feat(huggingface): setting for removing prompt from generated commit …

…message

Closes #294

CHANGELOG.md

@@ -7,6 +7,7 @@
 - More options for configuring LLM clients.
 - **Use the chosen LLM client icon as the generate commit message action's icon.**
 - Option to stop the commit message generation by clicking the action icon again.
+- Setting for HuggingFace client to automatically remove prompt from the generated commit message.
 
 ### Fixed
 

src/main/kotlin/com/github/blarc/ai/commits/intellij/plugin/settings/clients/LLMClientConfiguration.kt

@@ -40,6 +40,10 @@ abstract class LLMClientConfiguration(
         getSharedState().modelIds.add(modelId)
     }
 
+    open fun setCommitMessage(commitMessage: CommitMessage, prompt: String, result: String) {
+        commitMessage.setCommitMessage(result)
+    }
+
     abstract fun generateCommitMessage(commitWorkflowHandler: AbstractCommitWorkflowHandler<*, *>, commitMessage: CommitMessage, project: Project)
 
     abstract fun getGenerateCommitMessageJob(): Job?

src/main/kotlin/com/github/blarc/ai/commits/intellij/plugin/settings/clients/LLMClientService.kt

@@ -65,7 +65,7 @@ abstract class LLMClientService<C : LLMClientConfiguration>(private val cs: Coro
 
                 makeRequest(clientConfiguration, prompt, onSuccess = {
                     withContext(Dispatchers.EDT) {
-                        commitMessage.setCommitMessage(it)
+                        clientConfiguration.setCommitMessage(commitMessage, prompt, it)
                     }
                     AppSettings2.instance.recordHit()
                 }, onError = {

src/main/kotlin/com/github/blarc/ai/commits/intellij/plugin/settings/clients/huggingface/HuggingFaceClientConfiguration.kt

@@ -28,6 +28,8 @@ class HuggingFaceClientConfiguration : LLMClientConfiguration(
     var tokenIsStored: Boolean = false
     @Transient
     var token: String? = null
+    @Attribute
+    var removePrompt: Boolean = false
 
     companion object {
         const val CLIENT_NAME = "HuggingFace"
@@ -45,6 +47,15 @@ class HuggingFaceClientConfiguration : LLMClientConfiguration(
         return HuggingFaceClientSharedState.getInstance()
     }
 
+    override fun setCommitMessage(commitMessage: CommitMessage, prompt: String, result: String) {
+        var newResult = result
+        if (removePrompt) {
+            // https://github.com/Blarc/ai-commits-intellij-plugin/issues/294
+            newResult = result.substring(prompt.length+1)
+        }
+        super.setCommitMessage(commitMessage, prompt, newResult)
+    }
+
     override fun generateCommitMessage(commitWorkflowHandler: AbstractCommitWorkflowHandler<*, *>, commitMessage: CommitMessage, project: Project) {
         return HuggingFaceClientService.getInstance().generateCommitMessage(this, commitWorkflowHandler, commitMessage, project)
     }
@@ -65,6 +76,7 @@ class HuggingFaceClientConfiguration : LLMClientConfiguration(
         copy.timeout = timeout
         copy.waitForModel = waitForModel
         copy.maxNewTokens = maxNewTokens
+        copy.removePrompt = removePrompt
         return copy
     }
 

src/main/kotlin/com/github/blarc/ai/commits/intellij/plugin/settings/clients/huggingface/HuggingFaceClientPanel.kt

@@ -16,6 +16,7 @@ class HuggingFaceClientPanel private constructor(
     private val tokenPasswordField = JBPasswordField()
     private val maxNewTokensTextField = JBTextField()
     private val waitForModelCheckBox = JBCheckBox()
+    private val removePrompt = JBCheckBox()
 
     constructor(configuration: HuggingFaceClientConfiguration) : this(configuration, HuggingFaceClientService.getInstance())
 
@@ -27,6 +28,7 @@ class HuggingFaceClientPanel private constructor(
         temperatureRow()
         maxNewTokens()
         waitForModel()
+        removePrompt()
         verifyRow()
     }
 
@@ -40,6 +42,7 @@ class HuggingFaceClientPanel private constructor(
         clientConfiguration.token = String(tokenPasswordField.password)
         clientConfiguration.maxNewTokens = maxNewTokensTextField.text.toInt()
         clientConfiguration.waitForModel = waitForModelCheckBox.isSelected
+        clientConfiguration.removePrompt = removePrompt.isSelected
         service.verifyConfiguration(clientConfiguration, verifyLabel)
     }
 
@@ -83,4 +86,18 @@ class HuggingFaceClientPanel private constructor(
                 .align(AlignX.RIGHT)
         }
     }
+
+    private fun Panel.removePrompt() {
+        row {
+            label(message("settings.huggingface.removePrompt"))
+                .widthGroup("label")
+            cell(removePrompt)
+                .bindSelected(clientConfiguration::removePrompt)
+                .resizableColumn()
+                .align(Align.FILL)
+
+            contextHelp(message("settings.huggingface.removePrompt.comment"))
+                .align(AlignX.RIGHT)
+        }
+    }
 }

src/main/resources/messages/AiCommitsBundle.properties

@@ -135,4 +135,6 @@ settings.huggingface.token.example=hf_fKASPPYLkasgjasKwpSnAASRdasdCdAsddsASSDF
 settings.huggingface.maxNewTokens=Max new tokens
 settings.huggingface.waitForModel=Wait for model
 settings.huggingface.waitModel.comment=When a model is warm, it is ready to be used, and you will get a response relatively quickly. However, some models are cold and need to be loaded before they can be used. In that case, you will get a 503 error. Rather than doing many requests until it is loaded, you can wait for the model to be loaded.
+settings.huggingface.removePrompt=Remove prompt
+settings.huggingface.removePrompt.comment=Some models return the result prefixed with prompt. Checking this option will remove the prompt from result.