Merge pull request #30754 from vespa-engine/bratseth/improve-embedder-errors

Improve embedder error messages

5 files changed, 24 insertions, 17 deletions

diff --git a/config-model/src/main/java/com/yahoo/schema/document/SDField.java b/config-model/src/main/java/com/yahoo/schema/document/SDField.java
index 5cc51f9fedc..538cb56d210 100644
--- a/config-model/src/main/java/com/yahoo/schema/document/SDField.java
+++ b/config-model/src/main/java/com/yahoo/schema/document/SDField.java
@@ -636,7 +636,7 @@ public class SDField extends Field implements TypedKey, ImmutableSDField {
 
     /**
      * Returns the stemming setting of this field.
-     * Default is determined by the owning search definition.
+     * Default is determined by the owning schema.
      *
      * @return the stemming setting of this, or null, to use the default
      */
@@ -644,7 +644,7 @@ public class SDField extends Field implements TypedKey, ImmutableSDField {
     public Stemming getStemming() { return stemming; }
 
     /**
-     * Whether this field should be stemmed in this search definition
+     * Whether this field should be stemmed in this schema
      */
     @Override
     public Stemming getStemming(Schema schema) {
diff --git a/container-search/src/main/java/com/yahoo/prelude/querytransform/StemmingSearcher.java b/container-search/src/main/java/com/yahoo/prelude/querytransform/StemmingSearcher.java
index e8350831381..f27e9bbc281 100644
--- a/container-search/src/main/java/com/yahoo/prelude/querytransform/StemmingSearcher.java
+++ b/container-search/src/main/java/com/yahoo/prelude/querytransform/StemmingSearcher.java
@@ -383,8 +383,7 @@ public class StemmingSearcher extends Searcher {
     }
 
     private int getWeight(Item block) {
-        if (block instanceof AndSegmentItem
-                && ((AndSegmentItem) block).getItemCount() > 0) {
+        if (block instanceof AndSegmentItem && ((AndSegmentItem) block).getItemCount() > 0) {
             return ((AndSegmentItem) block).getItem(0).getWeight();
         } else {
             return block.getWeight();
diff --git a/container-search/src/main/java/com/yahoo/search/schema/internal/TensorConverter.java b/container-search/src/main/java/com/yahoo/search/schema/internal/TensorConverter.java
index 4f1fa44b982..4485e538e13 100644
--- a/container-search/src/main/java/com/yahoo/search/schema/internal/TensorConverter.java
+++ b/container-search/src/main/java/com/yahoo/search/schema/internal/TensorConverter.java
@@ -70,10 +70,13 @@ public class TensorConverter {
                 embedder = requireEmbedder(embedderId);
                 argument = matcher.group(2);
         } else if (embedders.isEmpty()) {
-            throw new IllegalStateException("No embedders provided");  // should never happen
+            throw new IllegalArgumentException("No embedders provided");  // should never happen
         } else if (embedders.size() > 1) {
-            throw new IllegalArgumentException("Multiple embedders are provided but no embedder id is given. " +
-                                               "Valid embedders are " + validEmbedders(embedders));
+            String usage = "Usage: embed(embedder-id, 'text'). " + embedderIds(embedders);
+            if (! argument.contains("\"") && ! argument.contains("'"))
+                throw new IllegalArgumentException("Multiple embedders are provided but the string to embed is not quoted. " + usage);
+            else
+                throw new IllegalArgumentException("Multiple embedders are provided but no embedder id is given. " + usage);
         } else {
             var entry = embedders.entrySet().stream().findFirst().get();
             embedderId = entry.getKey();
@@ -84,8 +87,7 @@ public class TensorConverter {
 
     private Embedder requireEmbedder(String embedderId) {
         if ( ! embedders.containsKey(embedderId))
-            throw new IllegalArgumentException("Can't find embedder '" + embedderId + "'. " +
-                                               "Valid embedders are " + validEmbedders(embedders));
+            throw new IllegalArgumentException("Can't find embedder '" + embedderId + "'. " + embedderIds(embedders));
         return embedders.get(embedderId);
     }
 
@@ -108,11 +110,11 @@ public class TensorConverter {
         return referencedValue.toString();
     }
 
-    private static String validEmbedders(Map<String, Embedder> embedders) {
+    private static String embedderIds(Map<String, Embedder> embedders) {
         List<String> embedderIds = new ArrayList<>();
-        embedders.forEach((key, value) -> embedderIds.add(key));
+        embedders.forEach((key, value) -> embedderIds.add("'" + key + "'"));
         embedderIds.sort(null);
-        return String.join(",", embedderIds);
+        return "Available embedder ids are " + String.join(", ", embedderIds) + ".";
     }
 
 }
diff --git a/container-search/src/test/java/com/yahoo/search/query/RankProfileInputTest.java b/container-search/src/test/java/com/yahoo/search/query/RankProfileInputTest.java
index 3cdfc95ea0e..f66d5395a02 100644
--- a/container-search/src/test/java/com/yahoo/search/query/RankProfileInputTest.java
+++ b/container-search/src/test/java/com/yahoo/search/query/RankProfileInputTest.java
@@ -161,7 +161,7 @@ public class RankProfileInputTest {
         assertEmbedQuery("embed(emb1, '" + text + "')", embedding1, embedders);
         assertEmbedQuery("embed(emb1, \"" + text + "\")", embedding1, embedders);
         assertEmbedQueryFails("embed(emb2, \"" + text + "\")", embedding1, embedders,
-                "Can't find embedder 'emb2'. Valid embedders are emb1");
+                "Can't find embedder 'emb2'. Available embedder ids are 'emb1'.");
 
         embedders = Map.of(
                 "emb1", new MockEmbedder(text, Language.UNKNOWN, embedding1),
@@ -170,7 +170,10 @@ public class RankProfileInputTest {
         assertEmbedQuery("embed(emb1, '" + text + "')", embedding1, embedders);
         assertEmbedQuery("embed(emb2, '" + text + "')", embedding2, embedders);
         assertEmbedQueryFails("embed(emb3, \"" + text + "\")", embedding1, embedders,
-                "Can't find embedder 'emb3'. Valid embedders are emb1,emb2");
+                "Can't find embedder 'emb3'. Available embedder ids are 'emb1', 'emb2'.");
+        assertEmbedQueryFails("embed(emb3, text)", embedding1, embedders,
+                              "Multiple embedders are provided but the string to embed is not quoted. " +
+                              "Usage: embed(embedder-id, 'text'). Available embedder ids are 'emb1', 'emb2'.");
 
         // And with specified language
         embedders = Map.of(
diff --git a/container-search/src/test/java/com/yahoo/search/query/profile/types/test/QueryProfileTypeTestCase.java b/container-search/src/test/java/com/yahoo/search/query/profile/types/test/QueryProfileTypeTestCase.java
index c08b2478f86..0ad04090ae6 100644
--- a/container-search/src/test/java/com/yahoo/search/query/profile/types/test/QueryProfileTypeTestCase.java
+++ b/container-search/src/test/java/com/yahoo/search/query/profile/types/test/QueryProfileTypeTestCase.java
@@ -30,7 +30,7 @@ import java.util.Map;
 import static org.junit.jupiter.api.Assertions.*;
 
 /**
- * tests query profiles with/and types
+ * Tests query profiles with/and types
  *
  * @author bratseth
  */
@@ -496,7 +496,7 @@ public class QueryProfileTypeTestCase {
         assertEmbedQuery("embed(emb1, '" + text + "')", embedding1, embedders);
         assertEmbedQuery("embed(emb1, \"" + text + "\")", embedding1, embedders);
         assertEmbedQueryFails("embed(emb2, \"" + text + "\")", embedding1, embedders,
-                "Can't find embedder 'emb2'. Valid embedders are emb1");
+                "Can't find embedder 'emb2'. Available embedder ids are 'emb1'.");
 
         embedders = Map.of(
                 "emb1", new MockEmbedder(text, Language.UNKNOWN, embedding1),
@@ -504,8 +504,11 @@ public class QueryProfileTypeTestCase {
         );
         assertEmbedQuery("embed(emb1, '" + text + "')", embedding1, embedders);
         assertEmbedQuery("embed(emb2, '" + text + "')", embedding2, embedders);
+        assertEmbedQueryFails("embed(emb2, text)", embedding1, embedders,
+                              "Multiple embedders are provided but the string to embed is not quoted. " +
+                              "Usage: embed(embedder-id, 'text'). Available embedder ids are 'emb1', 'emb2'.");
         assertEmbedQueryFails("embed(emb3, \"" + text + "\")", embedding1, embedders,
-                "Can't find embedder 'emb3'. Valid embedders are emb1,emb2");
+                "Can't find embedder 'emb3'. Available embedder ids are 'emb1', 'emb2'.");
 
         // And with specified language
         embedders = Map.of(