mikhail/llama.cpp-mtp-turboquant
1
0
Fork 0
Code Issues Pull Requests Actions 29 Packages Projects Releases Wiki Activity

cli : fix reasoning responses in CLI (#18961)

Browse Source 
* cli : fix reasoning responses in CLI

* fix build

* fix build (2)
This commit is contained in:
Xuan-Son Nguyen
2026-01-20 18:23:25 +01:00
committed by GitHub
parent d1e3556481
commit 2c1f199653
12 changed files with 417 additions and 390 deletions
Download Patch File Download Diff File Expand all files Collapse all files
common/chat-parser.cpp
+3 -3
View File
@@ -129,7 +129,7 @@ static void parse_json_tool_calls(
}
}
common_chat_msg_parser::common_chat_msg_parser(const std::string & input, bool is_partial, const common_chat_syntax & syntax)
common_chat_msg_parser::common_chat_msg_parser(const std::string & input, bool is_partial, const common_chat_parser_params & syntax)
: input_(input), is_partial_(is_partial), syntax_(syntax)
{
result_.role = "assistant";
@@ -1611,7 +1611,7 @@ static void common_chat_parse(common_chat_msg_parser & builder) {
builder.finish();
}
common_chat_msg common_chat_parse(const std::string & input, bool is_partial, const common_chat_syntax & syntax) {
common_chat_msg common_chat_parse(const std::string & input, bool is_partial, const common_chat_parser_params & syntax) {
if (syntax.format == COMMON_CHAT_FORMAT_PEG_SIMPLE ||
syntax.format == COMMON_CHAT_FORMAT_PEG_NATIVE ||
syntax.format == COMMON_CHAT_FORMAT_PEG_CONSTRUCTED) {
@@ -1635,7 +1635,7 @@ common_chat_msg common_chat_parse(const std::string & input, bool is_partial, co
return msg;
}
common_chat_msg common_chat_peg_parse(const common_peg_arena & parser, const std::string & input, bool is_partial, const common_chat_syntax & syntax) {
common_chat_msg common_chat_peg_parse(const common_peg_arena & parser, const std::string & input, bool is_partial, const common_chat_parser_params & syntax) {
if (parser.empty()) {
throw std::runtime_error("Failed to parse due to missing parser definition.");
}
common/chat-parser.h
+4 -4
View File
@@ -5,7 +5,7 @@
#include "json-partial.h"
#include "regex-partial.h"
#include <nlohmann/json.hpp>
#include <nlohmann/json_fwd.hpp>
#include <optional>
#include <string>
@@ -19,20 +19,20 @@ class common_chat_msg_partial_exception : public std::runtime_error {
class common_chat_msg_parser {
std::string input_;
bool is_partial_;
common_chat_syntax syntax_;
common_chat_parser_params syntax_; // TODO: rename to params
std::string healing_marker_;
size_t pos_ = 0;
common_chat_msg result_;
public:
common_chat_msg_parser(const std::string & input, bool is_partial, const common_chat_syntax & syntax);
common_chat_msg_parser(const std::string & input, bool is_partial, const common_chat_parser_params & syntax);
const std::string & input() const { return input_; }
size_t pos() const { return pos_; }
const std::string & healing_marker() const { return healing_marker_; }
const bool & is_partial() const { return is_partial_; }
const common_chat_msg & result() const { return result_; }
const common_chat_syntax & syntax() const { return syntax_; }
const common_chat_parser_params & syntax() const { return syntax_; }
void move_to(size_t pos) {
if (pos > input_.size()) {
common/chat.h
+16 -7
View File
@@ -145,7 +145,7 @@ struct common_chat_templates_inputs {
std::vector<common_chat_tool> tools;
common_chat_tool_choice tool_choice = COMMON_CHAT_TOOL_CHOICE_AUTO;
bool parallel_tool_calls = false;
common_reasoning_format reasoning_format = COMMON_REASONING_FORMAT_NONE;
common_reasoning_format reasoning_format = COMMON_REASONING_FORMAT_NONE; // TODO: refactor this to "bool enable_thinking"
bool enable_thinking = true;
std::chrono::system_clock::time_point now = std::chrono::system_clock::now();
std::map<std::string, std::string> chat_template_kwargs;
@@ -165,14 +165,21 @@ struct common_chat_params {
std::string parser;
};
struct common_chat_syntax {
// per-message parsing syntax
// should be derived from common_chat_params
struct common_chat_parser_params {
common_chat_format format = COMMON_CHAT_FORMAT_CONTENT_ONLY;
common_reasoning_format reasoning_format = COMMON_REASONING_FORMAT_NONE;
common_reasoning_format reasoning_format = COMMON_REASONING_FORMAT_NONE; // TODO: refactor this to "bool parse_reasoning"
// Whether reasoning_content should be inlined in the content (e.g. for reasoning_format=deepseek in stream mode)
bool reasoning_in_content = false;
bool thinking_forced_open = false;
bool parse_tool_calls = true;
common_peg_arena parser = {};
common_chat_parser_params() = default;
common_chat_parser_params(const common_chat_params & chat_params) {
format = chat_params.format;
thinking_forced_open = chat_params.thinking_forced_open;
}
};
// Check if the template supplied via "--chat-template" is supported or not. Returns true if it's valid
@@ -213,10 +220,12 @@ std::string common_chat_format_example(
const std::map<std::string, std::string> & chat_template_kwargs);
const char* common_chat_format_name(common_chat_format format);
const char* common_reasoning_format_name(common_reasoning_format format);
common_reasoning_format common_reasoning_format_from_name(const std::string & format);
common_chat_msg common_chat_parse(const std::string & input, bool is_partial, const common_chat_syntax & syntax);
common_chat_msg common_chat_peg_parse(const common_peg_arena & parser, const std::string & input, bool is_partial, const common_chat_syntax & syntax);
common_chat_msg common_chat_parse(const std::string & input, bool is_partial, const common_chat_parser_params & syntax);
common_chat_msg common_chat_peg_parse(const common_peg_arena & parser, const std::string & input, bool is_partial, const common_chat_parser_params & syntax);
// used by arg and server
const char * common_reasoning_format_name(common_reasoning_format format);
common_reasoning_format common_reasoning_format_from_name(const std::string & format);
common_chat_tool_choice common_chat_tool_choice_parse_oaicompat(const std::string & tool_choice);
common/common.h
+1
View File
@@ -284,6 +284,7 @@ struct common_params_diffusion {
};
// reasoning API response format (not to be confused as chat template's reasoning format)
// only used by server
enum common_reasoning_format {
COMMON_REASONING_FORMAT_NONE,
COMMON_REASONING_FORMAT_AUTO, // Same as deepseek, using `message.reasoning_content`
common/json-partial.h
+1
View File
@@ -1,5 +1,6 @@
#pragma once
// TODO: use json_fwd.hpp when possible
#include <nlohmann/json.hpp>
// Healing marker (empty if the JSON was fully parsed / wasn't healed).