From 125d2997dbd9d70a557ca3a92e2504f3e203a5e5 Mon Sep 17 00:00:00 2001
From: DonMoralez <donmoralez@gmail.com>
Date: Sat, 16 Dec 2023 14:12:06 +0200
Subject: [PATCH] Add prefixes sequence check for claude

---
 src/endpoints/backends/chat-completions.js | 28 ++++++++++++++++++++++
 src/endpoints/prompt-converters.js         | 18 +++++++++-----
 2 files changed, 40 insertions(+), 6 deletions(-)

diff --git a/src/endpoints/backends/chat-completions.js b/src/endpoints/backends/chat-completions.js
index 26a0e5d48..75531e766 100644
--- a/src/endpoints/backends/chat-completions.js
+++ b/src/endpoints/backends/chat-completions.js
@@ -40,6 +40,34 @@ async function sendClaudeRequest(request, response) {
         let requestPrompt = convertClaudePrompt(request.body.messages, !request.body.exclude_assistant, request.body.assistant_prefill, isSyspromptSupported, request.body.claude_use_sysprompt, request.body.human_sysprompt_message);
 
         console.log(chalk.green(`${divider}\nClaude request\n`) + chalk.cyan(`PROMPT\n${divider}\n${requestPrompt}\n${divider}`));
+
+        // Check Claude messages sequence and prefixes presence.
+        const sequence = requestPrompt.split('\n').filter(x => x.startsWith('Human:') || x.startsWith('Assistant:'));
+        let humanErrorCount = 0;
+        let assistantErrorCount = 0;
+        let humanFound = sequence.some(line => line.startsWith('Human:'));
+        let assistantFound = sequence.some(line => line.startsWith('Assistant:'));
+
+        for (let i = 0; i < sequence.length - 1; i++) {
+
+            if (sequence[i].startsWith(sequence[i + 1].split(':')[0])) {
+                if (sequence[i].startsWith('Human:')) {
+                    humanErrorCount++;
+                } else if (sequence[i].startsWith('Assistant:')) {
+                    assistantErrorCount++;
+                }
+            }
+        }
+
+        if (!humanFound) {
+            console.log(chalk.red(`${divider}\nWarning: No Human prefix found in the prompt.\n${divider}`));
+        }
+        if (!assistantFound) {
+            console.log(chalk.red(`${divider}\nWarning: No Assistant prefix found in the prompt.\n${divider}`));
+        }
+        if (humanErrorCount > 0 || assistantErrorCount > 0) {
+            console.log(chalk.red(`${divider}\nWarning: Detected incorrect Prefix sequence(s).\nIncorrect 'Human:' prefix(es): ${humanErrorCount}.\nIncorrect 'Assistant:' prefix(es): ${assistantErrorCount}.\nCheck the prompt above and fix it in the sillytavern.\nThe correct sequence should look like this:\nSystem prompt message  <--(for new sysprompt format only)\n       <------------------(Every message start with Assistant:/Human:prefix should have one empty line above)\nHuman:\n\nAssistant:\n\...\n\nHuman:\n\nAssistant:\n${divider}`));
+        }
         const stop_sequences = ['\n\nHuman:', '\n\nSystem:', '\n\nAssistant:'];
 
         // Add custom stop sequences
diff --git a/src/endpoints/prompt-converters.js b/src/endpoints/prompt-converters.js
index 7087ed77a..ce5be3a53 100644
--- a/src/endpoints/prompt-converters.js
+++ b/src/endpoints/prompt-converters.js
@@ -14,26 +14,31 @@ function convertClaudePrompt(messages, addAssistantPostfix, addAssistantPrefill,
     console.log(JSON.stringify(messages, null, 2));
     //Prepare messages for claude.
     if (messages.length > 0) {
+        messages[0].role = 'system';
         //Add the assistant's message to the end of messages.
         if (addAssistantPostfix) {
             messages.push({
                 role: 'assistant',
-                content: addAssistantPrefill || '4',
+                content: addAssistantPrefill || '',
             });
         }
         // Find the index of the first message with an assistant role and check for a "'user' role/Human:" before it.
         let hasUser = false;
+        let hasAssist = false;
         const firstAssistantIndex = messages.findIndex((message, i) => {
-            if (i > 0 && (message.role === 'user' || message.content.includes('Human:'))) {
+            if (i >= 0 && (message.role === 'user' || message.content.includes('\n\nHuman: '))) {
+                if (message.content.includes('\n\nAssistant: ')) {
+                    hasAssist = true;
+                }
                 hasUser = true;
             }
             return message.role === 'assistant' && i > 0;
         });
-        // When 2.1 and 'Use system prompt" checked, switches to system prompt format by setting the first message's role to 'system'.
-        // Also, insert the human's message before the first the assistant one, in case there are no such message or prefix found.
+        // When 2.1 and 'Use system prompt" checked, switches to the system prompt format by setting the first message's role to the 'system'.
+        // Inserts the human's message before the first the assistant one, if there are no such message or prefix found.
         if (withSyspromptSupport && useSystemPrompt) {
             messages[0].role = 'system';
-            if (firstAssistantIndex > 0 && !hasUser) {
+            if (firstAssistantIndex > 0 && (!hasUser || (hasUser && hasAssist))) { //addSysHumanMsg for test
                 messages.splice(firstAssistantIndex, 0, {
                     role: 'user',
                     content: addSysHumanMsg || 'Let\'s get started.',
@@ -50,6 +55,7 @@ function convertClaudePrompt(messages, addAssistantPostfix, addAssistantPrefill,
     }
 
     console.log(JSON.stringify(messages, null, 2));
+
     // Convert messages to requestPrompt.
     let requestPrompt = messages.map((v, i) => {
         // Claude doesn't support message names, so we'll just add them to the message content.
@@ -63,7 +69,7 @@ function convertClaudePrompt(messages, addAssistantPostfix, addAssistantPrefill,
             'user': '\n\nHuman: ',
             'system': i === 0 ? '' : v.name === 'example_assistant' ? '\n\nA: ' : v.name === 'example_user' ? '\n\nH: ' : '\n\n',
             'FixHumMsg': '\n\nFirst message: ',
-        }[v.role] ?? '\n\n4';
+        }[v.role] ?? '\n\n';
 
         return prefix + v.content;
     }).join('');