Fix "Invalid tool call arguments passed" in a rare case.

In a rare case, the model may emit a raw string that begins with a valid JSON string. This commit adds unit tests to cover that scenario and fixes the regression introduced during the Kimi-K2 adaptation.

Author: hksdpc255

common/chat-parser-xml-toolcall.cpp

@@ -545,15 +545,12 @@ inline bool parse_xml_tool_calls(common_chat_msg_parser & builder, const struct
                 }
                 builder.move_to(json_end);
                 auto [val_end_size, tc] = try_find_val_end();
-                if (tc && value_json->healing_marker.marker.empty()) {
+                if (tc && all_space(tc->prelude) && value_json->healing_marker.marker.empty()) {
                     if (tc->groups[0].end - tc->groups[0].begin != val_end_size) {
                         gen_partial_args([&](auto &, auto &needle) {arguments[key] = needle;});
                         LOG_DBG("Possible terminated JSON arg_value: %s\n", value_json->json.dump().c_str());
                         throw common_chat_msg_partial_exception("Partial literal: " + gbnf_format_literal(form.val_end) + (form.last_val_end ? gbnf_format_literal(*form.last_val_end) : ""));
-                    }
-                    if (all_space(tc->prelude)) {
-                        arguments[key] = value_json->json;
-                    }
+                    } else arguments[key] = value_json->json;
                 } else builder.move_to(val_start);
             }
 

tests/test-chat.cpp

@@ -2562,6 +2562,30 @@ Hey there!<|im_end|>
                     /*  .format = */ COMMON_CHAT_FORMAT_GLM_4_5,
                     /*  .reasoning_format = */ COMMON_REASONING_FORMAT_DEEPSEEK
             }); });
+            test_parser_with_streaming(
+                simple_assist_msg("", "", "complex_function", "{\"name\":\"John Doe\",\"age\":30,\"active\":true,\"score\":95.5}"),
+                "<tool_call>complex_function\n"
+                "<arg_key>name</arg_key>\n"
+                "<arg_value>John Doe</arg_value>\n"
+                "<arg_key>age</arg_key>\n"
+                "<arg_value>30</arg_value>\n"
+                "<arg_key>active</arg_key>\n"
+                "<arg_value>true</arg_value>\n"
+                "<arg_key>score</arg_key>\n"
+                "<arg_value>95.5</arg_value>\n"
+                "</tool_call>",
+            [&](const std::string &msg) { return common_chat_parse(msg, /* is_partial= */ true, {COMMON_CHAT_FORMAT_GLM_4_5}); });
+        test_parser_with_streaming(
+                simple_assist_msg("", "", "web_search", "{\"query\":\"\\\"From Zero\\\" Linkin Park album tracklist complete songs\",\"limit\":3,\"type\":\"text\"}"),
+                "<tool_call>web_search\n"
+                "<arg_key>query</arg_key>\n"
+                "<arg_value>\"From Zero\" Linkin Park album tracklist complete songs</arg_value>\n"
+                "<arg_key>limit</arg_key>\n"
+                "<arg_value>3</arg_value>\n"
+                "<arg_key>type</arg_key>\n"
+                "<arg_value>text</arg_value>\n"
+                "</tool_call>",
+            [&](const std::string &msg) { return common_chat_parse(msg, /* is_partial= */ true, {COMMON_CHAT_FORMAT_GLM_4_5}); });
 
         // Test interleaved thinking
         test_parser_with_streaming(simple_assist_msg("Hello, world!\n\nWhat's up?", "I'm\nthinkingThinking2", "special_function", "{\"arg1\": 1}"),