simplify logic by combining regex

Author: aldehir

common/chat.cpp

@@ -1351,13 +1351,11 @@ static common_chat_params common_chat_params_init_gpt_oss(const common_chat_temp
     return data;
 }
 static void common_chat_parse_gpt_oss(common_chat_msg_parser & builder) {
-    static const common_regex assistant_regex("assistant");
     static const common_regex message_regex("<\\|message\\|>");
-    static const common_regex channel_regex("<\\|channel\\|>");
-    static const common_regex start_regex("<\\|start\\|>");
+    static const common_regex channel_regex("<\\|channel\\|>(final|analysis|commentary)");
+    static const common_regex start_regex("<\\|start\\|>assistant");
     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 user_tool_call_regex(
         "functions\\.([a-zA-Z_][a-zA-Z0-9_]*)\\s*(?:(?:<\\|constrain\\|>)?([a-zA-Z]+))?\\s*<\\|message\\|>"
     );
@@ -1378,7 +1376,24 @@ static void common_chat_parse_gpt_oss(common_chat_msg_parser & builder) {
         return builder.consume_rest();
     };
 
+    auto try_consume_message = [&]() {
+        if (builder.try_consume_regex(message_regex)) {
+            if (!builder.try_find_regex(end_regex)) {
+                builder.add_content(builder.consume_rest());
+            }
+            return true;
+        }
+        return false;
+    };
+
     auto tool_call = [&]() {
+        if (!builder.syntax().parse_tool_calls) {
+            // Move back to the start and consume up to the next message
+            builder.move_to(message_start_pos);
+            builder.add_content(consume_until_next(message_start_pos + 1));
+            return;
+        }
+
         if (auto res = builder.try_consume_regex(user_tool_call_regex)) {
             auto name = builder.str(res->groups[1]);
             if (auto args = builder.try_consume_json_with_dumped_args({{}})) {
@@ -1396,30 +1411,12 @@ static void common_chat_parse_gpt_oss(common_chat_msg_parser & builder) {
 
     auto commentary = [&]() {
         if (builder.try_consume_regex(to_regex)) {
-            if (builder.syntax().parse_tool_calls) {
-                tool_call();
-            } else {
-                // Move back to the start and consume up to the next message
-                builder.move_to(message_start_pos);
-                builder.add_content(consume_until_next(message_start_pos + 1));
-            }
-        } else if (builder.try_consume_regex(message_regex)) {
-            if (!builder.try_find_regex(end_regex)) {
-                builder.add_content(builder.consume_rest());
-            }
-        } else {
+            tool_call();
+        } else if (!try_consume_message()) {
             throw common_chat_msg_parse_exception("expected: \" to=\" or <|message|>, got: " + consume_until_next());
         }
     };
 
-    auto final = [&]() {
-        if (builder.try_consume_regex(message_regex)) {
-            builder.add_content(builder.consume_rest());
-        } else {
-            throw common_chat_msg_parse_exception("expected: <|message|>, got: " + consume_until_next());
-        }
-    };
-
     auto analysis = [&]() {
         if (builder.try_consume_regex(to_regex)) {
             tool_call(); // built-in tools can be called in the analysis channel
@@ -1438,44 +1435,34 @@ static void common_chat_parse_gpt_oss(common_chat_msg_parser & builder) {
     };
 
     auto channel = [&]() {
-        if (auto channel = builder.try_consume_regex(channel_regex)) {
-            channel_start_pos = channel->groups[0].begin;
-            if (auto res = builder.try_consume_regex(channel_type_regexp)) {
-                auto type = builder.str(res->groups[0]);
-                if (type == "analysis") {
-                    analysis();
-                } else if (type == "final") {
-                    final();
-                } else if (type == "commentary") {
-                    commentary();
+        if (auto res = builder.try_consume_regex(channel_regex)) {
+            channel_start_pos = res->groups[0].begin;
+            auto type = builder.str(res->groups[1]);
+            if (type == "analysis") {
+                analysis();
+            } else if (type == "commentary") {
+                commentary();
+            } else if (type == "final") {
+                if (!try_consume_message()) {
+                    throw common_chat_msg_parse_exception("expected: <|message|>, got: " + consume_until_next());
                 }
-            } else {
-                throw common_chat_msg_parse_exception("expected one of: analysis, final, commentary, got: " + consume_until_next());
             }
         } else {
             throw common_chat_msg_parse_exception("expected: <|channel|>, got: " + consume_until_next());
         }
     };
 
-    auto start = [&]() {
-        if (builder.try_consume_regex(assistant_regex)) {
-            channel();
-        } else {
-            throw common_chat_msg_parse_exception("expected: <|assistant|>, got: " + consume_until_next());
-        }
-    };
-
     try {
         channel();
     } catch (const common_chat_msg_parse_exception & e) {
         LOG_DBG("Parse error: %s\n", e.what());
     }
 
     // Read in complete messages until done or partial exception raised
-    while (auto res = builder.try_find_literal("<|start|>")) {
+    while (auto res = builder.try_consume_regex(start_regex)) {
         message_start_pos = res->groups[0].begin;
         try {
-            start();
+            channel();
         } catch (const common_chat_msg_parse_exception & e) {
             LOG_DBG("Parse error: %s\n", e.what());
         }