Connector/NET commit: r784 - in trunk: . Driver/Source Driver/Source/docs TestSuite/Source

[rburnett]

Modified:
   trunk/CHANGES
   trunk/Driver/Source/Field.cs
   trunk/Driver/Source/MySqlConnectionStringBuilder.cs
   trunk/Driver/Source/NativeDriver.cs
   trunk/Driver/Source/docs/MySqlConnection.xml
   trunk/TestSuite/Source/CharacterSetTests.cs
Log:
Implemented TreatBlobsAsUTF8 mode.

Modified: trunk/CHANGES
===================================================================
--- trunk/CHANGES	2007-07-13 17:43:42 UTC (rev 783)
+++ trunk/CHANGES	2007-07-14 16:27:02 UTC (rev 784)
@@ -12,6 +12,7 @@
     overrides for Membership.CreateUser to work.    
   - Added 'Respect Binary Flags' connection string option to allow existing applications
     to select the old behavior of not always respecting the binary flags of a columns.
+  - Added ability to use blobs to store true UTF-8 data (see help)
     
 Version 5.1.2 - 6/12/2007
   - Fixed integration with the Website Administration Tool.  Before this fix, the test
link

Modified: trunk/Driver/Source/Field.cs
===================================================================
--- trunk/Driver/Source/Field.cs	2007-07-13 17:43:42 UTC (rev 783)
+++ trunk/Driver/Source/Field.cs	2007-07-14 16:27:02 UTC (rev 784)
@@ -22,6 +22,7 @@
 using MySql.Data.Common;
 using MySql.Data.Types;
 using System.Globalization;
+using System.Text.RegularExpressions;
 
 namespace MySql.Data.MySqlClient
 {
@@ -193,32 +194,55 @@
                 {
                     case MySqlDbType.Byte:
                         mySqlDbType = MySqlDbType.UByte;
-                        break;
+                        return;
                     case MySqlDbType.Int16:
                         mySqlDbType = MySqlDbType.UInt16;
-                        break;
+                        return;
                     case MySqlDbType.Int24:
                         mySqlDbType = MySqlDbType.UInt24;
-                        break;
+                        return;
                     case MySqlDbType.Int32:
                         mySqlDbType = MySqlDbType.UInt32;
-                        break;
+                        return;
                     case MySqlDbType.Int64:
                         mySqlDbType = MySqlDbType.UInt64;
-                        break;
+                        return;
                 }
             }
 
-            if (IsBlob && !IsBinary)
+            if (IsBlob)
             {
-                if (type == MySqlDbType.TinyBlob)
-                    mySqlDbType = MySqlDbType.TinyText;
-                else if (type == MySqlDbType.MediumBlob)
-                    mySqlDbType = MySqlDbType.MediumText;
-                else if (type == MySqlDbType.Blob)
-                    mySqlDbType = MySqlDbType.Text;
-                else if (type == MySqlDbType.LongBlob)
-                    mySqlDbType = MySqlDbType.LongText;
+                // handle blob to UTF8 conversion if requested.  This is only activated
+                // on binary blobs
+                if (IsBinary && connection.Settings.TreatBlobsAsUTF8)
+                {
+                    bool convertBlob = true;
+                    Regex includeRegex = connection.Settings.BlobAsUTF8IncludeRegex;
+                    Regex excludeRegex = connection.Settings.BlobAsUTF8ExcludeRegex;
+                    convertBlob &= (includeRegex != null &&
includeRegex.IsMatch(ColumnName));
+                    if (excludeRegex != null && excludeRegex.IsMatch(ColumnName))
+                        convertBlob = false;
+
+                    if (convertBlob)
+                    {
+                        binaryOk = false;
+                        Encoding = System.Text.Encoding.GetEncoding("UTF-8");
+                        charSetIndex = -1;  // lets driver know we are in charge of
encoding
+                        maxLength = 4;
+                    }
+                }
+
+                if (!IsBinary)
+                {
+                    if (type == MySqlDbType.TinyBlob)
+                        mySqlDbType = MySqlDbType.TinyText;
+                    else if (type == MySqlDbType.MediumBlob)
+                        mySqlDbType = MySqlDbType.MediumText;
+                    else if (type == MySqlDbType.Blob)
+                        mySqlDbType = MySqlDbType.Text;
+                    else if (type == MySqlDbType.LongBlob)
+                        mySqlDbType = MySqlDbType.LongText;
+                }
             }
 
             // now determine if we really should be binary

Modified: trunk/Driver/Source/MySqlConnectionStringBuilder.cs
===================================================================
--- trunk/Driver/Source/MySqlConnectionStringBuilder.cs	2007-07-13 17:43:42 UTC (rev 783)
+++ trunk/Driver/Source/MySqlConnectionStringBuilder.cs	2007-07-14 16:27:02 UTC (rev 784)
@@ -24,6 +24,7 @@
 using System.Data.Common;
 using System.Globalization;
 using System.Text;
+using System.Text.RegularExpressions;
 
 namespace MySql.Data.MySqlClient
 {
@@ -44,10 +45,10 @@
         bool oldSyntax, persistSI, usePerfMon, pooling;
         bool allowZeroDatetime, convertZeroDatetime;
         bool useUsageAdvisor, useSSL;
-        bool ignorePrepare;
-        bool useProcedureBodies;
-        bool autoEnlist;
-        bool respectBinaryFlags;
+        bool ignorePrepare, useProcedureBodies;
+        bool autoEnlist, respectBinaryFlags, treatBlobsAsUTF8;
+        string blobAsUtf8IncludePattern, blobAsUtf8ExcludePattern;
+        Regex blobAsUtf8ExcludeRegex, blobAsUtf8IncludeRegex;
 
         static MySqlConnectionStringBuilder()
         {
@@ -82,6 +83,9 @@
             defaultValues.Add(Keyword.UseProcedureBodies, true);
             defaultValues.Add(Keyword.AutoEnlist, true);
             defaultValues.Add(Keyword.RespectBinaryFlags, true);
+            defaultValues.Add(Keyword.BlobAsUTF8ExcludePattern, null);
+            defaultValues.Add(Keyword.BlobAsUTF8IncludePattern, null);
+            defaultValues.Add(Keyword.TreatBlobsAsUTF8, false);
         }
 
         /// <summary>
@@ -486,25 +490,6 @@
         }
 
         /// <summary>
-        /// Gets or sets the character set that should be used for sending queries to the
server.
-        /// </summary>
-#if !CF && !MONO
-        [DisplayName("Character Set")]
-        [Category("Advanced")]
-        [Description("Character set this connection should use")]
-        [RefreshProperties(RefreshProperties.All)]
-#endif
-            public string CharacterSet
-        {
-            get { return charSet; }
-            set
-            {
-                SetValue("Character Set", value); 
-                charSet = value;
-            }
-        }
-
-        /// <summary>
         /// Gets or sets a boolean value indicating if the Usage Advisor should be
enabled.
         /// </summary>
 #if !CF && !MONO
@@ -745,6 +730,88 @@
 
         #endregion
 
+        #region Language and Character Set Properties
+
+        /// <summary>
+        /// Gets or sets the character set that should be used for sending queries to the
server.
+        /// </summary>
+#if !CF && !MONO
+        [DisplayName("Character Set")]
+        [Category("Advanced")]
+        [Description("Character set this connection should use")]
+        [RefreshProperties(RefreshProperties.All)]
+#endif
+        public string CharacterSet
+        {
+            get { return charSet; }
+            set
+            {
+                SetValue("Character Set", value);
+                charSet = value;
+            }
+        }
+
+        /// <summary>
+        /// Indicates whether the driver should treat binary blobs as UTF8
+        /// </summary>
+#if !CF && !MONO
+        [DisplayName("Treat Blobs As UTF8")]
+        [Category("Advanced")]
+        [Description("Should binary blobs be treated as UTF8")]
+        [RefreshProperties(RefreshProperties.All)]
+#endif
+        public bool TreatBlobsAsUTF8
+        {
+            get { return treatBlobsAsUTF8; }
+            set
+            {
+                SetValue("TreatBlobsAsUTF8", value);
+                treatBlobsAsUTF8 = value;
+            }
+        }
+
+        /// <summary>
+        /// Gets or sets the pattern that matches the columns that should be treated as
UTF8
+        /// </summary>
+#if !CF && !MONO
+        [DisplayName("BlobAsUTF8IncludePattern")]
+        [Category("Advanced")]
+        [Description("Pattern that matches columns that should be treated as UTF8")]
+        [RefreshProperties(RefreshProperties.All)]
+#endif
+        public string BlobAsUTF8IncludePattern
+        {
+            get { return blobAsUtf8IncludePattern; }
+            set
+            {
+                SetValue("BlobAsUTF8IncludePattern", value);
+                blobAsUtf8IncludePattern = value;
+                blobAsUtf8IncludeRegex = null;
+            }
+        }
+
+        /// <summary>
+        /// Gets or sets the pattern that matches the columns that should not be treated
as UTF8
+        /// </summary>
+#if !CF && !MONO
+        [DisplayName("BlobAsUTF8ExcludePattern")]
+        [Category("Advanced")]
+        [Description("Pattern that matches columns that should not be treated as UTF8")]
+        [RefreshProperties(RefreshProperties.All)]
+#endif
+        public string BlobAsUTF8ExcludePattern
+        {
+            get { return blobAsUtf8ExcludePattern; }
+            set
+            {
+                SetValue("BlobAsUTF8ExcludePattern", value);
+                blobAsUtf8ExcludePattern = value;
+                blobAsUtf8ExcludeRegex = null;
+            }
+        }
+
+        #endregion
+
         #region Conversion Routines
 
         private static uint ConvertToUInt(object value)
@@ -818,6 +885,32 @@
 
         #endregion
 
+        #region Internal Properties
+
+        internal Regex BlobAsUTF8IncludeRegex
+        {
+            get
+            {
+                if (blobAsUtf8IncludePattern == null) return null;
+                if (blobAsUtf8IncludeRegex == null)
+                    blobAsUtf8IncludeRegex = new Regex(blobAsUtf8IncludePattern);
+                return blobAsUtf8IncludeRegex;
+            }
+        }
+
+        internal Regex BlobAsUTF8ExcludeRegex
+        {
+            get 
+            {
+                if (blobAsUtf8ExcludePattern == null) return null;
+                if (blobAsUtf8ExcludeRegex == null)
+                    blobAsUtf8ExcludeRegex = new Regex(blobAsUtf8ExcludePattern);
+                return blobAsUtf8ExcludeRegex;
+            }
+        }
+
+        #endregion
+
         /// <summary>
         /// Takes a given connection string and returns it, possible
         /// stripping out the password info
@@ -942,6 +1035,13 @@
                     return Keyword.AutoEnlist;
                 case "respect binary flags":
                     return Keyword.RespectBinaryFlags;
+                case "blobasutf8excludepattern":
+                    return Keyword.BlobAsUTF8ExcludePattern;
+                case "blobasutf8includepattern":
+                    return Keyword.BlobAsUTF8IncludePattern;
+                case "treatblobsasutf8":
+                case "treat blobs as utf8":
+                    return Keyword.TreatBlobsAsUTF8;
             }
             throw new ArgumentException(Resources.KeywordNotSupported, key);
         }
@@ -1012,6 +1112,12 @@
                     return AutoEnlist;
                 case Keyword.RespectBinaryFlags:
                     return RespectBinaryFlags;
+                case Keyword.TreatBlobsAsUTF8:
+                    return TreatBlobsAsUTF8;
+                case Keyword.BlobAsUTF8ExcludePattern:
+                    return blobAsUtf8ExcludePattern;
+                case Keyword.BlobAsUTF8IncludePattern:
+                    return blobAsUtf8IncludePattern;
                 default:
                     return null; /* this will never happen */
             }
@@ -1094,6 +1200,12 @@
                     autoEnlist = ConvertToBool(value); break;
                 case Keyword.RespectBinaryFlags:
                     respectBinaryFlags = ConvertToBool(value); break;
+                case Keyword.TreatBlobsAsUTF8:
+                    treatBlobsAsUTF8 = ConvertToBool(value); break;
+                case Keyword.BlobAsUTF8ExcludePattern:
+                    blobAsUtf8ExcludePattern = (string)value; break;
+                case Keyword.BlobAsUTF8IncludePattern:
+                    blobAsUtf8IncludePattern = (string)value; break;
             }
         }
 
@@ -1267,6 +1379,9 @@
         UseSSL,
         UseProcedureBodies,
         AutoEnlist,
-        RespectBinaryFlags
+        RespectBinaryFlags,
+        TreatBlobsAsUTF8,
+        BlobAsUTF8IncludePattern,
+        BlobAsUTF8ExcludePattern
     }
 }

Modified: trunk/Driver/Source/NativeDriver.cs
===================================================================
--- trunk/Driver/Source/NativeDriver.cs	2007-07-13 17:43:42 UTC (rev 783)
+++ trunk/Driver/Source/NativeDriver.cs	2007-07-14 16:27:02 UTC (rev 784)
@@ -713,7 +713,7 @@
                 stream.ReadInteger(2); // reserved
             }
 
-            if (charSets != null)
+            if (charSets != null && field.CharacterSetIndex != -1)
             {
                 CharacterSet cs = CharSetMap.GetChararcterSet(Version, (string)
charSets[field.CharacterSetIndex]);
                 field.MaxLength = cs.byteCount;

Modified: trunk/Driver/Source/docs/MySqlConnection.xml
===================================================================
--- trunk/Driver/Source/docs/MySqlConnection.xml	2007-07-13 17:43:42 UTC (rev 783)
+++ trunk/Driver/Source/docs/MySqlConnection.xml	2007-07-14 16:27:02 UTC (rev 784)
@@ -900,9 +900,23 @@
         Connector/Net 5.0 and earlier.
       </td>
     </tr>
+    <tr>
+      <td>BlobAsUTF8IncludePattern</td>
+      <td>null</td>
+      <td>
+        Pattern that should be used to indicate which blob columns should be treated as
UTF-8.
+      </td>
+    </tr>
+    <tr>
+      <td>BlobAsUTF8ExcludePattern</td>
+      <td>null</td>
+      <td>
+        Pattern that should be used to indicate which blob columns should not be treated
as UTF-8.
+      </td>
+    </tr>
   </table>
-		</div>
-	<para>
+</div>
+<para>
 	The following table lists the valid names for connection pooling values within 
 the <B>ConnectionString</B>. For more information about connection pooling,
see 
 Connection Pooling for the MySql Data Provider.</para>

Modified: trunk/TestSuite/Source/CharacterSetTests.cs
===================================================================
--- trunk/TestSuite/Source/CharacterSetTests.cs	2007-07-13 17:43:42 UTC (rev 783)
+++ trunk/TestSuite/Source/CharacterSetTests.cs	2007-07-14 16:27:02 UTC (rev 784)
@@ -24,6 +24,7 @@
 
 
 
+
 
 
 
@@ -189,6 +190,84 @@
 
 
 
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
 
 
 
\ No newline at end of file