Merge pull request #23878 from vespa-engine/bratseth/moodel-config

Bratseth/model config

4 files changed, 74 insertions, 93 deletions

diff --git a/config-model/src/main/java/com/yahoo/vespa/model/builder/xml/dom/DomConfigPayloadBuilder.java b/config-model/src/main/java/com/yahoo/vespa/model/builder/xml/dom/DomConfigPayloadBuilder.java
index 7ff01cbf82e..9390986c0c4 100644
--- a/config-model/src/main/java/com/yahoo/vespa/model/builder/xml/dom/DomConfigPayloadBuilder.java
+++ b/config-model/src/main/java/com/yahoo/vespa/model/builder/xml/dom/DomConfigPayloadBuilder.java
@@ -52,19 +52,19 @@ public class DomConfigPayloadBuilder {
 
     public static ConfigDefinitionKey parseConfigName(Element configE) {
         if (!configE.getNodeName().equals("config")) {
-            throw new ConfigurationRuntimeException("The root element must be 'config', but was '" + configE.getNodeName() + "'.");
+            throw new ConfigurationRuntimeException("The root element must be 'config', but was '" + configE.getNodeName() + "'");
         }
 
         if (!configE.hasAttribute("name")) {
             throw new ConfigurationRuntimeException
-                    ("The 'config' element must have a 'name' attribute that matches the name of the config definition.");
+                    ("The 'config' element must have a 'name' attribute that matches the name of the config definition");
         }
 
         String elementString = configE.getAttribute("name");
         if (!elementString.contains(".")) {
             throw new ConfigurationRuntimeException("The config name '" + elementString +
-                                                            "' contains illegal characters. Only names with the pattern " +
-                                                            namespacePattern.pattern() + "." + namePattern.pattern() + " are legal.");
+                                                    "' contains illegal characters. Only names with the pattern " +
+                                                    namespacePattern.pattern() + "." + namePattern.pattern() + " are legal.");
         }
 
         Tuple2<String, String> t = ConfigUtils.getNameAndNamespaceFromString(elementString);
@@ -73,28 +73,26 @@ public class DomConfigPayloadBuilder {
 
         if (!validName(xmlName)) {
             throw new ConfigurationRuntimeException("The config name '" + xmlName +
-                    "' contains illegal characters. Only names with the pattern " + namePattern.toString() + " are legal.");
+                                                    "' contains illegal characters. Only names with the pattern " +
+                                                    namePattern.toString() + " are legal.");
         }
 
         if (!validNamespace(xmlNamespace)) {
             throw new ConfigurationRuntimeException("The config namespace '" + xmlNamespace +
-                    "' contains illegal characters. Only namespaces with the pattern " + namespacePattern.toString() + " are legal.");
+                                                    "' contains illegal characters. Only namespaces with the pattern " +
+                                                    namespacePattern.toString() + " are legal.");
         }
         return new ConfigDefinitionKey(xmlName, xmlNamespace);
     }
 
     private static boolean validName(String name) {
         if (name == null) return false;
-
-        Matcher m = namePattern.matcher(name);
-        return m.matches();
+        return namePattern.matcher(name).matches();
     }
 
     private static boolean validNamespace(String namespace) {
         if (namespace == null) return false;
-
-        Matcher m = namespacePattern.matcher(namespace);
-        return m.matches();
+        return namespacePattern.matcher(namespace).matches();
     }
 
     private String extractName(Element element) {
@@ -118,12 +116,11 @@ public class DomConfigPayloadBuilder {
         return buf.toString();
     }
 
-    /**
-     * Parse leaf value in an xml tree
-     */
+    /** Parse leaf value in an xml tree. */
     private void parseLeaf(Element element, ConfigPayloadBuilder payloadBuilder, String parentName) {
         String name = extractName(element);
         String value = XML.getValue(element);
+        var definition = payloadBuilder.getConfigDefinition();
         if (value == null) {
             throw new ConfigurationRuntimeException("Element '" + name + "' must have either children or a value");
         }
@@ -136,8 +133,14 @@ public class DomConfigPayloadBuilder {
             } else {
                 payloadBuilder.getArray(parentName).append(value);
             }
-        } else {
-            // leaf scalar, e.g. <intVal>3</intVal>
+        }
+        else if (definition != null && definition.getModelDefs().containsKey(name)) { // model field special syntax
+            String modelString = XML.attribute("model-id", element).orElse("\"\"");
+            modelString += " " + XML.attribute("url", element).orElse("\"\"");
+            modelString += " " + XML.attribute("path", element).orElse("\"\"");
+            payloadBuilder.setField(name, modelString);
+        }
+        else { // leaf value: <myValueName>value</myValue>
             payloadBuilder.setField(name, value);
         }
     }
@@ -196,8 +199,8 @@ public class DomConfigPayloadBuilder {
                 parseComplex(currElem, children, payloadBuilder, parentName);
             }
         } catch (Exception exception) {
-            throw new ConfigurationRuntimeException("Error parsing element at " + XML.getNodePath(currElem, " > ") + ": " +
-                    Exceptions.toMessageString(exception));
+            throw new ConfigurationRuntimeException("Error parsing element at " + XML.getNodePath(currElem, " > ") +
+                                                    ": " + Exceptions.toMessageString(exception));
         }
     }
 
diff --git a/config-model/src/main/java/com/yahoo/vespa/model/container/xml/ContainerModelBuilder.java b/config-model/src/main/java/com/yahoo/vespa/model/container/xml/ContainerModelBuilder.java
index 1cf73c1ed03..0bf586a089f 100644
--- a/config-model/src/main/java/com/yahoo/vespa/model/container/xml/ContainerModelBuilder.java
+++ b/config-model/src/main/java/com/yahoo/vespa/model/container/xml/ContainerModelBuilder.java
@@ -957,7 +957,7 @@ public class ContainerModelBuilder extends ConfigModelBuilder<ContainerModel> {
     private static void addConfiguredComponents(DeployState deployState, ContainerCluster<? extends Container> cluster,
                                                 Element parent, String componentName) {
         for (Element component : XML.getChildren(parent, componentName)) {
-            component = ModelConfigTransformer.transform(deployState, component);
+            ModelIdResolver.resolveModelIds(component, deployState.isHosted());
             cluster.addComponent(new DomComponentBuilder().build(deployState, cluster, component));
         }
     }
diff --git a/config-model/src/main/java/com/yahoo/vespa/model/container/xml/ModelConfigTransformer.java b/config-model/src/main/java/com/yahoo/vespa/model/container/xml/ModelConfigTransformer.java
deleted file mode 100644
index 0065a582145..00000000000
--- a/config-model/src/main/java/com/yahoo/vespa/model/container/xml/ModelConfigTransformer.java
+++ /dev/null
@@ -1,73 +0,0 @@
-// Copyright Yahoo. Licensed under the terms of the Apache 2.0 license. See LICENSE in the project root.
-package com.yahoo.vespa.model.container.xml;
-
-import com.yahoo.config.model.deploy.DeployState;
-import com.yahoo.text.XML;
-import org.w3c.dom.Element;
-
-import java.util.Map;
-import java.util.stream.Collectors;
-
-/**
- * Translates model references in component configs.
- *
- * @author lesters
- * @author bratseth
- */
-public class ModelConfigTransformer {
-
-    private static final Map<String, String> providedModels =
-            Map.of("minilm-l6-v2", "https://data.vespa.oath.cloud/onnx_models/sentence_all_MiniLM_L6_v2.onnx",
-                   "bert-base-uncased", "https://data.vespa.oath.cloud/onnx_models/bert-base-uncased-vocab.txt");
-
-    // Until we have optional path parameters, use services.xml as it is guaranteed to exist
-    private final static String dummyPath = "services.xml";
-
-    /**
-     * Transforms the &lt;embedder ...&gt; element to component configuration.
-     *
-     * @param deployState the deploy state - as config generation can depend on context
-     * @param component the XML element containing the &lt;embedder ...&gt;
-     * @return a new XML element containting the &lt;component ...&gt; configuration
-     */
-    public static Element transform(DeployState deployState, Element component) {
-        for (Element config : XML.getChildren(component, "config")) {
-            for (Element value : XML.getChildren(config))
-                transformModelValue(value, config, deployState.isHosted());
-        }
-        return component;
-    }
-
-    /** Expans a model config value into regular config values. */
-    private static void transformModelValue(Element value, Element config, boolean hosted) {
-        if (value.hasAttribute("path")) {
-            addChild(value.getTagName() + "Url", "", config);
-            addChild(value.getTagName() + "Path", value.getAttribute("path"), config);
-            config.removeChild(value);
-        }
-        else if (value.hasAttribute("id") && hosted) {
-            addChild(value.getTagName() + "Url", modelIdToUrl(value.getAttribute("id")), config);
-            addChild(value.getTagName() + "Path", dummyPath, config);
-            config.removeChild(value);
-        }
-        else if (value.hasAttribute("url")) {
-            addChild(value.getTagName() + "Url", value.getAttribute("url"), config);
-            addChild(value.getTagName() + "Path", dummyPath, config);
-            config.removeChild(value);
-        }
-    }
-
-    private static void addChild(String name, String value, Element parent) {
-        Element element = parent.getOwnerDocument().createElement(name);
-        element.setTextContent(value);
-        parent.appendChild(element);
-    }
-
-    private static String modelIdToUrl(String id) {
-        if ( ! providedModels.containsKey(id))
-            throw new IllegalArgumentException("Unknown embedder model '" + id + "'. Available models are [" +
-                                               providedModels.keySet().stream().sorted().collect(Collectors.joining(", ")) + "]");
-        return providedModels.get(id);
-    }
-
-}
diff --git a/config-model/src/main/java/com/yahoo/vespa/model/container/xml/ModelIdResolver.java b/config-model/src/main/java/com/yahoo/vespa/model/container/xml/ModelIdResolver.java
new file mode 100644
index 00000000000..463832ea263
--- /dev/null
+++ b/config-model/src/main/java/com/yahoo/vespa/model/container/xml/ModelIdResolver.java
@@ -0,0 +1,51 @@
+// Copyright Yahoo. Licensed under the terms of the Apache 2.0 license. See LICENSE in the project root.
+package com.yahoo.vespa.model.container.xml;
+
+import com.yahoo.config.model.deploy.DeployState;
+import com.yahoo.text.XML;
+import org.w3c.dom.Element;
+
+import java.util.Map;
+import java.util.stream.Collectors;
+
+/**
+ * Replaces model id references in configs by their url.
+ *
+ * @author lesters
+ * @author bratseth
+ */
+public class ModelIdResolver {
+
+    private static final Map<String, String> providedModels =
+            Map.of("minilm-l6-v2", "https://data.vespa.oath.cloud/onnx_models/sentence_all_MiniLM_L6_v2.onnx",
+                   "bert-base-uncased", "https://data.vespa.oath.cloud/onnx_models/bert-base-uncased-vocab.txt");
+
+    /**
+     * Finds any config values of type 'model' below the given config element and
+     * supplies the url attribute of them if a model id is specified and hosted is true
+     * (regardless of whether an url is already specified).
+     *
+     * @param component the XML element of any component
+     */
+    public static void resolveModelIds(Element component, boolean hosted) {
+        if ( ! hosted) return;
+        for (Element config : XML.getChildren(component, "config")) {
+            for (Element value : XML.getChildren(config))
+                transformModelValue(value);
+        }
+    }
+
+    /** Expands a model config value into regular config values. */
+    private static void transformModelValue(Element value) {
+        if ( ! value.hasAttribute("model-id")) return;
+        value.setAttribute("url", modelIdToUrl(value.getTagName(), value.getAttribute("model-id")));
+    }
+
+    private static String modelIdToUrl(String valueName, String modelId) {
+        if ( ! providedModels.containsKey(modelId))
+            throw new IllegalArgumentException("Unknown model id '" + modelId + "' on '" + valueName + "'. Available models are [" +
+                                               providedModels.keySet().stream().sorted().collect(Collectors.joining(", ")) + "]");
+        return providedModels.get(modelId);
+    }
+
+}