Considerably improved roff_text parser.

[mandoc.git] / xml.c

diff --git a/xml.c b/xml.c
index 2e87ae44254bd8337fdccc0b4b776e0bc2056602..74afaa06d3de1663d6453e9cc3ee4bc92624c0dd 100644 (file)
--- a/xml.c
+++ b/xml.c
@@ -1,4 +1,4 @@
-/* $Id: xml.c,v 1.1 2008/11/30 21:41:35 kristaps Exp $ */
+/* $Id: xml.c,v 1.4 2008/12/01 15:32:36 kristaps Exp $ */

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


@@ -40,7 +40,8 @@ enum  md_tok {
        MD_BLKOUT,
        MD_IN,
        MD_OUT,
        MD_BLKOUT,
        MD_IN,
        MD_OUT,

-       MD_TEXT
+       MD_TEXT,
+       MD_OVERRIDE

 };
 
 struct md_xml {
 };
 
 struct md_xml {


@@ -74,6 +75,9 @@ static        int              mbuf_putstring(struct md_xml *,
                                const char *);
 static int              mbuf_nputstring(struct md_xml *, 
                                const char *, size_t);
                                const char *);
 static int              mbuf_nputstring(struct md_xml *, 
                                const char *, size_t);

+static int              mbuf_puts(struct md_xml *, const char *);
+static int              mbuf_nputs(struct md_xml *, 
+                               const char *, size_t);

 
 
 static int
 
 
 static int


@@ -86,6 +90,34 @@ mbuf_putstring(struct md_xml *p, const char *buf)
 
 static int
 mbuf_nputstring(struct md_xml *p, const char *buf, size_t sz)
 
 static int
 mbuf_nputstring(struct md_xml *p, const char *buf, size_t sz)

+{
+       size_t           i;
+
+       for (i = 0; i < sz; i++) {
+               switch (buf[i]) {
+               case ('&'):
+                       if ( ! md_buf_puts(p->mbuf, "&amp;", 5))
+                               return(0);
+                       p->pos += 5;
+                       break;
+               case ('"'):
+                       if ( ! md_buf_puts(p->mbuf, "&quot;", 6))
+                               return(0);
+                       p->pos += 6;
+                       break;
+               default:
+                       if ( ! md_buf_putchar(p->mbuf, buf[i]))
+                               return(0);
+                       p->pos++;
+                       break;
+               }
+       }
+       return(1);
+}
+
+
+static int
+mbuf_nputs(struct md_xml *p, const char *buf, size_t sz)

 {
 
        p->pos += sz;
 {
 
        p->pos += sz;


@@ -93,6 +125,14 @@ mbuf_nputstring(struct md_xml *p, const char *buf, size_t sz)
 }
 
 
 }
 
 

+static int
+mbuf_puts(struct md_xml *p, const char *buf)
+{
+
+       return(mbuf_nputs(p, buf, strlen(buf)));
+}
+
+

 static int
 mbuf_indent(struct md_xml *p)
 {
 static int
 mbuf_indent(struct md_xml *p)
 {


@@ -169,7 +209,7 @@ mbuf_data(struct md_xml *p, int space, char *buf)
                        if ( ! mbuf_indent(p))
                                return(0);
                } else if (space) {
                        if ( ! mbuf_indent(p))
                                return(0);
                } else if (space) {

-                       if ( ! mbuf_nputstring(p, " ", 1))
+                       if ( ! mbuf_nputs(p, " ", 1))

                                return(0);
                }
 
                                return(0);
                }
 


@@ -253,10 +293,12 @@ roffhead(void *arg)
        assert(arg);
        p = (struct md_xml *)arg;
 
        assert(arg);
        p = (struct md_xml *)arg;
 

-       if ( ! mbuf_putstring(p, "<?xml version=\"1.0\" "
+       if ( ! mbuf_puts(p, "<?xml version=\"1.0\" "

                                "encoding=\"UTF-8\"?>\n"))
                return(0);
                                "encoding=\"UTF-8\"?>\n"))
                return(0);

-       if ( ! mbuf_nputstring(p, "<block:mdoc>", 12))
+       if ( ! mbuf_puts(p, "<mdoc xmlns:block=\"block\" "
+                               "xmlns:special=\"special\" "
+                               "xmlns:inline=\"inline\">"))

                return(0);
 
        p->indent++;
                return(0);
 
        p->indent++;


@@ -276,7 +318,7 @@ rofftail(void *arg)
        if (0 != p->pos && ! mbuf_newline(p))
                return(0);
 
        if (0 != p->pos && ! mbuf_newline(p))
                return(0);
 

-       if ( ! mbuf_nputstring(p, "</block:mdoc>", 13))
+       if ( ! mbuf_puts(p, "</mdoc>"))

                return(0);
 
        p->last = MD_BLKOUT;
                return(0);
 
        p->last = MD_BLKOUT;


@@ -288,8 +330,19 @@ rofftail(void *arg)
 static int
 roffspecial(void *arg, int tok)
 {
 static int
 roffspecial(void *arg, int tok)
 {

+       struct md_xml   *p;
+
+       assert(arg);
+       p = (struct md_xml *)arg;
+
+       switch (tok) {
+       case (ROFF_Ns):
+               p->last = MD_OVERRIDE;
+               break;
+       default:
+               break;
+       }

 
 

-       /* FIXME */

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


@@ -311,27 +364,29 @@ roffblkin(void *arg, int tok, int *argc, char **argv)
        } else if ( ! mbuf_indent(p))
                return(0);
 
        } else if ( ! mbuf_indent(p))
                return(0);
 

-       if ( ! mbuf_nputstring(p, "<", 1))
+       if ( ! mbuf_nputs(p, "<", 1))

                return(0);
                return(0);

-       if ( ! mbuf_nputstring(p, "block:", 6))
+       if ( ! mbuf_nputs(p, "block:", 6))

                return(0);
                return(0);

-       if ( ! mbuf_putstring(p, toknames[tok]))
+       if ( ! mbuf_puts(p, toknames[tok]))

                return(0);
 
                return(0);
 

+       /* FIXME: xml won't like standards args (e.g., p1003.1-90). */
+

        for (i = 0; ROFF_ARGMAX != argc[i]; i++) {
        for (i = 0; ROFF_ARGMAX != argc[i]; i++) {

-               if ( ! mbuf_nputstring(p, " ", 1))
+               if ( ! mbuf_nputs(p, " ", 1))

                        return(0);
                        return(0);

-               if ( ! mbuf_putstring(p, tokargnames[argc[i]]))
+               if ( ! mbuf_puts(p, tokargnames[argc[i]]))

                        return(0);
                        return(0);

-               if ( ! mbuf_nputstring(p, "=\"", 2))
+               if ( ! mbuf_nputs(p, "=\"", 2))

                        return(0);
                if ( ! mbuf_putstring(p, argv[i] ? argv[i] : "true"))
                        return(0);
                        return(0);
                if ( ! mbuf_putstring(p, argv[i] ? argv[i] : "true"))
                        return(0);

-               if ( ! mbuf_nputstring(p, "\"", 1))
+               if ( ! mbuf_nputs(p, "\"", 1))

                        return(0);
        }
 
                        return(0);
        }
 

-       if ( ! mbuf_nputstring(p, ">", 1))
+       if ( ! mbuf_nputs(p, ">", 1))

                return(0);
 
        p->last = MD_BLKIN;
                return(0);
 
        p->last = MD_BLKIN;


@@ -358,13 +413,13 @@ roffblkout(void *arg, int tok)
        } else if ( ! mbuf_indent(p))
                return(0);
 
        } else if ( ! mbuf_indent(p))
                return(0);
 

-       if ( ! mbuf_nputstring(p, "</", 2))
+       if ( ! mbuf_nputs(p, "</", 2))

                return(0);
                return(0);

-       if ( ! mbuf_nputstring(p, "block:", 6))
+       if ( ! mbuf_nputs(p, "block:", 6))

                return(0);
                return(0);

-       if ( ! mbuf_putstring(p, toknames[tok]))
+       if ( ! mbuf_puts(p, toknames[tok]))

                return(0);
                return(0);

-       if ( ! mbuf_nputstring(p, ">", 1))
+       if ( ! mbuf_nputs(p, ">", 1))

                return(0);
 
        p->last = MD_BLKOUT;
                return(0);
 
        p->last = MD_BLKOUT;


@@ -396,7 +451,7 @@ roffin(void *arg, int tok, int *argc, char **argv)
                case (MD_TEXT):
                        /* FALLTHROUGH */
                case (MD_OUT):
                case (MD_TEXT):
                        /* FALLTHROUGH */
                case (MD_OUT):

-                       if ( ! mbuf_nputstring(p, " ", 1))
+                       if ( ! mbuf_nputs(p, " ", 1))

                                return(0);
                        break;
                default:
                                return(0);
                        break;
                default:


@@ -407,26 +462,26 @@ roffin(void *arg, int tok, int *argc, char **argv)
 
        p->last = MD_IN;
 
 
        p->last = MD_IN;
 

-       if ( ! mbuf_nputstring(p, "<", 1))
+       if ( ! mbuf_nputs(p, "<", 1))

                return(0);
                return(0);

-       if ( ! mbuf_nputstring(p, "inline:", 7))
+       if ( ! mbuf_nputs(p, "inline:", 7))

                return(0);
                return(0);

-       if ( ! mbuf_putstring(p, toknames[tok]))
+       if ( ! mbuf_puts(p, toknames[tok]))

                return(0);
 
        for (i = 0; ROFF_ARGMAX != argc[i]; i++) {
                return(0);
 
        for (i = 0; ROFF_ARGMAX != argc[i]; i++) {

-               if ( ! mbuf_nputstring(p, " ", 1))
+               if ( ! mbuf_nputs(p, " ", 1))

                        return(0);
                        return(0);

-               if ( ! mbuf_putstring(p, tokargnames[argc[i]]))
+               if ( ! mbuf_puts(p, tokargnames[argc[i]]))

                        return(0);
                        return(0);

-               if ( ! mbuf_nputstring(p, "=\"", 2))
+               if ( ! mbuf_nputs(p, "=\"", 2))

                        return(0);
                if ( ! mbuf_putstring(p, argv[i] ? argv[i] : "true"))
                        return(0);
                        return(0);
                if ( ! mbuf_putstring(p, argv[i] ? argv[i] : "true"))
                        return(0);

-               if ( ! mbuf_nputstring(p, "\"", 1))
+               if ( ! mbuf_nputs(p, "\"", 1))

                        return(0);
        }
                        return(0);
        }

-       return(mbuf_nputstring(p, ">", 1));
+       return(mbuf_nputs(p, ">", 1));

 }
 
 
 }
 
 


@@ -438,18 +493,20 @@ roffout(void *arg, int tok)
        assert(arg);
        p = (struct md_xml *)arg;
 
        assert(arg);
        p = (struct md_xml *)arg;
 

+       /* Continue with a regular out token. */
+

        if (0 == p->pos && ! mbuf_indent(p))
                return(0);
 
        p->last = MD_OUT;
 
        if (0 == p->pos && ! mbuf_indent(p))
                return(0);
 
        p->last = MD_OUT;
 

-       if ( ! mbuf_nputstring(p, "</", 2))
+       if ( ! mbuf_nputs(p, "</", 2))

                return(0);
                return(0);

-       if ( ! mbuf_nputstring(p, "inline:", 7))
+       if ( ! mbuf_nputs(p, "inline:", 7))

                return(0);
                return(0);

-       if ( ! mbuf_putstring(p, toknames[tok]))
+       if ( ! mbuf_puts(p, toknames[tok]))

                return(0);
                return(0);

-       return(mbuf_nputstring(p, ">", 1));
+       return(mbuf_nputs(p, ">", 1));

 }
 
 
 }
 
 


@@ -500,3 +557,4 @@ roffdata(void *arg, int space, char *buf)
        p->last = MD_TEXT;
        return(1);
 }
        p->last = MD_TEXT;
        return(1);
 }

+