snapshot of project "lynx", label v2-9-0dev_10h

6 files changed, 42 insertions, 17 deletions

diff --git a/CHANGES b/CHANGES
index 2b0d937d..4e816cc0 100644
--- a/CHANGES
+++ b/CHANGES
@@ -1,9 +1,11 @@
--- $LynxId: CHANGES,v 1.1112 2022/06/12 17:05:20 tom Exp $
+-- $LynxId: CHANGES,v 1.1113 2022/06/13 00:23:57 tom Exp $
 ===============================================================================
 Changes since Lynx 2.8 release
 ===============================================================================
 
 2022-06-12 (2.9.0dev.11)
+* add presentation type for xhtml, related state information to better handle
+  things such as "<script src=foo />" -Sholmi Fish, TD
 * improve line-breaks and whitespace for Japanese characters -KH
   + permit line breaks after any Japanese character (enabled by
     --enable-wcwidth-support configuration and only called on last byte of
diff --git a/WWW/Library/Implementation/SGML.c b/WWW/Library/Implementation/SGML.c
index d120644b..2534606e 100644
--- a/WWW/Library/Implementation/SGML.c
+++ b/WWW/Library/Implementation/SGML.c
@@ -1,5 +1,5 @@
 /*
- * $LynxId: SGML.c,v 1.182 2021/10/24 22:18:29 tom Exp $
+ * $LynxId: SGML.c,v 1.183 2022/06/13 00:20:50 tom Exp $
  *
  *			General SGML Parser code		SGML.c
  *			========================
@@ -1621,7 +1621,8 @@ static BOOL ignore_when_empty(HTTag * tag)
 	&& !(tag->flags & Tgf_mafse)
 	&& tag->contents != SGML_EMPTY
 	&& tag->tagclass != Tgc_Plike
-	&& (tag->tagclass == Tgc_SELECTlike
+	&& (tag->tagclass == Tgc_APPLETlike
+	    || tag->tagclass == Tgc_SELECTlike
 	    || (tag->contains && tag->icontains))) {
 	result = TRUE;
     }
@@ -3644,9 +3645,9 @@ static void SGML_character(HTStream *me, int c_in)
     case S_attr:
 	if (WHITE(c) || (c == '>') || (c == '=')) {	/* End of word */
 	    if ((c == '>')
-		&& (string->size == 1)
-		&& (string->data[0] == '/')) {
-		if (me->extended_html
+		&& (string->size >= 1)
+		&& (string->data[string->size - 1] == '/')) {
+		if ((LYxhtml_parsing || me->extended_html)
 		    && ignore_when_empty(me->current_tag)) {
 		    discard_empty(me);
 		}
@@ -4590,7 +4591,8 @@ const HTStreamClass SGMLParser =
 
 HTStream *SGML_new(const SGML_dtd * dtd,
 		   HTParentAnchor *anchor,
-		   HTStructured * target)
+		   HTStructured * target,
+		   int extended_html)
 {
     HTStream *me = typecalloc(struct _HTStream);
 
@@ -4660,6 +4662,10 @@ HTStream *SGML_new(const SGML_dtd * dtd,
 	sgml_in_psrc_was_initialized = TRUE;
     }
 #endif
+    if (extended_html)
+    {
+        me->extended_html = TRUE;
+    }
 
     sgml_offset = 0;
     return me;
diff --git a/WWW/Library/Implementation/SGML.h b/WWW/Library/Implementation/SGML.h
index 8b6445f6..7da79762 100644
--- a/WWW/Library/Implementation/SGML.h
+++ b/WWW/Library/Implementation/SGML.h
@@ -1,5 +1,5 @@
 /*
- * $LynxId: SGML.h,v 1.47 2021/07/22 23:34:13 tom Exp $
+ * $LynxId: SGML.h,v 1.48 2022/06/12 20:19:04 tom Exp $
  *			       SGML parse and stream definition for libwww
  *                             SGML AND STRUCTURED STREAMS
  *
@@ -281,7 +281,9 @@ Create an SGML parser
  */
     extern HTStream *SGML_new(const SGML_dtd * dtd,
 			      HTParentAnchor *anchor,
-			      HTStructured * target);
+			      HTStructured * target,
+			      int extended_html)
+        ;
 
     extern const HTStreamClass SGMLParser;
 
diff --git a/src/HTInit.c b/src/HTInit.c
index 4aa38e0c..466bc722 100644
--- a/src/HTInit.c
+++ b/src/HTInit.c
@@ -1,5 +1,5 @@
 /*
- * $LynxId: HTInit.c,v 1.97 2022/04/01 07:55:02 tom Exp $
+ * $LynxId: HTInit.c,v 1.98 2022/06/12 21:17:37 tom Exp $
  *
  *		Configuration-specific Initialization		HTInit.c
  *		----------------------------------------
@@ -178,7 +178,7 @@ void HTFormatInit(void)
      * application/xhtml+xml
      * text/html
      */
-    SET_INTERNL("application/xhtml+xml", STR_PRESENT, HTMLPresent, 1.0);
+    SET_INTERNL("application/xhtml+xml", STR_PRESENT, XHTMLPresent, 1.0);
     SET_INTERNL("application/xhtml+xml", STR_SOURCE, HTPlainPresent, 1.0);
     SET_INTERNL("text/css", STR_PRESENT, HTPlainPresent, 1.0);
     SET_INTERNL(STR_HTML, STR_PRESENT, HTMLPresent, 1.0);
diff --git a/src/HTML.c b/src/HTML.c
index 14dfbc56..75c2bf91 100644
--- a/src/HTML.c
+++ b/src/HTML.c
@@ -1,5 +1,5 @@
 /*
- * $LynxId: HTML.c,v 1.198 2022/06/12 16:48:05 tom Exp $
+ * $LynxId: HTML.c,v 1.199 2022/06/13 00:23:57 tom Exp $
  *
  *		Structured stream to Rich hypertext converter
  *		============================================
@@ -7963,7 +7963,7 @@ HTStream *HTMLToPlain(HTPresentation *pres,
     CTRACE((tfp, "HTMLToPlain calling CacheThru_new\n"));
     return CacheThru_new(anchor,
 			 SGML_new(&HTML_dtd, anchor,
-				  HTML_new(anchor, pres->rep_out, sink)));
+				  HTML_new(anchor, pres->rep_out, sink), FALSE));
 }
 
 /*	HTConverter for HTML source to plain text
@@ -8026,7 +8026,7 @@ HTStream *HTMLParsedPresent(HTPresentation *pres,
     CTRACE((tfp, "HTMLParsedPresent calling CacheThru_new\n"));
     return CacheThru_new(anchor,
 			 SGML_new(&HTML_dtd, anchor,
-				  HTMLGenerator(intermediate)));
+				  HTMLGenerator(intermediate), FALSE));
 }
 
 /*	HTConverter for HTML to C code
@@ -8054,7 +8054,7 @@ HTStream *HTMLToC(HTPresentation *pres GCC_UNUSED,
 	HTML_put_string(html, html->comment_start);
     CTRACE((tfp, "HTMLToC calling CacheThru_new\n"));
     return CacheThru_new(anchor,
-			 SGML_new(&HTML_dtd, anchor, html));
+			 SGML_new(&HTML_dtd, anchor, html, FALSE));
 }
 
 /*	Presenter for HTML
@@ -8073,7 +8073,17 @@ HTStream *HTMLPresent(HTPresentation *pres GCC_UNUSED,
     CTRACE((tfp, "HTMLPresent calling CacheThru_new\n"));
     return CacheThru_new(anchor,
 			 SGML_new(&HTML_dtd, anchor,
-				  HTML_new(anchor, WWW_PRESENT, NULL)));
+				  HTML_new(anchor, WWW_PRESENT, NULL), FALSE));
+}
+
+HTStream *XHTMLPresent(HTPresentation *pres GCC_UNUSED,
+		      HTParentAnchor *anchor,
+		      HTStream *sink GCC_UNUSED)
+{
+    CTRACE((tfp, "XHTMLPresent calling CacheThru_new\n"));
+    return CacheThru_new(anchor,
+			 SGML_new(&HTML_dtd, anchor,
+				  HTML_new(anchor, WWW_PRESENT, NULL), TRUE));
 }
 #endif /* !GUI */
 
diff --git a/src/HTML.h b/src/HTML.h
index 9f5d1d5d..2bc040fe 100644
--- a/src/HTML.h
+++ b/src/HTML.h
@@ -1,5 +1,5 @@
 /*
- * $LynxId: HTML.h,v 1.34 2019/08/23 23:18:06 tom Exp $
+ * $LynxId: HTML.h,v 1.35 2022/06/12 20:19:04 tom Exp $
  *
  *					HTML to rich text converter for libwww
  *
@@ -251,6 +251,11 @@ extern "C" {
 				 HTParentAnchor *anchor,
 				 HTStream *sink);
 
+    extern HTStream *XHTMLPresent(HTPresentation *pres,
+				 HTParentAnchor *anchor,
+				 HTStream *sink);
+
+
     extern HTStructured *HTML_new(HTParentAnchor *anchor,
 				  HTFormat format_out,
 				  HTStream *target);