Changeset 5177

Timestamp:

Nov 2, 2009, 10:52:06 AM (13 years ago)

Author:

Nicklas Nordborg

Message:

Fixes #1427: Raw data type lookup by name in rawbioassay batch importer

Location:

trunk/src

Files:

• plugins/core/net/sf/basedb/plugins/batchimport/AbstractItemImporter.java (modified) (2 diffs)
• plugins/core/net/sf/basedb/plugins/batchimport/RawBioAssayImporter.java (modified) (4 diffs)
• test/data/test.batchimport.rawbioassays.txt (modified) (1 diff)

trunk/src/plugins/core/net/sf/basedb/plugins/batchimport/AbstractItemImporter.java

@@ -71 +71 @@
 import net.sf.basedb.core.ProtocolType;
 import net.sf.basedb.core.RawDataType;
+import net.sf.basedb.core.RawDataTypes;
 import net.sf.basedb.core.RequestInformation;
 import net.sf.basedb.core.Sample;
@@ -1566 +1567 @@
 
   /**
+    Find a matching raw data type. This method will look in the following order:
+    <ol>
+    <li>The raw data type locked to the platform variant:
+      {@link PlatformVariant#getRawDataType()}
+    <li>The raw data type locked to the platform:
+      {@link Platform#getRawDataType()}
+    <li>Any registered raw data type with the given 'id':
+      {@link RawDataTypes#getRawDataType(String)}
+    <li>Any registered raw data type with the given 'name':
+      {@link RawDataTypes#getRawDataTypes()}
+    <li>The project default raw data type:
+      {@link #getProjectDefaultRawDataType(DbControl)}
+    </ol>
+    If a raw data type is found by a check, the remaining checks are 
+    ignored. If no item is found or if multiple items are found (in the 'name' lookup)
+    the settings ({@link #failIfMultipleFoundReferences}
+    and {@link #failIfNotFoundReference}) determine if an exception is thrown
+    or not.
+    @param platform The current platform or null to not use a platform for lookup
+    @param variant The current platform variant or null to not use a variant for lookup
+    @param identifier The 'id' or 'name' of a raw data type, or null to not use this
+      for lookup
+    @return A raw data type or null if not found
+    @since 2.15
+  */
+  protected RawDataType findRawDataType(DbControl dc, Platform platform, PlatformVariant variant, String identifier)
+  {
+    RawDataType rdt = null;
+    
+    // If a variant/platform is given, check them first
+    if (variant != null) rdt = variant.getRawDataType();
+    if (rdt == null && platform != null) rdt = platform.getRawDataType();
+    
+    // Find by 'id'
+    if (rdt == null && identifier != null) rdt = RawDataTypes.getRawDataType(identifier);
+      
+    // Find by 'name'
+    if (rdt == null && identifier != null)
+    {
+      for (RawDataType tmp : RawDataTypes.getRawDataTypes())
+      {
+        if (identifier.equals(tmp.getName()))
+        {
+          if (rdt == null)
+          {
+            rdt = tmp;
+            if (!failIfMultipleFoundReferences)
+            {
+              // It is safe to ignore the rest
+              break;
+            }
+          }
+          else 
+          {
+            // This is the second raw data type with the same name that we find
+            throw new BaseException("Fount >1 raw data type [name=" + identifier + "]");
+          }
+        }
+      }
+    }
+    
+    // Check the project default raw data type
+    if (rdt == null) rdt = getProjectDefaultRawDataType(dc);
+    if (rdt == null && failIfNotFoundReference)
+    {
+      throw new ItemNotFoundException("RawDataType[id or name=" + identifier + "]");
+    }
+    return rdt;
+  }
+  
+  /**
     Cache that holds loaded/created items. The cache key is item type + identifier,
     eq. SAMPLE:My sample, EXTRACT:An extract. The same item may be cached under more 

trunk/src/plugins/core/net/sf/basedb/plugins/batchimport/RawBioAssayImporter.java

@@ -42 +42 @@
 import net.sf.basedb.core.RawBioAssay;
 import net.sf.basedb.core.RawDataType;
-import net.sf.basedb.core.RawDataTypes;
 import net.sf.basedb.core.Scan;
 import net.sf.basedb.core.Software;
@@ -312 +311 @@
     Platform platform = null;
     PlatformVariant variant = null;
-    RawDataType rawDataType = null;
+    String rawDataTypeId = null;
     
     if (platformMapper != null) 
@@ -330 +329 @@
       variant = getProjectDefaultVariant(dc);
     }
-    if (variant != null && rawDataType == null) rawDataType = variant.getRawDataType();
-    if (platform != null && rawDataType == null) rawDataType = platform.getRawDataType();
-    if (rawDataType == null)
-    {
-      if (rawDataTypeMapper != null)
-      {
-        rawDataType = RawDataTypes.getSafeRawDataType(rawDataTypeMapper.getValue(data));
-      }
-      else
-      {
-        rawDataType = getProjectDefaultRawDataType(dc);
-      }
-    }
+    if (rawDataTypeMapper != null) rawDataTypeId = rawDataTypeMapper.getValue(data);
+    RawDataType rawDataType = findRawDataType(dc, platform, variant, rawDataTypeId);
     
     RawBioAssay rba = null;
@@ -399 +387 @@
         rba.setArrayDesign(designFromScan);
       }
-      Platform platform = null;
-      PlatformVariant variant = null;
-      RawDataType rawDataType = null;
-      if (platformMapper != null) 
+      // Only do this when updating since the same thing is done in 'createItem'
+      if (rba.isInDatabase())
       {
-        platform = findPlatform(dc, FallbackIdMethod.NAME_OR_EXTERNALID_OR_ID, platformMapper.getValue(data));
-      }
-      if (variantMapper != null) 
-      {
-        variant = findVariant(dc, platform, FallbackIdMethod.NAME_OR_EXTERNALID_OR_ID, variantMapper.getValue(data));
-      }
-      if (variant != null && rawDataType == null) rawDataType = variant.getRawDataType();
-      if (platform != null && rawDataType == null) rawDataType = platform.getRawDataType();
-      if (rawDataType == null && rawDataTypeMapper != null)
-      {
-        rawDataType = RawDataTypes.getSafeRawDataType(rawDataTypeMapper.getValue(data));
-      }
-      if (variant != null)
-      {
-        rba.setVariant(variant, rawDataType);
-      }
-      else if (platform != null)
-      {
-        rba.setPlatform(platform, rawDataType);
+        Platform platform = null;
+        PlatformVariant variant = null;
+        String rawDataTypeId = null;
+        if (platformMapper != null) 
+        {
+          platform = findPlatform(dc, FallbackIdMethod.NAME_OR_EXTERNALID_OR_ID, platformMapper.getValue(data));
+        }
+        else
+        {
+          platform = rba.getPlatform();
+        }
+        if (variantMapper != null) 
+        {
+          variant = findVariant(dc, platform, FallbackIdMethod.NAME_OR_EXTERNALID_OR_ID, variantMapper.getValue(data));
+        }
+        else
+        {
+          variant = rba.getVariant();
+        }
+        if (rawDataTypeMapper != null) 
+        {
+          rawDataTypeId = rawDataTypeMapper.getValue(data);
+        }
+        else
+        {
+          rawDataTypeId = rba.getRawDataType().getId();
+        }
+        RawDataType rawDataType = findRawDataType(dc, platform, variant, rawDataTypeId);
+        if (variant != null)
+        {
+          rba.setVariant(variant, rawDataType);
+        }
+        else if (platform != null)
+        {
+          rba.setPlatform(platform, rawDataType);
+        }
       }
     }

trunk/src/test/data/test.batchimport.rawbioassays.txt

@@ -1 +1 @@
 Name  Description Array index Platform  Variant Raw data type Scan  Protocol  Software  Array design  File type File
 Raw data 1  The first data set  1 Generic   genepix Scan 1  Feature Feature Generic.1 Generic raw data  test.gpr
-Raw data 2  The second  1 Generic Variant genepix Scan 2  Feature Feature Generic.2 Generic raw data  test.gpr
+Raw data 2  The second  1 Generic Variant GenePix Scan 2  Feature Feature Generic.2 Generic raw data  test.gpr
 Raw data 3.1  From same hyb 1 Affymetrix      Scan 3  Feature Feature Affy  CEL file  test.cel
 Raw data 3.2  From same hyb 2 Affymetrix      Scan 3  Feature Feature Affy  CEL file  test.cel