Fully skip first and last line for the purpose of cmp.

[mandoc.git] / html.c
diff --git a/html.c b/html.c

index 70641b72d4d62f52a993e3e4a25858ccb5580706..0ad84e4f99d3c7bef88ca28f9ff4d3595a1dee14 100644 (file)
--- a/html.c
+++ b/html.c
@@ -1,4 +1,4 @@
-/*     $Id: html.c,v 1.93 2010/01/29 14:39:37 kristaps Exp $ */
+/*     $Id: html.c,v 1.99 2010/04/12 19:45:39 kristaps Exp $ */
  /*
   * Copyright (c) 2008, 2009 Kristaps Dzonsons <kristaps@kth.se>
   *
@@ -54,7 +54,7 @@ static        const struct htmldata htmltags[TAG_MAX] = {
         {"h1",          0}, /* TAG_H1 */
         {"h2",          0}, /* TAG_H2 */
         {"span",        0}, /* TAG_SPAN */
-       {"link",        HTML_CLRLINE | HTML_NOSTACK}, /* TAG_LINK */
+       {"link",        HTML_CLRLINE | HTML_NOSTACK | HTML_AUTOCLOSE}, /* TAG_LINK */
         {"br",          HTML_CLRLINE | HTML_NOSTACK | HTML_AUTOCLOSE}, /* TAG_BR */
         {"a",           0}, /* TAG_A */
         {"table",       HTML_CLRLINE}, /* TAG_TABLE */
@@ -96,6 +96,8 @@ static        void              print_doctype(struct html *);
  static void              print_xmltype(struct html *);
  static int               print_encode(struct html *, const char *, int);
  static void              print_metaf(struct html *, enum roffdeco);
+static void              print_attr(struct html *, 
+                               const char *, const char *);
  static void             *ml_alloc(char *, enum htmltype);
  
  
@@ -351,6 +353,15 @@ print_encode(struct html *h, const char *p, int norecurse)
  }
  
  
+static void
+print_attr(struct html *h, const char *key, const char *val)
+{
+       printf(" %s=\"", key);
+       (void)print_encode(h, val, 1);
+       putchar('\"');
+}
+
+
  struct tag *
  print_otag(struct html *h, enum htmltag tag, 
                 int sz, const struct htmlpair *p)
@@ -358,6 +369,8 @@ print_otag(struct html *h, enum htmltag tag,
         int              i;
         struct tag      *t;
  
+       /* Push this tags onto the stack of open scopes. */
+
         if ( ! (HTML_NOSTACK & htmltags[tag].flags)) {
                 t = malloc(sizeof(struct tag));
                 if (NULL == t) {
@@ -374,14 +387,22 @@ print_otag(struct html *h, enum htmltag tag,
                 if ( ! (HTML_CLRLINE & htmltags[tag].flags))
                         putchar(' ');
  
+       /* Print out the tag name and attributes. */
+
         printf("<%s", htmltags[tag].name);
-       for (i = 0; i < sz; i++) {
-               printf(" %s=\"", htmlattrs[p[i].key]);
-               assert(p->val);
-               (void)print_encode(h, p[i].val, 1);
-               putchar('\"');
+       for (i = 0; i < sz; i++)
+               print_attr(h, htmlattrs[p[i].key], p[i].val);
+
+       /* Add non-overridable attributes. */
+
+       if (TAG_HTML == tag && HTML_XHTML_1_0_STRICT == h->type) {
+               print_attr(h, "xmlns", "http://www.w3.org/1999/xhtml");
+               print_attr(h, "xml:lang", "en");
+               print_attr(h, "lang", "en");
         }
  
+       /* Accomodate for XML "well-formed" singleton escaping. */
+
         if (HTML_AUTOCLOSE & htmltags[tag].flags)
                 switch (h->type) {
                 case (HTML_XHTML_1_0_STRICT):
@@ -445,19 +466,23 @@ print_doctype(struct html *h)
  {
         const char      *doctype;
         const char      *dtd;
+       const char      *name;
  
         switch (h->type) {
         case (HTML_HTML_4_01_STRICT):
+               name = "HTML";
                 doctype = "-//W3C//DTD HTML 4.01//EN";
                 dtd = "http://www.w3.org/TR/html4/strict.dtd";
                 break;
         default:
+               name = "html";
                 doctype = "-//W3C//DTD XHTML 1.0 Strict//EN";
                 dtd = "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd";
                 break;
         }
  
-       printf("<!DOCTYPE HTML PUBLIC \"%s\" \"%s\">\n", doctype, dtd);
+       printf("<!DOCTYPE %s PUBLIC \"%s\" \"%s\">\n", 
+                       name, doctype, dtd);
  }
  
  
@@ -482,8 +507,6 @@ print_text(struct html *h, const char *p)
                 case(')'):
                         /* FALLTHROUGH */
                 case(']'):
-                       /* FALLTHROUGH */
-               case('}'):
                         if ( ! (HTML_IGNDELIM & h->flags))
                                 h->flags |= HTML_NOSPACE;
                         break;
@@ -498,13 +521,15 @@ print_text(struct html *h, const char *p)
         if ( ! print_encode(h, p, 0))
                 h->flags &= ~HTML_NOSPACE;
  
+       /* 
+        * Note that we don't process the pipe: the parser sees it as
+        * punctuation, but we don't in terms of typography.
+        */
         if (*p && 0 == *(p + 1))
                 switch (*p) {
                 case('('):
                         /* FALLTHROUGH */
                 case('['):
-                       /* FALLTHROUGH */
-               case('{'):
                         h->flags |= HTML_NOSPACE;
                         break;
                 default: