Merge pull request #73 from seguler/dev2

Encode url to fix #51 Files named with spaces not supported + test

azure-storage-cpp-lite/include/storage_url.h

@@ -9,6 +9,8 @@
 namespace microsoft_azure {
     namespace storage {
 
+        std::string encode_url_path(const std::string& path);
+
         class storage_url {
         public:
             storage_url &set_domain(const std::string &domain) {
@@ -29,6 +31,10 @@ namespace microsoft_azure {
                 return m_path;
             }
 
+            std::string get_encoded_path() const {
+                return encode_url_path(m_path);
+            }
+
             storage_url &add_query(const std::string &name, const std::string &value) {
                 m_query[name].insert(value);
                 return *this;

azure-storage-cpp-lite/src/storage_credential.cpp

@@ -38,7 +38,7 @@ namespace microsoft_azure {
             }
 
             // Canonicalized resource
-            string_to_sign.append("/").append(m_account_name).append(url.get_path());
+            string_to_sign.append("/").append(m_account_name).append(url.get_encoded_path());
             for (const auto &name : url.get_query()) {
                 string_to_sign.append("\n").append(name.first);
                 bool first_value = true;

azure-storage-cpp-lite/src/storage_url.cpp

@@ -2,10 +2,97 @@
 
 namespace microsoft_azure {
     namespace storage {
+        bool is_alnum(char ch)        
+        {
+            return (ch >= 'A' && ch <= 'Z') || (ch >= 'a' && ch <= 'z') || (ch >= '0' && ch <= '9');
+        }
+
+        bool is_unreserved(char ch)
+        {
+            return is_alnum(ch) || ch == '-' || ch == '.' || ch == '_' || ch == '~';
+        }
+        bool is_sub_delim(char ch)
+        {
+            switch(ch)
+            {
+            case '!':
+            case '$':
+            case '&':
+            case '\'':
+            case '(':
+            case ')':
+            case '*':
+            case '+':
+            case ',':
+            case ';':
+            case '=':
+                return true;
+            default:
+                return false;
+            }
+        }
+
+        bool is_path_character(char ch)
+        {
+            return is_unreserved(ch) || is_sub_delim(ch) || ch == '%' || ch == '/' || ch == ':' || ch == '@';
+        }
+        
+        bool is_query_character(char ch)
+        {
+            return is_path_character(ch) || ch == '?';
+        }
+
+        std::string encode_url_path(const std::string& path)
+        {
+            const char* const hex = "0123456789ABCDEF";
+            std::string encoded;
+            for(size_t index = 0; index < path.size(); ++index)
+            {
+                char ch = path[index];
+                if(!is_path_character(ch)
+                    || ch == '%'
+                    || ch == '+'
+                    || ch == '&')
+                {
+                    encoded.push_back('%');
+                    encoded.push_back(hex[ (ch >> 4) & 0xF]);
+                    encoded.push_back(hex[ ch & 0xF ]);
+                }
+                else
+                {
+                    encoded.push_back(ch);
+                }
+            }
+            return encoded;
+        }
+
+        std::string encode_url_query(const std::string& path)
+        {
+            const char* const hex = "0123456789ABCDEF";
+            std::string encoded;
+            for(size_t index = 0; index < path.size(); ++index)
+            {
+                char ch = path[index];
+                if(!is_query_character(ch)
+                    || ch == '%'
+                    || ch == '+'
+                    || ch == '&')
+                {
+                    encoded.push_back('%');
+                    encoded.push_back(hex[ (ch >> 4) & 0xF]);
+                    encoded.push_back(hex[ ch & 0xF ]);
+                }
+                else
+                {
+                    encoded.push_back(ch);
+                }
+            }
+            return encoded;
+        }
 
         std::string storage_url::to_string() const {
             std::string url(m_domain);
-            url.append(m_path);
+            url.append(encode_url_path(m_path));
 
             bool first_query = true;
             for (const auto &q : m_query) {
@@ -17,7 +104,7 @@ namespace microsoft_azure {
                     url.append("&");
                 }
                 for (const auto &value : q.second) {
-                    url.append(q.first).append("=").append(value);
+                    url.append(encode_url_query(q.first)).append("=").append(encode_url_query(value));
                 }
             }
             return url;

blobfuse/tests.py

@@ -53,6 +53,18 @@ class TestFuse(unittest.TestCase):
         os.remove(filepath)
         self.assertEqual(False, os.path.exists(filepath))
 
+    def test_WriteReadSingleFileUnicode(self):
+        file1txt = "你好，世界！"
+        filepath = os.path.join(self.blobstage, "文本: hello?world&we^are%all~together1 .txt");
+        with open(filepath, 'w') as file1blob:
+            file1blob.write(file1txt)
+        self.assertEqual(True, os.path.exists(filepath))
+        with open(filepath, 'r') as file1blob:
+            file1txtrt = file1blob.read()
+            self.assertEqual(file1txt, file1txtrt)
+        os.remove(filepath)
+        self.assertEqual(False, os.path.exists(filepath))
+
     def test_medium_files(self):
         mediumBlobsSourceDir = os.path.join(self.blobstage, "mediumblobs")
         if not os.path.exists(mediumBlobsSourceDir):
@@ -79,6 +91,8 @@ class TestFuse(unittest.TestCase):
             
     def test_directory_operations(self):
         testDir = os.path.join(self.blobstage, "testDirectory")
+        subdir1 = "subDir1"
+        subdir2 = "Thİs!is-a%directory&name @1"
         self.assertFalse(os.path.exists(testDir))
         self.assertFalse(os.path.isdir(testDir))
 
@@ -96,9 +110,9 @@ class TestFuse(unittest.TestCase):
         with open(os.path.join(testDir, "file3"), 'w') as fileblob:
             fileblob.write(filetxt)
 
-        testSubDir1 = os.path.join(testDir, "subdir1")
+        testSubDir1 = os.path.join(testDir, subdir1)
         os.makedirs(testSubDir1)
-        testSubDir2 = os.path.join(testDir, "subdir2")
+        testSubDir2 = os.path.join(testDir, subdir2)
         os.makedirs(testSubDir2)
 
         children = os.listdir(testDir);
@@ -106,8 +120,8 @@ class TestFuse(unittest.TestCase):
         self.assertTrue("file1" in children)
         self.assertTrue("file2" in children)
         self.assertTrue("file3" in children)
-        self.assertTrue("subdir1" in children)
-        self.assertTrue("subdir2" in children)
+        self.assertTrue(subdir1 in children)
+        self.assertTrue(subdir2 in children)
 
         # Directory not empty should throw
         with self.assertRaises(OSError):
@@ -121,7 +135,7 @@ class TestFuse(unittest.TestCase):
 
         self.assertTrue("file1" in children)
         self.assertTrue("file3" in children)
-        self.assertTrue("subdir2" in children)
+        self.assertTrue(subdir2 in children)
 
         os.rmdir(testSubDir2)
         os.remove(os.path.join(testDir, "file1"))