add parse error checking

Author: aldehir

common/chat.cpp

@@ -1353,54 +1353,61 @@ 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) {
-    //harmony_msg_parser parser(builder);
-    //parser.parse();
-
+    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 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(
         "functions\\.([a-zA-Z_][a-zA-Z0-9_]*)\\s?(?:<\\|constrain\\|>([a-zA-Z]+))?<\\|message\\|>"
     );
 
-    auto user_function = [&]() {
+    auto user_function_call = [&]() {
         if (auto res = builder.try_consume_regex(tool_call_regex)) {
             auto name = builder.str(res->groups[1]);
             auto args = builder.consume_rest();
             builder.add_tool_call(name, "", args);
+        } else {
+            throw common_chat_msg_parse_exception("expected user function call");
         }
     };
 
     auto commentary = [&]() {
         if (builder.try_consume_regex(to_regex)) {
-            user_function();
-        }
-
-        if (builder.try_consume_literal("<|message|>")) {
+            user_function_call();
+        } else if (builder.try_consume_regex(message_regex)) {
             if (!builder.try_find_regex(end_regex)) {
                 builder.add_content(builder.consume_rest());
             }
+        } else {
+            throw common_chat_msg_parse_exception("expected: \" to=\" or <|message|>");
         }
     };
 
     auto final = [&]() {
-        if (builder.try_consume_literal("<|message|>")) {
+        if (builder.try_consume_regex(message_regex)) {
             builder.add_content(builder.consume_rest());
+        } else {
+            throw common_chat_msg_parse_exception("expected: <|message|>");
         }
     };
 
     auto analysis = [&]() {
-        if (builder.try_consume_literal("<|message|>")) {
+        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 {
                 builder.add_reasoning_content(builder.consume_rest());
             }
+        } else {
+            throw common_chat_msg_parse_exception("expected: <|message|>");
         }
     };
 
     auto channel = [&]() {
-        if (builder.try_consume_literal("<|channel|>")) {
+        if (builder.try_consume_regex(channel_regex)) {
             if (auto res = builder.try_consume_regex(channel_type_regexp)) {
                 auto type = builder.str(res->groups[0]);
                 if (type == "analysis") {
@@ -1411,12 +1418,18 @@ static void common_chat_parse_gpt_oss(common_chat_msg_parser & builder) {
                     commentary();
                 }
             }
+        } else {
+            throw common_chat_msg_parse_exception("expected: <|channel|>");
         }
     };
 
     auto start = [&]() {
-        if (builder.try_consume_literal("assistant")) {
-            channel();
+        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());
+            }
         }
     };
 
@@ -1427,14 +1440,10 @@ static void common_chat_parse_gpt_oss(common_chat_msg_parser & builder) {
     }
 
     while (builder.try_find_literal("<|start|>")) {
-        try {
-            start();
-        } catch(const common_chat_msg_parse_exception & e) {
-            LOG_ERR("Parse error: %s, skipping to next valid token", e.what());
-        }
+        start();
     }
 
-    builder.add_content(builder.consume_rest());
+    builder.consume_rest();
 }
 
 static common_chat_params common_chat_params_init_firefunction_v2(const common_chat_template & tmpl, const struct templates_params & inputs) {