De-chunking of text removed from parsers. This is a significant change (and I don...

[mandoc.git] / man_macro.c

diff --git a/man_macro.c b/man_macro.c
index bc39053fb864511278ee0632fdd5dc7f5a6a35b4..37534e091ffda2b02c1e3e42757c02cb1f481fe7 100644 (file)
--- a/man_macro.c
+++ b/man_macro.c
@@ -1,4 +1,4 @@
-/*     $Id: man_macro.c,v 1.24 2009/08/20 13:51:55 kristaps Exp $ */
+/*     $Id: man_macro.c,v 1.42 2010/03/29 10:10:35 kristaps Exp $ */

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


@@ -14,6 +14,10 @@
  * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
  * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
  */
  * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
  * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
  */

+#ifdef HAVE_CONFIG_H
+#include "config.h"
+#endif
+

 #include <assert.h>
 #include <ctype.h>
 #include <stdlib.h>
 #include <assert.h>
 #include <ctype.h>
 #include <stdlib.h>


@@ -21,26 +25,33 @@
 
 #include "libman.h"
 
 
 #include "libman.h"
 

-#define        REW_REWIND      (0)             /* See rew_scope(). */
-#define        REW_NOHALT      (1)             /* See rew_scope(). */
-#define        REW_HALT        (2)             /* See rew_scope(). */
+enum   rew {
+       REW_REWIND,
+       REW_NOHALT,
+       REW_HALT
+};

 
 

-static int              in_line_eoln(MACRO_PROT_ARGS);
-static int              blk_imp(MACRO_PROT_ARGS);

 static int              blk_close(MACRO_PROT_ARGS);
 static int              blk_close(MACRO_PROT_ARGS);

+static int              blk_dotted(MACRO_PROT_ARGS);
+static int              blk_exp(MACRO_PROT_ARGS);
+static int              blk_imp(MACRO_PROT_ARGS);
+static int              in_line_eoln(MACRO_PROT_ARGS);

 
 

-static int              rew_scope(enum man_type, struct man *, int);
-static int              rew_dohalt(int, enum man_type, 
+static int              rew_scope(enum man_type, 
+                               struct man *, enum mant);
+static enum rew         rew_dohalt(enum mant, enum man_type, 

                                const struct man_node *);
                                const struct man_node *);

-static int              rew_block(int, enum man_type, 
+static enum rew         rew_block(enum mant, enum man_type, 

                                const struct man_node *);
                                const struct man_node *);

+static int              rew_warn(struct man *, 
+                               struct man_node *, enum merr);

 
 const  struct man_macro __man_macros[MAN_MAX] = {
 
 const  struct man_macro __man_macros[MAN_MAX] = {

-       { in_line_eoln, 0 }, /* br */
+       { in_line_eoln, MAN_NSCOPED }, /* br */

        { in_line_eoln, 0 }, /* TH */
        { blk_imp, MAN_SCOPED }, /* SH */
        { blk_imp, MAN_SCOPED }, /* SS */
        { in_line_eoln, 0 }, /* TH */
        { blk_imp, MAN_SCOPED }, /* SH */
        { blk_imp, MAN_SCOPED }, /* SS */

-       { blk_imp, MAN_SCOPED }, /* TP */
+       { blk_imp, MAN_SCOPED | MAN_FSCOPED }, /* TP */

        { blk_imp, 0 }, /* LP */
        { blk_imp, 0 }, /* PP */
        { blk_imp, 0 }, /* P */
        { blk_imp, 0 }, /* LP */
        { blk_imp, 0 }, /* PP */
        { blk_imp, 0 }, /* P */


@@ -57,29 +68,64 @@ const       struct man_macro __man_macros[MAN_MAX] = {
        { in_line_eoln, MAN_SCOPED }, /* I */
        { in_line_eoln, 0 }, /* IR */
        { in_line_eoln, 0 }, /* RI */
        { in_line_eoln, MAN_SCOPED }, /* I */
        { in_line_eoln, 0 }, /* IR */
        { in_line_eoln, 0 }, /* RI */

-       { in_line_eoln, 0 }, /* na */
+       { in_line_eoln, MAN_NSCOPED }, /* na */

        { in_line_eoln, 0 }, /* i */
        { in_line_eoln, 0 }, /* i */

-       { in_line_eoln, 0 }, /* sp */
+       { in_line_eoln, MAN_NSCOPED }, /* sp */

        { in_line_eoln, 0 }, /* nf */
        { in_line_eoln, 0 }, /* fi */
        { in_line_eoln, 0 }, /* r */
        { blk_close, 0 }, /* RE */
        { in_line_eoln, 0 }, /* nf */
        { in_line_eoln, 0 }, /* fi */
        { in_line_eoln, 0 }, /* r */
        { blk_close, 0 }, /* RE */

-       { blk_imp, MAN_EXPLICIT }, /* RS */
+       { blk_exp, MAN_EXPLICIT }, /* RS */

        { in_line_eoln, 0 }, /* DT */
        { in_line_eoln, 0 }, /* DT */

+       { in_line_eoln, 0 }, /* UC */
+       { in_line_eoln, 0 }, /* PD */
+       { in_line_eoln, MAN_NSCOPED }, /* Sp */
+       { in_line_eoln, 0 }, /* Vb */
+       { in_line_eoln, 0 }, /* Ve */
+       { blk_exp, MAN_EXPLICIT | MAN_NOCLOSE}, /* de */
+       { blk_exp, MAN_EXPLICIT | MAN_NOCLOSE}, /* dei */
+       { blk_exp, MAN_EXPLICIT | MAN_NOCLOSE}, /* am */
+       { blk_exp, MAN_EXPLICIT | MAN_NOCLOSE}, /* ami */
+       { blk_exp, MAN_EXPLICIT | MAN_NOCLOSE}, /* ig */
+       { blk_dotted, 0 }, /* . */

 };
 
 const  struct man_macro * const man_macros = __man_macros;
 
 
 };
 
 const  struct man_macro * const man_macros = __man_macros;
 
 

+/*
+ * Warn when "n" is an explicit non-roff macro.
+ */
+static int
+rew_warn(struct man *m, struct man_node *n, enum merr er)
+{
+
+       if (er == WERRMAX || MAN_BLOCK != n->type)
+               return(1);
+       if (MAN_VALID & n->flags)
+               return(1);
+       if ( ! (MAN_EXPLICIT & man_macros[n->tok].flags))
+               return(1);
+       if (MAN_NOCLOSE & man_macros[n->tok].flags)
+               return(1);
+       return(man_nwarn(m, n, er));
+}
+
+
+/*
+ * Rewind scope.  If a code "er" != WERRMAX has been provided, it will
+ * be used if an explicit block scope is being closed out.
+ */

 int
 int

-man_unscope(struct man *m, const struct man_node *n)
+man_unscope(struct man *m, const struct man_node *n, enum merr er)

 {
 
        assert(n);
 {
 
        assert(n);

-       m->next = MAN_NEXT_SIBLING;

 
        /* LINTED */
        while (m->last != n) {
 
        /* LINTED */
        while (m->last != n) {

+               if ( ! rew_warn(m, m->last, er))
+                       return(0);

                if ( ! man_valid_post(m))
                        return(0);
                if ( ! man_action_post(m))
                if ( ! man_valid_post(m))
                        return(0);
                if ( ! man_action_post(m))


@@ -88,14 +134,22 @@ man_unscope(struct man *m, const struct man_node *n)
                assert(m->last);
        }
 
                assert(m->last);
        }
 

+       if ( ! rew_warn(m, m->last, er))
+               return(0);

        if ( ! man_valid_post(m))
                return(0);
        if ( ! man_valid_post(m))
                return(0);

-       return(man_action_post(m));
+       if ( ! man_action_post(m))
+               return(0);
+
+       m->next = MAN_ROOT == m->last->type ? 
+               MAN_NEXT_CHILD : MAN_NEXT_SIBLING;
+
+       return(1);

 }
 
 
 }
 
 

-static int
-rew_block(int ntok, enum man_type type, const struct man_node *n)
+static enum rew
+rew_block(enum mant ntok, enum man_type type, const struct man_node *n)

 {
 
        if (MAN_BLOCK == type && ntok == n->parent->tok && 
 {
 
        if (MAN_BLOCK == type && ntok == n->parent->tok && 


@@ -110,23 +164,52 @@ rew_block(int ntok, enum man_type type, const struct man_node *n)
  * section (all less sections), and scoped to subsections (all less
  * sections and subsections).
  */
  * section (all less sections), and scoped to subsections (all less
  * sections and subsections).
  */

-static int 
-rew_dohalt(int tok, enum man_type type, const struct man_node *n)
+static enum rew 
+rew_dohalt(enum mant tok, enum man_type type, const struct man_node *n)

 {
 {

-       int              c;
+       enum rew         c;

 
 

+       /* We cannot progress beyond the root ever. */

        if (MAN_ROOT == n->type)
                return(REW_HALT);
        if (MAN_ROOT == n->type)
                return(REW_HALT);

+

        assert(n->parent);
        assert(n->parent);

+
+       /* Normal nodes shouldn't go to the level of the root. */

        if (MAN_ROOT == n->parent->type)
                return(REW_REWIND);
        if (MAN_ROOT == n->parent->type)
                return(REW_REWIND);

+
+       /* Already-validated nodes should be closed out. */

        if (MAN_VALID & n->flags)
                return(REW_NOHALT);
 
        if (MAN_VALID & n->flags)
                return(REW_NOHALT);
 

-       /* Rewind to ourselves, first. */
+       /* First: rewind to ourselves. */

        if (type == n->type && tok == n->tok)
                return(REW_REWIND);
 
        if (type == n->type && tok == n->tok)
                return(REW_REWIND);
 

+       /*
+        * If we're a roff macro, then we can close out anything that
+        * stands between us and our parent context.
+        */
+       if (MAN_NOCLOSE & man_macros[tok].flags)
+               return(REW_NOHALT);
+
+       /* 
+        * Don't clobber roff macros: this is a bit complicated.  If the
+        * current macro is a roff macro, halt immediately and don't
+        * rewind.  If it's not, and the parent is, then close out the
+        * current scope and halt at the parent.
+        */
+       if (MAN_NOCLOSE & man_macros[n->tok].flags)
+               return(REW_HALT);
+       if (MAN_NOCLOSE & man_macros[n->parent->tok].flags)
+               return(REW_REWIND);
+
+       /* 
+        * Next follow the implicit scope-smashings as defined by man.7:
+        * section, sub-section, etc.
+        */
+

        switch (tok) {
        case (MAN_SH):
                break;
        switch (tok) {
        case (MAN_SH):
                break;


@@ -166,10 +249,10 @@ rew_dohalt(int tok, enum man_type type, const struct man_node *n)
  * scopes.  When a scope is closed, it must be validated and actioned.
  */
 static int
  * scopes.  When a scope is closed, it must be validated and actioned.
  */
 static int

-rew_scope(enum man_type type, struct man *m, int tok)
+rew_scope(enum man_type type, struct man *m, enum mant tok)

 {
        struct man_node *n;
 {
        struct man_node *n;

-       int              c;
+       enum rew         c;

 
        /* LINTED */
        for (n = m->last; n; n = n->parent) {
 
        /* LINTED */
        for (n = m->last; n; n = n->parent) {


@@ -185,18 +268,72 @@ rew_scope(enum man_type type, struct man *m, int tok)
                        break;
        }
 
                        break;
        }
 

-       /* Rewind until the current point. */
-
+       /* 
+        * Rewind until the current point.  Warn if we're a roff
+        * instruction that's mowing over explicit scopes.
+        */

        assert(n);
        assert(n);

-       return(man_unscope(m, n));
+       if (MAN_NOCLOSE & man_macros[tok].flags)
+               return(man_unscope(m, n, WROFFSCOPE));
+
+       return(man_unscope(m, n, WERRMAX));

 }
 
 
 }
 
 

+/*
+ * Closure for dotted macros (de, dei, am, ami, ign).  This must handle
+ * any of these as the parent node, so it needs special handling.
+ * Beyond this, it's the same as blk_close().
+ */
+/* ARGSUSED */
+int
+blk_dotted(MACRO_PROT_ARGS)
+{
+       enum mant        ntok;
+       struct man_node *nn;
+
+       /* Check for any of the following parents... */
+
+       for (nn = m->last->parent; nn; nn = nn->parent)
+               if (nn->tok == MAN_de || nn->tok == MAN_dei ||
+                               nn->tok == MAN_am ||
+                               nn->tok == MAN_ami ||
+                               nn->tok == MAN_ig) {
+                       ntok = nn->tok;
+                       break;
+               }
+
+       if (NULL == nn) {
+               if ( ! man_pwarn(m, line, ppos, WNOSCOPE))
+                       return(0);
+               return(1);
+       }
+
+       if ( ! rew_scope(MAN_BODY, m, ntok))
+               return(0);
+       if ( ! rew_scope(MAN_BLOCK, m, ntok))
+               return(0);
+
+       /*
+        * Restore flags set when we got here and also stipulate that we
+        * don't post-process the line when exiting the macro op
+        * function in man_pmacro().  See blk_exp().
+        */
+
+       m->flags = m->svflags | MAN_ILINE;
+       m->next = m->svnext;
+       return(1);
+}
+
+
+/*
+ * Close out a generic explicit macro.
+ */

 /* ARGSUSED */
 int
 blk_close(MACRO_PROT_ARGS)
 {
 /* ARGSUSED */
 int
 blk_close(MACRO_PROT_ARGS)
 {

-       int                      ntok;
+       enum mant                ntok;

        const struct man_node   *nn;
 
        switch (tok) {
        const struct man_node   *nn;
 
        switch (tok) {


@@ -220,11 +357,69 @@ blk_close(MACRO_PROT_ARGS)
                return(0);
        if ( ! rew_scope(MAN_BLOCK, m, ntok))
                return(0);
                return(0);
        if ( ! rew_scope(MAN_BLOCK, m, ntok))
                return(0);

-       m->next = MAN_NEXT_SIBLING;
+

        return(1);
 }
 
 
        return(1);
 }
 
 

+int
+blk_exp(MACRO_PROT_ARGS)
+{
+       int              w, la;
+       char            *p;
+
+       /* 
+        * Close out prior scopes.  "Regular" explicit macros cannot be
+        * nested, but we allow roff macros to be placed just about
+        * anywhere.
+        */
+
+       if ( ! (MAN_NOCLOSE & man_macros[tok].flags)) {
+               if ( ! rew_scope(MAN_BODY, m, tok))
+                       return(0);
+               if ( ! rew_scope(MAN_BLOCK, m, tok))
+                       return(0);
+       } else {
+               /*
+                * Save our state and next-scope indicator; we restore
+                * it when exiting from the roff instruction block.  See
+                * blk_dotted().
+                */
+               m->svflags = m->flags;
+               m->svnext = m->next;
+               
+               /* Make sure we drop any line modes. */
+               m->flags = 0;
+       }
+
+       if ( ! man_block_alloc(m, line, ppos, tok))
+               return(0);
+       if ( ! man_head_alloc(m, line, ppos, tok))
+               return(0);
+
+       for (;;) {
+               la = *pos;
+               w = man_args(m, line, pos, buf, &p);
+
+               if (-1 == w)
+                       return(0);
+               if (0 == w)
+                       break;
+
+               if ( ! man_word_alloc(m, line, la, p))
+                       return(0);
+       }
+
+       assert(m);
+       assert(tok != MAN_MAX);
+
+       if ( ! rew_scope(MAN_HEAD, m, tok))
+               return(0);
+       return(man_body_alloc(m, line, ppos, tok));
+}
+
+
+

 /*
  * Parse an implicit-block macro.  These contain a MAN_HEAD and a
  * MAN_BODY contained within a MAN_BLOCK.  Rules for closing out other
 /*
  * Parse an implicit-block macro.  These contain a MAN_HEAD and a
  * MAN_BODY contained within a MAN_BLOCK.  Rules for closing out other


@@ -236,6 +431,7 @@ blk_imp(MACRO_PROT_ARGS)
 {
        int              w, la;
        char            *p;
 {
        int              w, la;
        char            *p;

+       struct man_node *n;

 
        /* Close out prior scopes. */
 
 
        /* Close out prior scopes. */
 


@@ -251,6 +447,8 @@ blk_imp(MACRO_PROT_ARGS)
        if ( ! man_head_alloc(m, line, ppos, tok))
                return(0);
 
        if ( ! man_head_alloc(m, line, ppos, tok))
                return(0);
 

+       n = m->last;
+

        /* Add line arguments. */
 
        for (;;) {
        /* Add line arguments. */
 
        for (;;) {


@@ -269,11 +467,18 @@ blk_imp(MACRO_PROT_ARGS)
        /* Close out head and open body (unless MAN_SCOPE). */
 
        if (MAN_SCOPED & man_macros[tok].flags) {
        /* Close out head and open body (unless MAN_SCOPE). */
 
        if (MAN_SCOPED & man_macros[tok].flags) {

-               m->flags |= MAN_BLINE;
-               return(1);
-       } else if ( ! rew_scope(MAN_HEAD, m, tok))
-               return(0);
+               /* If we're forcing scope (`TP'), keep it open. */
+               if (MAN_FSCOPED & man_macros[tok].flags) {
+                       m->flags |= MAN_BLINE;
+                       return(1);
+               } else if (n == m->last) {
+                       m->flags |= MAN_BLINE;
+                       return(1);
+               }
+       }

 
 

+       if ( ! rew_scope(MAN_HEAD, m, tok))
+               return(0);

        return(man_body_alloc(m, line, ppos, tok));
 }
 
        return(man_body_alloc(m, line, ppos, tok));
 }
 


@@ -298,19 +503,33 @@ in_line_eoln(MACRO_PROT_ARGS)
                        return(0);
                if (0 == w)
                        break;
                        return(0);
                if (0 == w)
                        break;

-

                if ( ! man_word_alloc(m, line, la, p))
                        return(0);
        }
 
                if ( ! man_word_alloc(m, line, la, p))
                        return(0);
        }
 

-       if (n == m->last && (MAN_SCOPED & man_macros[tok].flags)) {
+       /*
+        * If no arguments are specified and this is MAN_SCOPED (i.e.,
+        * next-line scoped), then set our mode to indicate that we're
+        * waiting for terms to load into our context.
+        */
+
+       if (n == m->last && MAN_SCOPED & man_macros[tok].flags) {
+               assert( ! (MAN_NSCOPED & man_macros[tok].flags));

                m->flags |= MAN_ELINE;
                return(1);
        } 
 
                m->flags |= MAN_ELINE;
                return(1);
        } 
 

+       /* Set ignorable context, if applicable. */
+
+       if (MAN_NSCOPED & man_macros[tok].flags) {
+               assert( ! (MAN_SCOPED & man_macros[tok].flags));
+               m->flags |= MAN_ILINE;
+       }
+       

        /*
        /*

-        * Note that when TH is pruned, we'll be back at the root, so
-        * make sure that we don't clobber as its sibling.
+        * Rewind our element scope.  Note that when TH is pruned, we'll
+        * be back at the root, so make sure that we don't clobber as
+        * its sibling.

         */
 
        for ( ; m->last; m->last = m->last->parent) {
         */
 
        for ( ; m->last; m->last = m->last->parent) {


@@ -334,8 +553,9 @@ in_line_eoln(MACRO_PROT_ARGS)
                return(0);
        if (m->last->type != MAN_ROOT && ! man_action_post(m))
                return(0);
                return(0);
        if (m->last->type != MAN_ROOT && ! man_action_post(m))
                return(0);

-       if (m->last->type != MAN_ROOT)
-               m->next = MAN_NEXT_SIBLING;
+
+       m->next = MAN_ROOT == m->last->type ?
+               MAN_NEXT_CHILD : MAN_NEXT_SIBLING;

 
        return(1);
 }
 
        return(1);
 }


@@ -344,19 +564,7 @@ in_line_eoln(MACRO_PROT_ARGS)
 int
 man_macroend(struct man *m)
 {
 int
 man_macroend(struct man *m)
 {

-       struct man_node *n;
-
-       n = MAN_VALID & m->last->flags ?
-               m->last->parent : m->last;
-
-       for ( ; n; n = n->parent) {
-               if (MAN_BLOCK != n->type)
-                       continue;
-               if ( ! (MAN_EXPLICIT & man_macros[n->tok].flags))
-                       continue;
-               return(man_nerr(m, n, WEXITSCOPE));
-       }

 
 

-       return(man_unscope(m, m->first));
+       return(man_unscope(m, m->first, WEXITSCOPE));

 }
 
 }