tool_call_markers.h

Go to the documentation of this file.

// SPDX-License-Identifier: Apache-2.0
#pragma once
 
#include <entropic/types/config.h>  // GenerationParams
 
#include <chat.h>
 
#include <algorithm>
#include <string>
 
namespace entropic {
 
inline std::string close_marker_for_format(common_chat_format fmt) {
    std::string marker;
    switch (fmt) {
        case COMMON_CHAT_FORMAT_PEG_NATIVE:
        case COMMON_CHAT_FORMAT_PEG_SIMPLE:
            marker = "</tool_call>";
            break;
        case COMMON_CHAT_FORMAT_PEG_GEMMA4:
            // gh#103 (v2.8.2): gemma4 wraps a call as
            // `<|tool_call>call:NAME{...}<tool_call|>` — the per-call CLOSE is
            // `<tool_call|>` (confirmed against a live gemma4_e4b transcript +
            // chat.cpp:1178). Distinct from the open `<|tool_call>`, so the
            // ends-with stop check fires on the close, never the open. This is
            // the family where the missing hard-stop is most severe: the
            // runaway past the call defeats extraction, so the terminal
            // directive registers as ZERO tool calls (gh#103 consumer report).
            marker = "<tool_call|>";
            break;
        default:
            // CONTENT_ONLY / unknown: no confirmed per-call close marker —
            // return "" so no stop is injected (batch-safe).
            break;
    }
    return marker;
}
 
inline void append_sequential_stop(GenerationParams& params,
                                   const std::string& marker) {
    if (params.tool_call_mode != "sequential" || marker.empty()) { return; }
    if (std::find(params.stop.begin(), params.stop.end(), marker)
            == params.stop.end()) {
        params.stop.push_back(marker);
    }
}
 
}  // namespace entropic