Skip to content

Index

desktop

Desktop helpers for Diapason (welcome sequence, monitors, voice commands).

Functions:

run_welcome_sequence 

run_welcome_sequence(
    settings: WelcomeClapSettings | None = None,
    *,
    skip_tts: bool = False,
    skip_chrome: bool = False,
    skip_cursor: bool = False,
) -> Mapping[str, Any]

Execute welcome actions via registered tools. Returns step results.

Source code in src/diapason/desktop/welcome_runner.py 
 91
 92
 93
 94
 95
 96
 97
 98
 99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
def run_welcome_sequence(
    settings: WelcomeClapSettings | None = None,
    *,
    skip_tts: bool = False,
    skip_chrome: bool = False,
    skip_cursor: bool = False,
) -> Mapping[str, Any]:
    """Execute welcome actions via registered tools. Returns step results."""
    import diapason.speech  # noqa: F401 — register TTS
    import diapason.tools  # noqa: F401 — register tools
    from diapason.tools.desktop_tools import (
        FocusAppTool,
        OpenBrowserOnMonitorTool,
        OpenUriTool,
        PlayAudioFileTool,
    )
    from diapason.tools.text_to_speech import TextToSpeechTool

    cfg = settings or load_welcome_settings()
    results: dict[str, Any] = {}

    logger.info("Welcome sequence: opening song")
    results["song"] = OpenUriTool().execute(uri=cfg.song_uri)

    if not skip_chrome:
        if cfg.open_claude:
            results["claude"] = OpenBrowserOnMonitorTool().execute(
                url=cfg.claude_url,
                monitor=cfg.claude_monitor,
                fullscreen=cfg.fullscreen,
                label="Claude",
            )
        if cfg.open_binance:
            results["binance"] = OpenBrowserOnMonitorTool().execute(
                url=cfg.binance_url,
                monitor=cfg.binance_monitor,
                fullscreen=cfg.fullscreen,
                label="Binance",
            )

    delay = max(0.0, cfg.after_song_delay_s)
    if delay:
        time.sleep(delay)

    if not skip_tts and cfg.welcome_phrase.strip():
        tts = TextToSpeechTool().execute(
            text=cfg.welcome_phrase.strip(),
            backend="elevenlabs",
            voice_id=cfg.elevenlabs_voice_id,
        )
        results["tts"] = tts
        if tts.success and tts.content:
            results["play"] = PlayAudioFileTool().execute(path=tts.content)

    if not skip_cursor:
        results["cursor"] = FocusAppTool().execute(
            app_name=cfg.focus_app_name,
            fullscreen=cfg.fullscreen,
        )

    ok = all(getattr(r, "success", False) for r in results.values() if r is not None)
    logger.info("Welcome sequence finished (all_ok=%s)", ok)
    return results