mikhail/llama.cpp-mtp-turboquant
1
0
Fork 0
Code Issues Pull Requests Actions 28 Packages Projects Releases Wiki Activity

autoparser: support case of JSON_NATIVE with per-call markers (test case: Reka-Edge) (#21892)

Browse Source
This commit is contained in:
Piotr Wilkin (ilintar)
2026-04-15 10:51:50 +02:00
committed by GitHub
parent e39eba26f3
commit e1a9a6dcbe
6 changed files with 314 additions and 14 deletions
Download Patch File Download Diff File Expand all files Collapse all files
common/chat-auto-parser-generator.cpp
+13 -4
View File
@@ -198,10 +198,19 @@ common_peg_parser analyze_tools::build_tool_parser_json_native(parser_build_cont
args_field = format.function_field + "." + args_field;
}
auto tools_parser = p.standard_json_tools(
format.section_start, format.section_end, inputs.tools, inputs.parallel_tool_calls,
inputs.tool_choice == COMMON_CHAT_TOOL_CHOICE_REQUIRED, name_field, args_field, format.tools_array_wrapped,
format.fun_name_is_key, format.id_field, format.gen_id_field, format.parameter_order);
auto tools_parser = p.eps();
if (format.section_start.empty() && !format.per_call_start.empty()) {
auto single_tool_parser = p.standard_json_tools(
format.per_call_start, format.per_call_end, inputs.tools, inputs.parallel_tool_calls,
inputs.tool_choice == COMMON_CHAT_TOOL_CHOICE_REQUIRED, name_field, args_field, format.tools_array_wrapped,
format.fun_name_is_key, format.id_field, format.gen_id_field, format.parameter_order);
tools_parser = p.trigger_rule("tool-calls", p.one_or_more(single_tool_parser + p.space()));
} else {
tools_parser = p.standard_json_tools(
format.section_start, format.section_end, inputs.tools, inputs.parallel_tool_calls,
inputs.tool_choice == COMMON_CHAT_TOOL_CHOICE_REQUIRED, name_field, args_field, format.tools_array_wrapped,
format.fun_name_is_key, format.id_field, format.gen_id_field, format.parameter_order);
}
// Handle content wrappers if present
if (ctx.content && ctx.content->is_always_wrapped()) {
common/chat-auto-parser.h
+6 -2
View File
@@ -308,19 +308,23 @@ struct analyze_tools : analyze_base {
private:
// Extract tool calling 'haystack' for further analysis and delegate further analysis based on format
void analyze_tool_calls(const analyze_reasoning & reasoning);
void analyze_tool_calls(const analyze_reasoning & reasoning, bool supports_parallel_tool_calls);
// Analyze format based on position of function and argument name in needle
void analyze_tool_call_format(const std::string & haystack,
const std::string & fun_name_needle,
const std::string & arg_name_needle,
const analyze_reasoning & reasoning);
const analyze_reasoning & reasoning,
bool supports_parallel_tool_calls);
// Analyze specifics of JSON native format (entire tool call is a JSON object)
void analyze_tool_call_format_json_native(const std::string & clean_haystack,
const std::string & fun_name_needle,
const std::string & arg_name_needle);
// Check if parallel calls in JSON native format array wrapped or tag wrapped
void analyze_json_native_parallel_calls();
// Analyze specifics of non-JSON native format (tags for function name or for function name and arguments)
void analyze_tool_call_format_non_json(const std::string & clean_haystack,
const std::string & fun_name_needle);
common/chat-diff-analyzer.cpp
+44 -4
View File
@@ -558,7 +558,7 @@ analyze_tools::analyze_tools(const common_chat_template & tmpl,
: analyze_base(tmpl) {
LOG_DBG(ANSI_ORANGE "Phase 3: Tool call analysis\n" ANSI_RESET);
analyze_tool_calls(reasoning);
analyze_tool_calls(reasoning, caps.supports_parallel_tool_calls);
if (format.mode != tool_format::NONE && format.mode != tool_format::JSON_NATIVE) {
if (caps.supports_parallel_tool_calls) {
@@ -577,7 +577,7 @@ analyze_tools::analyze_tools(const common_chat_template & tmpl,
}
}
void analyze_tools::analyze_tool_calls(const analyze_reasoning & reasoning) {
void analyze_tools::analyze_tool_calls(const analyze_reasoning & reasoning, bool supports_parallel_tool_calls) {
json assistant_no_tools = json{
{ "role", "assistant" },
{ "content", ASSISTANT_MSG }
@@ -611,13 +611,14 @@ void analyze_tools::analyze_tool_calls(const analyze_reasoning & reasoning) {
return;
}
analyze_tool_call_format(tool_section, FUN_FIRST, ARG_FIRST, reasoning);
analyze_tool_call_format(tool_section, FUN_FIRST, ARG_FIRST, reasoning, supports_parallel_tool_calls);
}
void analyze_tools::analyze_tool_call_format(const std::string & haystack,
const std::string & fun_name_needle,
const std::string & arg_name_needle,
const analyze_reasoning & reasoning) {
const analyze_reasoning & reasoning,
bool supports_parallel_tool_calls) {
if (fun_name_needle.empty() || arg_name_needle.empty() || haystack.empty()) {
return;
}
@@ -660,6 +661,9 @@ void analyze_tools::analyze_tool_call_format(const std::string & haystack,
if (format.mode == tool_format::JSON_NATIVE) {
analyze_tool_call_format_json_native(clean_haystack, fun_name_needle, arg_name_needle);
if (supports_parallel_tool_calls) {
analyze_json_native_parallel_calls();
}
} else {
analyze_tool_call_format_non_json(clean_haystack, fun_name_needle);
}
@@ -668,6 +672,42 @@ void analyze_tools::analyze_tool_call_format(const std::string & haystack,
format.per_call_end = trim_whitespace(format.per_call_end);
}
void analyze_tools::analyze_json_native_parallel_calls() {
json assistant_one_tool = json{
{ "role", "assistant" },
{ "content", "" },
{ "tool_calls", json::array({ first_tool_call }) }
};
json assistant_two_tools = json{
{ "role", "assistant" },
{ "content", "" },
{ "tool_calls", json::array({ first_tool_call, second_tool_call }) }
};
template_params params;
params.messages = json::array({ user_msg, assistant_one_tool });
params.tools = tools;
params.add_generation_prompt = false;
params.enable_thinking = true;
auto comparison = compare_variants(
*tmpl, params, [&](template_params & p) { p.messages = json::array({ user_msg, assistant_two_tools }); });
if (!comparison) {
LOG_DBG(ANSI_ORANGE "%s: Template application failed\n" ANSI_RESET, __func__);
return;
}
std::string & second_call = comparison->diff.right;
if (!format.section_start.empty() && second_call.find(format.section_start) != std::string::npos) {
format.per_call_start = format.section_start;
format.per_call_end = format.section_end;
format.section_start.clear();
format.section_end.clear();
}
}
void analyze_tools::analyze_tool_call_format_json_native(const std::string & clean_haystack,
const std::string & fun_name_needle,
const std::string & arg_name_needle) {
common/chat-peg-parser.cpp
+2 -2
View File
@@ -676,7 +676,7 @@ common_peg_parser common_chat_peg_builder::build_json_tools_nested_keys(
ordered_json params = function.contains("parameters") ? function.at("parameters") : ordered_json::object();
auto nested_name = literal("\"" + nested_name_field + "\"") + space() + literal(":") + space() +
literal("\"") + tool_name(literal(name)) + literal("\"");
atomic(literal("\"") + tool_name(literal(name)) + literal("\""));
auto nested_args = literal("\"" + nested_args_field + "\"") + space() + literal(":") + space() +
tool_args(schema(json(), "tool-" + name + "-schema", params));
@@ -744,7 +744,7 @@ common_peg_parser common_chat_peg_builder::build_json_tools_flat_keys(
ordered_json params = function.contains("parameters") ? function.at("parameters") : ordered_json::object();
auto tool_name_ = name_key_parser + space() + literal(":") + space() +
literal("\"") + tool_name(literal(name)) + literal("\"");
atomic(literal("\"") + tool_name(literal(name)) + literal("\""));
auto tool_args_ = args_key_parser + space() + literal(":") + space() +
tool_args(schema(json(), "tool-" + name + "-schema", params));