During import from url take filename from content-disposition

If the http header "Content-Disposition:"  is present take the
filename specified in there as the name of the imported profile,
falling back to scanning the file contents for metadata.

If both filename= and filename*= attributes are present, the
latter takes precedence provided the character set is utf-8.
(Extended attributes as defined in RFC 5987).

In case of import from AS, the behaviour is unchanaged.

Issue: #450.
Signed-off-by: Selva Nair <selva.nair@gmail.com>

as.c

@@ -22,6 +22,7 @@
 #include <windows.h>
 #include <wininet.h>
 #include <stdlib.h>
+#include <shlwapi.h>
 
 #include "config.h"
 #include "localization.h"
@@ -38,6 +39,21 @@
 #define PROFILE_NAME_TOKEN L"# OVPN_ACCESS_SERVER_PROFILE="
 #define FRIENDLY_NAME_TOKEN L"# OVPN_ACCESS_SERVER_FRIENDLY_NAME="
 
+/** Replace characters not allowed in Windows filenames with '_' */
+void
+SanitizeFilename(wchar_t *fname)
+{
+    const wchar_t *reserved = L"<>:\"/\\|?*;"; /* remap these and ascii 1 to 31 */
+    while (*fname) {
+        wchar_t c = *fname;
+        if (c < 32 || wcschr(reserved, c))
+        {
+            *fname = L'_';
+        }
+        ++fname;
+    }
+}
+
 /**
  * Extract profile name from profile content.
  *
@@ -87,14 +103,7 @@ ExtractProfileName(const WCHAR *profile, const WCHAR *default_name, WCHAR *out_n
 
     out_name[out_name_length - 1] = L'\0';
 
-    /* sanitize profile name */
+    SanitizeFilename(out_name);
-    const WCHAR *reserved = L"<>:\"/\\|?*;"; /* remap these and ascii 1 to 31 */
-    while (*out_name) {
-        wchar_t c = *out_name;
-        if (c < 32 || wcschr(reserved, c))
-            *out_name = L'_';
-        ++out_name;
-    }
 
     free(buf);
 }
@@ -295,6 +304,77 @@ GetASUrl(const WCHAR *host, bool autologin, struct UrlComponents *comps)
     comps->path[URL_LEN - 1] = L'\0';
 }
 
+/**
+ * Read content-disposition header and extract file name if any.
+ * Returns true on success, false otherwise.
+ */
+bool
+ExtractFilenameFromHeader(HINTERNET hRequest, wchar_t *name, size_t len)
+{
+    DWORD index = 0;
+    char *buf = NULL;
+    DWORD buflen = 256;
+    bool res = false;
+    UINT codepage = 28591; /* ISO 8859_1 -- the default char set for http header */
+
+    buf = malloc(buflen);
+    if (!buf
+        || (!HttpQueryInfoA(hRequest, HTTP_QUERY_CONTENT_DISPOSITION, buf, &buflen, &index)
+            && GetLastError() != ERROR_INSUFFICIENT_BUFFER))
+    {
+        goto done;
+    }
+
+    if (GetLastError() == ERROR_INSUFFICIENT_BUFFER)
+    {
+        /* try again with more space */
+        free(buf);
+        buf = malloc(buflen);
+        if (!buf
+            || !HttpQueryInfoA(hRequest, HTTP_QUERY_CONTENT_DISPOSITION, buf, &buflen, &index))
+        {
+            goto done;
+        }
+    }
+
+    /* look for filename=<name> */
+    char *p = strtok(buf, ";");
+    char *fn = NULL;
+    for ( ; p; p = strtok(NULL, ";"))
+    {
+        if ((fn = strstr(p, "filename=")) != NULL)
+        {
+            fn += 9;
+            continue;
+        }
+        else if ((fn = strstr(p, "filename*=utf-8''")) != NULL)
+        {
+            fn += 17;
+            UrlUnescapeA(fn, NULL, NULL, URL_UNESCAPE_INPLACE);
+            codepage = CP_UTF8;
+            break; /* we prefer filename*= value */
+        }
+    }
+
+    if (fn && strlen(fn))
+    {
+        StrTrimA(fn, " \""); /* strip leading and trailing spaces and quotes */
+        wchar_t *wfn = WidenEx(codepage, fn);
+        if (wfn)
+        {
+            wcsncpy_s(name, len, wfn, _TRUNCATE);
+            res = true;
+            free(wfn);
+        }
+    }
+
+    SanitizeFilename(name);
+
+done:
+    free(buf);
+    return res;
+}
+
 /**
  * Download profile from a generic URL and save it to a temp file
  *
@@ -467,6 +547,10 @@ again:
     }
 
     WCHAR name[MAX_PATH] = {0};
+    /* read filename from header or from the profile metadata */
+    if (strlen(comps->content_type) == 0 /* AS profile */
+        || !ExtractFilenameFromHeader(hRequest, name, MAX_PATH))
+    {
         WCHAR* wbuf = Widen(buf);
         if (!wbuf) {
             MessageBoxW(hWnd, L"Failed to convert profile content to wchar", _T(PACKAGE_NAME), MB_OK);
@@ -474,6 +558,7 @@ again:
         }
         ExtractProfileName(wbuf, comps->host, name, MAX_PATH);
         free(wbuf);
+    }
 
     /* save profile content into tmp file */
     DWORD res = GetTempPathW((DWORD)out_path_size, out_path);

misc.c

@@ -466,22 +466,23 @@ CheckFileAccess (const TCHAR *path, int access)
     return ret;
 }
 
-/*
+/**
- * Convert a NUL terminated utf8 string to widechar. The caller must free
+ * Convert a NUL terminated narrow string to wide string using
+ * specified codepage. The caller must free
  * the returned pointer. Return NULL on error.
  */
 WCHAR *
-Widen(const char *utf8)
+WidenEx(UINT codepage, const char *str)
 {
     WCHAR *wstr = NULL;
-    if (!utf8)
+    if (!str)
         return wstr;
 
-    int nch = MultiByteToWideChar(CP_UTF8, 0, utf8, -1, NULL, 0);
+    int nch = MultiByteToWideChar(codepage, 0, str, -1, NULL, 0);
     if (nch > 0)
         wstr = malloc(sizeof(WCHAR) * nch);
     if (wstr)
-        nch =  MultiByteToWideChar(CP_UTF8, 0, utf8, -1, wstr, nch);
+        nch =  MultiByteToWideChar(codepage, 0, str, -1, wstr, nch);
 
     if (nch == 0 && wstr)
     {
@@ -492,6 +493,15 @@ Widen(const char *utf8)
     return wstr;
 }
 
+/*
+ * Same as WidenEx with codepage = UTF8
+ */
+WCHAR *
+Widen(const char *utf8)
+{
+    return WidenEx(CP_UTF8, utf8);
+}
+
 /* Return false if input contains any characters in exclude */
 BOOL
 validate_input(const WCHAR *input, const WCHAR *exclude)

misc.h

@@ -46,6 +46,7 @@ BOOL CheckFileAccess (const TCHAR *path, int access);
 BOOL Base64Encode(const char *input, int input_len, char **output);
 int Base64Decode(const char *input, char **output);
 WCHAR *Widen(const char *utf8);
+WCHAR *WidenEx(UINT codepage, const char *utf8);
 BOOL validate_input(const WCHAR *input, const WCHAR *exclude);
 /* Concatenate two wide strings with a separator */
 void wcs_concat2(WCHAR *dest, int len, const WCHAR *src1, const WCHAR *src2, const WCHAR *sep);