lynx-dev Re: resolving weird URL types (patch)

[Leonid Pauzner]

13-May-99 14:58 I wrote:
> Example:  press 'g' and enter xyz://aaa.bbb.cc URL.

> Apparently, lynx will not reject weird URL schemes
> at LYEnsureAbsoluteURL() / LYConvertURL() stage,
> it will try to resolve DNS first.

> Such URL prefixes normally rejected by is_url() in getfile()
> with "Badly formed address" alert message
> but user defined URLs (startfile, etc.) does serve another way.


>                 Lynx Trace Log (2.8.2pre.3)

> User message: Trace ON!
>  'xyz://aaa.bbb.cc' is not a URL
> Converted 'xyz://aaa.bbb.cc' to 'c:/lynx/dist28/pre3/xyz://aaa.bbb.cc'
> Can't stat() or fopen() 'c:/lynx/dist28/pre3/xyz://aaa.bbb.cc'
> Looking up xyz: first.


> In fact, there are two unsupported url types:
> one without "://" so guessing is possible
> and the other with unsupported prefix before "://" string (no guessing),
> but is_url() use a single UNKNOWN_URL_TYPE.

> IMO, LYEnsureAbsoluteURL() should filter out such urls itself.

The patch:

* avoid URL guessing (LYConvertToURL() calls) when URL has '://' string
  and unsupported prefix.
* LYFillLocalFileURL() moved from LYCharUtils.c to LYUtils.c
  (where all such functions located).
* cosmetic changes in getfile().


diff -u old/lycharut.c ./lycharut.c
--- old/lycharut.c      Wed May  5 17:34:00 1999
+++ ./lycharut.c        Thu May 13 15:07:24 1999
@@ -294,112 +294,6 @@
 }

 /*
-**  If an HREF, itself or if resolved against a base,
-**  represents a file URL, and the host is defaulted,
-**  force in "//localhost".  We need this until
-**  all the other Lynx code which performs security
-**  checks based on the "localhost" string is changed
-**  to assume "//localhost" when a host field is not
-**  present in file URLs - FM
-*/
-PUBLIC void LYFillLocalFileURL ARGS2(
-       char **,        href,
-       CONST char *,   base)
-{
-    char * temp = NULL;
-
-    if (*href == NULL || *(*href) == '\0')
-       return;
-
-    if (!strcmp(*href, "//") || !strncmp(*href, "///", 3)) {
-       if (base != NULL && !strncmp(base, "file:", 5)) {
-           StrAllocCopy(temp, "file:");
-           StrAllocCat(temp, *href);
-           StrAllocCopy(*href, temp);
-       }
-    }
-    if (!strncmp(*href, "file:", 5)) {
-       if (*(*href+5) == '\0') {
-           StrAllocCat(*href, "//localhost");
-       } else if (!strcmp(*href, "file://")) {
-           StrAllocCat(*href, "localhost");
-       } else if (!strncmp(*href, "file:///", 8)) {
-           StrAllocCopy(temp, (*href+7));
-           LYLocalFileToURL (href, temp);
-       } else if (!strncmp(*href, "file:/", 6) && !LYIsHtmlSep(*(*href+6))) {
-           StrAllocCopy(temp, (*href+5));
-           LYLocalFileToURL (href, temp);
-       }
-    }
-
-#if defined(DOSPATH) || defined(__EMX__)
-    if (isalpha(*(*href)) && (*(*href+1) == ':'))  {
-       /*
-        * If it's a local DOS path beginning with drive letter,
-        * add file://localhost/ prefix and go ahead.
-        */
-       StrAllocCopy(temp, *href);
-       LYLocalFileToURL (href, temp);
-    }
-
-    /* use below: strlen("file://localhost/") = 17 */
-    if (!strncmp(*href, "file://localhost/", 17)
-         && (strlen(*href) == 19)
-         && isalpha(*(*href+17))
-         && (*(*href+18) == ':')) {
-       /*
-        * Terminate DOS drive letter with a slash to surf root successfully.
-        * Here seems a proper place to do so.
-        */
-       StrAllocCat(*href, "/");
-    }
-#endif /* DOSPATH */
-
-    /*
-     * No path in a file://localhost URL means a
-     * directory listing for the current default. - FM
-     */
-    if (!strcmp(*href, "file://localhost")) {
-       char *temp2;
-#ifdef VMS
-       temp2 = HTVMS_wwwName(getenv("PATH"));
-#else
-       char curdir[LY_MAXPATH];
-#if HAVE_GETCWD
-       getcwd (curdir, sizeof(curdir));
-#else
-       getwd (curdir);
-#endif /* NO_GETCWD */
-       temp2 = wwwName(curdir);
-#endif /* VMS */
-       LYAddHtmlSep(href);
-       /*
-        *  Check for pathological cases - current dir has chars which
-        *  MUST BE URL-escaped - kw
-        */
-       if (strchr(temp2, '%') != NULL || strchr(temp2, '#') != NULL) {
-           FREE(temp);
-           temp = HTEscape(temp2, URL_PATH);
-           StrAllocCat(*href, temp);
-       } else {
-           StrAllocCat(*href, temp2);
-       }
-    }
-
-#ifdef VMS
-    /*
-     * On VMS, a file://localhost/ URL means
-     * a listing for the login directory. - FM
-     */
-    if (!strcmp(*href, "file://localhost/"))
-       StrAllocCat(*href, (HTVMS_wwwName((char *)Home_Dir())+1));
-#endif /* VMS */
-
-    FREE(temp);
-    return;
-}
-
-/*
 **  This function writes a line with a META tag to an open file,
 **  which will specify a charset parameter to use when the file is
 **  read back in.  It is meant for temporary HTML files used by the
diff -u old/lycharut.h ./lycharut.h
--- old/lycharut.h      Fri Jul 24 20:01:04 1998
+++ ./lycharut.h        Thu May 13 15:08:06 1999
@@ -36,9 +36,6 @@
        char *          str));
 extern char *LYFindEndOfComment PARAMS((
        char *          str));
-extern void LYFillLocalFileURL PARAMS((
-       char **         href,
-       CONST char *    base));
 extern void LYAddMETAcharsetToFD PARAMS((
        FILE *          fd,
        int             disp_chndl));
diff -u old/lygetfil.c ./lygetfil.c
--- old/lygetfil.c      Wed May  5 17:34:00 1999
+++ ./lygetfil.c        Thu May 13 13:10:22 1999
@@ -140,11 +140,19 @@
         *
         *  Some special URL's we handle ourselves. :)
         */
-       if ((url_type = is_url(doc->address)) != 0) {
+       if ((url_type = is_url(doc->address)) == 0) {
+               CTRACE_SLEEP(MessageSecs);
+               HTUserMsg2(WWW_BAD_ADDR_MESSAGE, doc->address);
+               CTRACE(tfp,"\n");
+               return(NULLFILE);
+         } else {
+               /* url_type recognized, do all the checks: */
+
                if (LYValidate && !LYPermitURL) {
                    if (!(url_type == HTTP_URL_TYPE ||
                          url_type == HTTPS_URL_TYPE ||
                          url_type == LYNXHIST_URL_TYPE ||
+                         /* registered with LYRegisterLynxProtocols() in 
LYMain.c: */
                          url_type == LYNXKEYMAP_URL_TYPE ||
                          url_type == LYNXIMGMAP_URL_TYPE ||
                          url_type == LYNXCOOKIE_URL_TYPE ||
@@ -181,6 +189,7 @@
                          url_type == LYNXOPTIONS_URL_TYPE ||
                          url_type == LYNXCFG_URL_TYPE ||
                          url_type == LYNXCOMPILE_OPTS_URL_TYPE ||
+                         url_type == LYNXMESSAGES_URL_TYPE ||
                          url_type == LYNXDOWNLOAD_URL_TYPE ||
                          url_type == MAILTO_URL_TYPE ||
                          url_type == NEWSPOST_URL_TYPE ||
@@ -474,6 +483,7 @@
                 */
                } else if (local_host_only &&
                           url_type != NEWS_URL_TYPE &&
+                          /* any LYNXfoo:/ URL not handled above is allowed */
                           url_type != LYNXKEYMAP_URL_TYPE &&
                           url_type != LYNXIMGMAP_URL_TYPE &&
                           url_type != LYNXCOOKIE_URL_TYPE &&
@@ -726,6 +736,7 @@
                                url_type == LYNXOPTIONS_URL_TYPE ||
                                url_type == LYNXCFG_URL_TYPE ||
                                url_type == LYNXCOMPILE_OPTS_URL_TYPE ||
+                               url_type == LYNXMESSAGES_URL_TYPE ||
                                url_type == LYNXHIST_URL_TYPE ||
                                url_type == LYNXCOOKIE_URL_TYPE ||
                                (LYValidate &&
@@ -932,12 +943,7 @@
                        }
                    }
                }
-         } else {
-             CTRACE_SLEEP(MessageSecs);
-             HTUserMsg2(WWW_BAD_ADDR_MESSAGE, doc->address);
-             CTRACE(tfp,"\n");
-             return(NULLFILE);
-         }
+         } /* url_type handled */
 }

 /*
diff -u old/lyutils.c ./lyutils.c
--- old/lyutils.c       Sat May  8 10:46:28 1999
+++ ./lyutils.c Thu May 13 16:26:20 1999
@@ -3782,6 +3782,112 @@
 #endif /* VMS */

 /*
+**  If an HREF, itself or if resolved against a base,
+**  represents a file URL, and the host is defaulted,
+**  force in "//localhost".  We need this until
+**  all the other Lynx code which performs security
+**  checks based on the "localhost" string is changed
+**  to assume "//localhost" when a host field is not
+**  present in file URLs - FM
+*/
+PUBLIC void LYFillLocalFileURL ARGS2(
+       char **,        href,
+       CONST char *,   base)
+{
+    char * temp = NULL;
+
+    if (*href == NULL || *(*href) == '\0')
+       return;
+
+    if (!strcmp(*href, "//") || !strncmp(*href, "///", 3)) {
+       if (base != NULL && !strncmp(base, "file:", 5)) {
+           StrAllocCopy(temp, "file:");
+           StrAllocCat(temp, *href);
+           StrAllocCopy(*href, temp);
+       }
+    }
+    if (!strncmp(*href, "file:", 5)) {
+       if (*(*href+5) == '\0') {
+           StrAllocCat(*href, "//localhost");
+       } else if (!strcmp(*href, "file://")) {
+           StrAllocCat(*href, "localhost");
+       } else if (!strncmp(*href, "file:///", 8)) {
+           StrAllocCopy(temp, (*href+7));
+           LYLocalFileToURL (href, temp);
+       } else if (!strncmp(*href, "file:/", 6) && !LYIsHtmlSep(*(*href+6))) {
+           StrAllocCopy(temp, (*href+5));
+           LYLocalFileToURL (href, temp);
+       }
+    }
+
+#if defined(DOSPATH) || defined(__EMX__)
+    if (isalpha(*(*href)) && (*(*href+1) == ':'))  {
+       /*
+        * If it's a local DOS path beginning with drive letter,
+        * add file://localhost/ prefix and go ahead.
+        */
+       StrAllocCopy(temp, *href);
+       LYLocalFileToURL (href, temp);
+    }
+
+    /* use below: strlen("file://localhost/") = 17 */
+    if (!strncmp(*href, "file://localhost/", 17)
+         && (strlen(*href) == 19)
+         && isalpha(*(*href+17))
+         && (*(*href+18) == ':')) {
+       /*
+        * Terminate DOS drive letter with a slash to surf root successfully.
+        * Here seems a proper place to do so.
+        */
+       StrAllocCat(*href, "/");
+    }
+#endif /* DOSPATH */
+
+    /*
+     * No path in a file://localhost URL means a
+     * directory listing for the current default. - FM
+     */
+    if (!strcmp(*href, "file://localhost")) {
+       char *temp2;
+#ifdef VMS
+       temp2 = HTVMS_wwwName(getenv("PATH"));
+#else
+       char curdir[LY_MAXPATH];
+#if HAVE_GETCWD
+       getcwd (curdir, sizeof(curdir));
+#else
+       getwd (curdir);
+#endif /* NO_GETCWD */
+       temp2 = wwwName(curdir);
+#endif /* VMS */
+       LYAddHtmlSep(href);
+       /*
+        *  Check for pathological cases - current dir has chars which
+        *  MUST BE URL-escaped - kw
+        */
+       if (strchr(temp2, '%') != NULL || strchr(temp2, '#') != NULL) {
+           FREE(temp);
+           temp = HTEscape(temp2, URL_PATH);
+           StrAllocCat(*href, temp);
+       } else {
+           StrAllocCat(*href, temp2);
+       }
+    }
+
+#ifdef VMS
+    /*
+     * On VMS, a file://localhost/ URL means
+     * a listing for the login directory. - FM
+     */
+    if (!strcmp(*href, "file://localhost/"))
+       StrAllocCat(*href, (HTVMS_wwwName((char *)Home_Dir())+1));
+#endif /* VMS */
+
+    FREE(temp);
+    return;
+}
+
+/*
 **  This function ensures that an href will be
 **  converted to a fully resolved, absolute URL,
 **  with guessing of the host or expansions of
@@ -3812,7 +3918,22 @@
     if (!is_url(*href)) {
        CTRACE(tfp, "%s%s'%s' is not a URL\n",
                    (name ? name : ""), (name ? " " : ""), *href);
-       LYConvertToURL(href);
+       if (strstr(*href, "://") != NULL) {
+            /*
+             * Weird prefix before "://" - no guessing possible.
+             * Drop trace message and fall down to getfile() silently: small
+             * overhead but does not require LYMain.c/LYMainLoop.c coding.
+             */
+            CTRACE(tfp, " unsupported URL prefix before '://' %s\n",
+                        "- no recovery (sleep till getfile warning)");
+            /* this warning will be issued from getfile() stage:
+               CTRACE_SLEEP(MessageSecs);
+               HTUserMsg2(WWW_BAD_ADDR_MESSAGE, doc->address);
+               CTRACE(tfp,"\n");
+             */
+       } else {
+            LYConvertToURL(href);
+       }
     }
     if ((temp = HTParse(*href, "", PARSE_ALL)) != NULL && *temp != '\0')
        StrAllocCopy(*href, temp);
@@ -3824,6 +3945,8 @@
  *  as a file URL if the string resolves to a file or
  *  directory on the local system, otherwise as an
  *  http URL. - FM
+ *
+ *  Called by LYEnsureAbsoluteURL() only.
  */
 PUBLIC void LYConvertToURL ARGS1(
        char **,        AllocatedString)
diff -u old/lyutils.h ./lyutils.h
--- old/lyutils.h       Wed May  5 17:34:00 1999
+++ ./lyutils.h Thu May 13 16:15:10 1999
@@ -100,6 +100,7 @@
 extern void LYCloseTempFP PARAMS((FILE *fp));
 extern void LYConvertToURL PARAMS((char **AllocatedString));
 extern void LYDoCSI PARAMS((char *url, CONST char *comment, char **csi));
+extern void LYFillLocalFileURL PARAMS((char **href, CONST char *base));
 extern void LYEnsureAbsoluteURL PARAMS((char **href, CONST char *name));
 extern void LYFakeZap PARAMS((BOOL set));
 extern void LYLocalFileToURL PARAMS((char **target, CONST char *source));