snapshot of project "lynx", label v2-7-1ac_0-93

11 files changed, 381 insertions, 117 deletions

diff --git a/WWW/Library/Implementation/HTAccess.c b/WWW/Library/Implementation/HTAccess.c
index 6dedbeb9..879b34c9 100644
--- a/WWW/Library/Implementation/HTAccess.c
+++ b/WWW/Library/Implementation/HTAccess.c
@@ -632,6 +632,8 @@ PUBLIC HTStream *HTSaveStream ARGS1(
     return (*p->saveStream)(anchor);
 }
 
+extern char LYinternal_flag;		       /* from LYMainLoop.c */
+
 /*	Load a document - with logging etc		HTLoadDocument()
 **	----------------------------------
 **
@@ -695,6 +697,26 @@ PRIVATE BOOL HTLoadDocument ARGS4(
     }
 
     /*
+     *  If this is marked as an internal link but we don't have the
+     *  document loaded any more, and we haven't explicitly flagged
+     *  that we want to reload with LYforce_no_cache, then something
+     *  has disappeared from the cache when we expected it to be still
+     *  there.  The user probably doesn't expect a new network access.
+     *  So if we have POST data and safe is not set in the anchor,
+     *  ask for confirmation, and fail if not granted.  The exception
+     *  are LYNXIMGMAP documents, for which we defer to LYLoadIMGmap
+     *  for prompting if necessary. - kw
+     */
+    if (LYinternal_flag && !LYforce_no_cache &&
+	anchor->post_data && !anchor->safe &&
+	(text = (HText *)HTAnchor_document(anchor)) == NULL &&
+	strncmp(full_address, "LYNXIMGMAP:", 11) &&
+	HTConfirm("Document with POST content not found in cache.  Resubmit?")
+	!= TRUE) {
+	return NO;
+    }
+
+    /*
     **  If we don't have POST content, check whether this is a previous
     **  redirecting URL, and keep re-checking until we get to the final
     **  destination or redirection limit.  If we do have POST content,
@@ -783,13 +805,16 @@ PRIVATE BOOL HTLoadDocument ARGS4(
 	*    testing whether we are just repositioning.  For an internal
 	*    link, the potential callers of this function from mainloop()
 	*    down will either avoid making the call (and do the repositioning
-	*    differently) or set LYoverride_no_cache.
+	*    differently) or set LYinternal_flag (or LYoverride_no_cache).
 	*    Note that (a) LYNXIMGMAP pseudo-documents and (b) The "List Page"
 	*    document are treated logically as being part of the document on
 	*    which they are based, for the purpose of whether to treat a link
-	*    as internal, but the logic for this (by setting LYoverride_no_cache
-	*    as necessary) is implemented elsewhere.  For LYNXIMGMAP the same
-	*    caveat as above applies.
+	*    as internal, but the logic for this (by setting LYinternal_flag
+	*    as necessary) is implemented elsewhere.  There is a specific
+	*    test for LYNXIMGMAP here so that the generated pseudo-document
+	*    will not be re-used unless LYoverride_no_cache is set.  The same
+	*    caveat as above applies w.r.t. reloading of the underlying
+	*    resource.
 	*
 	**  We also should be checking other aspects of cache
 	**  regulation (e.g., based on an If-Modified-Since check,
@@ -800,7 +825,9 @@ PRIVATE BOOL HTLoadDocument ARGS4(
 	if (LYoverride_no_cache || !HText_hasNoCacheSet(text) ||
 	    !HText_AreDifferent(anchor, full_address))
 #else
-	if (LYoverride_no_cache || !HText_hasNoCacheSet(text))
+	if (LYoverride_no_cache ||
+	    ((LYinternal_flag || !HText_hasNoCacheSet(text)) &&
+	     strncmp(full_address, "LYNXIMGMAP:", 11)))
 #endif /* TRACK_INTERNAL_LINKS */
 	{
             if (TRACE)
@@ -814,7 +841,12 @@ PRIVATE BOOL HTLoadDocument ARGS4(
 	    redirection_attempts = 0;
 	    return YES;
 	} else {
+#if NOT_USED_CODE
+	    /* disabled 1997-10-28 - kw
+	       callers already do this when requested
+	    */
 	    reloading = TRUE;
+#endif
 	    ForcingNoCache = YES;
             if (TRACE)
 	        fprintf(stderr, "HTAccess: Auto-reloading document.\n");
diff --git a/WWW/Library/Implementation/HTAnchor.c b/WWW/Library/Implementation/HTAnchor.c
index d3c3f34e..c2cf273c 100644
--- a/WWW/Library/Implementation/HTAnchor.c
+++ b/WWW/Library/Implementation/HTAnchor.c
@@ -238,11 +238,21 @@ PUBLIC HTChildAnchor * HTAnchor_findChildAndLink ARGS4(
         HTAnchor * dest;
 
         parsed_doc.address = HTParse(href, relative_to, PARSE_ALL);
-        parsed_doc.post_data = NULL;
-        parsed_doc.post_content_type = NULL;
-        parsed_doc.bookmark = NULL;
-        parsed_doc.isHEAD = FALSE;
-        parsed_doc.safe = FALSE;
+#ifndef DONT_TRACK_INTERNAL_LINKS
+	if (ltype && parent->post_data && ltype == LINK_INTERNAL) {
+	    /* for internal links, find a destination with the same
+	       post data if the source of the link has post data. - kw */
+	    parsed_doc.post_data = parent->post_data;
+	    parsed_doc.post_content_type = parent->post_content_type;
+	} else
+#endif
+	{
+	    parsed_doc.post_data = NULL;
+	    parsed_doc.post_content_type = NULL;
+	}
+	parsed_doc.bookmark = NULL;
+	parsed_doc.isHEAD = FALSE;
+	parsed_doc.safe = FALSE;
         dest = HTAnchor_findAddress(&parsed_doc);
 
         HTAnchor_link((HTAnchor *)child, dest, ltype);
@@ -690,6 +700,8 @@ PUBLIC BOOL HTAnchor_delete ARGS1(
 #ifdef EXP_CHARTRANS
     FREE (me->UCStages);
 #endif
+    ImageMapList_free(me->imaps);
+
 
     /*
      *  Finally, kill the parent anchor passed in.
diff --git a/WWW/Library/Implementation/HTAnchor.h b/WWW/Library/Implementation/HTAnchor.h
index 458cac40..b86c372c 100644
--- a/WWW/Library/Implementation/HTAnchor.h
+++ b/WWW/Library/Implementation/HTAnchor.h
@@ -125,9 +125,8 @@ struct _HTParentAnchor {
   char *	expires;		/* Expires */
   char *	last_modified;		/* Last-Modified */
   char *	server;			/* Server */
-#ifdef EXP_CHARTRANS
-  UCAnchorInfo *	UCStages;
-#endif
+  UCAnchorInfo *	UCStages; 	/* chartrans stages */
+  HTList *	imaps;			/* client side image maps */
 };
 
 typedef struct {
@@ -411,7 +410,6 @@ extern void HTAnchor_setPhysical PARAMS((
 
 #endif /* HTANCHOR_H */
 
-#ifdef EXP_CHARTRANS
 
 extern LYUCcharset * HTAnchor_getUCInfoStage PARAMS((HTParentAnchor * me,
 						     int which_stage));
@@ -433,7 +431,8 @@ extern LYUCcharset * HTAnchor_copyUCInfoStage PARAMS((HTParentAnchor * me,
 						     int to_stage,
 						     int from_stage,
 						     int set_by));
-#endif
+
+extern void ImageMapList_free PARAMS((HTList * list));
 /*
 
     */
diff --git a/WWW/Library/Implementation/HTDOS.c b/WWW/Library/Implementation/HTDOS.c
index c5687808..1e67a1a0 100644
--- a/WWW/Library/Implementation/HTDOS.c
+++ b/WWW/Library/Implementation/HTDOS.c
@@ -51,7 +51,7 @@ char * HTDOS_wwwName (char *dosname)
 ** ON EXIT:
 **      returns         dos file specification
 **
-** Bug:	Returns pointer to static -- non-reentrant
+** Bug(?):	Returns pointer to input string, which is modified
 */
 char * HTDOS_name(char *dosname)	{
 
@@ -82,13 +82,15 @@ char * HTDOS_name(char *dosname)	{
 		printf("\n\n%s = i%\n\n",cp_url,strlen(cp_url));
 		sleep(5);
 /**/
-		return(cp_url);
+      strcpy(dosname, cp_url);
+      return(dosname); /* return(cp_url); */
 	} else {
 /*
 		printf("\n\n%s = %i\n\n",cp_url+1,strlen(cp_url));
 		sleep(5);
 /**/
-		return(cp_url+1);
+		strcpy(dosname, cp_url+1);
+      return(dosname); /* return(cp_url+1);  */
 	}
 }
 
diff --git a/WWW/Library/Implementation/HTFile.c b/WWW/Library/Implementation/HTFile.c
index 3ec307fc..45bd6464 100644
--- a/WWW/Library/Implementation/HTFile.c
+++ b/WWW/Library/Implementation/HTFile.c
@@ -74,11 +74,9 @@
 #include "HTBTree.h"
 #include "HTAlert.h"
 #include "HTCJK.h"
-#ifdef EXP_CHARTRANS
 #include "UCDefs.h"
 #include "UCMap.h"
 #include "UCAux.h"
-#endif /* EXP_CHARTRANS */
 
 #include "LYexit.h"
 #include "LYLeaks.h"
@@ -130,11 +128,9 @@ PUBLIC int HTDirReadme = HT_DIR_README_TOP;
 
 extern int current_char_set;
 extern CONST char *LYchar_set_names[];
+extern BOOLEAN LYRawMode;
 extern BOOL HTPassEightBitRaw;
 extern HTCJKlang HTCJK;
-#ifndef EXP_CHARTRANS
-#define UCLYhndl_HTFile_for_unspec 0 /* a dummy define */
-#endif
 
 PRIVATE char *HTMountRoot = "/Net/";		/* Where to find mounts */
 #ifdef VMS
@@ -573,7 +569,16 @@ PUBLIC char * HTLocalName ARGS1(
 	    if (TRACE)
 	        fprintf(stderr, "Node `%s' means path `%s'\n", name, path);
 #ifdef DOSPATH
-		 return(HTDOS_name(path));
+	    {
+		char *ret_path = NULL;
+		StrAllocCopy(ret_path, HTDOS_name(path));
+		if (TRACE) {
+		    fprintf(stderr, "HTDOS_name changed `%s' to `%s'\n",
+			    path, ret_path);
+		}
+		FREE(path);
+		return(ret_path);
+	    }
 #else
 	    return(path);
 #endif /* DOSPATH */
@@ -686,12 +691,12 @@ PUBLIC CONST char * HTFileSuffix ARGS2(
 	suff = (HTSuffix *)HTList_objectAt(HTSuffixes, i);
 	if (suff->rep == rep &&
 #if defined(VMS) || defined(FNAMES_8_3)
-	    /*  Don't return a suffix whose first char is a dot and which
-		has more dots or with asterisks, for
+	    /*  Don't return a suffix whose first char is a dot, and which
+		has more dots or asterisks after that, for
 		these systems - kw */
 	    (!suff->suffix || !suff->suffix[0] || suff->suffix[0] != '.' ||
 	     (strchr(suff->suffix + 1, '.') == NULL &&
-	      strchr(suff->suffix + 1, '.') == NULL)) &&
+	      strchr(suff->suffix + 1, '*') == NULL)) &&
 #endif
 	    ((trivial_enc && IsUnityEnc(suff->encoding)) ||
 	     (!trivial_enc && !IsUnityEnc(suff->encoding) &&
@@ -859,7 +864,6 @@ PUBLIC HTFormat HTCharsetFormat ARGS3(
 	cp2 += 7;
 	while (*cp2 == ' ' || *cp2 == '=')
 	    cp2++;
-#ifdef EXP_CHARTRANS
 	StrAllocCopy(cp3, cp2); /* copy to mutilate more */
 	for (cp4 = cp3; (*cp4 != '\0' && *cp4 != '"' &&
 			 *cp4 != ';'  && *cp4 != ':' &&
@@ -885,7 +889,8 @@ PUBLIC HTFormat HTCharsetFormat ARGS3(
 	    format = HTAtom_for(cp);
 	    StrAllocCopy(anchor->charset, cp4);
 	    HTAnchor_setUCInfoStage(anchor, chndl,
-				    UCT_STAGE_MIME, UCT_SETBY_MIME);
+				    UCT_STAGE_MIME,
+				    UCT_SETBY_MIME);
 	} else if (chndl < 0) {
 	    /*
 	    **  Got something but we don't recognize it.
@@ -894,10 +899,10 @@ PUBLIC HTFormat HTCharsetFormat ARGS3(
 	    if (UCCanTranslateFromTo(chndl, current_char_set)) {
 		chartrans_ok = YES;
 		HTAnchor_setUCInfoStage(anchor, chndl,
-					UCT_STAGE_MIME, UCT_SETBY_DEFAULT);
+					UCT_STAGE_MIME,
+					UCT_SETBY_DEFAULT);
 	    }
 	}
-	FREE(cp3);
 	if (chartrans_ok) {
 	    LYUCcharset *p_in = HTAnchor_getUCInfoStage(anchor,
 							UCT_STAGE_MIME);
@@ -927,18 +932,58 @@ PUBLIC HTFormat HTCharsetFormat ARGS3(
 					UCT_STAGE_HTEXT,
 					UCT_SETBY_DEFAULT);
 	    }
-	    if ((p_in->enc != UCT_ENC_CJK) &&
-		(p_in->codepoints & UCT_CP_SUBSETOF_LAT1)) {
+	    if (p_in->enc != UCT_ENC_CJK) {
 		HTCJK = NOCJK;
-	    } else if (chndl == current_char_set) {
-		HTPassEightBitRaw = TRUE;
+		if (!(p_in->codepoints &
+		      UCT_CP_SUBSETOF_LAT1) &&
+		    chndl == current_char_set) {
+		    HTPassEightBitRaw = TRUE;
+		}
+	    } else if (p_out->enc == UCT_ENC_CJK) {
+		if (LYRawMode) {
+		    if ((!strcmp(p_in->MIMEname, "euc-jp") ||
+			 !strcmp(p_in->MIMEname, "shift_jis")) &&
+			(!strcmp(p_out->MIMEname, "euc-jp") ||
+			 !strcmp(p_out->MIMEname, "shift_jis"))) {
+			HTCJK = JAPANESE;
+		    } else if (!strcmp(p_in->MIMEname, "euc-cn") &&
+			       !strcmp(p_out->MIMEname, "euc-cn")) {
+			HTCJK = CHINESE;
+		    } else if (!strcmp(p_in->MIMEname, "big-5") &&
+			       !strcmp(p_out->MIMEname, "big-5")) {
+			HTCJK = TAIPEI;
+		    } else if (!strcmp(p_in->MIMEname, "euc-kr") &&
+			       !strcmp(p_out->MIMEname, "euc-kr")) {
+			HTCJK = KOREAN;
+		    } else {
+			HTCJK = NOCJK;
+		    }
+		} else {
+		    HTCJK = NOCJK;
+		}
 	    }
 	/*
-	**  Fall through to old behavior.
+	**  Check for an iso-8859-# we don't know. - FM
 	*/
-	} else
-#endif /* EXP_CHARTRANS */
-	    if (!strncmp(cp2, "us-ascii", 8) ||
+	} else if (!strncmp(cp4, "iso-8859-", 9) &&
+		   isdigit((unsigned char)cp4[9]) &&
+		   !strncmp(LYchar_set_names[current_char_set],
+			    "Other ISO Latin", 15)) {
+	    /*
+	    **  Hope it's a match, for now. - FM
+	    */
+	    *cp1 = '\0';
+	    format = HTAtom_for(cp);
+	    cp1 = &cp4[10];
+	    while (*cp1 &&
+		   isdigit((unsigned char)(*cp1)))
+		cp1++;
+	    *cp1 = '\0';
+	    StrAllocCopy(anchor->charset, cp4);
+	    HTPassEightBitRaw = TRUE;
+	    HTAlert(anchor->charset);
+#ifdef NOT_USED			/* pre chartrans */
+	} else if (!strncmp(cp2, "us-ascii", 8) ||
 	    !strncmp(cp2, "iso-8859-1", 10)) {
 	    *cp1 = '\0';
 	    format = HTAtom_for(cp);
@@ -1022,7 +1067,9 @@ PUBLIC HTFormat HTCharsetFormat ARGS3(
 	    *cp1 = '\0';
 	    format = HTAtom_for(cp);
 	    StrAllocCopy(anchor->charset, "iso-2022-cn");
+#endif /* NOT_USED */
 	}
+	FREE(cp3);
     } else if (cp1 != NULL) {
 	/*
 	**  No charset parameter is present.
@@ -1034,15 +1081,14 @@ PUBLIC HTFormat HTCharsetFormat ARGS3(
     }
     FREE(cp);
 
-#ifdef EXP_CHARTRANS
     if (!chartrans_ok && !anchor->charset && default_LYhndl >= 0) {
 	HTAnchor_setUCInfoStage(anchor, default_LYhndl,
-				UCT_STAGE_MIME, UCT_SETBY_DEFAULT);
+				UCT_STAGE_MIME,
+				UCT_SETBY_DEFAULT);
     }
     HTAnchor_copyUCInfoStage(anchor,
 			    UCT_STAGE_PARSER, UCT_STAGE_MIME,
 			    -1);
-#endif
 
     return format;
 }
@@ -1628,6 +1674,7 @@ PUBLIC int HTLoadFile ARGS4(
 	   	    HTList_addObject(methods, put);
 	        }
 	    }
+
 	    /*
 	    **  Trim vmsname at semicolon if a version number was
 	    **  included, so it doesn't interfere with the check
@@ -1638,7 +1685,34 @@ PUBLIC int HTLoadFile ARGS4(
 	    /*
 	    **  Fake a Content-Encoding for compressed files. - FM
 	    */
-	    if ((len = strlen(vmsname)) > 2) {
+	    if (!IsUnityEnc(myEncoding)) {
+		/*
+		 *  We already know from the call to HTFileFormat above
+		 *  that this is a compressed file, no need to look at
+		 *  the filename again. - kw
+		 */
+#ifdef USE_ZLIB
+		if (strcmp(format_out->name, "www/download") != 0 &&
+		    (!strcmp(HTAtom_name(myEncoding), "gzip") ||
+		     !strcmp(HTAtom_name(myEncoding), "x-gzip"))) {
+		    fclose(fp);
+		    if (semicolon != NULL)
+			*semicolon = ';';
+		    gzfp = gzopen(vmsname, "rb");
+
+		    if (TRACE)
+			fprintf(stderr,
+				"HTLoadFile: gzopen of `%s' gives %p\n",
+				vmsname, (void*)gzfp);
+		    use_gzread = YES;
+		} else
+#endif  /* USE_ZLIB */
+		{
+		    StrAllocCopy(anchor->content_type, format->name);
+		    StrAllocCopy(anchor->content_encoding, HTAtom_name(myEncoding));
+		    format = HTAtom_for("www/compressed");
+		}
+	    } else if ((len = strlen(vmsname)) > 2) {
 	        if ((vmsname[len - 1] == 'Z') &&
 		    (vmsname[len - 2] == '.' ||
 		     vmsname[len - 2] == '-' ||
@@ -1670,12 +1744,14 @@ PUBLIC int HTLoadFile ARGS4(
 #ifdef USE_ZLIB
 			if (strcmp(format_out->name, "www/download") != 0) {
 			    fclose(fp);
-			    gzfp = gzopen(localname, "rb");
+			    if (semicolon != NULL)
+				*semicolon = ';';
+			    gzfp = gzopen(vmsname, "rb");
 
 			    if (TRACE)
 				fprintf(stderr,
 				       "HTLoadFile: gzopen of `%s' gives %p\n",
-					localname, (void*)gzfp);
+					vmsname, (void*)gzfp);
 			    use_gzread = YES;
 			}
 #else  /* USE_ZLIB */
@@ -1990,15 +2066,14 @@ PUBLIC int HTLoadFile ARGS4(
 		    }
     		}
     		FREE(pathname);
-
-#ifdef EXP_CHARTRANS
+		
 		if (UCLYhndl_HTFile_for_unspec >= 0) {
 		    HTAnchor_setUCInfoStage(anchor,
 					    UCLYhndl_HTFile_for_unspec,
 					    UCT_STAGE_PARSER,
 					    UCT_SETBY_DEFAULT);
 		}
-#endif		
+
 		target = HTML_new(anchor, format_out, sink);
 		targetClass = *target->isa;	/* Copy routine entry points */
 		    
diff --git a/WWW/Library/Implementation/HTMIME.c b/WWW/Library/Implementation/HTMIME.c
index 67795a22..ecd61ac2 100644
--- a/WWW/Library/Implementation/HTMIME.c
+++ b/WWW/Library/Implementation/HTMIME.c
@@ -15,11 +15,9 @@
 #include "HTMIME.h"		/* Implemented here */
 #include "HTAlert.h"
 #include "HTCJK.h"
-#ifdef EXP_CHARTRANS
 #include "UCMap.h"
 #include "UCDefs.h"
 #include "UCAux.h"
-#endif
 
 #include "LYLeaks.h"
 
@@ -27,6 +25,7 @@
 
 extern int current_char_set;
 extern CONST char *LYchar_set_names[];
+extern BOOLEAN LYRawMode;
 extern BOOL HTPassEightBitRaw;
 extern HTCJKlang HTCJK;
 
@@ -370,28 +369,26 @@ PRIVATE void HTMIME_put_character ARGS2(
 			    cp2 += 7;
 			    while (*cp2 == ' ' || *cp2 == '=' || *cp2 == '\"')
 			        cp2++;
-#ifdef EXP_CHARTRANS
 			    StrAllocCopy(cp3, cp2); /* copy to mutilate more */
-			    for (cp4=cp3; (*cp4 != '\0' && *cp4 != '\"' &&
-					   *cp4 != ';'  && *cp4 != ':' &&
-					   !WHITE(*cp4));	cp4++)
-				/* nothing */ ;
+			    for (cp4 = cp3; (*cp4 != '\0' && *cp4 != '\"' &&
+					     *cp4 != ';'  && *cp4 != ':' &&
+					     !WHITE(*cp4));	cp4++)
+				; /* do nothing */
 			    *cp4 = '\0';
 			    cp4 = cp3;
 			    chndl = UCGetLYhndl_byMIME(cp3);
 			    if (chndl < 0) {
-				if (0==strcmp(cp4, "cn-big5")) {
+				if (!strcmp(cp4, "cn-big5")) {
 				    cp4 += 3;
 				    chndl = UCGetLYhndl_byMIME(cp4);
-				}
-				else if (0==strncmp(cp4, "cn-gb", 5)) {
+				} else if (!strncmp(cp4, "cn-gb", 5)) {
 				    StrAllocCopy(cp3, "gb2312");
 				    cp4 = cp3;
 				    chndl = UCGetLYhndl_byMIME(cp4);
 				}
 			    }
-			    if (UCCanTranslateFromTo(chndl, current_char_set))
-			    {
+			    if (UCCanTranslateFromTo(chndl,
+						     current_char_set)) {
 				chartrans_ok = YES;
 				*cp1 = '\0';
 				me->format = HTAtom_for(cp);
@@ -403,8 +400,7 @@ PRIVATE void HTMIME_put_character ARGS2(
 						 recognize it */
 				chndl = UCLYhndl_for_unrec;
 				if (UCCanTranslateFromTo(chndl,
-							 current_char_set))
-				{
+							 current_char_set)) {
 				    chartrans_ok = YES;
 				    *cp1 = '\0';
 				    me->format = HTAtom_for(cp);
@@ -416,42 +412,97 @@ PRIVATE void HTMIME_put_character ARGS2(
 			    if (chartrans_ok) {
 				LYUCcharset * p_in =
 				    HTAnchor_getUCInfoStage(me->anchor,
-							     UCT_STAGE_MIME);
+							    UCT_STAGE_MIME);
 				LYUCcharset * p_out =
 				    HTAnchor_setUCInfoStage(me->anchor,
 							    current_char_set,
-					 UCT_STAGE_HTEXT, UCT_SETBY_DEFAULT);
-				if (!p_out) /* try again */
+					 		    UCT_STAGE_HTEXT,
+							    UCT_SETBY_DEFAULT);
+				if (!p_out)
+				    /*
+				    **  Try again.
+				    */
 				    p_out =
 				      HTAnchor_getUCInfoStage(me->anchor,
-							     UCT_STAGE_HTEXT);
+							      UCT_STAGE_HTEXT);
 
-				if (0==strcmp(p_in->MIMEname,"x-transparent"))
-				{
+				if (!strcmp(p_in->MIMEname,
+					    "x-transparent")) {
 				    HTPassEightBitRaw = TRUE;
 				    HTAnchor_setUCInfoStage(me->anchor,
 				       HTAnchor_getUCLYhndl(me->anchor,
 							    UCT_STAGE_HTEXT),
 				       UCT_STAGE_MIME, UCT_SETBY_DEFAULT);
 				}
-				if (0==strcmp(p_out->MIMEname,"x-transparent"))
-				{
+				if (!strcmp(p_out->MIMEname,
+					    "x-transparent")) {
 				    HTPassEightBitRaw = TRUE;
 				    HTAnchor_setUCInfoStage(me->anchor,
-				       HTAnchor_getUCLYhndl(me->anchor,
-							    UCT_STAGE_MIME),
-				       UCT_STAGE_HTEXT, UCT_SETBY_DEFAULT);
+					 HTAnchor_getUCLYhndl(me->anchor,
+							      UCT_STAGE_MIME),
+							    UCT_STAGE_HTEXT,
+							    UCT_SETBY_DEFAULT);
 				}
-				if ((p_in->enc != UCT_ENC_CJK) &&
-				    (p_in->codepoints &
-				     UCT_CP_SUBSETOF_LAT1)) {
+				if (p_in->enc != UCT_ENC_CJK) {
 				    HTCJK = NOCJK;
-				} else if (chndl == current_char_set) {
-				HTPassEightBitRaw = TRUE;
+				    if (!(p_in->codepoints &
+					  UCT_CP_SUBSETOF_LAT1) &&
+					chndl == current_char_set) {
+					HTPassEightBitRaw = TRUE;
+				    }
+				} else if (p_out->enc == UCT_ENC_CJK) {
+				    if (LYRawMode) {
+					if ((!strcmp(p_in->MIMEname,
+						     "euc-jp") ||
+					     !strcmp(p_in->MIMEname,
+						     "shift_jis")) &&
+					    (!strcmp(p_out->MIMEname,
+						     "euc-jp") ||
+					     !strcmp(p_out->MIMEname,
+						     "shift_jis"))) {
+					    HTCJK = JAPANESE;
+					} else if (!strcmp(p_in->MIMEname,
+							   "euc-cn") &&
+						   !strcmp(p_out->MIMEname,
+							   "euc-cn")) {
+					    HTCJK = CHINESE;
+					} else if (!strcmp(p_in->MIMEname,
+							   "big-5") &&
+						   !strcmp(p_out->MIMEname,
+							   "big-5")) {
+					    HTCJK = TAIPEI;
+					} else if (!strcmp(p_in->MIMEname,
+							   "euc-kr") &&
+						   !strcmp(p_out->MIMEname,
+							   "euc-kr")) {
+					    HTCJK = KOREAN;
+					} else {
+					    HTCJK = NOCJK;
+					}
+				    } else {
+					HTCJK = NOCJK;
+				    }
 				}
-			} else  /* Fall through to old behavior */
-#endif /* EXP_CHARTRANS */
-			    if (!strncmp(cp2, "us-ascii", 8) ||
+			} else if
+			       (!strncmp(cp4, "iso-8859-", 9) &&
+				isdigit((unsigned char)cp4[9]) &&
+				!strncmp(LYchar_set_names[current_char_set],
+				 	 "Other ISO Latin", 15)) {
+			        /*
+			        **  Hope it's a match, for now. - FM
+			        */
+				*cp1 = '\0';
+				me->format = HTAtom_for(cp);
+				cp1 = &cp4[10];
+				while (*cp1 &&
+				       isdigit((unsigned char)(*cp1)))
+				    cp1++;
+				*cp1 = '\0';
+				StrAllocCopy(me->anchor->charset, cp4);
+				HTPassEightBitRaw = TRUE;
+				HTAlert(me->anchor->charset);
+#ifdef NOT_USED			/* pre-chartrans */
+			} else if (!strncmp(cp2, "us-ascii", 8) ||
 			        !strncmp(cp2, "iso-8859-1", 10)) {
 				*cp1 = '\0';
 				me->format = HTAtom_for(cp);
@@ -562,7 +613,9 @@ PRIVATE void HTMIME_put_character ARGS2(
 				me->format = HTAtom_for(cp);
 				StrAllocCopy(me->anchor->charset,
 					     "iso-2022-cn");
+#endif /* NOT_USED */
 			    }
+			    FREE(cp3);
 			} else {
 			    /*
 			    **  No charset parameter is present.
@@ -1326,6 +1379,16 @@ PRIVATE void HTMIME_put_character ARGS2(
                    "HTMIME: Was CONTENT_L, found E, checking for 'ngth:'\n");
 	    break;
 	    
+	case 'o':
+	case 'O':
+	    me->check_pointer = "cation:";
+	    me->if_ok = miCONTENT_LOCATION;
+	    me->state = miCHECK;
+            if (TRACE)
+                fprintf(stderr,
+                   "HTMIME: Was CONTENT_L, found O, checking for 'cation:'\n");
+	    break;
+	    
 	default:
           if (TRACE)
             fprintf (stderr,
@@ -2207,6 +2270,8 @@ PUBLIC HTStream* HTNetMIME ARGS3(
 **
 **	Written by S. Ichikawa,
 **	partially inspired by encdec.c of <jh@efd.lth.se>.
+**
+**	Generalized HTmmdecode for chartrans - K. Weide 1997-03-06
 */
 #define	BUFLEN	1024
 #ifdef ESC
@@ -2292,8 +2357,6 @@ PUBLIC void HTmmdec_quote ARGS2(
     strcpy(t, buf);
 }
 
-/* Generalized HTmmdecode for chartrans - kweide 1997-03-06 */
-
 PUBLIC void HTmmdecode ARGS2(
 	char *,		trg,
 	char *,		str)
@@ -2304,8 +2367,10 @@ PUBLIC void HTmmdecode ARGS2(
 
     buf[0] = '\0';
 
-/* encoded-words look like  =?ISO-8859-1?B?SWYgeW91IGNhbiByZWFkIHRoaXMgeW8=?=  */
-
+    /*
+    **  Encoded-words look like
+    **		=?ISO-8859-1?B?SWYgeW91IGNhbiByZWFkIHRoaXMgeW8=?=
+    */
     for (s = str, u = buf; *s; ) {
 	base64 = quote = 0;
 	if (*s == '=' && s[1] == '?' &&
@@ -2313,7 +2378,8 @@ PUBLIC void HTmmdecode ARGS2(
 	{ /* must be beginning of word */
 	    qm2 = strchr(s+2, '?'); /* 2nd question mark */
 	    if (qm2 &&
-		(qm2[1] == 'B' || qm2[1] == 'b' || qm2[1] == 'Q' || qm2[1] == 'q') &&
+		(qm2[1] == 'B' || qm2[1] == 'b' || qm2[1] == 'Q' ||
+		 qm2[1] == 'q') &&
 		qm2[2] == '?') { /* 3rd question mark */
 		char * qm4 = strchr(qm2 + 3, '?'); /* 4th question mark */
 		if (qm4 && qm4 - s < 74 &&  /* RFC 2047 length restriction */
@@ -2327,15 +2393,13 @@ PUBLIC void HTmmdecode ARGS2(
 			}
 		    if (!invalid) {
 			int LYhndl;
+
 			*qm2 = '\0';
-#ifdef EXP_CHARTRANS
 			for (p = s+2; *p; p++)
 			    *p = TOLOWER(*p);
 			invalid = ((LYhndl = UCGetLYhndl_byMIME(s+2)) < 0 ||
-				   !UCCanTranslateFromTo(LYhndl, current_char_set));
-#else
-			invalid = (0!=strncasecomp(s+2, "ISO-2022-JP", 11));
-#endif
+				   !UCCanTranslateFromTo(LYhndl,
+							 current_char_set));
 			*qm2 = '?';
 		    }
 		    if (!invalid) {
diff --git a/WWW/Library/Implementation/HTMLDTD.c b/WWW/Library/Implementation/HTMLDTD.c
index a519fb82..cbeb35f9 100644
--- a/WWW/Library/Implementation/HTMLDTD.c
+++ b/WWW/Library/Implementation/HTMLDTD.c
@@ -1430,7 +1430,7 @@ static attr ulist_attr[] = {			/* UL attributes */
  /* { "BLOCKQUOTE", bq_attr,	HTML_BQ_ATTRIBUTES,	SGML_MIXED }, */
 #define T_BLOCKQUOTE	0x0200,	0xAFBCF,0xAFFFF,0xB6680,0xB6FAF,0x8031F,0x00000
  /* { "BODY"	, body_attr,	HTML_BODY_ATTRIBUTES,	SGML_MIXED }, */
-#define	T_BODY		0x20000,0x2FB8F,0x2FFFF,0x30000,0x30000,0xDFFFF,0x00003
+#define	T_BODY		0x20000,0x2FB8F,0x2FFFF,0x30000,0x30000,0xDFF7F,0x00003
  /* { "BODYTEXT", bodytext_attr,HTML_BODYTEXT_ATTRIBUTES, SGML_MIXED }, */
 #define T_BODYTEXT	0x20000,0x0FB8F,0xAFFFF,0x30200,0xB7FAF,0x8F17F,0x00003
  /* { "BQ"	, bq_attr,	HTML_BQ_ATTRIBUTES,	SGML_MIXED }, */
@@ -1501,7 +1501,7 @@ static attr ulist_attr[] = {			/* UL attributes */
  /* { "H6"	, h_attr,	HTML_H_ATTRIBUTES,	SGML_MIXED }, */
 #define T_H6  		0x0100,	0x0B04F,0x0B05F,0x36680,0x37FAF,0x80317,0x00000
  /* { "HEAD"	, gen_attr,	HTML_GEN_ATTRIBUTES,	SGML_MIXED }, */
-#define T_HEAD		0x40000,0x4F000,0x47000,0x10000,0x10000,0x9FFFF,0x00006
+#define T_HEAD		0x40000,0x4F000,0x47000,0x10000,0x10000,0x9FF7F,0x00006
  /* { "HR"	, hr_attr,	HTML_HR_ATTRIBUTES,	SGML_EMPTY }, */
 #define	T_HR		0x4000,	0x00000,0x00000,0x3FE80,0x3FFBF,0x87F37,0x00001
  /* { "HTML"	, gen_attr,	HTML_GEN_ATTRIBUTES,	SGML_MIXED }, */
@@ -1574,7 +1574,7 @@ static attr ulist_attr[] = {			/* UL attributes */
  /* { "SCRIPT"	, script_attr,	HTML_SCRIPT_ATTRIBUTES,	SGML_LITTERAL }, */
 #define T_SCRIPT	0x2000,	0x00000,0x00000,0x77F9F,0x77FFF,0x87F5F,0x00000
  /* { "SELECT"	, select_attr,	HTML_SELECT_ATTRIBUTES,	SGML_MIXED }, */
-#define T_SELECT	0x0040,	0x08000,0x08000,0x03FAF,0x13FBF,0x80F5F,0x00000
+#define T_SELECT	0x0040,	0x08000,0x08000,0x03FAF,0x13FBF,0x80F5F,0x00008
 #define T_SHY		0x1000, 0x00000,0x00000,0x3779F,0x77FBF,0x8101F,0x00001
  /* { "SMALL"	, gen_attr,	HTML_GEN_ATTRIBUTES,	SGML_MIXED }, */
 #define T_SMALL		0x0001, 0x8B04F,0x8FFFF,0xA778F,0xF7FBF,0x00001,0x00004
diff --git a/WWW/Library/Implementation/HTNews.c b/WWW/Library/Implementation/HTNews.c
index bfd64e5a..adc9568b 100644
--- a/WWW/Library/Implementation/HTNews.c
+++ b/WWW/Library/Implementation/HTNews.c
@@ -54,6 +54,10 @@ struct _HTStructured {
 	CONST HTStructuredClass *	isa;
 	/* ... */
 };
+struct _HTStream 
+{
+  HTStreamClass * isa;
+};
 
 #define LINE_LENGTH 512			/* Maximum length of line of ARTICLE etc */
 #define GROUP_NAME_LENGTH	256	/* Maximum length of group name */
@@ -62,6 +66,7 @@ extern BOOLEAN LYListNewsNumbers;
 extern BOOLEAN LYListNewsDates;
 extern HTCJKlang HTCJK;
 extern int interrupted_in_htgetcharacter;
+extern BOOL keep_mime_headers;   /* Include mime headers and force raw text */
 extern BOOL using_proxy;	/* Are we using an NNTP proxy? */
 
 /*
@@ -76,16 +81,24 @@ PRIVATE char response_text[LINE_LENGTH+1];	/* Last response */
 /* PRIVATE HText *	HT;	*/		/* the new hypertext */
 PRIVATE HTStructured * target;			/* The output sink */
 PRIVATE HTStructuredClass targetClass;		/* Copy of fn addresses */
+PRIVATE HTStream * rawtarget = NULL;		/* The output sink for rawtext */
+PRIVATE HTStreamClass rawtargetClass;		/* Copy of fn addresses */
 PRIVATE HTParentAnchor *node_anchor;		/* Its anchor */
 PRIVATE int	diagnostic;			/* level: 0=none 2=source */
+PRIVATE BOOL rawtext = NO;			/* Flag: HEAD or -mime_headers */
 PRIVATE HTList *NNTP_AuthInfo = NULL;		/* AUTHINFO database */
 
 #define PUTC(c) (*targetClass.put_character)(target, c)
 #define PUTS(s) (*targetClass.put_string)(target, s)
+#define RAW_PUTS(s) (*rawtargetClass.put_string)(rawtarget, s)
 #define START(e) (*targetClass.start_element)(target, e, 0, 0, -1, 0)
 #define END(e) (*targetClass.end_element)(target, e, 0)
 #define MAYBE_END(e) if (HTML_dtd.tags[e].contents != SGML_EMPTY) \
                         (*targetClass.end_element)(target, e, 0)
+#define FREE_TARGET if (rawtext) (*rawtargetClass._free)(rawtarget); \
+    			else (*targetClass._free)(target)
+#define ABORT_TARGET if (rawtext) (*rawtargetClass._abort)(rawtarget, NULL); \
+    			else (*targetClass._abort)(target, NULL)
 
 typedef struct _NNTPAuth {
    char * host;
@@ -958,7 +971,7 @@ PRIVATE int read_article NOARGS
     **  The header fields are either ignored,
     **  or formatted and put into the text.
     */
-    if (!diagnostic) {
+    if (!diagnostic && !rawtext) {
 	while (!done) {
 	    char ch = *p++ = NEXT_CHAR;
 	    if (ch == (char)EOF) {
@@ -1217,20 +1230,26 @@ PRIVATE int read_article NOARGS
 	FREE(href);
     }
 
-    if (diagnostic) {
+    if (rawtext) {
+	/*
+	 *  No tags - kw
+	 */
+	;
+    } else if (diagnostic) {
         /*
 	**  Read in the HEAD and BODY of the Article
 	**  as XMP formatted text. - FM
 	*/
 	START(HTML_XMP);
+	PUTC('\n');
     } else {
         /*
 	**  Read in the BODY of the Article
 	**  as PRE formatted text. - FM
 	*/
 	START(HTML_PRE);
+	PUTC('\n');
     }
-    PUTC('\n');
 
     p = line;
     while (!done) {
@@ -1258,10 +1277,15 @@ PRIVATE int read_article NOARGS
 		    done = YES;
 		    break;
 		} else {			/* Line starts with dot */
-		    PUTS(&line[1]);	/* Ignore first dot */
+		    if (rawtext)
+			RAW_PUTS(&line[1]);
+		    else
+			PUTS(&line[1]);	/* Ignore first dot */
 		}
 	    } else {
-	        if (diagnostic || !scan_for_buried_news_references) {
+		if (rawtext) {
+		    RAW_PUTS(line);
+	        } else if (diagnostic || !scan_for_buried_news_references) {
 		    /*
 		    **  All lines are passed as unmodified source. - FM
 		    */
@@ -1361,7 +1385,10 @@ PRIVATE int read_article NOARGS
 	    p = line;				/* Restart at beginning */
 	} /* if end of line */
     } /* Loop over characters */
-    
+
+    if (rawtext)
+	return(HT_LOADED);
+
     if (diagnostic)
 	END(HTML_XMP);
     else
@@ -1955,6 +1982,7 @@ PUBLIC int HTLoadNews ARGS4(
     BOOL reply_wanted;		/* Flag: followup post was asked for */
     BOOL spost_wanted;		/* Flag: new SSL post to group was asked for */
     BOOL sreply_wanted;		/* Flag: followup SSL post was asked for */
+    BOOL head_wanted = NO;	/* Flag: want HEAD of single article */
     int first, last;		/* First and last articles asked for */
     char *cp = 0;
     char *ListArg = NULL;
@@ -1965,6 +1993,7 @@ PUBLIC int HTLoadNews ARGS4(
     diagnostic = (format_out == WWW_SOURCE ||	/* set global flag */
     		  format_out == HTAtom_for("www/download") ||
 		  format_out == HTAtom_for("www/dump"));
+    rawtext = NO;
     
     if (TRACE) fprintf(stderr, "HTNews: Looking for %s\n", arg);
     
@@ -2260,6 +2289,34 @@ PUBLIC int HTLoadNews ARGS4(
         return NO;			/* Ignore if no name */
     }
 
+    if (!(post_wanted || reply_wanted || spost_wanted || sreply_wanted ||
+	  (group_wanted && last != -1) || list_wanted)) {
+	head_wanted = anAnchor->isHEAD;
+	if (head_wanted && !strncmp(command, "ARTICLE_", 8)) {
+	    /* overwrite "ARTICLE" - hack... */
+	    strcpy(command, "HEAD ");
+	    for (cp = command + 5; ; cp++)
+		if ((*cp = *(cp + 3)) == '\0')
+		    break;
+	}
+	rawtext = (head_wanted || keep_mime_headers);
+    }
+    if (rawtext) {
+        node_anchor = anAnchor;
+	rawtarget = HTStreamStack(WWW_PLAINTEXT,
+				  format_out,
+				  stream, anAnchor);
+	if (!rawtarget) {
+	    FREE(NewsHost);
+	    FREE(NewsHREF);
+	    FREE(ProxyHost);
+	    FREE(ProxyHREF);
+	    FREE(ListArg);
+	    HTAlert("No target for raw text!");
+	    return(HT_NOT_LOADED);
+	}	/* Copy routine entry points */
+	rawtargetClass = *rawtarget->isa;
+    } else
     /*
     **  Make a hypertext object with an anchor list.
     */
@@ -2298,8 +2355,9 @@ PUBLIC int HTLoadNews ARGS4(
 		     "HTNews: Interrupted on connect; recovering cleanly.\n");
 		_HTProgress("Connection interrupted.");
 		if (!(post_wanted || reply_wanted ||
-		      spost_wanted || sreply_wanted))
-		    (*targetClass._abort)(target, NULL);
+		      spost_wanted || sreply_wanted)) {
+		    ABORT_TARGET;
+		}
 		FREE(NewsHost);
 		FREE(NewsHREF);
 		FREE(ProxyHost);
@@ -2360,8 +2418,9 @@ PUBLIC int HTLoadNews ARGS4(
 			if (status == HT_INTERRUPTED) {
 			    _HTProgress("Connection interrupted.");
 			    if (!(post_wanted || reply_wanted ||
-				  spost_wanted || sreply_wanted))
-			        (*targetClass._abort)(target, NULL);
+				  spost_wanted || sreply_wanted)) {
+			        ABORT_TARGET;
+			    }
 			    FREE(NewsHost);
 			    FREE(NewsHREF);
 			    FREE(ProxyHost);
@@ -2609,7 +2668,9 @@ Send_NNTP_command:
 		**  the number (first) as the command and go back
 		**  to send it and check the response. - FM
 		*/
-		sprintf(command, "ARTICLE %d%c%c", first, CR, LF);
+		sprintf(command, "%s %d%c%c",
+			head_wanted ? "HEAD" : "ARTICLE",
+			first, CR, LF);
 		group_wanted = FALSE;
 		retries = 2;
 		goto Send_NNTP_command;
@@ -2630,9 +2691,9 @@ Send_NNTP_command:
 	if (!(post_wanted || reply_wanted ||
 	      spost_wanted || sreply_wanted)) {
 	    if (status == HT_NOT_LOADED) {
-		(*targetClass._abort)(target, NULL);
+		ABORT_TARGET;
 	    } else {
-		(*targetClass._free)(target);
+		FREE_TARGET;
 	    }
 	}
 	FREE(NewsHREF);
@@ -2660,8 +2721,9 @@ Send_NNTP_command:
 	    NULL,NULL,NULL, arg);No -- message earlier wil have covered it */
 
     if (!(post_wanted || reply_wanted ||
-	  spost_wanted || sreply_wanted))
-        (*targetClass._abort)(target, NULL);
+	  spost_wanted || sreply_wanted)) {
+        ABORT_TARGET;
+    }
     FREE(NewsHREF);
     if (ProxyHREF) {
 	StrAllocCopy(NewsHost, ProxyHost);
diff --git a/WWW/Library/Implementation/HTParse.c b/WWW/Library/Implementation/HTParse.c
index 3c5fafc5..6f558ef5 100644
--- a/WWW/Library/Implementation/HTParse.c
+++ b/WWW/Library/Implementation/HTParse.c
@@ -98,6 +98,7 @@ PRIVATE void scan ARGS2(
 	if (*p =='#') {
 	    parts->anchor = (p + 1);
 	    *p = '\0';			/* terminate the rest */
+	    break;		/* leave things after first # alone - kw */
 	}
     }
 
@@ -137,8 +138,12 @@ PRIVATE void scan ARGS2(
 	     *  or it's an nntp or snews URL, or news URL with a host.
 	     *  Restore the '#' in the address.
 	     */
-	    *(parts->anchor - 1) = '#';
-	    parts->anchor = NULL;
+	    /* but only if we have found a path component of which this will
+	     * become part. - kw  */
+	    if (parts->relative || parts->absolute) {
+		*(parts->anchor - 1) = '#';
+		parts->anchor = NULL;
+	    }
 	}
     }
 
@@ -693,12 +698,13 @@ PUBLIC char * HTUnEscape ARGS1(
         return str;
 
     while (*p != '\0') {
-        if (*p == HEX_ESCAPE) {
+        if (*p == HEX_ESCAPE &&
+	    p[1] && p[2] &&	/* tests shouldn't be needed, but.. */
+	    isxdigit((unsigned char)p[1]) &&
+	    isxdigit((unsigned char)p[2])) {
 	    p++;
-	    if (*p)
-	        *q = from_hex(*p++) * 16;
-	    if (*p)
-	        *q = FROMASCII(*q + from_hex(*p++));
+	    *q = from_hex(*p++) * 16;
+	    *q = FROMASCII(*q + from_hex(*p++));
 	    q++;
 	} else {
 	    *q++ = *p++; 
diff --git a/WWW/Library/Implementation/SGML.c b/WWW/Library/Implementation/SGML.c
index c2cadecd..6e64714f 100644
--- a/WWW/Library/Implementation/SGML.c
+++ b/WWW/Library/Implementation/SGML.c
@@ -792,6 +792,16 @@ PRIVATE void start_element ARGS1(
 			    new_tag->name);
 	    }
 	}
+	if (context->element_stack && !valid &&
+	    (context->element_stack->tag->flags & Tgf_strict) &&
+	    !(valid = element_valid_within(new_tag, context->element_stack->tag,
+					   direct_container))) {
+	    if (TRACE)
+		fprintf(stderr, "SGML: Still open %s \t<- ignoring start <%s>\n",
+			context->element_stack->tag->name,
+			new_tag->name);
+	    return;
+	}
 
 	if (context->element_stack && !extra_action_taken &&
 	    canclose_check == close_NO && !valid && (new_tag->flags & Tgf_mafse)) {
diff --git a/WWW/Library/Implementation/SGML.h b/WWW/Library/Implementation/SGML.h
index d8f90c19..645bf266 100644
--- a/WWW/Library/Implementation/SGML.h
+++ b/WWW/Library/Implementation/SGML.h
@@ -90,8 +90,10 @@ typedef enum _TagClass {
 typedef enum _TagFlags {
     Tgf_endO		= 0x00001, /* end tag can be Omitted */
     Tgf_startO		= 0x00002, /* start tag can be Omitted */
-    Tgf_mafse   	= 0x00004  /* Make Attribute-Free Start-tag End instead
+    Tgf_mafse   	= 0x00004, /* Make Attribute-Free Start-tag End instead
 				      (if found invalid) */
+    Tgf_strict		= 0x00008  /* Ignore contained invalid elements,
+				      don't pass them on */
 } TagFlags;
 
 /*              A tag structure describes an SGML element.