

+const  char *const __man_merrnames[WERRMAX] = {                 
+       "invalid character", /* WNPRINT */
+       "system: malloc error", /* WNMEM */
+       "invalid manual section", /* WMSEC */
+       "invalid date format", /* WDATE */
+       "scope of prior line violated", /* WLNSCOPE */
+       "trailing whitespace", /* WTSPACE */
+       "unterminated quoted parameter", /* WTQUOTE */
+       "document has no body", /* WNODATA */
+       "document has no title/section", /* WNOTITLE */
+       "invalid escape sequence", /* WESCAPE */
+       "invalid number format", /* WNUMFMT */
+       "expected block head arguments", /* WHEADARGS */
+       "expected block body arguments", /* WBODYARGS */
+       "expected empty block head", /* WNHEADARGS */
+       "unknown macro", /* WMACRO */
+       "ill-formed macro", /* WMACROFORM */
+       "scope open on exit", /* WEXITSCOPE */
+       "no scope context", /* WNOSCOPE */
+       "literal context already open", /* WOLITERAL */
+       "no literal context open" /* WNLITERAL */
+};
+

 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",
 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",           "na",           "i"
+       "RI",           "na",           "i",            "sp",
+       "nf",           "fi",           "r",            "RE",
+       "RS",           "DT"

        };
 
 const  char * const *man_macronames = __man_macronames;
        };
 
 const  char * const *man_macronames = __man_macronames;


@@ -42,6 +67,12 @@ static       int              man_ptext(struct man *, int, char *);
 static int              man_pmacro(struct man *, int, char *);
 static void             man_free1(struct man *);
 static int              man_alloc1(struct man *);
 static int              man_pmacro(struct man *, int, char *);
 static void             man_free1(struct man *);
 static int              man_alloc1(struct man *);

+static int              pstring(struct man *, int, int, 
+                               const char *, size_t);
+
+#ifdef __linux__
+extern size_t            strlcpy(char *, const char *, size_t);
+#endif

 
 
 const struct man_node *
 
 
 const struct man_node *


@@ -182,6 +213,24 @@ man_node_append(struct man *man, struct man_node *p)
                abort();
                /* NOTREACHED */
        }
                abort();
                /* NOTREACHED */
        }

+       
+       p->parent->nchild++;
+
+       if ( ! man_valid_pre(man, p))
+               return(0);
+
+       switch (p->type) {
+       case (MAN_HEAD):
+               assert(MAN_BLOCK == p->parent->type);
+               p->parent->head = p;
+               break;
+       case (MAN_BODY):
+               assert(MAN_BLOCK == p->parent->type);
+               p->parent->body = p;
+               break;
+       default:
+               break;
+       }

 
        man->last = p;
 
 
        man->last = p;
 


@@ -218,29 +267,99 @@ man_node_alloc(int line, int pos, enum man_type type, int tok)
 
 
 int
 
 
 int

-man_elem_alloc(struct man *man, int line, int pos, int tok)
+man_elem_alloc(struct man *m, int line, int pos, int tok)

 {
        struct man_node *p;
 
        p = man_node_alloc(line, pos, MAN_ELEM, tok);
        if (NULL == p)
                return(0);
 {
        struct man_node *p;
 
        p = man_node_alloc(line, pos, MAN_ELEM, tok);
        if (NULL == p)
                return(0);

-       return(man_node_append(man, p));
+       if ( ! man_node_append(m, p))
+               return(0);
+       m->next = MAN_NEXT_CHILD;
+       return(1);

 }
 
 
 int
 }
 
 
 int

-man_word_alloc(struct man *man, 
-               int line, int pos, const char *word)
+man_head_alloc(struct man *m, int line, int pos, int tok)

 {
 {

-       struct man_node *p;
+       struct man_node *p;

 
 

-       p = man_node_alloc(line, pos, MAN_TEXT, -1);
+       p = man_node_alloc(line, pos, MAN_HEAD, tok);

        if (NULL == p)
                return(0);
        if (NULL == p)
                return(0);

-       if (NULL == (p->string = strdup(word)))
+       if ( ! man_node_append(m, p))

                return(0);
                return(0);

-       return(man_node_append(man, p));
+       m->next = MAN_NEXT_CHILD;
+       return(1);
+}
+
+
+int
+man_body_alloc(struct man *m, int line, int pos, int tok)
+{
+       struct man_node *p;
+
+       p = man_node_alloc(line, pos, MAN_BODY, tok);
+       if (NULL == p)
+               return(0);
+       if ( ! man_node_append(m, p))
+               return(0);
+       m->next = MAN_NEXT_CHILD;
+       return(1);
+}
+
+
+int
+man_block_alloc(struct man *m, int line, int pos, int tok)
+{
+       struct man_node *p;
+
+       p = man_node_alloc(line, pos, MAN_BLOCK, tok);
+       if (NULL == p)
+               return(0);
+       if ( ! man_node_append(m, p))
+               return(0);
+       m->next = MAN_NEXT_CHILD;
+       return(1);
+}
+
+
+static int
+pstring(struct man *m, int line, int pos, 
+               const char *p, size_t len)
+{
+       struct man_node *n;
+       size_t           sv;
+
+       n = man_node_alloc(line, pos, MAN_TEXT, -1);
+       if (NULL == n)
+               return(0);
+
+       n->string = malloc(len + 1);
+       if (NULL == n->string) {
+               free(n);
+               return(0);
+       }
+
+       sv = strlcpy(n->string, p, len + 1);
+
+       /* Prohibit truncation. */
+       assert(sv < len + 1);
+
+       if ( ! man_node_append(m, n))
+               return(0);
+       m->next = MAN_NEXT_SIBLING;
+       return(1);
+}
+
+
+int
+man_word_alloc(struct man *m, int line, int pos, const char *word)
+{
+
+       return(pstring(m, line, pos, word, strlen(word)));

 }
 
 
 }
 
 


@@ -250,6 +369,8 @@ man_node_free(struct man_node *p)
 
        if (p->string)
                free(p->string);
 
        if (p->string)
                free(p->string);

+       if (p->parent)
+               p->parent->nchild--;

        free(p);
 }
 
        free(p);
 }
 


@@ -263,6 +384,7 @@ man_node_freelist(struct man_node *p)
        if (p->next)
                man_node_freelist(p->next);
 
        if (p->next)
                man_node_freelist(p->next);
 

+       assert(0 == p->nchild);

        man_node_free(p);
 }
 
        man_node_free(p);
 }
 


@@ -270,44 +392,83 @@ man_node_freelist(struct man_node *p)
 static int
 man_ptext(struct man *m, int line, char *buf)
 {
 static int
 man_ptext(struct man *m, int line, char *buf)
 {

+       int              i, j;
+
+       /* Literal free-form text whitespace is preserved. */
+
+       if (MAN_LITERAL & m->flags) {
+               if ( ! man_word_alloc(m, line, 0, buf))
+                       return(0);
+               goto descope;
+       }
+
+       /* First de-chunk and allocate words. */
+
+       for (i = 0; ' ' == buf[i]; i++)
+               /* Skip leading whitespace. */ ;
+       if (0 == buf[i]) {
+               if ( ! pstring(m, line, 0, &buf[i], 0))
+                       return(0);
+               goto descope;
+       }
+
+       for (j = i; buf[i]; i++) {
+               if (' ' != buf[i])
+                       continue;
+
+               /* Escaped whitespace. */
+               if (i && ' ' == buf[i] && '\\' == buf[i - 1])
+                       continue;
+
+               buf[i++] = 0;
+               if ( ! pstring(m, line, j, &buf[j], (size_t)(i - j)))
+                       return(0);
+
+               for ( ; ' ' == buf[i]; i++)
+                       /* Skip trailing whitespace. */ ;
+
+               j = i;
+               if (0 == buf[i])
+                       break;
+       }

 
 

-       if ( ! man_word_alloc(m, line, 0, buf))
+       if (j != i && ! pstring(m, line, j, &buf[j], (size_t)(i - j)))

                return(0);
                return(0);

-       m->next = MAN_NEXT_SIBLING;
+
+descope:

 
        /*
 
        /*

-        * If this is one of the zany NLINE macros that consumes the
-        * next line of input as being influenced, then close out the
-        * existing macro "scope" and continue processing.
+        * 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).

         */
 
         */
 

-       if ( ! (MAN_NLINE & m->flags))
-               return(1);
+       if (MAN_ELINE & m->flags) {
+               m->flags &= ~MAN_ELINE;
+               if ( ! man_unscope(m, m->last->parent))
+                       return(0);
+       }

 
 

-       m->flags &= ~MAN_NLINE;
-       m->last = m->last->parent;
+       if ( ! (MAN_BLINE & m->flags))
+               return(1);
+       m->flags &= ~MAN_BLINE;

 
 

-       assert(MAN_ROOT != m->last->type);
-       if ( ! man_valid_post(m))
+       if ( ! man_unscope(m, m->last->parent))

                return(0);
                return(0);

-       if ( ! man_action_post(m))
-               return(0);
-
-       return(1);
+       return(man_body_alloc(m, line, 0, m->last->tok));

 }
 
 
 int
 man_pmacro(struct man *m, int ln, char *buf)
 {
 }
 
 
 int
 man_pmacro(struct man *m, int ln, char *buf)
 {

-       int               i, j, c, ppos, fl;
-       char              mac[5];
-       struct man_node  *n;
+       int              i, j, c, ppos, fl;
+       char             mac[5];
+       struct man_node *n;

 
        /* Comments and empties are quickly ignored. */
 
 
        /* Comments and empties are quickly ignored. */
 

-       n = m->last;
-       fl = MAN_NLINE & m->flags;
+       fl = m->flags;

 
        if (0 == buf[1])
                goto out;
 
        if (0 == buf[1])
                goto out;


@@ -337,24 +498,20 @@ man_pmacro(struct man *m, int ln, char *buf)
 
        if (j == 4 || j < 1) {
                if ( ! (MAN_IGN_MACRO & m->pflags)) {
 
        if (j == 4 || j < 1) {
                if ( ! (MAN_IGN_MACRO & m->pflags)) {

-                       (void)man_verr(m, ln, ppos, 
-                               "ill-formed macro: %s", mac);
+                       (void)man_perr(m, ln, ppos, WMACROFORM);

                        goto err;
                } 
                        goto err;
                } 

-               if ( ! man_vwarn(m, ln, ppos, 
-                               "ill-formed macro: %s", mac))
+               if ( ! man_pwarn(m, ln, ppos, WMACROFORM))

                        goto err;
                return(1);
        }
        
        if (MAN_MAX == (c = man_hash_find(m->htab, mac))) {
                if ( ! (MAN_IGN_MACRO & m->pflags)) {
                        goto err;
                return(1);
        }
        
        if (MAN_MAX == (c = man_hash_find(m->htab, mac))) {
                if ( ! (MAN_IGN_MACRO & m->pflags)) {

-                       (void)man_verr(m, ln, ppos, 
-                               "unknown macro: %s", mac);
+                       (void)man_perr(m, ln, ppos, WMACRO);

                        goto err;
                } 
                        goto err;
                } 

-               if ( ! man_vwarn(m, ln, ppos, 
-                               "unknown macro: %s", mac))
+               if ( ! man_pwarn(m, ln, ppos, WMACRO))

                        goto err;
                return(1);
        }
                        goto err;
                return(1);
        }


@@ -364,34 +521,56 @@ man_pmacro(struct man *m, int ln, char *buf)
        while (buf[i] && ' ' == buf[i])
                i++;
 
        while (buf[i] && ' ' == buf[i])
                i++;
 

+       /* Remove prior ELINE macro, if applicable. */
+
+       if (m->flags & MAN_ELINE) {
+               n = m->last;
+               assert(NULL == n->child);
+               if ( ! man_nwarn(m, n, WLNSCOPE))
+                       return(0);
+
+               if (n->prev) {
+                       assert(n != n->parent->child);
+                       assert(n == n->prev->next);
+                       n->prev->next = NULL;
+                       m->last = n->prev;
+               } else {
+                       assert(n == n->parent->child);
+                       n->parent->child = NULL;
+                       m->last = n->parent;
+               }
+
+               man_node_free(n);
+               m->flags &= ~MAN_ELINE;
+       }
+

        /* Begin recursive parse sequence. */
 
        /* Begin recursive parse sequence. */
 

-       if ( ! man_macro(m, c, ln, ppos, &i, buf))
+       assert(man_macros[c].fp);
+
+       if ( ! (*man_macros[c].fp)(m, c, ln, ppos, &i, buf))

                goto err;
 
 out:
                goto err;
 
 out:

-       if (fl) {
-               /*
-                * A NLINE macro has been immediately followed with
-                * another.  Close out the preceding macro's scope, and
-                * continue.
-                */
-               assert(MAN_ROOT != m->last->type);
-               assert(m->last->parent);
-               assert(MAN_ROOT != m->last->parent->type);
-
-               if (n != m->last)
-                       m->last = m->last->parent;
-
-               if ( ! man_valid_post(m))
-                       return(0);
-               if ( ! man_action_post(m))
-                       return(0);
-               m->next = MAN_NEXT_SIBLING;
-               m->flags &= ~MAN_NLINE;
-       } 
+       if ( ! (MAN_BLINE & fl))
+               return(1);

 
 

-       return(1);
+       /* 
+        * If we've opened a new next-line element scope, then return
+        * now, as the next line will close out the block scope.
+        */
+
+       if (MAN_ELINE & m->flags)
+               return(1);
+
+       /* Close out the block scope opened in the prior line.  */
+
+       assert(MAN_BLINE & m->flags);
+       m->flags &= ~MAN_BLINE;
+
+       if ( ! man_unscope(m, m->last->parent))
+               return(0);
+       return(man_body_alloc(m, ln, 0, m->last->tok));

 
 err:   /* Error out. */
 
 
 err:   /* Error out. */
 


@@ -432,3 +611,16 @@ man_vwarn(struct man *man, int ln, int pos, const char *fmt, ...)
 }
 
 
 }
 
 

+int
+man_err(struct man *m, int line, int pos, int iserr, enum merr type)
+{
+       const char       *p;
+       
+       p = __man_merrnames[(int)type];
+       assert(p);
+
+       if (iserr)
+               return(man_verr(m, line, pos, p));
+
+       return(man_vwarn(m, line, pos, p));
+}