[PATCH v5 4/4] shlwapi: fixup leading and trailing typos in UrlFixupW

[Thomas Csovscity]

From: Thomas Csovscity

---
 dlls/kernelbase/path.c   | 76 +++++++++++++++++++++++++++++++++++++++-
 dlls/shlwapi/tests/url.c | 62 ++++++++++++++++----------------
 2 files changed, 106 insertions(+), 32 deletions(-)

diff --git a/dlls/kernelbase/path.c b/dlls/kernelbase/path.c
index 86b15944102..17bcd4f8d31 100644
--- a/dlls/kernelbase/path.c
+++ b/dlls/kernelbase/path.c
@@ -4091,7 +4091,7 @@ INT WINAPI UrlCompareW(const WCHAR *url1, const WCHAR *url2, BOOL ignore_slash)
 HRESULT WINAPI UrlFixupW(const WCHAR *url, WCHAR *translatedUrl, DWORD maxChars)
 {
     DWORD src_len, dst_len, len;
-    DWORD pos = 0;
+    DWORD colon_pos = 0, pos = 0;
     WCHAR *helper_str = NULL;
     WCHAR *save_str = translatedUrl;

@@ -4133,7 +4133,81 @@ HRESULT WINAPI UrlFixupW(const WCHAR *url, WCHAR *translatedUrl, DWORD maxChars)
     if (!helper_str)
         return S_FALSE;

+    /*
+     * Colon should not be at the beginning
+     */
+    colon_pos = helper_str-url;
+    if (helper_str && (1 >= colon_pos))
+        return S_FALSE;
+    /*
+     * Check if string fits into maxChars
+     */
+    // TODO: Use own buffer for adjustments, output buffer may not fit before fix
+    if (colon_pos >= maxChars && maxChars < 3)
+        return S_FALSE;
+
+    /*
+     * Set potential scheme
+     */
+    lstrcpynW(save_str, url, pos+2);
+    url = url + colon_pos + 1;
+
+    /*
+     * Find schemes with trailing or leading typos
+     */
+    for (pos=1; pos<ARRAY_SIZE(url_scheme); pos++)
+    {
+        /* http is always prefered before https, if string has typos */
+        len = wcslen(url_scheme[pos]);
+        if (!_wcsnicmp(url_scheme[pos], L"https", len+1))
+        {
+            continue;
+        }
+        if ( (len <= wcslen(save_str)) && (StrStrIW(save_str, url_scheme[pos])) )
+        {
+            /*
+             * check if string fits into maxChars
+             */
+            if (len+1 >= maxChars)
+                return S_FALSE;
+
+            lstrcpynW(save_str, url_scheme[pos], len+1);
+            lstrcatW(save_str, L":");
+            goto scheme_done;
+        }
+    }
+
+    /* Return false in most remaining cases should be safe */
+    return S_FALSE;
+
+    /* Concat scheme + rest */
 scheme_done:
+    /*
+     * Concat L"://"" for ftp, http, https scheme 
+     * else ":"
+     */
+    if ( 0 == lstrcmpW(save_str, L"ftp:") ||
+         0 == lstrcmpW(save_str, L"http:") ||
+         0 == lstrcmpW(save_str, L"https:") )
+    {
+        if ( url[0] == L'\'  || url[0] == L'/' )
+        {
+            url++;
+            if ( url[0] == L'\'  || url[0] == L'/' )
+                url++;
+        }
+        lstrcatW(save_str, L"//");
+    }
+    /*
+     * Remove leading "/", "" and ":" from url.
+     * Output already have them if needed
+     */
+    while ( url[0] == L'\'  || url[0] == L'/' )
+    {
+        lstrcatW(save_str, L"/");
+        url++;
+    }
+
     /* Add the URL path */
     src_len = lstrlenW(url) + 1;
     dst_len = maxChars - lstrlenW(save_str);
diff --git a/dlls/shlwapi/tests/url.c b/dlls/shlwapi/tests/url.c
index 154c0b47bef..660f80a4a27 100644
--- a/dlls/shlwapi/tests/url.c
+++ b/dlls/shlwapi/tests/url.c
@@ -1618,8 +1618,8 @@ static const struct parse_urlfixupw_test_t {
     HRESULT res;
     BOOL todo;
 } parse_urlfixupw_tests[] = {
-    {L"ftp:",              L"ftp://",            S_OK, TRUE},
-    {L"http:",             L"http://",           S_OK, TRUE},
+    {L"ftp:",              L"ftp://",            S_OK},
+    {L"http:",             L"http://",           S_OK},
     {L"gopher:",           L"gopher:",           S_OK},
     {L"mailto:",           L"mailto:",           S_OK},
     {L"news:",             L"news:",             S_OK},
@@ -1628,7 +1628,7 @@ static const struct parse_urlfixupw_test_t {
     {L"wais:",             L"wais:",             S_OK},
     {L"file:",             L"file:",             S_OK},
     {L"mk:",               L"mk:",               S_OK},
-    {L"https:",            L"https://",          S_OK, TRUE},
+    {L"https:",            L"https://",          S_OK},
     {L"shell:",            L"shell:",            S_OK},
     {L"snews:",            L"snews:",            S_OK},
     {L"local:",            L"local:",            S_OK},
@@ -1650,27 +1650,27 @@ static const struct parse_urlfixupw_test_t {
     {L"ftp://",               L"ftp://",             S_OK},

/* typos in front one letter */
-    {L"aftp://",              L"ftp://",             S_OK, TRUE},
-    {L"ahttp://",             L"http://",            S_OK, TRUE},
-    {L"agopher://",           L"gopher://",          S_OK, TRUE},
-    {L"amailto://",           L"mailto://",          S_OK, TRUE},
+    {L"aftp://",              L"ftp://",             S_OK},
+    {L"ahttp://",             L"http://",            S_OK},
+    {L"agopher://",           L"gopher://",          S_OK},
+    {L"amailto://",           L"mailto://",          S_OK},
     {L"anews://",             L"snews://",           S_OK, TRUE},
-    {L"anntp://",             L"nntp://",            S_OK, TRUE},
-    {L"atelnet://",           L"telnet://",          S_OK, TRUE},
-    {L"awais://",             L"wais://",            S_OK, TRUE},
-    {L"afile://",             L"file://",            S_OK, TRUE},
-    {L"amk://",               L"mk://",              S_OK, TRUE},
-    {L"ahttps://",            L"http://",            S_OK, TRUE},
-    {L"ashell://",            L"shell://",           S_OK, TRUE},
+    {L"anntp://",             L"nntp://",            S_OK},
+    {L"atelnet://",           L"telnet://",          S_OK},
+    {L"awais://",             L"wais://",            S_OK},
+    {L"afile://",             L"file://",            S_OK},
+    {L"amk://",               L"mk://",              S_OK},
+    {L"ahttps://",            L"http://",            S_OK},
+    {L"ashell://",            L"shell://",           S_OK},
     {L"asnews://",            L"snews://",           S_OK, TRUE},
-    {L"alocal://",            L"local://",           S_OK, TRUE},
-    {L"ajavascript://",       L"javascript://",      S_OK, TRUE},
-    {L"avbscript://",         L"vbscript://",        S_OK, TRUE},
-    {L"aabout://",            L"about://",           S_OK, TRUE},
-    {L"ares://",              L"res://",             S_OK, TRUE},
+    {L"alocal://",            L"local://",           S_OK},
+    {L"ajavascript://",       L"javascript://",      S_OK},
+    {L"avbscript://",         L"vbscript://",        S_OK},
+    {L"aabout://",            L"about://",           S_OK},
+    {L"ares://",              L"res://",             S_OK},
     {L"ams-shell-rooted://",  L"ms-shell-rooted://", S_OK, TRUE},
     {L"ams-shell-idlist://",  L"ms-shell-idlist://", S_OK, TRUE},
-    {L"ahcp://",              L"hcp://",             S_OK, TRUE},
+    {L"ahcp://",              L"hcp://",             S_OK},

/* typos in front two letters */
     {L"bahttp://",             L"about://",          S_OK, TRUE},
@@ -1712,11 +1712,11 @@ static const struct parse_urlfixupw_test_t {
     {L"newsssss://", L"snews://",  S_OK, TRUE},

/* typos letters behind*/
-    {L"mks://",        L"mk://",       S_OK, TRUE},
-    {L"mkss://",       L"mk://",       S_OK, TRUE},
-    {L"mailtooo://",   L"mailto://",   S_OK, TRUE},
-    {L"mailtoooo://",  L"mailto://",   S_OK, TRUE},
-    {L"mailtooooo://", L"mailto://",   S_OK, TRUE},
+    {L"mks://",        L"mk://",       S_OK},
+    {L"mkss://",       L"mk://",       S_OK},
+    {L"mailtooo://",   L"mailto://",   S_OK},
+    {L"mailtoooo://",  L"mailto://",   S_OK},
+    {L"mailtooooo://", L"mailto://",   S_OK},

/* typos fixing two letters */
     {L"cc:", L"hcp:",      S_OK, TRUE},
@@ -1837,11 +1837,11 @@ static const struct parse_urlfixupw_test_t {

/* taken from example in documentation */
     {L"http://www.microsoft.com", L"http://www.microsoft.com",   S_OK},
-    {L"http:www.microsoft.com",   L"http://www.microsoft.com",   S_OK, TRUE},
+    {L"http:www.microsoft.com",   L"http://www.microsoft.com",   S_OK},
     {L"mail:[e-mail]", L"mailto:[e-mail]", S_OK, TRUE},
     {L"htpp:wwwmicrosoft.com",    L"http://wwwmicrosoft.com",    S_OK, TRUE},
     {L"htps:\www.microsoft.com", L"http://www.microsoft.com",   S_OK, TRUE},
-    {L"http:[e-mail]", L"http://[e-mail]", S_OK, TRUE},
+    {L"http:[e-mail]", L"http://[e-mail]", S_OK},
     {L"abc:def",                  NULL,                          S_FALSE, TRUE},
     {L"[e-mail]",      NULL,                          S_FALSE, TRUE},

@@ -1871,10 +1871,10 @@ static const struct parse_urlfixupw_test_t {
     {L"hftpz:", L"http://", S_OK, TRUE},
     {L"hztpz:", L"http://", S_OK, TRUE},

-    {L"httpz:",     L"http://", S_OK, TRUE},
-    {L"httpsz:",    L"http://", S_OK, TRUE},
-    {L"httpzz:",    L"http://", S_OK, TRUE},
-    {L"httpzzz:",   L"http://", S_OK, TRUE},
+    {L"httpz:",     L"http://", S_OK},
+    {L"httpsz:",    L"http://", S_OK},
+    {L"httpzz:",    L"http://", S_OK},
+    {L"httpzzz:",   L"http://", S_OK},
     {L"httpzzzz:",  NULL,       S_FALSE, TRUE},
     {L"httpzzzzz:", NULL,       S_FALSE, TRUE},