Felipe97/llama-cpp-compiled

1#include "parsers.h"2 3// MiniCPM5 format:4// - Reasoning: <think>{reasoning}</think> (optional)5// - Tool calls: <function name="foo"><param name="bar">value</param></function>6common_chat_params common_chat_params_init_minicpm5(const common_chat_template &          tmpl,7                                                           const autoparser::generation_params & inputs) {8    common_chat_params data;9 10    data.prompt            = common_chat_template_direct_apply_impl(tmpl, inputs);11    data.generation_prompt = common_chat_template_generation_prompt_impl(tmpl, inputs);12    data.format            = COMMON_CHAT_FORMAT_PEG_NATIVE;13    data.supports_thinking = true;14    data.preserved_tokens  = {15        "<function",16        "<param",17        "</function>",18        "</param>",19        "<think>",20        "</think>",21    };22 23    data.thinking_start_tag = "<think>";24    data.thinking_end_tags  = {"</think>"};25 26    data.message_delimiters = {27        { COMMON_CHAT_ROLE_ASSISTANT, "<|im_start|>assistant"             },28        { COMMON_CHAT_ROLE_TOOL,      "<|im_start|>user\n<tool_response>" },29        { COMMON_CHAT_ROLE_USER,      "<|im_start|>user"                  },30        { COMMON_CHAT_ROLE_SYSTEM,    "<|im_start|>system"                },31    };32 33    auto has_tools           = inputs.tools.is_array() && !inputs.tools.empty();34    auto has_response_format = inputs.json_schema.is_object() && !inputs.json_schema.empty();35    auto extract_reasoning   = inputs.reasoning_format != COMMON_REASONING_FORMAT_NONE;36    auto include_grammar     = has_response_format || (has_tools && inputs.tool_choice != COMMON_CHAT_TOOL_CHOICE_NONE);37 38    if (inputs.has_continuation()) {39        const auto & msg = inputs.continue_msg;40 41        data.generation_prompt = "<|im_start|>assistant\n<think>\n" + msg.reasoning_content;42        if (inputs.continue_final_message == COMMON_CHAT_CONTINUATION_CONTENT) {43            data.generation_prompt += "\n</think>\n\n" + msg.render_content();44        }45 46        data.prompt += data.generation_prompt;47    }48 49    auto parser = build_chat_peg_parser([&](common_chat_peg_builder & p) {50        auto generation_prompt = p.literal("<|im_start|>assistant\n");51 52        auto reasoning = p.eps();53        if (extract_reasoning) {54            reasoning = ("<think>" << p.reasoning(p.until("</think>")) << "</think>") + p.space();55        }56 57        // Response format parser58        if (has_response_format) {59            return generation_prompt + reasoning + p.content(p.schema(p.json(), "response-format", inputs.json_schema));60        }61 62        if (has_tools && inputs.tool_choice != COMMON_CHAT_TOOL_CHOICE_NONE) {63            // CDATA lets a value carry characters that would otherwise close the tag (e.g.64            // </param>); capture the inner text only, excluding the CDATA markers.65            auto string_value = p.choice({66                p.literal("<![CDATA[") + p.ac(p.tool_arg_string_value(p.until("]]>")) + p.literal("]]>"), "]]>") + p.tool_arg_close(p.literal("</param>")),67                p.negate(p.literal("<![CDATA[")) + p.ac(p.tool_arg_string_value(p.until("</param>")) + p.tool_arg_close(p.literal("</param>")), "</param>")68            });69 70            auto tool_choice = p.choice();71            foreach_function(inputs.tools, [&](const json & tool) {72                const auto &      function = tool.at("function");73                const std::string name     = function.at("name");74 75                std::vector<common_peg_parser> arg_rules;76                foreach_parameter(function, [&](const common_chat_schema_property & prop, const common_chat_schema_document_ptr & doc) {77                    auto value_parser = p.eps();78                    if (prop.schema->may_be_string()) {79                        value_parser = string_value;80                    } else {81                        value_parser = p.tool_arg_json_value(82                                p.schema(p.json(), "tool-" + name + "-arg-" + prop.name + "-schema", doc, *prop.schema)83                            ) + p.tool_arg_close(p.literal("</param>"));84                    }85 86                    arg_rules.push_back(p.tool_arg(87                        p.tool_arg_open(p.literal("<param name=\"") + p.tool_arg_name(p.literal(prop.name)) + p.literal("\">")) +88                        value_parser89                    ));90                });91 92                auto args = p.eps();93                if (!arg_rules.empty()) {94                    args = p.zero_or_more(p.choice(arg_rules) + p.space());95                }96 97                auto tool_parser = p.tool(98                    p.tool_open(p.literal("<function name=\"") + p.tool_name(p.literal(name)) + p.literal("\">"))99                    << p.tool_args(args)100                    << p.tool_close(p.literal("</function>")));101 102                tool_choice |= p.rule("tool-" + name, tool_parser);103            });104 105            auto max_calls  = inputs.parallel_tool_calls ? -1 : 1;106            auto tool_calls = p.trigger_rule("tool-call", p.repeat(tool_choice + p.space(), 1, max_calls));107 108            auto content = p.content(p.until("<function"));109 110            return generation_prompt + reasoning + content + tool_calls + p.end();111        }112 113        return generation_prompt + reasoning + p.content(p.rest()) + p.end();114    });115 116    data.parser = parser.save();117 118    if (include_grammar) {119        data.grammar_lazy = !(has_response_format || (has_tools && inputs.tool_choice == COMMON_CHAT_TOOL_CHOICE_REQUIRED));120        data.grammar      = build_grammar([&](const common_grammar_builder & builder) {121            parser.build_grammar(builder, data.grammar_lazy);122        });123 124        data.grammar_triggers = {125            { COMMON_GRAMMAR_TRIGGER_TYPE_WORD, "<function" },126        };127    }128 129    return data;130}131