Add in builtin tools

Author: aldehir

common/chat.cpp

@@ -1360,23 +1360,32 @@ static void common_chat_parse_gpt_oss(common_chat_msg_parser & builder) {
     static const common_regex end_regex("<\\|end\\|>");
     static const common_regex to_regex(" to=");
     static const common_regex channel_type_regexp("(final|analysis|commentary)");
-    static const common_regex tool_call_regex(
+    static const common_regex user_tool_call_regex(
         "functions\\.([a-zA-Z_][a-zA-Z0-9_]*)\\s?(?:<\\|constrain\\|>([a-zA-Z]+))?<\\|message\\|>"
     );
+    static const common_regex browser_tool_call_regex("browser\\.(search|open|find)[\\s\\S]*<\\|message\\|>");
+    static const common_regex python_tool_call_regex("python\\s?(?:<|constrain|>code)?<\\|message\\|>");
 
-    auto user_function_call = [&]() {
-        if (auto res = builder.try_consume_regex(tool_call_regex)) {
+    auto tool_call = [&]() {
+        if (auto res = builder.try_consume_regex(user_tool_call_regex)) {
             auto name = builder.str(res->groups[1]);
             auto args = builder.consume_rest();
             builder.add_tool_call(name, "", args);
+        } else if (auto res = builder.try_consume_regex(browser_tool_call_regex)) {
+            auto code = builder.consume_rest();
+            LOG_DBG("builtin tool call to python code: %s", code.c_str());
+        } else if (auto res = builder.try_consume_regex(python_tool_call_regex)) {
+            auto name = builder.str(res->groups[1]);
+            auto args = builder.consume_rest();
+            LOG_DBG("builtin tool call to browser.%s %s", name.c_str(), args.c_str());
         } else {
-            throw common_chat_msg_parse_exception("expected user function call");
+            throw common_chat_msg_parse_exception("expected function call");
         }
     };
 
     auto commentary = [&]() {
         if (builder.try_consume_regex(to_regex)) {
-            user_function_call();
+            tool_call();
         } else if (builder.try_consume_regex(message_regex)) {
             if (!builder.try_find_regex(end_regex)) {
                 builder.add_content(builder.consume_rest());
@@ -1395,7 +1404,9 @@ static void common_chat_parse_gpt_oss(common_chat_msg_parser & builder) {
     };
 
     auto analysis = [&]() {
-        if (builder.try_consume_regex(message_regex)) {
+        if (builder.try_consume_regex(to_regex)) {
+            tool_call(); // built-in tools can be called in the analysis channel
+        } else if (builder.try_consume_regex(message_regex)) {
             if (auto res = builder.try_find_regex(end_regex, std::string::npos, false)) {
                 builder.add_reasoning_content(res->prelude);
             } else {
@@ -1425,11 +1436,9 @@ static void common_chat_parse_gpt_oss(common_chat_msg_parser & builder) {
 
     auto start = [&]() {
         if (builder.try_consume_regex(assistant_regex)) {
-            try {
-                channel();
-            } catch (const common_chat_msg_parse_exception & e) {
-                LOG_ERR("Parse error: %s, skipping to next valid start", e.what());
-            }
+            channel();
+        } else {
+            throw common_chat_msg_parse_exception("expected: <|assistant|>");
         }
     };
 
@@ -1439,8 +1448,13 @@ static void common_chat_parse_gpt_oss(common_chat_msg_parser & builder) {
         LOG_ERR("Parse error: %s", e.what());
     }
 
+    // Read in complete messages until done or partial exception raised
     while (builder.try_find_literal("<|start|>")) {
-        start();
+        try {
+            start();
+        } catch (const common_chat_msg_parse_exception & e) {
+            LOG_ERR("Parse error: %s", e.what());
+        }
     }
 
     builder.consume_rest();