Andrew Eikum : shlwapi: Ignore the hash of HTML URLs in UrlCombine.

[Alexandre Julliard]

Module: wine
Branch: master
Commit: 386cb659df0f7e7616d62332f6a116b8a0c74a8b
URL:    http://source.winehq.org/git/wine.git/?a=commit;h=386cb659df0f7e7616d62332f6a116b8a0c74a8b

Author: Andrew Eikum <aeikum at codeweavers.com>
Date:   Fri Feb 26 19:20:42 2010 -0600

shlwapi: Ignore the hash of HTML URLs in UrlCombine.

---

 dlls/shlwapi/tests/url.c |    5 ++++
 dlls/shlwapi/url.c       |   48 +++++++++++++++++++++++++++++++++++++++------
 2 files changed, 46 insertions(+), 7 deletions(-)

diff --git a/dlls/shlwapi/tests/url.c b/dlls/shlwapi/tests/url.c
index 3d86ae7..31df64e 100644
--- a/dlls/shlwapi/tests/url.c
+++ b/dlls/shlwapi/tests/url.c
@@ -278,6 +278,11 @@ static const TEST_URL_COMBINE TEST_COMBINE[] = {
     {"http://www.winehq.org/tests/../tests/", "/tests10/..", URL_DONT_SIMPLIFY, S_OK, "http://www.winehq.org/tests10/.."},
     {"http://www.winehq.org/tests/../", "tests11", URL_DONT_SIMPLIFY, S_OK, "http://www.winehq.org/tests/../tests11"},
     {"file:///C:\\dir\\file.txt", "test.txt", 0, S_OK, "file:///C:/dir/test.txt"},
+    {"file:///C:\\dir\\file.txt#hash\\hash", "test.txt", 0, S_OK, "file:///C:/dir/file.txt#hash/test.txt"},
+    {"file:///C:\\dir\\file.html#hash\\hash", "test.html", 0, S_OK, "file:///C:/dir/test.html"},
+    {"file:///C:\\dir\\file.htm#hash\\hash", "test.htm", 0, S_OK, "file:///C:/dir/test.htm"},
+    {"file:///C:\\dir\\file.hTmL#hash\\hash", "test.hTmL", 0, S_OK, "file:///C:/dir/test.hTmL"},
+    {"file:///C:\\dir.html\\file.txt#hash\\hash", "test.txt", 0, S_OK, "file:///C:/dir.html/file.txt#hash/test.txt"},
     {"C:\\winehq\\winehq.txt", "C:\\Test\\test.txt", 0, S_OK, "file:///C:/Test/test.txt"},
     {"http://www.winehq.org/test/", "test%20file.txt", 0, S_OK, "http://www.winehq.org/test/test%20file.txt"},
     {"http://www.winehq.org/test/", "test%20file.txt", URL_FILE_USE_PATHURL, S_OK, "http://www.winehq.org/test/test%20file.txt"},
diff --git a/dlls/shlwapi/url.c b/dlls/shlwapi/url.c
index 4a93b23..2fdae3d 100644
--- a/dlls/shlwapi/url.c
+++ b/dlls/shlwapi/url.c
@@ -630,6 +630,8 @@ HRESULT WINAPI UrlCombineW(LPCWSTR pszBase, LPCWSTR pszRelative,
 	process_case = 1;
     }
     else do {
+        BOOL manual_search = FALSE;
+
         /* mk is a special case */
         if(base.nScheme == URL_SCHEME_MK) {
             static const WCHAR wsz[] = {':',':',0};
@@ -659,13 +661,45 @@ HRESULT WINAPI UrlCombineW(LPCWSTR pszBase, LPCWSTR pszRelative,
             }
         }
 
-        /* Change .sizep2 to not have the last leaf in it,
-         * Note: we need to start after the location (if it exists)
-         */
-        work = strrchrW((base.pszSuffix+sizeloc), '/');
-        if (work) {
-            len = (DWORD)(work - base.pszSuffix + 1);
-            base.cchSuffix = len;
+        /* If there is a '#' and the characters immediately preceeding it are
+         * ".htm[l]", then begin looking for the last leaf starting from
+         * the '#'. Otherwise the '#' is not meaningful and just start
+         * looking from the end. */
+        if ((work = strchrW(base.pszSuffix + sizeloc, '#'))) {
+            const WCHAR htmlW[] = {'.','h','t','m','l',0};
+            const int len_htmlW = 5;
+            const WCHAR htmW[] = {'.','h','t','m',0};
+            const int len_htmW = 4;
+
+            if (work - base.pszSuffix > len_htmW * sizeof(WCHAR)) {
+                work -= len_htmW;
+                if (strncmpiW(work, htmW, len_htmW) == 0)
+                    manual_search = TRUE;
+                work += len_htmW;
+            }
+
+            if (!manual_search &&
+                    work - base.pszSuffix > len_htmlW * sizeof(WCHAR)) {
+                work -= len_htmlW;
+                if (strncmpiW(work, htmlW, len_htmlW) == 0)
+                    manual_search = TRUE;
+                work += len_htmlW;
+            }
+        }
+
+        if (manual_search) {
+            /* search backwards starting from the current position */
+            while (*work != '/' && work > base.pszSuffix + sizeloc)
+                --work;
+            if (work > base.pszSuffix + sizeloc)
+                base.cchSuffix = work - base.pszSuffix + 1;
+        }else {
+            /* search backwards starting from the end of the string */
+            work = strrchrW((base.pszSuffix+sizeloc), '/');
+            if (work) {
+                len = (DWORD)(work - base.pszSuffix + 1);
+                base.cchSuffix = len;
+            }
         }
 
 	/*