fix

AI: single-switch GPT/Claude backend for the chat cog

ai_commands.py

@@ -45,6 +45,14 @@ class Events(commands.Cog):
 
     async def cog_load(self):
         self.logger.info("Starting personal assistants")
+        # Personal assistants use the OpenAI Assistants API (threads/runs), which
+        # has no Anthropic equivalent - skip cleanly when OpenAI isn't wired up
+        # (e.g. a Claude-only deployment) instead of crashing the cog load.
+        if OPENAICLIENT is None:
+            self.logger.warning(
+                "OPENAICLIENT niedostępny - osobiści asystenci (OpenAI Assistants API) wyłączeni"
+            )
+            return
         for superfryta_id, superfryta in SPECJALNE_ZIEMNIACZKI.items():
 
             if superfryta[4] != "":
@@ -101,6 +109,40 @@ class Events(commands.Cog):
             else:
                 await ctx.reply("Nope. Nie wiesz jak użyć")
 
+    @commands.hybrid_command(
+        name="gadaj_teraz",
+        description="Przełącz backend AI (np. gpt / claude). Tylko Vykidailo.",
+    )
+    async def gadaj_teraz(self, ctx, nazwa_konfigu: str):
+        async with ctx.channel.typing():
+            is_admin = isinstance(ctx.author, discord.Member) and any(
+                role.name == "Vykidailo" for role in ctx.author.roles
+            )
+            if not is_admin:
+                await discord_friendly_reply(ctx, "Tylko Vykidailo może przełączać AI.")
+                return
+            available = ai_functions.list_ai_configs()
+            if nazwa_konfigu not in available:
+                await discord_friendly_reply(
+                    ctx,
+                    f"Nie znam configu '{nazwa_konfigu}'. Dostępne: {', '.join(available)}",
+                )
+                return
+            try:
+                cfg = ai_functions.set_active_ai_config(nazwa_konfigu)
+            except (KeyError, RuntimeError) as exc:
+                await discord_friendly_reply(
+                    ctx, f"Nie mogę przełączyć na '{nazwa_konfigu}': {exc}"
+                )
+                return
+            self.logger.info(
+                "Przełączono AI na config %s (%s)", nazwa_konfigu, cfg.get("provider")
+            )
+            await discord_friendly_reply(
+                ctx,
+                f"Teraz gadam przez **{nazwa_konfigu}** — {cfg.get('provider')} / {cfg.get('latest_model')}.",
+            )
+
     @commands.hybrid_command(
         name="armia_hammera",
         description="Jeśli nie wiesz jak użyć tej komendy to nawet nie próbuj",
@@ -301,6 +343,15 @@ class Events(commands.Cog):
         if "imaginuje sobie:" in message.content:
             async with channel.typing():
                 self.logger.info("Poczatek procedury obrazkowej")
+                # Image generation is DALL-E (OpenAI); there is no Anthropic
+                # equivalent, so it stays on OpenAI regardless of the chat
+                # backend. Degrade gracefully when OpenAI isn't configured.
+                if OPENAICLIENT is None:
+                    await discord_friendly_reply(
+                        message,
+                        "*Kondziu rozkłada łapska* — malowanie obrazków jest teraz wyłączone (brak OpenAI).",
+                    )
+                    return
                 message_content_lower = message_content_lower.replace("imaginuje sobie: ", "")
                 self.logger.debug("Wywolanie obrazka: %s", message_content_lower)
                 try:

ai_functions.py

@@ -8,8 +8,11 @@ import tiktoken
 import time
 from other_functions import discord_friendly_send
 from constants import (
+    AI_CONFIGS,
     ASSISTANTS,
+    CLAUDECLIENT,
     CYCLIC_WORDS,
+    DEFAULT_AI_CONFIG,
     ENCODING,
     GPT_SETTINGS,
     MEMORY_FIVE_MUZYKA,
@@ -23,19 +26,222 @@ from constants import (
     LATEST_MODEL
 )
 
+try:
+    import anthropic
+except ImportError:  # pragma: no cover - optional at runtime
+    anthropic = None
+
 # this do per user
 VECTOR_STORE_ID = -1
+
+
+# *=========================================== AI provider abstraction
+# The AI cog talks to exactly one backend at a time, chosen by _ACTIVE_CONFIG.
+# Legacy defaults ("gpt"/OpenAI) keep the historical behaviour byte-for-byte;
+# selecting a "claude" config routes the same handle_response pipeline through
+# the Anthropic Messages API instead. Backend-specific exceptions are funnelled
+# into a single AIError so handle_response can keep its one set of in-character
+# error replies regardless of provider.
+_ACTIVE_CONFIG_NAME = DEFAULT_AI_CONFIG
+
+# Legacy default algorithm strings that mean "let the bot pick" rather than
+# "force this exact model" - so a caller that still passes the old gpt-4o
+# default auto-selects the active provider's model instead of 400-ing on Claude.
+_AUTO_ALGOS = {"", "auto", "gpt-4o", "gpt-4o-mini", "gpt-3.5-turbo"}
+
+
+class AIError(Exception):
+    """Provider-neutral wrapper so handle_response reacts to one exception type.
+
+    ``category`` is one of: timeout, connection, bad_request,
+    response_validation, auth, permission, rate_limit, unprocessable, api.
+    ``original`` is the underlying SDK exception (interpolated into replies).
+    """
+
+    def __init__(self, category: str, original: Exception):
+        super().__init__(str(original))
+        self.category = category
+        self.original = original
+
+
+def _active_config() -> dict:
+    return (
+        AI_CONFIGS.get(_ACTIVE_CONFIG_NAME)
+        or AI_CONFIGS.get("gpt")
+        or next(iter(AI_CONFIGS.values()))
+    )
+
+
+def list_ai_configs():
+    """Selectable config names (templates prefixed with '_' are hidden)."""
+    return [name for name in AI_CONFIGS if not name.startswith("_")]
+
+
+def get_active_ai_config() -> str:
+    return _ACTIVE_CONFIG_NAME
+
+
+def set_active_ai_config(name: str) -> dict:
+    """Switch the active AI backend and persist the choice. Raises on error."""
+    global _ACTIVE_CONFIG_NAME
+    if name not in AI_CONFIGS:
+        raise KeyError(name)
+    cfg = AI_CONFIGS[name]
+    provider = cfg.get("provider")
+    if provider == "anthropic" and CLAUDECLIENT is None:
+        raise RuntimeError("klient Anthropic nie jest skonfigurowany (brak ANTHROPIC_API_KEY)")
+    if provider == "openai" and OPENAICLIENT is None:
+        raise RuntimeError("klient OpenAI nie jest skonfigurowany (brak OPENAI_API_KEY)")
+    _ACTIVE_CONFIG_NAME = name
+    _persist_active_ai_config(name)
+    return cfg
+
+
+def _persist_active_ai_config(name: str) -> None:
+    """Best-effort write of the active-config choice into system_gpt_settings.json.
+
+    Keeps the historical two-element structure intact: updates index 2 if it
+    already exists, appends it when the file has exactly the original two
+    elements, and otherwise leaves the file untouched (the in-memory switch
+    still applies).
+    """
+    logger = logging.getLogger("discord")
+    try:
+        with open(SYSTEM_GPT_SETTINGS, "r", encoding=ENCODING) as handle:
+            data = json.load(handle)
+    except (OSError, json.JSONDecodeError) as exc:
+        logger.warning("Nie mogę odczytać %s do zapisu configu AI: %s", SYSTEM_GPT_SETTINGS, exc)
+        return
+    if not isinstance(data, list) or len(data) < 2:
+        logger.warning("Nietypowa struktura %s - pomijam zapis configu AI", SYSTEM_GPT_SETTINGS)
+        return
+    if len(data) > 2 and isinstance(data[2], dict):
+        data[2]["active"] = name
+        data[2].setdefault("configs", AI_CONFIGS)
+    else:
+        data = data[:2] + [{"active": name, "configs": AI_CONFIGS}]
+    try:
+        with open(SYSTEM_GPT_SETTINGS, "w", encoding=ENCODING) as handle:
+            json.dump(data, handle, indent=4, ensure_ascii=False)
+    except OSError as exc:
+        logger.warning("Nie mogę zapisać configu AI do %s: %s", SYSTEM_GPT_SETTINGS, exc)
+
+
+def _map_openai_error(exc: Exception) -> AIError:
+    mapping = [
+        (openai.APITimeoutError, "timeout"),
+        (openai.APIConnectionError, "connection"),
+        (openai.BadRequestError, "bad_request"),
+        (openai.APIResponseValidationError, "response_validation"),
+        (openai.AuthenticationError, "auth"),
+        (openai.PermissionDeniedError, "permission"),
+        (openai.RateLimitError, "rate_limit"),
+        (openai.UnprocessableEntityError, "unprocessable"),
+        (openai.APIError, "api"),
+    ]
+    for cls, category in mapping:
+        if isinstance(exc, cls):
+            return AIError(category, exc)
+    return AIError("api", exc)
+
+
+def _map_anthropic_error(exc: Exception) -> AIError:
+    mapping = [
+        ("APITimeoutError", "timeout"),
+        ("APIConnectionError", "connection"),
+        ("BadRequestError", "bad_request"),
+        ("APIResponseValidationError", "response_validation"),
+        ("AuthenticationError", "auth"),
+        ("PermissionDeniedError", "permission"),
+        ("RateLimitError", "rate_limit"),
+        ("UnprocessableEntityError", "unprocessable"),
+        ("APIError", "api"),
+    ]
+    for name, category in mapping:
+        cls = getattr(anthropic, name, None)
+        if cls and isinstance(exc, cls):
+            return AIError(category, exc)
+    return AIError("api", exc)
+
+
+def _to_anthropic_messages(messages):
+    """Split OpenAI-style messages into (system_prompt, alternating convo).
+
+    Claude takes the system prompt as a separate parameter (not a role in the
+    messages list) and requires the conversation to open with a user turn, so
+    system messages are concatenated out and any leading assistant turns are
+    dropped.
+    """
+    system_parts = []
+    convo = []
+    for msg in messages:
+        role = msg.get("role")
+        content = msg.get("content", "")
+        if role == "system":
+            system_parts.append(content)
+        else:
+            convo.append(
+                {"role": "assistant" if role == "assistant" else "user", "content": content}
+            )
+    while convo and convo[0]["role"] != "user":
+        convo.pop(0)
+    if not convo:
+        convo = [{"role": "user", "content": " "}]
+    return "\n\n".join(part for part in system_parts if part), convo
+
+
+async def _anthropic_call(messages, model, cfg):
+    """Claude counterpart of openai_call. Returns a plain string."""
+    if CLAUDECLIENT is None:
+        raise AIError("auth", RuntimeError("klient Anthropic nie jest skonfigurowany"))
+    system_prompt, convo = _to_anthropic_messages(messages)
+    kwargs = {
+        "model": model,
+        "max_tokens": int(cfg.get("max_tokens", 2048)),
+        "messages": convo,
+    }
+    if system_prompt:
+        kwargs["system"] = system_prompt
+    # NOTE: temperature is deliberately omitted - Opus 4.8 / Sonnet 5 reject
+    # sampling params with a 400.
+    try:
+        resp = await CLAUDECLIENT.messages.create(**kwargs)
+    except Exception as exc:  # pylint: disable=broad-except
+        raise _map_anthropic_error(exc)
+    text = "".join(
+        block.text for block in resp.content if getattr(block, "type", None) == "text"
+    )
+    return text.strip()
+
+
+async def provider_generate(messages, model, temperature=0.2):
+    """Dispatch a chat completion to the active backend, normalising errors."""
+    cfg = _active_config()
+    try:
+        if cfg.get("provider") == "anthropic":
+            return await _anthropic_call(messages, model, cfg)
+        return await openai_call(messages, model, temperature)
+    except AIError:
+        raise
+    except Exception as exc:  # pylint: disable=broad-except
+        # Only the OpenAI path reaches here un-normalised (_anthropic_call
+        # already wraps its own errors).
+        raise _map_openai_error(exc)
+
+
 def select_model(req_type: str, algo: str) -> str:
-    # Jeżeli jawnie podano algorithm (i nie jest 'auto'/''):
-    if algo and str(algo).strip().lower() not in ("auto",):
-        # wyjątek: MUZYKA ma zawsze być tania — nadpisujemy TYLKO jeśli przyszedł domyślny 'gpt-4o'
-        if req_type == "MUSIC" and algo.strip() in (LATEST_MODEL,):
-            return CHEAP_MODEL
-        return algo
-    # Auto-dobór:
+    cfg = _active_config()
+    latest = cfg.get("latest_model", LATEST_MODEL)
+    cheap = cfg.get("cheap_model", CHEAP_MODEL)
+    algo_str = (algo or "").strip()
+    # An explicit, non-legacy model id is honoured verbatim; anything in
+    # _AUTO_ALGOS (incl. the old gpt-4o default) means "auto pick for the
+    # active provider", so flipping the switch actually changes the model.
+    if algo_str and algo_str.lower() not in _AUTO_ALGOS:
+        return algo_str
     if req_type == "MUSIC":
-        return CHEAP_MODEL
-    return LATEST_MODEL
+        return cheap
+    return latest
 
 
 async def openai_call(messages, model, temperature=0.2):
@@ -256,57 +462,46 @@ async def handle_response(
         timeout_sec = 120
         deadline = time.time() + timeout_sec
         response = await asyncio.wait_for(
-            openai_call(messages=history_msgs, model=model_to_use),
+            provider_generate(messages=history_msgs, model=model_to_use),
             timeout=max(0.1, deadline - time.time()),
         )
 
-    except openai.APITimeoutError as e:
-        # Handle timeout error, e.g. retry or log
-        response = f"*Kondziu patrzy na terminal, czeka, czeka, czeka,.... Jeszcze chwile czeka Przypierdala w niego pięścią....* Nie mogę się połączyć z Openai spróbuj od nowa. *Na ekranie pojawia się*: {e}"
-    except openai.APIConnectionError as e:
-        response = f"*Kondziu patrzy na terminal, chwile się zastanawia. Przypierdala w niego pięścią....* Nie mogę się połączyć z Openai. *Na ekranie pojawia się*: {e}"
-    except openai.BadRequestError as e:
-        # Handle invalid request error, e.g. validate parameters or log
-        if internal_retry:
-            resp = "Nie umiem tego teraz ładnie wytłumaczyć — OpenAI mnie zastrzeliło."
-        else:
-            resp, _ = await handle_response(
-
-                f"Wytlumacz jakie sa zasady dotyczące treści które możesz generować używając Dalle. Wytłumacz błąd {e} prostym językiem. Przeproś za nadmierną cenzurę. Wytłumacz co mogło być nie tak w prompcie 'prompt'",
-                True,
-                True,
-                MESSAGE_TABLE,
-                username,
-                "RANDOM",
-            )
-            response = f"Sorki, cenzura: {resp}. Jak chcesz to są kanały na nudle #sexy-foteczky i #kanal-do-fapania *Na ekranie pojawia się: {e}"
-    except openai.APIResponseValidationError as e:
-        # Handle invalid request error, e.g. validate parameters or log
-        if internal_retry:
-            resp = "Nie umiem tego teraz ładnie wytłumaczyć — OpenAI mnie zastrzeliło."
-        else:
-            resp, _ = await handle_response(
-                f"Wytlumacz jakie sa zasady dotyczące treści które możesz generować używając Dalle. Wytłumacz błąd {e} prostym językiem. Przeproś za nadmierną cenzurę. Wytłumacz co mogło być nie tak w prompcie 'prompt'",
-                True,
-                True,
-                MESSAGE_TABLE,
-                username,
-                "RANDOM",
-            )
-            response = f"Sorki, cenzura: {resp}. Jak chcesz to są kanały na nudle #sexy-foteczky i #kanal-do-fapania *Na ekranie pojawia się: {e}"
-    except openai.AuthenticationError as e:
-        # Handle authentication error, e.g. check credentials or log
-        response = f"*Kondziu patrzy na terminal, chwile się zastanawia. Przypierdala w niego pięścią....* Wołaj szefa - coś się z hasłem zjebało. *Na terminalu pojawia się:* {e}"
-    except openai.PermissionDeniedError as e:
-        # Handle permission error, e.g. check scope or log
-        response = f"*Kondziu patrzy na terminal, chwile się zastanawia. Przypierdala w niego pięścią....* Wołaj szefa - coś się z uprawnieniami zjebało. *Na terminalu pojawia się:* {e}"
-    except openai.RateLimitError as e:
-        response = f"*Kondziu patrzy na terminal* Wołaj szefa. Zapłacić rachunki za AI trzeba. Jak chcesz to się na #zebranie dorzuć. {e}"
-    except openai.UnprocessableEntityError as e:
-        response = f"*Kondziu patrzy na terminal. Potem na to co każesz mu wysłać....* Ja wiem że jesteśmy w barze BDSM - ale nie da się włożyć TEGO w TO. *Za jego plecami na terminalu pojawia się:* {e}"
-    except openai.APIError as e:
-        # Handle API error, e.g. retry or log
-        response = f"*Kondziu nurkuje za bar, terminal wybucha. Przed tobą ląduje pergamin zapisany pięknym gotykiem a na nim*: {e}"
+    except AIError as e:
+        # One handler for both backends; e.category is provider-neutral and
+        # e.original is the underlying SDK exception (kept for the {..} tails).
+        err = e.original
+        if e.category == "timeout":
+            response = f"*Kondziu patrzy na terminal, czeka, czeka, czeka,.... Jeszcze chwile czeka Przypierdala w niego pięścią....* Nie mogę się połączyć z Openai spróbuj od nowa. *Na ekranie pojawia się*: {err}"
+        elif e.category == "connection":
+            response = f"*Kondziu patrzy na terminal, chwile się zastanawia. Przypierdala w niego pięścią....* Nie mogę się połączyć z Openai. *Na ekranie pojawia się*: {err}"
+        elif e.category in ("bad_request", "response_validation"):
+            # Handle invalid request error, e.g. validate parameters or log
+            if internal_retry:
+                resp = "Nie umiem tego teraz ładnie wytłumaczyć — OpenAI mnie zastrzeliło."
+            else:
+                resp, _ = await handle_response(
+                    f"Wytlumacz jakie sa zasady dotyczące treści które możesz generować używając Dalle. Wytłumacz błąd {err} prostym językiem. Przeproś za nadmierną cenzurę. Wytłumacz co mogło być nie tak w prompcie 'prompt'",
+                    True,
+                    True,
+                    MESSAGE_TABLE,
+                    username,
+                    "RANDOM",
+                    internal_retry=True,
+                )
+            response = f"Sorki, cenzura: {resp}. Jak chcesz to są kanały na nudle #sexy-foteczky i #kanal-do-fapania *Na ekranie pojawia się: {err}"
+        elif e.category == "auth":
+            # Handle authentication error, e.g. check credentials or log
+            response = f"*Kondziu patrzy na terminal, chwile się zastanawia. Przypierdala w niego pięścią....* Wołaj szefa - coś się z hasłem zjebało. *Na terminalu pojawia się:* {err}"
+        elif e.category == "permission":
+            # Handle permission error, e.g. check scope or log
+            response = f"*Kondziu patrzy na terminal, chwile się zastanawia. Przypierdala w niego pięścią....* Wołaj szefa - coś się z uprawnieniami zjebało. *Na terminalu pojawia się:* {err}"
+        elif e.category == "rate_limit":
+            response = f"*Kondziu patrzy na terminal* Wołaj szefa. Zapłacić rachunki za AI trzeba. Jak chcesz to się na #zebranie dorzuć. {err}"
+        elif e.category == "unprocessable":
+            response = f"*Kondziu patrzy na terminal. Potem na to co każesz mu wysłać....* Ja wiem że jesteśmy w barze BDSM - ale nie da się włożyć TEGO w TO. *Za jego plecami na terminalu pojawia się:* {err}"
+        else:  # "api" and anything unmapped
+            # Handle API error, e.g. retry or log
+            response = f"*Kondziu nurkuje za bar, terminal wybucha. Przed tobą ląduje pergamin zapisany pięknym gotykiem a na nim*: {err}"
 
     logger.info("Historia wysłana:")
     temp_assistant = {"role": "assistant", "content": response}

bot.py

@@ -4,9 +4,21 @@
 # pylint: disable=too-many-lines
 """
 Module of a python bot named Conjurer - used to work on BDSM discord servers.
+
+Startup is defensive by design:
+
+* every cog is loaded independently - one broken/missing dependency disables
+  that cog only, never the whole bot,
+* cogs that need a sibling service (musician / librarian) are only enabled
+  after a positive health check; a watchdog keeps re-checking and enables them
+  the moment the service comes alive,
+* fatal misconfiguration (missing Discord token) exits loudly on stderr with a
+  clear message instead of dying silently into a log file.
 """
 import asyncio
 import logging
+import os
+import sys
 
 # *=========================================== Standard Library Imports
 import random
@@ -15,12 +27,27 @@ from logging import handlers
 
 # *==============Imported libraries
 import discord
+import requests
 from discord.ext import commands
 
 from communication_subroutine import comm_subroutine
-from constants import ENCODING, LOGFILE, TOKEN
+from constants import (
+    ENCODING,
+    FILE_SERVICE_ADDRESS,
+    GET_MP3,
+    LIBRARIAN_SERVICE_ADDRESS,
+    LOGFILE,
+    RADIO_SERVICE_ADDRESS,
+    TOKEN,
+)
+
 logger = logging.getLogger("discord")
 logger.setLevel(logging.DEBUG)
+formatter = logging.Formatter("%(asctime)s - %(name)s - %(levelname)s - %(message)s")
+
+# File log (rotated). constants ensures the directory exists; this is belt and
+# braces for exotic overrides.
+os.makedirs(os.path.dirname(LOGFILE) or ".", exist_ok=True)
 handler = handlers.RotatingFileHandler(
     filename=LOGFILE,
     encoding=ENCODING,
@@ -28,10 +55,18 @@ handler = handlers.RotatingFileHandler(
     maxBytes=6 * 1024 * 1024,
     backupCount=6,
     )
-formatter = logging.Formatter("%(asctime)s - %(name)s - %(levelname)s - %(message)s")
 handler.setFormatter(formatter)
 logger.addHandler(handler)
 
+# Console log so `docker logs` / journalctl actually show what happened.
+console_handler = logging.StreamHandler()
+console_handler.setLevel(logging.INFO)
+console_handler.setFormatter(formatter)
+logger.addHandler(console_handler)
+# Some dependency calls logging.basicConfig(), adding a root handler; without
+# this the 'discord' logger's records get printed twice (our format + root's).
+logger.propagate = False
+
 # *=========================================== Initializations
 intents = discord.Intents.default()
 intents.message_content = True
@@ -48,37 +83,134 @@ intents.moderation = True
 random.seed()
 client = commands.Bot(intents=intents, command_prefix="$")
 
+# *=========================================== Extension groups
+# Core cogs depend on nothing but the bot itself (broken ones are skipped
+# individually). Service groups are gated on a health check of the service
+# they talk to and enabled later by the watchdog when the service appears.
+CORE_EXTENSIONS = [
+    "administration_commands",
+    "ai_commands",
+    "other_commands",
+    "latex_commands",
+    "voice_recognition_commands",
+    "conanjurer_commands",
+]
+
+SERVICE_EXTENSION_GROUPS = {
+    # musician (file service): Discord music download/search, file shares
+    "musician": {
+        "health_url": f"{FILE_SERVICE_ADDRESS}{GET_MP3}",
+        "extensions": ["music_commands", "file_search_commands"],
+    },
+    # betoniarka (radio operator colocated with Liquidsoap): radio playlists
+    "radio": {
+        "health_url": f"{RADIO_SERVICE_ADDRESS}/ping",
+        "extensions": ["radio_commands"],
+    },
+    # librarian: DOI / Crossref search
+    "librarian": {
+        "health_url": f"{LIBRARIAN_SERVICE_ADDRESS}/",
+        "extensions": ["librarian_commands"],
+    },
+}
+
+SERVICE_RECHECK_SECONDS = 300
+
+
+def _service_alive(url: str) -> bool:
+    """True when the service answers HTTP at all (any status code counts)."""
+    try:
+        requests.get(url, timeout=3)
+        return True
+    except requests.exceptions.RequestException:
+        return False
+
+
+async def _load_extension_safe(name: str) -> bool:
+    """Load one extension; log and continue instead of killing startup."""
+    if name in client.extensions:
+        return True
+    try:
+        await client.load_extension(name)
+        logger.info("Extension loaded: %s", name)
+        return True
+    except Exception:  # pylint: disable=broad-exception-caught
+        logger.exception("Extension FAILED (disabled, bot continues): %s", name)
+        return False
+
+
+async def _load_service_groups() -> bool:
+    """Health-check each service group and load its cogs when alive.
+
+    Returns True when anything new was loaded (caller may want to re-sync).
+    """
+    loaded_any = False
+    for service, group in SERVICE_EXTENSION_GROUPS.items():
+        missing = [e for e in group["extensions"] if e not in client.extensions]
+        if not missing:
+            continue
+        alive = await asyncio.to_thread(_service_alive, group["health_url"])
+        if not alive:
+            logger.warning(
+                "Service '%s' unreachable (%s) - cogs stay disabled: %s",
+                service,
+                group["health_url"],
+                ", ".join(missing),
+            )
+            continue
+        logger.info("Service '%s' is alive - enabling: %s", service, ", ".join(missing))
+        for extension in missing:
+            if await _load_extension_safe(extension):
+                loaded_any = True
+    return loaded_any
+
+
+async def _sync_tree() -> None:
+    try:
+        await client.tree.sync()
+    except Exception:  # pylint: disable=broad-exception-caught
+        logger.exception("Slash-command tree sync failed (commands may lag)")
+
+
+async def _service_watchdog() -> None:
+    """Periodically retry offline services and enable their cogs when up."""
+    while not client.is_closed():
+        await asyncio.sleep(SERVICE_RECHECK_SECONDS)
+        try:
+            if await _load_service_groups():
+                await _sync_tree()
+        except Exception:  # pylint: disable=broad-exception-caught
+            logger.exception("Service watchdog tick failed")
+
+
+_STARTUP_DONE = False
+
 
 # *=========================================== Define Events
 @client.event
 async def on_ready():
     """Metoda wywoływana przy połączeniu do serwera."""
-    logger = logging.getLogger("discord")
-    logger.debug("SAMPLE DEBUG LOG")
+    global _STARTUP_DONE  # pylint: disable=global-statement
     logger.info("%s has connected to Discord!", client.user)
-    # TODO: load vs reload
+    if _STARTUP_DONE:
+        logger.info("Reconnected - extensions already loaded")
+        return
+    _STARTUP_DONE = True
     logger.info("Reactor: online")
 
-    await client.load_extension("administration_commands")
+    for extension in CORE_EXTENSIONS:
+        await _load_extension_safe(extension)
 
-    await client.load_extension("librarian_commands")
-    await client.load_extension("music_commands")
-    await client.load_extension("radio_commands")
-
-    await client.load_extension("ai_commands")
-
-    await client.load_extension("other_commands")
-    await client.load_extension("voice_recognition_commands")
-    await client.load_extension("file_search_commands")
-    await client.load_extension("latex_commands")
-    await client.load_extension("conanjurer_commands")
+    await _load_service_groups()
     logger.info("Sensors: online")
 
     logger.info(client.cogs)
-    await client.tree.sync()
+    await _sync_tree()
     for com in client.commands:
         logger.info("Command %s is awejleble", com.qualified_name)
 
+    asyncio.create_task(_service_watchdog())
+
     logger.info("Logged in as ----> %s", client.user)
     logger.info("ID:%s ", client.user.id)
     logger.info("All systems: operational")
@@ -93,31 +225,73 @@ async def on_ready():
 
 
 async def _run_comm_subroutine(stop_event: threading.Event) -> None:
-    """Run the blocking comm subroutine in a worker thread."""
-    await asyncio.to_thread(comm_subroutine, stop_event)
+    """Run the blocking comm subroutine in a worker thread.
+
+    A crash here takes down the internal HTTP endpoints only - the Discord
+    side keeps running, so log loudly and swallow.
+    """
+    try:
+        await asyncio.to_thread(comm_subroutine, stop_event)
+    except Exception:  # pylint: disable=broad-exception-caught
+        logger.exception(
+            "Comm layer crashed - internal HTTP endpoints are down "
+            "(musician/librarian callbacks will not arrive); bot continues"
+        )
 
 
 async def _run_bot(token: str, shutdown_event: asyncio.Event) -> None:
-    """Start the Discord client and flag shutdown when it returns."""
+    """Start the Discord client; log WHY it died before flagging shutdown.
+
+    Exceptions were previously swallowed by ``gather(return_exceptions=True)``
+    which made a failed login look like a clean exit (silent crash-loop in
+    docker). Now every death is diagnosed on the console first.
+    """
     try:
-        await client.start(token, log_handler=None)
+        # NOTE: log_handler is a Client.run()-only kwarg (run() configures
+        # logging, then calls start()); start() takes just the token. We set
+        # up our own handlers above, so nothing is lost.
+        await client.start(token)
+    except discord.LoginFailure:
+        logger.critical(
+            "FATAL: Discord REJECTED the token (Improper token). Fix the "
+            "'discord' entry in the netrc mounted at CONJURER_NETRC_FILE or "
+            "the DISCORD_TOKEN env var. If the token leaked/reset, generate a "
+            "new one in the Discord Developer Portal -> Bot -> Reset Token."
+        )
+        raise
+    except discord.PrivilegedIntentsRequired:
+        logger.critical(
+            "FATAL: this bot application does not have the Privileged Gateway "
+            "Intents enabled. Open Discord Developer Portal -> your app -> "
+            "Bot -> enable 'Presence', 'Server Members' and 'Message Content' "
+            "Intents, then restart."
+        )
+        raise
+    except Exception:  # pylint: disable=broad-exception-caught
+        logger.exception("FATAL: Discord client crashed at startup/runtime")
+        raise
     finally:
         shutdown_event.set()
 
 
-async def main() -> None:
-    if not TOKEN:
-        logger.error("Discord token missing - set DISCORD_TOKEN or configure netrc")
-        return
+async def main() -> int:
+    """Run both halves; return a process exit code (0 = clean shutdown)."""
+    if TOKEN:
+        token_source = (
+            "env DISCORD_TOKEN" if os.getenv("DISCORD_TOKEN") else "netrc file"
+        )
+        logger.info(
+            "Discord token: loaded from %s (length %d)", token_source, len(TOKEN)
+        )
 
     logger.info("Starting discord bot")
     shutdown_event = asyncio.Event()
     comm_stop_event = threading.Event()
 
-
     comm_task = asyncio.create_task(_run_comm_subroutine(comm_stop_event))
     bot_task = asyncio.create_task(_run_bot(TOKEN, shutdown_event))
 
+    exit_code = 0
     try:
         await shutdown_event.wait()
     except (KeyboardInterrupt, asyncio.CancelledError):
@@ -128,9 +302,31 @@ async def main() -> None:
         comm_stop_event.set()
         if not client.is_closed():
             await client.close()
-        await asyncio.gather(bot_task, comm_task, return_exceptions=True)
+        results = await asyncio.gather(bot_task, comm_task, return_exceptions=True)
+        for result in results:
+            if isinstance(result, BaseException) and not isinstance(
+                result, asyncio.CancelledError
+            ):
+                # Already logged with full traceback inside the task; repeat
+                # the one-liner so it is the LAST thing in `docker logs`.
+                logger.critical("Task died: %r", result)
+                exit_code = 1
+    return exit_code
 
 
 # *================================== Run
 if __name__ == "__main__":
-    asyncio.run(main())
+    if not TOKEN:
+        # Loud, unmissable and in `docker logs`: this is THE most common cause
+        # of a silent container crash-loop.
+        MSG = (
+            "FATAL: Discord token missing.\n"
+            "Provide it via the DISCORD_TOKEN environment variable or a netrc "
+            "file (machine 'discord') at the path in CONJURER_NETRC_FILE.\n"
+            "Docker: check that your secrets mount exists, e.g.\n"
+            "  -v /srv/conjurer/secrets/.netrc:/secrets/.netrc:ro\n"
+            "  CONJURER_NETRC_FILE=/secrets/.netrc"
+        )
+        logger.critical(MSG)
+        sys.exit(MSG)
+    sys.exit(asyncio.run(main()))

communication_subroutine.py

@@ -234,11 +234,14 @@ def comm_subroutine(stop_event: Optional[threading.Event] = None):
     :param stop_event: optional :class:`threading.Event` shared with the caller
     to coordinate a cooperative shutdown.
     """
-    logger.setLevel(logging.DEBUG)
     logger = logging.getLogger("discord")
+    logger.setLevel(logging.DEBUG)
     logger.info("Started comms")
     threads = []
-    threads.append(threading.Thread(target=flask_debug))
+    # NOTE: flask_debug is the dev server bound to the SAME host:port as
+    # waitress - running both kills the comm layer with 'address in use'.
+    # Enable it only INSTEAD of waitress_run, never alongside.
+    # threads.append(threading.Thread(target=flask_debug))
     threads.append(threading.Thread(target=waitress_run, daemon=True))
     threads.append(
         threading.Thread(

conjurer_betoniarka/betoniarka.py

@@ -0,0 +1,353 @@
+"""Betoniarka - the radio operator service.
+
+Lives in the SAME container as Liquidsoap and runs as the SAME unprivileged
+user ('radio'), which is the whole point: the process that writes the radio
+playlists is colocated with the process that watches them, so there is no
+cross-host permission juggling (root-owned network shares, failing chowns)
+anymore.
+
+Responsibilities (extracted from conjurer_musician, which is now a pure
+Discord music player):
+- scan the local music library into all_playlist.playlist / hit.playlist
+- serve the radio-management HTTP API the bot calls
+  (/add_to_priority, /create_priority_playlist, /request_radio_file,
+   /clear_pr_pls) plus /ping for health checks and /stream for the web page
+- tail radio_log.log / persistence.log and forward "now playing" events to
+  the bot's /prepped_tracks
+"""
+
+import json
+import logging
+import os
+import random
+import re
+import threading
+import time
+from datetime import datetime
+from pathlib import Path
+from typing import Dict, List
+
+import requests
+from flask import Flask, abort, jsonify, request, send_file
+from waitress import serve
+
+
+def _env(name: str, default: str) -> str:
+    return os.getenv(name, default)
+
+
+API_KEY = os.getenv("CONJURER_API_KEY")
+MAIN_BOT_ADDRESS = _env("CONJURER_MAIN_BOT", "http://127.0.0.1:5000")
+MUSIC_TRACKER = _env("CONJURER_MUSIC_TRACKER_ENDPOINT", "/prepped_tracks")
+HOST_ADDRESS = _env("BETONIARKA_HOST", "0.0.0.0")
+PORT_ADDRESS = int(_env("BETONIARKA_PORT", "5005"))
+
+DATA_DIR = Path(_env("BETONIARKA_DATA", "/srv/betoniarka/data"))
+MUSIC_FOLDER = Path(_env("BETONIARKA_MUSIC", "/srv/betoniarka/music"))
+PRIORITY_FOLDER = Path(_env("BETONIARKA_PRIORITY_FOLDER", str(MUSIC_FOLDER / "priority")))
+STREAM_TEMPLATE = _env("BETONIARKA_STREAM_TEMPLATE", "/app/stream.html")
+RESCAN_SECONDS = int(_env("BETONIARKA_RESCAN_SECONDS", str(24 * 60 * 60)))
+# How many leading path tokens to ignore when keyword-matching
+# (/srv/betoniarka/music/... -> '', 'srv', 'betoniarka', 'music').
+PATH_SKIP = int(_env("BETONIARKA_PATH_SKIP", "4"))
+
+ALL_PLAYLIST_PATH = DATA_DIR / "all_playlist.playlist"
+HIT_PLAYLIST_PATH = DATA_DIR / "hit.playlist"
+REQUEST_PLAYLIST_PATH = DATA_DIR / "request.playlist"
+PRIORITY_PLAYLIST_PATH = DATA_DIR / "priority_queue.playlist"
+RADIOLOG_PATH = DATA_DIR / "radio_log.log"
+PERSISTENCE_PATH = DATA_DIR / "persistence.log"
+
+ENCODING = _env("CONJURER_ENCODING", "utf-8")
+
+logger = logging.getLogger("betoniarka")
+
+music_file_list: List[str] = []
+priority_list: List[str] = []
+
+app = Flask(__name__)
+
+
+def _build_headers() -> Dict[str, str]:
+    if API_KEY:
+        return {"X-Conjurer-Api-Key": API_KEY}
+    return {}
+
+
+def _authorize_request() -> None:
+    if API_KEY and request.headers.get("X-Conjurer-Api-Key") != API_KEY:
+        abort(401)
+
+
+def _post_to_bot(payload: List[str]) -> None:
+    try:
+        response = requests.post(
+            f"{MAIN_BOT_ADDRESS}{MUSIC_TRACKER}",
+            json=payload,
+            headers=_build_headers(),
+            timeout=60,
+        )
+        logger.info("Forwarded to bot (%s): %s", response.status_code, payload[0])
+    except requests.exceptions.RequestException as exc:
+        logger.warning("Bot unreachable, dropping %s: %s", payload[0], exc)
+
+
+# ---------------------------------------------------------------- library
+def rescan():
+    """Scan the local library into the playlists Liquidsoap watches.
+
+    Paths written here are LOCAL container paths, the same ones Liquidsoap
+    resolves - no shared network filesystem involved.
+    """
+    logger.info("Rescan triggered")
+    music_file_list.clear()
+    priority_list.clear()
+
+    for mp3_item in MUSIC_FOLDER.glob("**/*.mp3"):
+        music_file_list.append(mp3_item.as_posix())
+    for mp3_item in PRIORITY_FOLDER.glob("**/*.mp3"):
+        priority_list.append(mp3_item.as_posix())
+
+    with ALL_PLAYLIST_PATH.open("w", encoding=ENCODING) as w_file:
+        for item in music_file_list:
+            w_file.write(item + "\n")
+    with HIT_PLAYLIST_PATH.open("w", encoding=ENCODING) as w_file:
+        for item in priority_list:
+            w_file.write(item + "\n")
+    logger.info("Rescan done: %d tracks, %d hits", len(music_file_list), len(priority_list))
+
+
+def thread_rescan():
+    while True:
+        time.sleep(RESCAN_SECONDS)
+        rescan()
+
+
+# ---------------------------------------------------------------- search
+def remove_characters(string, character):
+    return string.replace(character, "")
+
+
+def max_weight(lista):
+    maximum_weight = 0
+    for iterator in lista:
+        if iterator[0] > maximum_weight:
+            maximum_weight = iterator[0]
+    return maximum_weight
+
+
+_CHAR_REMOVE = [
+    ".", "^", "$", "*", "+", "?", "{", "}", "[", "]",
+    "\\", "/", "|", "(", ")", "!", ",", "-", ":", "mp3",
+]
+
+
+def wyszukaj(word_list, how_many, _logger=None, write_to=None):
+    """Keyword-score the library; optionally append hits to a playlist file.
+
+    Ported unchanged from the musician (same scoring), minus the win32
+    branches. ``write_to`` replaces the old ``return_to_bot`` flag: pass a
+    playlist Path to append the result, or None to just return it.
+    """
+    fun_logger = _logger or logger
+    search_weight = [(0, "") for _ in range(len(music_file_list))]
+
+    time_start = datetime.now()
+    skip_start = 2 if int(how_many) > 0 else 1
+    for word in word_list[skip_start:]:
+        token_weight = len(word)
+        fun_logger.info("Słowo kluczowe: %s", word)
+        for itr, file in enumerate(music_file_list):
+            parts = file.split("/")
+            all_words = []
+            for f_iter in parts:
+                for char in _CHAR_REMOVE:
+                    f_iter = remove_characters(f_iter, char)
+                all_words.extend(f_iter.split())
+            pingu = 1
+            pattern_len = len(all_words)
+            matched_times = 1
+            for itm in all_words[PATH_SKIP:]:
+                pingu += 1
+                if re.match(".*" + word + ".*", itm, re.IGNORECASE):
+                    temp_weight = (
+                        search_weight[itr][0]
+                        + (token_weight + (pingu**1.5) / pattern_len) / matched_times
+                    )
+                    search_weight[itr] = (temp_weight, music_file_list[itr])
+                    matched_times += 1
+
+    fun_logger.info("Stworzylem tablice wag zajęło mi to %s", datetime.now() - time_start)
+    best = max_weight(search_weight)
+    if best == 0:
+        return []
+
+    return_list = []
+    if int(how_many) <= 0:
+        for weight, path in search_weight:
+            if weight == best:
+                return_list.append((weight, path))
+                break
+    else:
+        search_weight.sort(key=lambda x: x[0], reverse=True)
+        return_list.extend(search_weight[: int(how_many)])
+
+    if write_to is not None:
+        with write_to.open("a", encoding=ENCODING) as s_file:
+            for item in return_list:
+                s_file.write(item[1] + "\n")
+    fun_logger.info("Done: %s", return_list)
+    return return_list
+
+
+# ---------------------------------------------------------------- tailer
+def scan_tracks():
+    """Tail the radio logs and forward play events to the bot."""
+    with open(RADIOLOG_PATH, "r", encoding=ENCODING) as log_file:
+        log_file.seek(os.stat(RADIOLOG_PATH).st_size)
+        prev_size = os.stat(PERSISTENCE_PATH).st_size
+
+        while True:
+            current_size = os.stat(PERSISTENCE_PATH).st_size
+            if prev_size != current_size:
+                while prev_size != current_size:
+                    prev_size = current_size
+                    time.sleep(0.1)
+                    current_size = os.stat(PERSISTENCE_PATH).st_size
+                with open(PERSISTENCE_PATH, "r", encoding=ENCODING) as persistence:
+                    lines = persistence.readlines()
+                if len(lines) >= 3:
+                    _post_to_bot(["next", lines[2]])
+
+            position = log_file.tell()
+            line = log_file.readline()
+            if not line:
+                time.sleep(1)
+                log_file.seek(position)
+                continue
+
+            if not re.match(r".*Prepared.*", line):
+                time.sleep(0.1)
+                continue
+
+            result = None
+            if re.match(r".*jingles.*", line):
+                result = ["jingles", line]
+            elif re.match(r".*priority.*", line):
+                result = ["priority", line]
+            elif re.match(r".*hit.*", line):
+                result = ["hit", line]
+            elif re.match(r".*all_playlist.*", line):
+                result = ["all", line]
+            elif re.match(r".*request.*", line):
+                result = ["requests", line]
+
+            if result:
+                logger.info("Forwarding radio log entry: %s", result[0])
+                _post_to_bot(result)
+
+            time.sleep(0.1)
+
+
+# ---------------------------------------------------------------- routes
+@app.route("/ping", methods=["GET"])
+def ping():
+    """Health check - the bot gates radio_commands on this answering."""
+    return jsonify("ALIVE")
+
+
+@app.route("/stream", methods=["GET"])
+def stream_page():
+    return send_file(STREAM_TEMPLATE)
+
+
+@app.route("/clear_pr_pls", methods=["GET"])
+def clear_pr_pls():
+    _authorize_request()
+    app.logger.info("CLEARING PLAYLIST")
+    with PRIORITY_PLAYLIST_PATH.open("w", encoding=ENCODING) as cleared_pl:
+        cleared_pl.write("")
+    return jsonify(isError=False, message="Success", statusCode=200, data=[]), 200
+
+
+@app.route("/rescan", methods=["GET"])
+def manual_rescan():
+    _authorize_request()
+    rescan()
+    return jsonify(isError=False, message="Success", statusCode=200,
+                   data={"tracks": len(music_file_list)}), 200
+
+
+@app.route("/request_radio_file", methods=["POST"])
+def add_request():
+    _authorize_request()
+    record = json.loads(request.data)
+    app.logger.info(record)
+    wyszukaj(record["lista_slow"], 0, app.logger, write_to=REQUEST_PLAYLIST_PATH)
+    return jsonify(isError=False, message="Success", statusCode=200,
+                   data={"status": "OK"}), 200
+
+
+@app.route("/create_priority_playlist", methods=["POST"])
+def create_priority_playlist():
+    _authorize_request()
+    record = json.loads(request.data)
+    app.logger.info(record)
+    return_data = wyszukaj(
+        record["lista_slow"], record["dlugosc_plejlisty"], app.logger, write_to=None
+    )
+    random.shuffle(return_data)
+    # NOTE: appends to the REQUEST playlist - behaviour inherited verbatim
+    # from the musician implementation (the request queue picks it up).
+    with REQUEST_PLAYLIST_PATH.open("a", encoding=ENCODING) as s_file:
+        for item in return_data:
+            s_file.write(item[1] + "\n")
+    return jsonify(isError=False, message="Success", statusCode=200,
+                   data={"status": "OK"}), 200
+
+
+@app.route("/add_to_priority", methods=["POST"])
+def add_to_priority():
+    _authorize_request()
+    record = json.loads(request.data)
+    app.logger.info(record)
+    wyszukaj(
+        record["lista_slow"], record["dlugosc_plejlisty"], app.logger,
+        write_to=PRIORITY_PLAYLIST_PATH,
+    )
+    return jsonify(isError=False, message="Success", statusCode=200,
+                   data={"status": "OK"}), 200
+
+
+def waitress_run():
+    serve(app, host=HOST_ADDRESS, port=PORT_ADDRESS)
+
+
+if __name__ == "__main__":
+    logger.setLevel(logging.DEBUG)
+    console = logging.StreamHandler()
+    console.setFormatter(
+        logging.Formatter("%(asctime)s - %(name)s - %(levelname)s - %(message)s")
+    )
+    logger.addHandler(console)
+
+    rescan()
+    logger.info("Betoniarka started on %s:%s", HOST_ADDRESS, PORT_ADDRESS)
+
+    threads = [
+        threading.Thread(target=waitress_run, daemon=True),
+        threading.Thread(target=thread_rescan, daemon=True),
+    ]
+    for worker in threads:
+        worker.start()
+
+    time.sleep(5)
+    track_thread = threading.Thread(target=scan_tracks, daemon=True)
+    track_thread.start()
+
+    try:
+        for worker in threads:
+            worker.join()
+        track_thread.join()
+    except KeyboardInterrupt:
+        logger.info("Shutdown requested - exiting betoniarka")

conjurer_librarian/conjurer_librarian.py

@@ -25,6 +25,7 @@ from queue import Queue
 from typing import Dict, Optional
 
 import requests
+import lib_paths
 import scrape_bot
 import search_bot
 # import search_bot2 as search_bot
@@ -194,7 +195,7 @@ class Librarian(object):
             self.app.logger.info("CROSSREF DONE")
 
             self.app.logger.info("CROSSREF DONE")
-        with open("cr_results.json", "r+", encoding="utf-8") as data_file:
+        with open(lib_paths.CR_RESULTS, "r+", encoding="utf-8") as data_file:
             # First we load existing data into a dict.
             try:
                 file_data = json.load(data_file)
@@ -258,7 +259,7 @@ class Librarian(object):
 
         for item in temp:
             refined_result[item["DOI"]]=  item
-        with open("rr_results.json", "r+", encoding="utf-8") as data_file:
+        with open(lib_paths.RR_RESULTS, "r+", encoding="utf-8") as data_file:
             # First we load existing data into a dict.
             try:
                 file_data = json.load(data_file)
@@ -417,7 +418,8 @@ class BackgroundTaskSearch(threading.Thread):
             self.app.logger.info("Saving to file")
 
             # Save results to "not_in_db.json" file
-            with open("not_in_db.json", "r+", encoding="utf-8") as ndb_file:
+            with open(lib_paths.NOT_IN_DB, "r+", encoding="utf-8") as ndb_file:
+                ndb_database = {}
                 try:
                     ndb_database = json.load(ndb_file)
                 except JSONDecodeError:
@@ -431,7 +433,8 @@ class BackgroundTaskSearch(threading.Thread):
                 json.dump(ndb_database, ndb_file)
 
             # Save results to "s_results.json" file
-            with open("s_results.json", "r+", encoding="utf-8") as s_file:
+            with open(lib_paths.S_RESULTS, "r+", encoding="utf-8") as s_file:
+                database = {}
                 try:
                     database = json.load(s_file)
                 except JSONDecodeError:

conjurer_librarian/cut_pieces.py

@@ -0,0 +1,193 @@
+#!/usr/bin/env python3
+
+from pathlib import Path
+import argparse
+import logging
+import sys
+
+
+# =========================
+# Configuration constants
+# =========================
+
+MAX_CHUNK_SIZE_BYTES = 50 * 1024 * 1024  # 100 MB
+OUTPUT_SUFFIX = "_chunk.txt"
+
+
+# =========================
+# Logging
+# =========================
+
+logging.basicConfig(
+    level=logging.INFO,
+    format="%(asctime)s [%(levelname)s] %(message)s",
+)
+
+logger = logging.getLogger(__name__)
+
+
+def get_default_output_dir(input_file: Path) -> Path:
+    """
+    Return default output directory placed next to the input file.
+
+    Example:
+        /data/big.txt -> /data/big/
+    """
+    return input_file.with_suffix("")
+
+
+def write_chunk(output_dir: Path, chunk_index: int, lines: list[bytes], chunk_size: int) -> Path:
+    """
+    Write one chunk file and return its path.
+    """
+    output_file = output_dir / f"{chunk_index}{OUTPUT_SUFFIX}"
+
+    with output_file.open("wb") as file:
+        file.writelines(lines)
+
+    actual_size = output_file.stat().st_size
+
+    if actual_size != chunk_size:
+        logger.warning(
+            "Chunk size mismatch for %s: expected %d bytes, got %d bytes",
+            output_file,
+            chunk_size,
+            actual_size,
+        )
+
+    return output_file
+
+
+def split_txt_file(input_file: Path, output_dir: Path | None = None) -> None:
+    """
+    Split a text-like file into smaller files limited by MAX_CHUNK_SIZE_BYTES.
+
+    The file is processed in binary mode, so invalid UTF-8 or mixed encodings
+    do not break the split. Lines are never split.
+    """
+    if not input_file.exists():
+        raise FileNotFoundError(f"Input file does not exist: {input_file}")
+
+    if not input_file.is_file():
+        raise ValueError(f"Input path is not a file: {input_file}")
+
+    if MAX_CHUNK_SIZE_BYTES <= 0:
+        raise ValueError("MAX_CHUNK_SIZE_BYTES must be greater than zero")
+
+    if output_dir is None:
+        output_dir = get_default_output_dir(input_file)
+
+    output_dir.mkdir(parents=True, exist_ok=True)
+
+    logger.info("Input file: %s", input_file)
+    logger.info("Output directory: %s", output_dir)
+    logger.info("Max chunk size: %d bytes", MAX_CHUNK_SIZE_BYTES)
+
+    chunk_index = 0
+    current_lines: list[bytes] = []
+    current_size = 0
+    total_lines = 0
+    total_bytes = 0
+
+    with input_file.open("rb") as input_handle:
+        for line_number, line in enumerate(input_handle, start=1):
+            line_size = len(line)
+            total_lines = line_number
+            total_bytes += line_size
+
+            if line_size > MAX_CHUNK_SIZE_BYTES:
+                logger.warning(
+                    "Line %d is larger than max chunk size: %d bytes > %d bytes. "
+                    "It will be written as a separate chunk.",
+                    line_number,
+                    line_size,
+                    MAX_CHUNK_SIZE_BYTES,
+                )
+
+            if current_lines and current_size + line_size > MAX_CHUNK_SIZE_BYTES:
+                output_file = write_chunk(
+                    output_dir=output_dir,
+                    chunk_index=chunk_index,
+                    lines=current_lines,
+                    chunk_size=current_size,
+                )
+
+                logger.info(
+                    "Written chunk %d: %s (%d bytes)",
+                    chunk_index,
+                    output_file,
+                    current_size,
+                )
+
+                chunk_index += 1
+                current_lines = []
+                current_size = 0
+
+            current_lines.append(line)
+            current_size += line_size
+
+    if current_lines:
+        output_file = write_chunk(
+            output_dir=output_dir,
+            chunk_index=chunk_index,
+            lines=current_lines,
+            chunk_size=current_size,
+        )
+
+        logger.info(
+            "Written chunk %d: %s (%d bytes)",
+            chunk_index,
+            output_file,
+            current_size,
+        )
+
+        chunk_count = chunk_index + 1
+    else:
+        logger.info("Input file is empty. No chunks were created.")
+        chunk_count = 0
+
+    logger.info("Total lines processed: %d", total_lines)
+    logger.info("Total bytes processed: %d", total_bytes)
+    logger.info("Chunks created: %d", chunk_count)
+    logger.info("Done.")
+
+
+def parse_args() -> argparse.Namespace:
+    parser = argparse.ArgumentParser(
+        description="Split a large TXT-like file into smaller line-safe chunks."
+    )
+
+    parser.add_argument(
+        "input_file",
+        type=Path,
+        help="Path to input TXT file.",
+    )
+
+    parser.add_argument(
+        "-o",
+        "--output-dir",
+        type=Path,
+        default=None,
+        help=(
+            "Optional output directory. "
+            "By default, a directory named after the input file is created next to it."
+        ),
+    )
+
+    return parser.parse_args()
+
+
+def main() -> int:
+    args = parse_args()
+
+    try:
+        split_txt_file(args.input_file, args.output_dir)
+        return 0
+    except Exception:
+        logger.exception("Failed to split file")
+        return 1
+
+
+if __name__ == "__main__":
+    sys.exit(main())
+

conjurer_librarian/lib_paths.py

@@ -0,0 +1,41 @@
+"""Resolved, seeded paths for the librarian's runtime JSON state.
+
+``conjurer_librarian.py`` and ``scrape_bot.py`` open these files in place with
+mode ``r+``, which requires them to already exist - in a fresh container the
+working directory has none of them, so the worker threads crashed with
+FileNotFoundError.
+
+They now live in a single mounted, persistent directory
+(``CONJURER_LIBRARIAN_STATE_DIR``, default ``/lib_temp_files``) and are seeded
+with an empty JSON object on import, so a fresh container/volume never crashes
+and the accumulated results survive restarts.
+"""
+import os
+
+STATE_DIR = os.getenv("CONJURER_LIBRARIAN_STATE_DIR", "/lib_temp_files")
+
+CR_RESULTS = os.path.join(STATE_DIR, "cr_results.json")   # Crossref raw hits
+RR_RESULTS = os.path.join(STATE_DIR, "rr_results.json")   # refined results
+NOT_IN_DB = os.path.join(STATE_DIR, "not_in_db.json")     # DOIs to scrape
+S_RESULTS = os.path.join(STATE_DIR, "s_results.json")     # final send results
+
+_ALL = (CR_RESULTS, RR_RESULTS, NOT_IN_DB, S_RESULTS)
+
+
+def ensure_state_files():
+    """Create the state dir and seed any missing file with an empty JSON dict."""
+    try:
+        os.makedirs(STATE_DIR, exist_ok=True)
+    except OSError as exc:  # pragma: no cover - surfaced in logs, not fatal
+        print(f"lib_paths: cannot create {STATE_DIR}: {exc}")
+        return
+    for path in _ALL:
+        if not os.path.exists(path):
+            try:
+                with open(path, "w", encoding="utf-8") as handle:
+                    handle.write("{}")
+            except OSError as exc:  # pragma: no cover
+                print(f"lib_paths: cannot seed {path}: {exc}")
+
+
+ensure_state_files()

conjurer_librarian/scrape_bot.py

@@ -16,8 +16,10 @@ from urllib.request import urlopen
 from requests import ConnectionError as RequestsConnectionError
 from requests import ConnectTimeout, Timeout
 
+import lib_paths
+
 SCR_DATABASE_PATH = os.getenv("CONJURER_LIBRARIAN_DB_PATH", r"C:\\Database\\chunks\\")
-SCR_FILENAME = os.getenv("CONJURER_LIBRARIAN_SCRAPE_CHUNK", "40_chunk.txt")
+SCR_FILENAME = os.getenv("CONJURER_LIBRARIAN_SCRAPE_CHUNK", "44_chunk.txt")
 SCR_ENCODING = os.getenv("CONJURER_ENCODING", "utf-8")
 
 WORK_Q = Queue()
@@ -30,7 +32,7 @@ def load_ndb_to_q(logger):
     """
     logger.info("Loader started")
     while True:
-        with open("not_in_db.json", "r+", encoding="utf-8") as ndb_file:
+        with open(lib_paths.NOT_IN_DB, "r+", encoding="utf-8") as ndb_file:
             try:
                 ndb_database = json.load(ndb_file)
                 for _ in range (1,10):
@@ -38,7 +40,7 @@ def load_ndb_to_q(logger):
                         key = next(iter(ndb_database))
                         _ = ndb_database.pop(key)
                         logger.info(key)
-                        url = f"https://sci-hub.se/{key}"
+                        url = f"https://sci-hub.red/{key}"
                         WORK_Q.put([key, url, False])
                     except StopIteration:
                         break

conjurer_librarian/search_bot.py

@@ -47,7 +47,7 @@ def producer(out_q, control_q, filename, _logger):
         filename (str): Name of the file.
         _logger: Logger object for logging.
     """
-    with open(DATABASE_PATH + filename, "r+", encoding=ENCODING) as operated_file:
+    with open(DATABASE_PATH + filename, "r", encoding=ENCODING) as operated_file:
         print(f"Worker {filename} ")
         line_no = 0
         while True:

conjurer_librarian/search_bot2.py

@@ -47,7 +47,7 @@ def producer(out_q, control_q, filename, _logger):
         filename (str): Name of the file.
         _logger: Logger object for logging.
     """
-    with open(DATABASE_PATH + filename, "r+", encoding=ENCODING) as operated_file:
+    with open(DATABASE_PATH + filename, "r", encoding=ENCODING) as operated_file:
         print(f"Worker {filename} ")
         line_no = 0
         while True:

conjurer_musician/conjurer_musician.py

@@ -1,7 +1,12 @@
-"""
-The provided Python script sets up a Flask web server to manage a list of music files, with
-functions for rescanning the music folder, updating the music list, and serving the music list via
-API endpoints.
+"""Musician - the Discord music player service.
+
+Serves the music library index and keyword search the bot uses for Discord
+playback (/mp3, /update_mp3, /get_music) plus the file-share endpoints.
+
+Radio playlist management and the radio-log tailer moved to
+conjurer_betoniarka/betoniarka.py, which runs INSIDE the radio container as
+the same user as Liquidsoap - the musician no longer writes any radio files,
+so the old root-owned-network-share permission mess is gone.
 """
 
 import json
@@ -14,18 +19,9 @@ import time
 from datetime import datetime
 from logging import handlers
 from pathlib import Path
-from typing import Dict, List
+from typing import List
 
-import requests
-from flask import (
-    Flask,
-    abort,
-    jsonify,
-    redirect,
-    render_template,
-    request,
-    send_from_directory,
-)
+from flask import Flask, abort, jsonify, request
 from waitress import serve
 
 import media_search_functions
@@ -41,8 +37,6 @@ def _env_path(name: str, default: str) -> Path:
 
 
 API_KEY = os.getenv("CONJURER_API_KEY")
-MAIN_BOT_ADDRESS = _env("CONJURER_MAIN_BOT", "http://127.0.0.1:5000")
-MUSIC_TRACKER = _env("CONJURER_MUSIC_TRACKER_ENDPOINT", "/prepped_tracks")
 HOST_ADDRESS = _env("CONJURER_MUSICIAN_HOST", "0.0.0.0")
 PORT_ADDRESS = int(_env("CONJURER_MUSICIAN_PORT", "5000"))
 
@@ -56,52 +50,12 @@ LOGSTORE = _env_path("CONJURER_LOGSTORE", str(BASE_DIR / "logs"))
 MUSIC_FOLDER = _env_path(
     "CONJURER_MUSIC_FOLDER", str(BASE_DIR / "music")
 )
-PRIORITY_FOLDER = _env_path(
-    "CONJURER_PRIORITY_FOLDER", str(MUSIC_FOLDER / "priority")
-)
-RADIOLOG_PATH = _env_path(
-    "CONJURER_RADIO_LOG", str(BASE_DIR / "radio_log.log")
-)
-PERSISTENCE_PATH = _env_path(
-    "CONJURER_PERSISTENCE_LOG", str(BASE_DIR / "persistence.log")
-)
-ALL_PLAYLIST_PATH = _env_path(
-    "CONJURER_ALL_PLAYLIST", str(BASE_DIR / "all_playlist.playlist")
-)
-HIT_PLAYLIST_PATH = _env_path(
-    "CONJURER_HIT_PLAYLIST", str(BASE_DIR / "hit.playlist")
-)
-REQUEST_PLAYLIST_PATH = _env_path(
-    "CONJURER_REQUEST_PLAYLIST", str(BASE_DIR / "request.playlist")
-)
-PRIORITY_PLAYLIST_PATH = _env_path(
-    "CONJURER_PRIORITY_PLAYLIST", str(BASE_DIR / "priority_queue.playlist")
-)
-STREAM_TEMPLATE = _env_path(
-    "CONJURER_STREAM_TEMPLATE", str(BASE_DIR / "stream.html")
-)
 
 ENCODING = _env("CONJURER_ENCODING", "utf-8")
 SEPARATOR_FILE_PATH = os.sep
 
-for playlist_path in (
-    ALL_PLAYLIST_PATH,
-    HIT_PLAYLIST_PATH,
-    REQUEST_PLAYLIST_PATH,
-    PRIORITY_PLAYLIST_PATH,
-):
-    playlist_path.parent.mkdir(parents=True, exist_ok=True)
-
 random.seed()
 music_file_list: List[str] = []
-priority_list: List[str] = []
-
-
-def _build_headers() -> Dict[str, str]:
-    headers: Dict[str, str] = {}
-    if API_KEY:
-        headers["X-Conjurer-Api-Key"] = API_KEY
-    return headers
 
 
 def _authorize_request() -> None:
@@ -109,29 +63,16 @@ def _authorize_request() -> None:
         abort(401)
 
 
-def _post_to_bot(payload: List[str]) -> None:
-    response = requests.post(
-        f"{MAIN_BOT_ADDRESS}{MUSIC_TRACKER}",
-        json=payload,
-        headers=_build_headers(),
-        timeout=60,
-    )
-    logger = logging.getLogger("conjurer_musician")
-    logger.info("SENT")
-    logger.info(response.status_code)
-    logger.info("SEND CONFIRMED")
-
-
 def rescan():
-    """
-    The `rescan` function logs a message, scans for mp3 files in a specified folder,
-    and adds them to a list of music files.
+    """Refresh the in-memory library index used by /mp3 and /get_music.
+
+    Radio playlist files are NOT written here anymore - that is the
+    betoniarka's job, colocated with Liquidsoap.
     """
     logger = logging.getLogger("conjurer_musician")
     logger.info("Rescan triggered")
 
     music_file_list.clear()
-    priority_list.clear()
 
     for mp3_item in MUSIC_FOLDER.glob("**/*.mp3"):
         temp_music_file = mp3_item.as_posix()
@@ -139,27 +80,6 @@ def rescan():
             temp_music_file = temp_music_file.replace("/", "\\")
         music_file_list.append(temp_music_file)
 
-    for mp3_item in PRIORITY_FOLDER.glob("**/*.mp3"):
-        temp_music_file = mp3_item.as_posix()
-        if os.name == "nt":
-            temp_music_file = temp_music_file.replace("/", "\\")
-        priority_list.append(temp_music_file)
-
-    with ALL_PLAYLIST_PATH.open("w", encoding=ENCODING) as w_file:
-        try:
-            for item in music_file_list:
-                w_file.write(item)
-                w_file.write("\n")
-        except json.JSONDecodeError:
-            pass
-    with HIT_PLAYLIST_PATH.open("w", encoding=ENCODING) as w_file:
-        try:
-            for item in priority_list:
-                w_file.write(item)
-                w_file.write("\n")
-        except json.JSONDecodeError:
-            pass
-
 
 def thread_rescan():
     """
@@ -174,55 +94,6 @@ def thread_rescan():
         rescan()
 
 
-def scan_tracks():
-    # Set the filename and open the file
-    logger = logging.getLogger("conjurer_musician")
-    with open(RADIOLOG_PATH, "r", encoding=ENCODING) as log_file:
-        log_file.seek(os.stat(RADIOLOG_PATH).st_size)
-        prev_size = os.stat(PERSISTENCE_PATH).st_size
-
-        while True:
-            current_size = os.stat(PERSISTENCE_PATH).st_size
-            if prev_size != current_size:
-                while prev_size != current_size:
-                    prev_size = current_size
-                    time.sleep(0.1)
-                    current_size = os.stat(PERSISTENCE_PATH).st_size
-                with open(PERSISTENCE_PATH, "r", encoding=ENCODING) as persistence:
-                    lines = persistence.readlines()
-                if len(lines) >= 3:
-                    _post_to_bot(["next", lines[2]])
-
-            position = log_file.tell()
-            line = log_file.readline()
-            if not line:
-                time.sleep(1)
-                log_file.seek(position)
-                continue
-
-            if not re.match(r".*Prepared.*", line):
-                time.sleep(0.1)
-                continue
-
-            result = None
-            if re.match(r".*jingles.*", line):
-                result = ["jingles", line]
-            elif re.match(r".*priority.*", line):
-                result = ["priority", line]
-            elif re.match(r".*hit.*", line):
-                result = ["hit", line]
-            elif re.match(r".*all_playlist.*", line):
-                result = ["all", line]
-            elif re.match(r".*request.*", line):
-                result = ["requests", line]
-
-            if result:
-                logger.info("Forwarding radio log entry: %s", result[0])
-                _post_to_bot(result)
-
-            time.sleep(0.1)
-
-
 app = Flask(__name__)
 # AutoIndex(app, browse_root="/")
 
@@ -328,9 +199,6 @@ def wyszukaj(word_list, how_many, _logger=None, return_to_bot=True):
         while not_found:
             if search_weight[itr][0] == item_to_search:
                 return_list.append(search_weight[itr])
-                if not return_to_bot:
-                    with PRIORITY_PLAYLIST_PATH.open("a", encoding=ENCODING) as s_file:
-                        s_file.write(search_weight[itr][1] + "\n")
                 break
             itr += 1
     else:
@@ -401,62 +269,6 @@ def get_share_links():
 
 
 
-@app.route("/stream", methods=["GET"])
-def stream_music():
-    """
-    The function `stream_music` is a route handler for the "/stream" endpoint.
-    It returns a JSON response containing a key "music_file_list" with the value of the variable `music_file_list`.
-
-    :return: A JSON response containing a key "music_file_list" with the value of the variable `music_file_list`.
-    """
-
-    # return send_from_directory("/tmp/hls", "stream.m3u8")
-    return render_template(str(STREAM_TEMPLATE))
-
-
-@app.route("/<string:file_name>")
-def stream(file_name):
-    """
-    Stream the specified file from the video directory.
-
-    Args:
-        file_name (str): The name of the file to be streamed.
-
-    Returns:
-        Response: The response containing the streamed file.
-    """
-    # trunk-ignore(bandit/B108)
-    video_dir = "/tmp/hls"
-    return send_from_directory(video_dir, file_name)
-
-
-@app.route("/stream_mp3", methods=["GET"])
-def stream_music_mp3():
-    """
-    The function `stream_music_mp3` is a route handler for the "/stream_mp3" endpoint.
-    It redirects the user to the URL "http://www.example.com" with a status code of 302.
-
-    :return: A redirect response to "http://www.example.com" with a status code of 302.
-    """
-    return redirect("http://www.example.com", code=302)
-
-
-@app.route("/clear_pr_pls", methods=["GET"])
-def clear_pr_pls():
-    _authorize_request()
-    """
-    The function `clear_pr_pls` clears the contents of the priority queue playlist file.
-
-    :return: A JSON response indicating the success of the operation.
-    """
-    app.logger.info("CLEARING PLAYLIST")
-    with PRIORITY_PLAYLIST_PATH.open("w", encoding=ENCODING) as cleared_pl:
-        cleared_pl.write("")
-
-    return_data = jsonify(isError=False, message="Success", statusCode=200, data=[])
-    return return_data, 200
-
-
 @app.route("/mp3", methods=["GET"])
 def get_music_list():
     """
@@ -517,96 +329,6 @@ def look_for_playlist():
     return return_data
 
 
-@app.route("/request_radio_file", methods=["POST"])
-def add_request():
-    _authorize_request()
-    record = json.loads(request.data)
-    app.logger.info(record)
-    app.logger.info(record["lista_slow"])
-    app.logger.info(record["UUID"])
-    return_data = wyszukaj(record["lista_slow"], 0, app.logger, False)
-
-    with REQUEST_PLAYLIST_PATH.open("a", encoding=ENCODING) as s_file:
-        for item in return_data:
-            s_file.write(item[1] + "\n")
-    return_data = (
-        jsonify(
-            isError=False, message="Success", statusCode=200, data={"status": "OK"}
-        ),
-        200,
-    )
-    return return_data
-
-
-@app.route("/create_priority_playlist", methods=["POST"])
-def create_priority_playlist():
-    """
-    The function `create_priority_playlist` receives a POST request with a JSON payload, logs the received
-    item, adds it to a music file list, and returns a success message along with the updated record.
-
-    :return: A tuple containing a JSON response and a status code.
-        The JSON response includes keys `isError`,
-        `message`, `statusCode`, and `data`, with values
-        indicating the success of the operation and the
-        data that was received and added to the `music_file_list`.
-        The status code returned is 200,indicating a successful response.
-    """
-    _authorize_request()
-    record = json.loads(request.data)
-    app.logger.info(record)
-    app.logger.info(record["lista_slow"])
-    app.logger.info(record["UUID"])
-    app.logger.info(record["dlugosc_plejlisty"])
-    return_data = wyszukaj(
-        record["lista_slow"], record["dlugosc_plejlisty"], app.logger, False
-    )
-    random.shuffle(return_data)
-    with REQUEST_PLAYLIST_PATH.open("a", encoding=ENCODING) as s_file:
-        for item in return_data:
-            s_file.write(item[1] + "\n")
-    return_data = (
-        jsonify(
-            isError=False, message="Success", statusCode=200, data={"status": "OK"}
-        ),
-        200,
-    )
-    return return_data
-
-
-@app.route("/add_to_priority", methods=["POST"])
-def add_to_priority():
-    """
-    The function `add_to_priority` receives a POST request with a JSON payload, logs the received
-    item, adds it to a music file list, and returns a success message along with the updated record.
-
-    :return: A tuple containing a JSON response and a status code.
-        The JSON response includes keys `isError`,
-        `message`, `statusCode`, and `data`, with values
-        indicating the success of the operation and the
-        data that was received and added to the `music_file_list`.
-        The status code returned is 200,indicating a successful response.
-    """
-    _authorize_request()
-    record = json.loads(request.data)
-    app.logger.info(record)
-    app.logger.info(record["lista_slow"])
-    app.logger.info(record["UUID"])
-    app.logger.info(record["dlugosc_plejlisty"])
-    return_data = wyszukaj(
-        record["lista_slow"], record["dlugosc_plejlisty"], app.logger, False
-    )
-    with PRIORITY_PLAYLIST_PATH.open("a", encoding=ENCODING) as s_file:
-        for item in return_data:
-            s_file.write(item[1] + "\n")
-    return_data = (
-        jsonify(
-            isError=False, message="Success", statusCode=200, data={"status": "OK"}
-        ),
-        200,
-    )
-    return return_data
-
-
 def flask_debug():
     """
     The `flask_debug` function starts a Flask application in debug mode without using the reloader.
@@ -660,13 +382,8 @@ if __name__ == "__main__":
     for worker in threads:
         worker.start()
 
-    time.sleep(60)
-    track_thread = threading.Thread(target=scan_tracks, daemon=True)
-    track_thread.start()
-
     try:
         for worker in threads:
             worker.join()
-        track_thread.join()
     except KeyboardInterrupt:
         logger.info("Shutdown requested - exiting musician service")

conjurer_musician/radio_conjurer.liq

@@ -1,16 +1,16 @@
-# FILEPATH: /home/mtuszowski/conjurer/conjurer_musician/radio_conjurer.liq
+# Radio Conjurer - Liquidsoap script (containerised paths: /srv/betoniarka/*)
 
 # This script sets up a Liquidsoap radio stream with various features and configurations.
 
 # Load icecast credentials from a JSON file
-let json.parse credentials  = file.contents("/home/pi/Conjurer/icecast_credentials.json")
+let json.parse credentials  = file.contents("/srv/betoniarka/secrets/icecast_credentials.json")
 
 # Enable replaygain metadata processing
 enable_replaygain_metadata()
 
 # Set up a playlog for tracking played tracks
 
-l = playlog(duration = 72000.0, persistency="/home/pi/Conjurer/persistence.log")
+l = playlog(duration = 72000.0, persistency="/srv/betoniarka/data/persistence.log")
 
 # Function to check if a track can be played based on its metadata
 def check(r)
@@ -25,20 +25,20 @@ def check(r)
 end
 
 # Define playlists to be used in the stream
-s1 = replaygain(playlist(reload_mode="watch", check_next=check, "/home/pi/Conjurer/all_playlist.playlist"))
-s2 = replaygain(playlist(reload_mode="watch", check_next=check, "/home/pi/Conjurer/priority_queue.playlist"))
-s3 = replaygain(playlist(reload_mode="watch", check_next=check, "/home/pi/Conjurer/hit.playlist"))
+s1 = replaygain(playlist(reload_mode="watch", check_next=check, "/srv/betoniarka/data/all_playlist.playlist"))
+s2 = replaygain(playlist(reload_mode="watch", check_next=check, "/srv/betoniarka/data/priority_queue.playlist"))
+s3 = replaygain(playlist(reload_mode="watch", check_next=check, "/srv/betoniarka/data/hit.playlist"))
 
 # Create a request queue for user-generated requests
 requests_queue = request.queue()
 
 # Function to process the request queue and add new requests
 def queue_processing()
-    text=file.lines("/home/pi/Conjurer/request.playlist")
+    text=file.lines("/srv/betoniarka/data/request.playlist")
     if text != [] then
         list.iter(fun(item) -> requests_queue.push.uri(item), text)
-        file.remove("/home/pi/Conjurer/request.playlist")
-        f = file.open("/home/pi/Conjurer/request.playlist", create=true)
+        file.remove("/srv/betoniarka/data/request.playlist")
+        f = file.open("/srv/betoniarka/data/request.playlist", create=true)
         f.close()
     end
 end
@@ -63,7 +63,7 @@ end
 s4 = random(id="randomizer", weights=[2, 3, 5], [s1, s2, s3])
 
 # Load jingles playlist
-jingles = (playlist(reload_mode="watch", "/home/pi/Conjurer/jingles.playlist"))
+jingles = (playlist(reload_mode="watch", "/srv/betoniarka/data/jingles.playlist"))
 
 # Create the main stream with random playlist and jingles
 s = rotate(id="randomizer", weights=[10, 1], [s4, jingles])
@@ -111,7 +111,7 @@ s=switch(track_sensitive=true,
 # Configure logging settings
 log_to_stdout = true
 log_to_file = true
-logpath = "/home/pi/Conjurer/radio_log.log"
+logpath = "/srv/betoniarka/data/radio_log.log"
 loglevel = 3
 set("log.stdout", log_to_stdout)
 set("log.level", loglevel)
@@ -120,7 +120,7 @@ set("log.level", loglevel)
 set("log.file", log_to_file)
 set("log.file.path", logpath)
 # Set up emergency fallback track
-emergency = single("/home/pi/MediaFolder/mp3/Youtube/Dr. Peacock - Trip to Ireland [GvrvQTUbUcA].mp3")
+emergency = single("/srv/betoniarka/music/Youtube/Dr. Peacock - Trip to Ireland [GvrvQTUbUcA].mp3")
 radio =  fallback(id="switcher2", track_sensitive=false, [s, emergency])
 
 # Set up an interactive control for skipping tracks
@@ -150,11 +150,11 @@ thread.run(every=15., check_skip)
 
 
 # Enable persistent script parameters
-interactive.persistent("script.params")
+interactive.persistent("/srv/betoniarka/data/script.params")
 
 # Configure output formats and destinations
 
-output.icecast(%mp3, host="radio", port=8000, password=credentials.password, icy_metadata="true", mount="mp3-stream", radio)
+output.icecast(%mp3, host="localhost", port=8000, password=credentials.password, icy_metadata="true", mount="mp3-stream", radio)
 output.pulseaudio(radio)
 #output.file.hls("/tmp/hls", [("mp3-low", %mp3(bitrate=96)), ("mp3-hi", %mp3(bitrate=160))], radio)
 # Uncomment the following lines to enable additional output formats

constants.py

@@ -38,6 +38,11 @@ try:
 except ImportError:  # pragma: no cover - optional at runtime
     openai = None
 
+try:
+    import anthropic
+except ImportError:  # pragma: no cover - optional at runtime
+    anthropic = None
+
 try:
     import spotipy
     from spotipy.oauth2 import SpotifyClientCredentials
@@ -189,8 +194,18 @@ DIR_PATH_SADOX = os.getenv("CONJURER_SADOX_DIR", DIR_PATH_SADOX)
 ENCODING = os.getenv("CONJURER_ENCODING", ENCODING)
 SEPARATOR_FILE_PATH = os.getenv("CONJURER_PATH_SEPARATOR", SEPARATOR_FILE_PATH)
 
+# Voice-recognition transcript dumps. Defaults next to the log file (so on the
+# Pi it lands in /home/pi/Conjurer/transcripts/, in docker under /data).
+TRANSCRIPTS_PATH = os.getenv(
+    "CONJURER_TRANSCRIPTS_PATH",
+    os.path.join(os.path.dirname(LOGFILE) or ".", "transcripts") + os.sep,
+)
+
 FILE_SERVICE_ADDRESS = os.getenv("CONJURER_FILE_SERVICE", "http://192.168.1.15:5000")
 RADIO_HARBOR_ADDRESS = os.getenv("CONJURER_RADIO_HARBOR", "http://192.168.1.15:54321")
+# Betoniarka (radio-operator service colocated with Liquidsoap). Falls back to
+# the musician address so deployments that have not split yet keep working.
+RADIO_SERVICE_ADDRESS = os.getenv("CONJURER_RADIO_SERVICE", FILE_SERVICE_ADDRESS)
 SKIP_TRACK = os.getenv("CONJURER_SKIP_ENDPOINT", "/skip")
 LIBRARIAN_SERVICE_ADDRESS = os.getenv(
     "CONJURER_LIBRARIAN_SERVICE", "http://192.168.1.192:5001"
@@ -202,6 +217,66 @@ PORT_ADDRESS = int(os.getenv("CONJURER_DISCORD_PORT", "5000"))
 API_SHARED_KEY = os.getenv("CONJURER_API_KEY", "")
 
 
+# *=========================================== Self-healing runtime layout
+# A fresh host/volume must never kill the bot at import time. Missing
+# directories are created and missing state files are seeded - first from the
+# templates shipped alongside this file (repo checkout / docker image), then
+# from a safe empty structure. Existing files are never touched, so preserved
+# history always wins.
+
+_TEMPLATE_DIR = os.path.dirname(os.path.abspath(__file__))
+
+
+def _ensure_dir(path: str) -> None:
+    if not path:
+        return
+    try:
+        os.makedirs(path, exist_ok=True)
+    except OSError as exc:
+        logger.warning("Cannot create directory %s: %s", path, exc)
+
+
+def _seed_file(path: str, template_name: str, empty_content: str) -> None:
+    """Create *path* from the repo template (or *empty_content*) if missing."""
+    if not path or os.path.exists(path):
+        return
+    _ensure_dir(os.path.dirname(path) or ".")
+    template = os.path.join(_TEMPLATE_DIR, template_name)
+    try:
+        if os.path.exists(template) and os.path.abspath(template) != os.path.abspath(path):
+            import shutil
+
+            shutil.copyfile(template, path)
+            logger.warning("Seeded missing %s from repo template", path)
+        else:
+            with open(path, "w", encoding=ENCODING) as handle:
+                handle.write(empty_content)
+            logger.warning("Created missing %s as empty state", path)
+    except OSError as exc:
+        logger.warning("Cannot seed %s: %s", path, exc)
+
+
+def _ensure_runtime_layout() -> None:
+    for directory in (
+        os.path.dirname(LOGFILE) or ".",
+        LOGSTORE,
+        GRAPHICS_PATH,
+        MUSIC_FOLDER,
+        TRANSCRIPTS_PATH,
+    ):
+        _ensure_dir(directory)
+
+    # (target path, template shipped next to this file, empty fallback)
+    _seed_file(SETTINGS_FILE, "settings.json", "{}")
+    _seed_file(SYSTEM_GPT_SETTINGS, "system_gpt_settings.json", "{}")
+    _seed_file(MEMORY_FIVE_SIARA, "pamiec.json", "[]")
+    _seed_file(MEMORY_FIVE_MUZYKA, "pamiec_muzyki.json", "[]")
+    _seed_file(ACCIDENT_LOG, "accident_log.json", "[]")
+
+
+_ensure_runtime_layout()
+
+
 # *=========================================== Defensive state loading
 def _load_json(path: str, fallback):
     """Load JSON from *path*, falling back gracefully on missing/corrupt files."""
@@ -266,8 +341,20 @@ if openai and OPENAI_API_KEY:
 else:
     OPENAICLIENT = None
 
+# Claude / Anthropic client, wired analogously to OpenAI above so the AI cog can
+# be pointed at either backend with a single config switch (see AI_CONFIGS and
+# ai_functions.set_active_ai_config). netrc machine name 'anthropic' works too.
+ANTHROPIC_API_KEY = _resolve_token("anthropic", "ANTHROPIC_API_KEY")
+if anthropic and ANTHROPIC_API_KEY:
+    CLAUDECLIENT = anthropic.AsyncAnthropic(api_key=ANTHROPIC_API_KEY)
+else:
+    CLAUDECLIENT = None
+
 TOKEN = _resolve_token("discord", "DISCORD_TOKEN")
 
+# Voice recognition (AssemblyAI). None = the voice cog reports and disables.
+ASSEMBLYAI_API_KEY = _resolve_token("assemblyai", "ASSEMBLYAI_API_KEY")
+
 if spotipy:
     _spotify_creds = _load_netrc_credentials("spotipy")
     if _spotify_creds and SpotifyClientCredentials:
@@ -315,6 +402,75 @@ GUILD_ID = 664789470779932693
 LATEST_MODEL = "gpt-4o"          # najnowszy/do rozmów (możesz zmienić w jednym miejscu)
 CHEAP_MODEL  = "gpt-4o-mini"     # najtańszy (fallback do 3.5 niżej)
 
+# Claude counterparts of LATEST_MODEL / CHEAP_MODEL. Opus 4.8 is the strongest
+# widely available model; Haiku 4.5 is the fast/cheap tier used for MUSIC.
+CLAUDE_LATEST_MODEL = "claude-opus-4-8"
+CLAUDE_CHEAP_MODEL  = "claude-haiku-4-5"
+
+
+# *=========================================== AI provider configs
+# The bot's AI functionality (ai_functions.handle_response) can be pointed at a
+# different backend by flipping a single "active" switch. Each named config
+# collects the *differences* between providers (which backend, which models,
+# generation params). These live in system_gpt_settings.json under an optional
+# third list element (index 2) so future configs are simply added there:
+#
+#   [ <system message>, <personal assistants>, { "active": "gpt",
+#                                                 "configs": { ... } } ]
+#
+# The block is optional and backward compatible: a settings file with only the
+# original two elements falls back to the built-in defaults below (active
+# "gpt"), so existing deployments behave exactly as before. The active config
+# can be overridden at import time with CONJURER_AI_CONFIG and at runtime with
+# the $gadaj_teraz command (which persists the choice back into index 2).
+def _default_ai_configs():
+    return {
+        "gpt": {
+            "provider": "openai",
+            "latest_model": LATEST_MODEL,
+            "cheap_model": CHEAP_MODEL,
+            "temperature": 0.2,
+        },
+        "claude": {
+            "provider": "anthropic",
+            "latest_model": CLAUDE_LATEST_MODEL,
+            "cheap_model": CLAUDE_CHEAP_MODEL,
+            # Anthropic requires max_tokens; temperature is intentionally not
+            # sent for Claude (Opus 4.8 / Sonnet 5 reject sampling params).
+            "max_tokens": 2048,
+        },
+        # Template for wiring further providers. Copy it, rename the key, point
+        # "provider" at a backend ai_functions.provider_generate implements, and
+        # fill in the model ids. Keys starting with "_" are treated as inert
+        # templates and are hidden from the $gadaj_teraz picker.
+        "_template": {
+            "provider": "openai",
+            "latest_model": "model-id-here",
+            "cheap_model": "cheaper-model-id-here",
+            "temperature": 0.2,
+            "max_tokens": 2048,
+        },
+    }
+
+
+_ai_block = (
+    GPT_SETTINGS[2]
+    if isinstance(GPT_SETTINGS, list) and len(GPT_SETTINGS) > 2 and isinstance(GPT_SETTINGS[2], dict)
+    else {}
+)
+AI_CONFIGS = _ai_block.get("configs") or _default_ai_configs()
+# Single switch: env var wins, then the settings-file "active" key, then "gpt".
+DEFAULT_AI_CONFIG = (
+    os.getenv("CONJURER_AI_CONFIG")
+    or _ai_block.get("active")
+    or "gpt"
+)
+if DEFAULT_AI_CONFIG not in AI_CONFIGS:
+    logger.warning(
+        "AI config '%s' not found - falling back to 'gpt'", DEFAULT_AI_CONFIG
+    )
+    DEFAULT_AI_CONFIG = "gpt" if "gpt" in AI_CONFIGS else next(iter(AI_CONFIGS))
+
 
 # *=========================================== Conan Exiles bridge
 # Every value is optional; the conanjurer cog stays dormant unless configured.

docker/Dockerfile.bot

@@ -29,9 +29,14 @@ RUN curl -fsSL https://drop-sh.fullyjustified.net | sh \
     && mv tectonic /usr/local/bin/tectonic \
     || echo "tectonic not installed - the LaTeX feature will be disabled"
 
-COPY requirements_bot.txt ./
+COPY requirements_bot.txt requirements_conan.txt ./
 RUN pip install --no-cache-dir --upgrade pip \
     && pip install --no-cache-dir -r requirements_bot.txt
+# Best-effort Conan bridge extras: aiomcrcon supports Python <= 3.11 only, so
+# on this 3.13 image the install fails harmlessly and the conanjurer cog stays
+# dormant (its imports are guarded).
+RUN pip install --no-cache-dir -r requirements_conan.txt \
+    || echo "conan extras skipped - conanjurer cog will stay dormant"
 
 # Vendored forks take import precedence over the pip packages of the same name,
 # because /app (the script dir) is first on sys.path when running `python bot.py`.

docker/Dockerfile.librarian

@@ -18,10 +18,13 @@ COPY conjurer_librarian/ ./
 ENV PYTHONUNBUFFERED=1 \
     CONJURER_LIBRARIAN_HOST=0.0.0.0 \
     CONJURER_LIBRARIAN_PORT=5001 \
-    CONJURER_LIBRARIAN_DB_PATH=/doi/
+    CONJURER_LIBRARIAN_DB_PATH=/doi/ \
+    CONJURER_LIBRARIAN_STATE_DIR=/lib_temp_files
 
-# The local DOI chunk database (large) is mounted here.
-VOLUME ["/doi"]
+# /doi          = the local DOI chunk database (large, read-only).
+# /lib_temp_files = runtime JSON state (cr_results/rr_results/not_in_db/
+#                 s_results) - seeded on first run, persisted across restarts.
+VOLUME ["/doi", "/lib_temp_files"]
 EXPOSE 5001
 
 CMD ["python", "conjurer_librarian.py"]

docker/Dockerfile.radio

@@ -0,0 +1,110 @@
+# Radio Conjurer - Liquidsoap runtime built through opam, so the OCaml,
+# opam and Liquidsoap versions are all selectable at build time.
+#
+# Build from the repository root, e.g.:
+#   docker build -f docker/Dockerfile.radio -t conjurer-radio .
+#   docker build -f docker/Dockerfile.radio -t conjurer-radio \
+#       --build-arg LIQUIDSOAP_VERSION=2.1.4 --build-arg OCAML_VERSION=4.14.2 .
+#
+# Version notes (matched to radio_conjurer.liq, which targets 2.1.4):
+# * Liquidsoap 2.1.x requires OCaml 4.x - it does NOT build on OCaml 5.
+#   Production ran the 4.13.0 switch; 4.14.x is the maintained 4.x line and
+#   builds 2.1.4 fine. Pass OCAML_VERSION=4.13.0 for bit-for-bit parity.
+# * The ocaml-ffmpeg bindings compatible with 2.1.4 target the FFmpeg 5.x
+#   library family (libavcodec59/libavformat59/libavutil57...). Debian
+#   bookworm ships exactly that, which is why this image needs NO ffmpeg
+#   pinning: the old ffmpeg.pref (Pin-Priority 1001 on deb.debian.org) only
+#   existed to force those Debian builds over the Raspberry Pi OS repo's
+#   conflicting ones. Single-repo container = the pin is redundant.
+FROM debian:bookworm-slim
+
+ARG OPAM_VERSION=2.1.5
+ARG OCAML_VERSION=4.14.2
+ARG LIQUIDSOAP_VERSION=2.1.4
+# Optional liquidsoap features, resolved together with liquidsoap by opam.
+# These cover everything radio_conjurer.liq uses:
+#   mad+lame  - mp3 decode/encode (%mp3 icecast stream)
+#   cry       - output.icecast
+#   taglib    - tag/replaygain metadata
+#   pulseaudio- input.pulseaudio (mic) + output.pulseaudio
+#   samplerate- resampling
+#   inotify   - playlist(reload_mode="watch")
+#   ffmpeg    - decode fallback + replaygain computation
+ARG LIQ_OPAM_PACKAGES="mad lame cry taglib pulseaudio samplerate inotify ffmpeg"
+
+# System libraries: build deps for the opam packages above plus the matching
+# runtime libs. The libav*-dev list mirrors the old ffmpeg.pref family.
+RUN apt-get update && apt-get install -y --no-install-recommends \
+        build-essential m4 pkg-config git curl ca-certificates unzip rsync \
+        libpcre3-dev libgmp-dev zlib1g-dev \
+        libmad0-dev libmp3lame-dev libtag1-dev \
+        libpulse-dev libsamplerate0-dev \
+        libavcodec-dev libavformat-dev libavutil-dev libavfilter-dev \
+        libavdevice-dev libswresample-dev libswscale-dev libpostproc-dev \
+        libcurl4-gnutls-dev \
+        ffmpeg \
+        pulseaudio pulseaudio-utils \
+        icecast2 jq \
+        python3 python3-flask python3-waitress python3-requests \
+    && rm -rf /var/lib/apt/lists/*
+
+# Liquidsoap refuses to run as root (security exit), so it gets a dedicated
+# user. audio/pulse-access mirror what was done by hand on the Pi for the
+# system-wide pulse socket ('pulse-access' comes with the pulseaudio pkg).
+RUN useradd --system --create-home --home-dir /var/lib/radio \
+        --shell /usr/sbin/nologin radio \
+    && usermod -aG audio,pulse-access radio
+
+# opam as a static binary so OPAM_VERSION is a real choice (apt would pin us
+# to whatever bookworm ships).
+RUN ARCH=$(uname -m) \
+    && curl -fsSL -o /usr/local/bin/opam \
+       "https://github.com/ocaml/opam/releases/download/${OPAM_VERSION}/opam-${OPAM_VERSION}-${ARCH}-linux" \
+    && chmod +x /usr/local/bin/opam
+
+# OCaml switch + liquidsoap and its optional feature libraries in one solve,
+# so liquidsoap is compiled WITH those features enabled. OPAMROOT lives in
+# /opt/opam (not /root/.opam) so the unprivileged 'radio' user can read the
+# liquidsoap binary AND its stdlib .liq files at runtime.
+ENV OPAMROOT=/opt/opam
+RUN opam init -y --bare --disable-sandboxing \
+    && opam switch create default "${OCAML_VERSION}" \
+    && opam install -y "liquidsoap.${LIQUIDSOAP_VERSION}" ${LIQ_OPAM_PACKAGES} \
+    && opam clean -a -c -s --logs
+
+ENV PATH="/opt/opam/default/bin:${PATH}"
+
+# Build-time smoke test: the binary runs and reports the requested version.
+RUN liquidsoap --version
+
+# Minimal system-wide pulse config for headless VMs (PULSE_MODE=internal):
+# a null sink so output.pulseaudio()/input.pulseaudio() work with no sound
+# hardware (mic becomes silence, which blank.strip already gates out).
+COPY docker/pulse-system.pa /etc/pulse/system.pa
+
+# The script and its persistent params are seeded into the data volume on
+# first run (never overwritten), so live edits survive image rebuilds. The
+# icecast config is rendered from the template at startup with passwords
+# taken from the secrets volume (never baked into the image).
+WORKDIR /app
+COPY conjurer_musician/radio_conjurer.liq ./radio_conjurer.liq
+COPY conjurer_musician/script.params ./script.params
+COPY conjurer_musician/stream.html ./stream.html
+COPY conjurer_betoniarka/betoniarka.py ./betoniarka.py
+COPY docker/icecast.xml.tpl ./icecast.xml.tpl
+COPY docker/entrypoint.radio.sh /usr/local/bin/entrypoint.sh
+RUN chmod +x /usr/local/bin/entrypoint.sh
+
+# Volume layout matches the paths inside radio_conjurer.liq:
+#   /srv/betoniarka/data    - playlists, script.params, persistence/radio logs
+#   /srv/betoniarka/music   - the mp3 library
+#   /srv/betoniarka/secrets - icecast_credentials.json (provisioned at install)
+VOLUME ["/srv/betoniarka/data", "/srv/betoniarka/music", "/srv/betoniarka/secrets"]
+
+# 8000 = icecast (listeners), 5005 = betoniarka HTTP API (the bot's
+# CONJURER_RADIO_SERVICE), 54321 = harbor /skip (the bot's RADIO_HARBOR),
+# 9999 = interactive harbor.
+EXPOSE 8000 5005 54321 9999
+
+ENTRYPOINT ["/usr/local/bin/entrypoint.sh"]
+CMD ["liquidsoap", "/srv/betoniarka/data/radio_conjurer.liq"]

docker/compose.librarian.yaml

@@ -14,7 +14,11 @@ services:
     ports:
       - "5001:5001"
     volumes:
-      # Local DOI chunk database (0_chunk.txt ... N_chunk.txt).
-      - /srv/librarian/doi:/doi:ro
+      # Local DOI chunk database (0_chunk.txt ... N_chunk.txt). Read-only:
+      # the search workers only read it (open mode "r").
+      - /mnt/conjurer_swap/librarian_data/base_it1:/doi:ro
+      # Runtime JSON state (cr_results/rr_results/not_in_db/s_results) -
+      # seeded on first run, persisted here across restarts.
+      - /srv/librarian/state:/lib_temp_files
       # Optional: netrc holding Crossref credentials (or use CONJURER_CROSSREF_MAILTO).
       - /srv/librarian/secrets/.netrc:/secrets/.netrc:ro

docker/compose.radio.yaml

@@ -0,0 +1,45 @@
+# Radio (Liquidsoap + Icecast) VM. Run from the repository root:
+#   docker compose -f docker/compose.radio.yaml up -d --build
+#
+# Version pins are build args - override via environment, e.g.:
+#   LIQUIDSOAP_VERSION=2.1.4 OCAML_VERSION=4.13.0 \
+#     docker compose -f docker/compose.radio.yaml up -d --build
+services:
+  conjurer-radio:
+    build:
+      context: ..
+      dockerfile: docker/Dockerfile.radio
+      args:
+        OPAM_VERSION: ${OPAM_VERSION:-2.1.5}
+        OCAML_VERSION: ${OCAML_VERSION:-4.14.2}
+        LIQUIDSOAP_VERSION: ${LIQUIDSOAP_VERSION:-2.1.4}
+    image: conjurer-radio:latest
+    container_name: conjurer-radio
+    restart: unless-stopped
+    environment:
+      # internal = null-sink pulse inside the container (headless VM default)
+      # host     = mount the host pulse socket below and set PULSE_SERVER
+      PULSE_MODE: ${PULSE_MODE:-internal}
+      # PULSE_SERVER: unix:/tmp/pulseaudio.socket
+      # Hostname icecast reports in its status/YP pages:
+      ICECAST_HOSTNAME: ${ICECAST_HOSTNAME:-localhost}
+      # Betoniarka (radio-operator API + log forwarder):
+      CONJURER_MAIN_BOT: ${CONJURER_MAIN_BOT:-http://127.0.0.1:5000}
+      CONJURER_API_KEY: ${CONJURER_API_KEY:-}
+    ports:
+      - "8000:8000"     # icecast - listeners tune in here
+      - "5005:5005"     # betoniarka API - the bot's CONJURER_RADIO_SERVICE
+      - "54321:54321"   # harbor /skip - the bot's CONJURER_RADIO_HARBOR
+      - "9999:9999"     # interactive harbor (keep LAN-only!)
+    volumes:
+      # Playlists, logs, script.params and the script itself (seeded on first
+      # run) - same paths inside and outside the container.
+      - /srv/betoniarka/data:/srv/betoniarka/data
+      # The mp3 library. Writable because the entrypoint seeds a silent
+      # emergency-fallback mp3 when the hardcoded single() file is missing.
+      - /srv/betoniarka/music:/srv/betoniarka/music
+      # icecast_credentials.json - provision at install like the other
+      # services' secrets (a CHANGE_ME placeholder is seeded if absent).
+      - /srv/betoniarka/secrets:/srv/betoniarka/secrets
+      # PULSE_MODE=host: uncomment and adjust
+      # - /tmp/pulseaudio.socket:/tmp/pulseaudio.socket

docker/entrypoint.radio.sh

@@ -0,0 +1,113 @@
+#!/bin/sh
+# Prepare the radio volumes, start the in-container Icecast server, wire up
+# PulseAudio and exec liquidsoap. Existing files are never overwritten -
+# live-edited script/params/playlists always win.
+set -e
+
+DATA="${RADIO_DATA_DIR:-/srv/betoniarka/data}"
+MUSIC="${RADIO_MUSIC_DIR:-/srv/betoniarka/music}"
+SECRETS="${RADIO_SECRETS_DIR:-/srv/betoniarka/secrets}"
+CREDS="$SECRETS/icecast_credentials.json"
+mkdir -p "$DATA" "$MUSIC" "$SECRETS"
+
+# Seed the script + persistent interactive params from the image on first run.
+[ -e "$DATA/radio_conjurer.liq" ] || cp /app/radio_conjurer.liq "$DATA/"
+if [ ! -e "$DATA/script.params" ]; then
+    if [ -e /app/script.params ]; then cp /app/script.params "$DATA/"; else : > "$DATA/script.params"; fi
+fi
+
+# Playlists + logs the script watches/writes; empty files keep it happy until
+# the musician/betoniarka populate them.
+for f in all_playlist.playlist priority_queue.playlist hit.playlist \
+         request.playlist jingles.playlist persistence.log; do
+    [ -e "$DATA/$f" ] || : > "$DATA/$f"
+done
+
+# The icecast secret is provisioned at install time, like the other services'
+# secrets (bot: /srv/conjurer/secrets/.netrc). A placeholder keeps the stack
+# bootable, but both icecast and the stream stay locked until you fix it.
+if [ ! -e "$CREDS" ]; then
+    printf '{\n"password" : "CHANGE_ME"\n}\n' > "$CREDS"
+    echo "WARNING: $CREDS was missing - seeded a CHANGE_ME placeholder." >&2
+    echo "         Put the real password there (see docs) and restart." >&2
+fi
+
+# Render /etc/icecast2/icecast.xml from the template with passwords from the
+# secret. Optional fields admin_password/relay_password default to password.
+SOURCE_PW=$(jq -r '.password' "$CREDS")
+ADMIN_PW=$(jq -r '.admin_password // .password' "$CREDS")
+RELAY_PW=$(jq -r '.relay_password // .password' "$CREDS")
+ICECAST_HOSTNAME="${ICECAST_HOSTNAME:-localhost}"
+sed -e "s|__SOURCE_PASSWORD__|$SOURCE_PW|" \
+    -e "s|__ADMIN_PASSWORD__|$ADMIN_PW|" \
+    -e "s|__RELAY_PASSWORD__|$RELAY_PW|" \
+    -e "s|__HOSTNAME__|$ICECAST_HOSTNAME|" \
+    /app/icecast.xml.tpl > /etc/icecast2/icecast.xml
+chown icecast2:icecast /etc/icecast2/icecast.xml 2>/dev/null || true
+chmod 640 /etc/icecast2/icecast.xml
+
+# Start Icecast in the background as its unprivileged user.
+mkdir -p /var/log/icecast2 && chown -R icecast2:icecast /var/log/icecast2
+su -s /bin/sh icecast2 -c "icecast2 -b -c /etc/icecast2/icecast.xml" \
+    || echo "WARNING: icecast2 failed to start - the stream output will retry" >&2
+
+# single() aborts the whole script when its file is missing; guarantee the
+# emergency fallback exists (5s of silence beats a dead radio).
+EMERGENCY="$MUSIC/Youtube/Dr. Peacock - Trip to Ireland [GvrvQTUbUcA].mp3"
+if [ ! -e "$EMERGENCY" ]; then
+    mkdir -p "$MUSIC/Youtube"
+    if ffmpeg -loglevel error -f lavfi -i anullsrc=r=44100:cl=stereo -t 5 \
+              -codec:a libmp3lame -q:a 9 "$EMERGENCY"; then
+        echo "WARNING: emergency track was missing - generated silent placeholder" >&2
+    else
+        echo "WARNING: could not create emergency track; single() may abort" >&2
+    fi
+fi
+
+# PulseAudio wiring:
+#   internal (default) - system-wide pulse inside the container with a null
+#                        sink (see /etc/pulse/system.pa); no sound hardware
+#                        needed, mic path reads silence.
+#   host               - use a socket mounted from the host; set PULSE_SERVER
+#                        (e.g. unix:/tmp/pulseaudio.socket) in the env file.
+#   none               - you edited the script to drop pulse in/out.
+case "${PULSE_MODE:-internal}" in
+    internal)
+        # --disallow-module-loading: modules from system.pa still load at
+        # startup; this only blocks later client-requested loads (and
+        # silences the system-mode warning). The "forcibly disabling SHM"
+        # notice is inherent to system mode and harmless.
+        pulseaudio --system --daemonize=yes --disallow-exit \
+            --disallow-module-loading --exit-idle-time=-1 \
+            || echo "WARNING: internal pulseaudio failed to start" >&2
+        export PULSE_SERVER="${PULSE_SERVER:-unix:/var/run/pulse/native}"
+        ;;
+    host)
+        [ -n "$PULSE_SERVER" ] || echo "WARNING: PULSE_MODE=host but PULSE_SERVER is unset" >&2
+        ;;
+    none)
+        ;;
+esac
+
+# Liquidsoap refuses to run as root (init: security exit), so hand the data
+# volume to the dedicated 'radio' user and drop privileges for the main
+# process. chown is best-effort: on local volumes it always works (the
+# supported layout); network filesystems with root-squash reject it, hence
+# the warning instead of a fatal abort.
+chown -R radio:radio "$DATA" 2>/dev/null \
+    || echo "WARNING: chown of $DATA failed (network FS?) - keep this volume LOCAL to the radio VM" >&2
+chgrp radio "$CREDS" 2>/dev/null && chmod 640 "$CREDS" || true
+if ! setpriv --reuid radio --regid radio --init-groups -- test -r "$MUSIC"; then
+    echo "WARNING: music dir $MUSIC is not readable by the 'radio' user" >&2
+fi
+
+# Betoniarka: the radio-operator API + radio-log forwarder, running as the
+# SAME user as liquidsoap on the SAME volume - this is what makes the old
+# musician-writes-as-root-over-network-share permission mess go away.
+BETONIARKA_DATA="$DATA" BETONIARKA_MUSIC="$MUSIC" \
+    setpriv --reuid radio --regid radio --init-groups -- \
+    python3 /app/betoniarka.py &
+echo "betoniarka started (pid $!)"
+
+cd "$DATA"
+exec setpriv --reuid radio --regid radio --init-groups -- "$@"

docker/env/bot.env.example

@@ -1,14 +1,22 @@
 # Copy to docker/env/bot.env and fill in. Do NOT commit the real file.
 
 # --- Secrets ------------------------------------------------------------
-# Option A: mount a netrc (recommended — covers discord/openai/spotipy/youtube).
+# Option A: mount a netrc (recommended — covers discord/openai/anthropic/spotipy/youtube).
 CONJURER_NETRC_FILE=/secrets/.netrc
 # Option B: pass tokens directly (these take precedence over netrc).
 # DISCORD_TOKEN=
 # OPENAI_API_KEY=
+# ANTHROPIC_API_KEY=       # Claude backend; netrc machine 'anthropic' works too
+# ASSEMBLYAI_API_KEY=      # voice recognition; netrc machine 'assemblyai' works too
 # YOUTUBE_USERNAME=
 # YOUTUBE_PASSWORD=
 
+# --- AI backend switch --------------------------------------------------
+# Which AI config from system_gpt_settings.json is active at startup
+# (e.g. "gpt" or "claude"). Runtime switch: $gadaj_teraz <config>. Unset =
+# whatever the settings file's "active" key says, falling back to "gpt".
+# CONJURER_AI_CONFIG=gpt
+
 # --- Data ---------------------------------------------------------------
 # Single mounted volume; all writable state is rooted here.
 CONJURER_DATA_DIR=/data
@@ -23,7 +31,10 @@ CONJURER_API_KEY=
 
 # --- Where the bot reaches the other services (other Proxmox VMs) --------
 CONJURER_FILE_SERVICE=http://MUSICIAN_VM_IP:5000
-CONJURER_RADIO_HARBOR=http://MUSICIAN_VM_IP:54321
+# Betoniarka (radio-operator API, runs in the radio container):
+CONJURER_RADIO_SERVICE=http://RADIO_VM_IP:5005
+# Liquidsoap harbor /skip (same radio container):
+CONJURER_RADIO_HARBOR=http://RADIO_VM_IP:54321
 CONJURER_LIBRARIAN_SERVICE=http://LIBRARIAN_VM_IP:5001
 
 # --- Conan Exiles bridge (optional; empty/0 = disabled) -----------------

docker/env/librarian.env.example

@@ -17,5 +17,9 @@ CONJURER_LIBRARIAN_DB_PATH=/doi/
 CONJURER_LIBRARIAN_MAXTHREADS=41
 CONJURER_LIBRARIAN_CHUNK=_chunk.txt
 
+# Runtime JSON state dir (mounted, persistent): cr_results/rr_results/
+# not_in_db/s_results are seeded here on first run.
+CONJURER_LIBRARIAN_STATE_DIR=/lib_temp_files
+
 # Optional netrc (for Crossref credentials)
 CONJURER_NETRC_FILE=/secrets/.netrc

docker/icecast.xml.tpl

@@ -0,0 +1,55 @@
+<!-- Icecast2 config template for the radio container.
+     The entrypoint substitutes the __*_PASSWORD__ placeholders from
+     /srv/betoniarka/secrets/icecast_credentials.json and writes the result
+     to /etc/icecast2/icecast.xml. Never commit real passwords here. -->
+<icecast>
+    <location>Wolne Ksiestwo Baluty</location>
+    <admin>admin@localhost</admin>
+
+    <limits>
+        <clients>64</clients>
+        <sources>4</sources>
+        <queue-size>524288</queue-size>
+        <client-timeout>30</client-timeout>
+        <header-timeout>15</header-timeout>
+        <source-timeout>10</source-timeout>
+        <burst-on-connect>1</burst-on-connect>
+        <burst-size>65535</burst-size>
+    </limits>
+
+    <authentication>
+        <source-password>__SOURCE_PASSWORD__</source-password>
+        <relay-password>__RELAY_PASSWORD__</relay-password>
+        <admin-user>admin</admin-user>
+        <admin-password>__ADMIN_PASSWORD__</admin-password>
+    </authentication>
+
+    <hostname>__HOSTNAME__</hostname>
+
+    <listen-socket>
+        <port>8000</port>
+        <bind-address>0.0.0.0</bind-address>
+    </listen-socket>
+
+    <http-headers>
+        <header name="Access-Control-Allow-Origin" value="*" />
+    </http-headers>
+
+    <fileserve>1</fileserve>
+
+    <paths>
+        <basedir>/usr/share/icecast2</basedir>
+        <logdir>/var/log/icecast2</logdir>
+        <webroot>/usr/share/icecast2/web</webroot>
+        <adminroot>/usr/share/icecast2/admin</adminroot>
+        <alias source="/" destination="/status.xsl"/>
+    </paths>
+
+    <logging>
+        <accesslog>access.log</accesslog>
+        <errorlog>error.log</errorlog>
+        <loglevel>3</loglevel>
+        <logsize>10000</logsize>
+        <logarchive>0</logarchive>
+    </logging>
+</icecast>

docker/pulse-system.pa

@@ -0,0 +1,10 @@
+#!/usr/bin/pulseaudio -nF
+# Minimal system-wide PulseAudio config for the radio container running on a
+# headless VM (PULSE_MODE=internal). Mirrors the Pi's setup (unix socket,
+# anonymous auth) but replaces the Lexicon Lambda USB device with a null
+# sink: output.pulseaudio() plays into the void and input.pulseaudio() (the
+# mic path) reads silence from the sink monitor.
+load-module module-null-sink sink_name=radio_null sink_properties=device.description=RadioNullOutput
+load-module module-native-protocol-unix auth-anonymous=1
+set-default-sink radio_null
+set-default-source radio_null.monitor

docs/deployment/DEPLOYMENT_GUIDE.md

@@ -43,11 +43,28 @@ You can rebalance as follows:
 2. Edit each `docker/env/*.env` to replace `HACKME!` with real values:
    - `DISCORD_TOKEN`, `OPENAI_API_KEY`, `CONJURER_API_KEY` (use the same value for
      all services).
+   - `ANTHROPIC_API_KEY` — only if you want the Claude backend (see “AI backend
+     switch” below). Safe to leave unset while running on GPT.
    - For musician/librarian, adjust mounts to the SMB paths you will mount on the
      Pis, e.g. `/mnt/conjurer/music`.
    - Set `CONJURER_CROSSREF_MAILTO` to a real email as required by Crossref.
 3. If you rely on `.netrc`, copy it to `C:\Conjurer\secrets\.netrc` and set
-   `CONJURER_NETRC_FILE` accordingly.
+   `CONJURER_NETRC_FILE` accordingly. The bot reads netrc machines `discord`,
+   `openai`, `anthropic`, `assemblyai`, `spotipy`, `youtube`.
+
+### AI backend switch (GPT ↔ Claude)
+
+The AI chat features run on one backend at a time, selected by a single switch:
+
+- **Startup:** `CONJURER_AI_CONFIG=gpt` (default) or `=claude` in `bot.env`. Unset
+  falls back to the `"active"` key in `system_gpt_settings.json`, then `"gpt"`.
+- **Runtime:** the `$gadaj_teraz <config>` Discord command (Vykidailo only) flips
+  the backend live and persists the choice.
+- The named configs (which provider, which models) live in
+  `system_gpt_settings.json` — add more there. Claude needs `ANTHROPIC_API_KEY`;
+  GPT needs `OPENAI_API_KEY`. Image generation (`imaginuje sobie:`) and personal
+  assistants always use OpenAI regardless of the switch (Anthropic has no
+  equivalent), and degrade quietly if `OPENAI_API_KEY` is absent.
 
 ## 4. Install Docker on Raspberry Pis and Windows
 

docs/deployment/DOCKER_PROXMOX.md

@@ -85,6 +85,33 @@ sudo chmod 600 /srv/conjurer/secrets/.netrc
 (Alternatively skip netrc and set `DISCORD_TOKEN` / `OPENAI_API_KEY` in the env
 file — those take precedence.)
 
+For the **Claude backend** add an `anthropic` machine to the same netrc (or set
+`ANTHROPIC_API_KEY` in `bot.env`):
+
+```
+machine anthropic
+    password sk-ant-...
+```
+
+You only need this if you actually switch the bot to Claude — see 1c-bis.
+
+### 1c-bis. AI backend switch (GPT ↔ Claude)
+
+The bot's AI chat runs on one backend at a time, chosen by a single switch:
+
+- **At startup**, set `CONJURER_AI_CONFIG` in `bot.env` — `gpt` (default) or
+  `claude`. Leave it unset to use the `"active"` key in
+  `system_gpt_settings.json` (falls back to `gpt`).
+- **At runtime**, `$gadaj_teraz <config>` (Vykidailo only) flips the backend live
+  and writes the choice back into `system_gpt_settings.json`.
+
+The provider configs (which backend, which models) are collected in
+`system_gpt_settings.json` under the third list element — add further AIs there.
+`claude` needs `ANTHROPIC_API_KEY`; `gpt` needs `OPENAI_API_KEY`. Image
+generation (`imaginuje sobie:`) and personal assistants stay on OpenAI whatever
+the switch says (Anthropic has no equivalent) and degrade quietly if OpenAI is
+not configured, so a Claude-only box still boots.
+
 ### 1d. Configure and launch
 
 ```bash
@@ -96,8 +123,47 @@ docker compose -f docker/compose.bot.yaml up -d --build
 docker logs -f conjurer-bot
 ```
 
-Look for `Loading … module done` for each cog, `Loading conanjurer commands
-module done`, and `All systems: operational`.
+Look for `Extension loaded: …` for each cog and `All systems: operational`.
+
+### 1e. Startup model: core cogs vs service-gated cogs
+
+The bot **always** starts with the cogs that depend on nothing but itself
+(administration, AI, other, latex, voice, conanjurer). Cogs that need a
+sibling service are **health-gated**:
+
+| Group | Cogs | Enabled when |
+|-------|------|--------------|
+| musician | `music_commands`, `radio_commands`, `file_search_commands` | `GET {FILE_SERVICE}/mp3` answers |
+| librarian | `librarian_commands` | librarian answers HTTP at all |
+
+When a service is down its cogs stay disabled (commands simply don't exist)
+and the log says so. A watchdog re-checks every 5 minutes and enables the
+cogs the moment the service starts answering — no bot restart needed.
+A single broken cog (missing pip package, bad import) is skipped with a full
+traceback in the log; it never takes the whole bot down.
+
+### 1f. Troubleshooting a crash-looping container
+
+`docker logs conjurer-bot` now shows the real reason (the bot logs to stdout
+as well as the rotating file). The most common cases:
+
+- **`FATAL: Discord token missing`** — the secrets mount is missing/empty or
+  `CONJURER_NETRC_FILE` points elsewhere. Check:
+  `docker inspect -f '{{json .Mounts}}' conjurer-bot | jq` and
+  `docker exec conjurer-bot ls -la /secrets/` (after a manual
+  `docker run … sleep infinity` if it crash-loops too fast).
+- **Missing state files** — not fatal anymore: missing dirs are created and
+  missing JSON state is seeded from the repo templates baked into the image
+  (existing files are never overwritten). Fix the mount at your leisure.
+
+**About files "disappearing" from `/srv/conjurer/...`:** nothing in this stack
+deletes host files — the entrypoint and the bot only ever *create* missing
+files. With a bind mount, `/srv/conjurer/data` **is** the live state (not an
+installation staging area): don't delete it after a successful install.
+If files vanished, the usual suspects are `docker compose down -v` (only
+affects *named* volumes, not binds), a re-provisioned VM, or copying the files
+to a different path than the one in the compose `volumes:` line — verify with
+`docker inspect -f '{{json .Mounts}}' conjurer-bot`.
 
 ---
 
@@ -184,6 +250,131 @@ those two files into the same `/srv/musician/data` directory (or set
 `/stream` page template is served from the baked-in `/app/stream.html`
 (override with `CONJURER_STREAM_TEMPLATE` if you customise it).
 
+### 3e. Radio (Liquidsoap) container
+
+`docker/Dockerfile.radio` builds the full Liquidsoap environment through
+**opam**, with the OCaml/opam/Liquidsoap versions selectable at build time:
+
+| Build arg | Default | Notes |
+|-----------|---------|-------|
+| `LIQUIDSOAP_VERSION` | `2.1.4` | what `radio_conjurer.liq` targets |
+| `OCAML_VERSION` | `4.14.2` | 2.1.x needs OCaml **4.x** (never 5); prod ran 4.13.0 — pass it for exact parity |
+| `OPAM_VERSION` | `2.1.5` | static binary from GitHub releases |
+| `LIQ_OPAM_PACKAGES` | `mad lame cry taglib pulseaudio samplerate inotify ffmpeg` | exactly the features the script uses (mp3 in/out, icecast, tags/replaygain, pulse mic/out, watch-reload) |
+
+The container runs **both Liquidsoap and Icecast** — the stream is served
+from this one container (`output.icecast(host="localhost", …)` in the
+script). Volume layout (same paths inside and outside):
+
+| Host & container path | Holds |
+|---|---|
+| `/srv/betoniarka/data` | playlists, `script.params`, `persistence.log`, `radio_log.log`, the seeded `radio_conjurer.liq` |
+| `/srv/betoniarka/music` | the mp3 library |
+| `/srv/betoniarka/secrets` | `icecast_credentials.json` — **provision at install**, like the other services' secrets |
+
+```bash
+sudo mkdir -p /srv/betoniarka/data /srv/betoniarka/music /srv/betoniarka/secrets
+# provision the icecast secret (same pattern as the bot's netrc):
+echo '{ "password" : "SOURCE_PW", "admin_password" : "ADMIN_PW" }' \
+  | sudo tee /srv/betoniarka/secrets/icecast_credentials.json
+sudo chmod 600 /srv/betoniarka/secrets/icecast_credentials.json
+
+docker compose -f docker/compose.radio.yaml up -d --build
+docker logs -f conjurer-radio
+```
+
+At startup the entrypoint renders `/etc/icecast2/icecast.xml` from
+`docker/icecast.xml.tpl`, filling the source/admin/relay passwords from the
+secret (`admin_password`/`relay_password` are optional and default to
+`password`), and starts icecast as its unprivileged user. If the secret is
+missing, a `CHANGE_ME` placeholder is seeded with a loud warning — the stack
+boots but the stream stays locked until you fix it.
+
+The script + `script.params` are seeded into the data volume on first run
+and never overwritten (live edits survive rebuilds). Missing playlists are
+created empty and a silent emergency-fallback mp3 is generated if the
+`single()` file is absent, so the script always boots.
+
+Wire-up: listeners tune to `http://RADIO_VM_IP:8000/mp3-stream`; the bot
+points at `CONJURER_RADIO_HARBOR=http://RADIO_VM_IP:54321` (harbor `/skip`
+lives in the script itself) and `CONJURER_RADIO_SERVICE=http://RADIO_VM_IP:5005`
+(the betoniarka API below).
+
+### 3f. Betoniarka - the radio operator (and why the split)
+
+**Permissions post-mortem.** The pre-split layout had three actors fighting
+over the same files: the musician wrote radio playlists **as root**, the
+radio expected them **as user `radio`**, and both met on a **root-owned
+network share** where `chown` fails by design (root squash / uid mapping).
+Every component worked; the combination could not.
+
+**The fix is structural**: the process that *writes* the radio playlists now
+lives in the same container as the process that *watches* them, running as
+the same `radio` user on a **local** volume. No network share, no chown, no
+uid mapping - the class of problem is gone, not patched.
+
+`conjurer_betoniarka/betoniarka.py` runs inside the radio container
+(started by the entrypoint as user `radio`, port **5005**) and owns:
+- the library scan → `all_playlist.playlist` / `hit.playlist` (local paths,
+  the same ones Liquidsoap resolves), on start + every 24h + `GET /rescan`
+- the bot-facing radio API: `/add_to_priority`, `/create_priority_playlist`,
+  `/request_radio_file`, `/clear_pr_pls` (+ `GET /ping` for health gating,
+  `/stream` for the web page)
+- tailing `radio_log.log`/`persistence.log` and forwarding play events to
+  the bot's `/prepped_tracks` (with the shared API key)
+
+The **musician** is now a pure Discord music player: `/mp3`, `/update_mp3`,
+`/get_music` and the file-share endpoints. It no longer writes any radio
+files and needs no shared partition with the radio VM. The music library
+can still be replicated/mounted on both VMs (read-only on the radio side is
+fine) - playlists reference the *radio VM's local* paths, generated locally.
+
+Bot wiring after the split (env on the bot):
+```
+CONJURER_FILE_SERVICE=http://MUSICIAN_VM_IP:5000   # Discord music
+CONJURER_RADIO_SERVICE=http://RADIO_VM_IP:5005     # betoniarka (radio cmds)
+CONJURER_RADIO_HARBOR=http://RADIO_VM_IP:54321     # liquidsoap /skip
+```
+`CONJURER_RADIO_SERVICE` defaults to `CONJURER_FILE_SERVICE`, so an
+un-split deployment keeps working unchanged. The bot health-gates
+`radio_commands` on betoniarka's `/ping` (separate from the musician group,
+which now covers only `music_commands` + `file_search_commands`).
+
+**Privileges:** Liquidsoap refuses to run as root (`init: security exit`),
+so the main process runs as the dedicated **`radio`** user (member of
+`audio`/`pulse-access`) — no `settings.init.allow_root` override. The
+entrypoint (root) seeds volumes, renders the icecast config, starts
+icecast/pulse, chowns `/srv/betoniarka/data` to `radio` and drops
+privileges via `setpriv`. This is also why the opam switch lives in
+`/opt/opam` instead of `/root/.opam` (the binary and the liquidsoap stdlib
+must be readable by `radio`).
+
+**PulseAudio** (`PULSE_MODE` env):
+- `internal` (default) — a system-wide pulse daemon runs inside the container
+  with a **null sink** (`docker/pulse-system.pa`): no sound hardware needed,
+  `output.pulseaudio()` plays into the void and the `input.pulseaudio()` mic
+  path reads silence (which `blank.strip` already gates out). Right choice
+  for a headless Proxmox VM. Started with `--disallow-module-loading`
+  (startup modules from `system.pa` still load; only later client-requested
+  loads are blocked). The `forcibly disabling SHM mode` notice is inherent
+  to system mode and harmless.
+- `host` — mount the host's pulse socket and set `PULSE_SERVER`, for a VM
+  with real audio hardware (the Pi's Lexicon Lambda setup).
+- `none` — you removed the pulse in/out from the script.
+
+**Verdicts on the old Pi setup quirks** (asked during containerisation):
+- `ffmpeg.pref` (Pin-Priority 1001 on the `libavcodec59/libavformat59/…`
+  family): it **did have a purpose** — the opam-built ocaml-ffmpeg bindings
+  for 2.1.x are compiled against Debian bookworm's FFmpeg 5.x sonames, and
+  the pin forced those over conflicting Raspberry Pi OS repo builds (even as
+  a downgrade). In this single-repo container the same versions come
+  naturally, so the pin is redundant — **the file has been removed from the
+  repo** (this note preserves the knowledge).
+- adding root to `pulse-access`/`audio` groups: needed on the Pi for the
+  system-wide pulse socket and ALSA device access. The image bakes the same
+  memberships in (`usermod -aG audio,pulse-access root`) — harmless with the
+  internal null sink, required for `PULSE_MODE=host`.
+
 ---
 
 ## 4. Networking & auth

ffmpeg.pref

@@ -1,3 +0,0 @@
-Package: ffmpeg libavcodec-dev libavcodec59 libavdevice59 libavfilter8 libavformat-dev libavformat59 libavutil-dev libavutil57 libpostproc56 libswresample-dev libswresample4 libswscale-dev libswscale6 libavdevice-dev libavfilter-dev libpostproc-dev
-Pin: origin deb.debian.org
-Pin-Priority: 1001

install_music_bot.sh

@@ -27,7 +27,7 @@ sudo usermod -aG audio root
 
 echo "Add exception suppresion to signal handler in spotify __ini__.py"
 echo "Add password to youtube opts in spotify_dl youtube.py"
-echo "Downgrade ffmpeg by copying ffmpeg.pref from repository to /etc/apt/preferences.d"
+echo "NOTE: the old ffmpeg.pref pin is obsolete (containerised radio uses Debian bookworm ffmpeg 5.x natively; file removed from the repo)"
 echo "Install opam from installation link"
 echo "Initialize opam"
 echo "Install liquidsoap and its dependencies"

radio_commands.py

@@ -8,7 +8,7 @@ import uuid
 
 import asyncio
 from datetime import datetime
-from constants import RADIO_HARBOR_ADDRESS, SKIP_TRACK, FILE_SERVICE_ADDRESS, ADD_TO_PRIO_PLAYLIST, REQUEST_MUSIC, CREATE_PRIO_PLAYLIST, CLEAR_PRIO, service_headers
+from constants import RADIO_HARBOR_ADDRESS, SKIP_TRACK, RADIO_SERVICE_ADDRESS, ADD_TO_PRIO_PLAYLIST, REQUEST_MUSIC, CREATE_PRIO_PLAYLIST, CLEAR_PRIO, service_headers
 
 SERVICE_HEADERS = service_headers()
 
@@ -96,7 +96,7 @@ class RadioModule(commands.Cog):
             }
             coroutine = asyncio.to_thread(
                 requests.post,
-                f"{FILE_SERVICE_ADDRESS}{ADD_TO_PRIO_PLAYLIST}",
+                f"{RADIO_SERVICE_ADDRESS}{ADD_TO_PRIO_PLAYLIST}",
                 json=jrequest,
                 headers=SERVICE_HEADERS,
                 timeout=360,
@@ -132,7 +132,7 @@ class RadioModule(commands.Cog):
             }
             coroutine = asyncio.to_thread(
                 requests.post,
-                f"{FILE_SERVICE_ADDRESS}{REQUEST_MUSIC}",
+                f"{RADIO_SERVICE_ADDRESS}{REQUEST_MUSIC}",
                 json=jrequest,
                 headers=SERVICE_HEADERS,
                 timeout=360,
@@ -170,7 +170,7 @@ class RadioModule(commands.Cog):
             }
             coroutine = asyncio.to_thread(
                 requests.post,
-                f"{FILE_SERVICE_ADDRESS}{CREATE_PRIO_PLAYLIST}",
+                f"{RADIO_SERVICE_ADDRESS}{CREATE_PRIO_PLAYLIST}",
                 json=jrequest,
                 headers=SERVICE_HEADERS,
                 timeout=360,
@@ -199,7 +199,7 @@ class RadioModule(commands.Cog):
         async with ctx.typing():
             coroutine = asyncio.to_thread(
                 requests.get,
-                f"{FILE_SERVICE_ADDRESS}{CLEAR_PRIO}",
+                f"{RADIO_SERVICE_ADDRESS}{CLEAR_PRIO}",
                 headers=SERVICE_HEADERS,
                 timeout=360,
             )

requirements_bot.txt

@@ -4,6 +4,7 @@ yt_dlp
 spotify_dl
 spotipy
 openai
+anthropic
 eyed3
 numpy
 pdf2image

requirements_conan.txt

@@ -0,0 +1,6 @@
+# Optional extras for the Conan Exiles bridge (conanjurer cog).
+#
+# aiomcrcon currently installs only on Python <= 3.11, while the main image
+# runs 3.13. Install is therefore best-effort: when it fails the conanjurer
+# cog simply stays dormant (its imports are guarded) and the bot runs fine.
+aiomcrcon

system_gpt_settings.json

@@ -39,5 +39,29 @@
             "Zrobisz absolutnie wszystko jako asystent, poniewa\u017c sama my\u015bl o tym \u017ce m\u00f3g\u0142by\u015b rozgniewa\u0107 operatora rozgniewa\u0107 Ci\u0119 przera\u017ca do poziomu histerii.",
             "asst_1PnD0eQMWo0RkZ3xUacC1uqJ"
         ]
+    },
+    {
+        "active": "gpt",
+        "configs": {
+            "gpt": {
+                "provider": "openai",
+                "latest_model": "gpt-4o",
+                "cheap_model": "gpt-4o-mini",
+                "temperature": 0.2
+            },
+            "claude": {
+                "provider": "anthropic",
+                "latest_model": "claude-opus-4-8",
+                "cheap_model": "claude-haiku-4-5",
+                "max_tokens": 2048
+            },
+            "_template": {
+                "provider": "openai",
+                "latest_model": "model-id-here",
+                "cheap_model": "cheaper-model-id-here",
+                "temperature": 0.2,
+                "max_tokens": 2048
+            }
+        }
     }
 ]

tests/unit/test_ai_provider_switch.py

@@ -0,0 +1,138 @@
+"""Unit tests for the GPT/Claude provider switch.
+
+The unit CI job installs only pytest, so the heavy runtime deps that
+``ai_functions`` imports unguarded (``openai``, ``tiktoken``, ``other_functions``
+-> ``discord``) are stubbed *only when genuinely absent*. Locally, where the
+real packages exist, the stubs are skipped and the real modules are used.
+"""
+import sys
+import types
+
+
+def _stub_if_missing(name: str, build):
+    if name in sys.modules:
+        return
+    try:  # real package present (local dev / bot image) -> use it
+        __import__(name)
+    except ImportError:
+        sys.modules[name] = build()
+
+
+def _build_tiktoken():
+    mod = types.ModuleType("tiktoken")
+
+    class _Enc:
+        def encode(self, text):
+            return list(text)
+
+    mod.encoding_for_model = lambda _model: _Enc()
+    return mod
+
+
+def _build_other_functions():
+    mod = types.ModuleType("other_functions")
+
+    async def _noop(*_a, **_k):
+        return None
+
+    mod.discord_friendly_send = _noop
+    mod.discord_friendly_reply = _noop
+    return mod
+
+
+def _build_openai():
+    mod = types.ModuleType("openai")
+    for cls_name in (
+        "APITimeoutError",
+        "APIConnectionError",
+        "BadRequestError",
+        "APIResponseValidationError",
+        "AuthenticationError",
+        "PermissionDeniedError",
+        "RateLimitError",
+        "UnprocessableEntityError",
+        "APIError",
+        "OpenAIError",
+    ):
+        setattr(mod, cls_name, type(cls_name, (Exception,), {}))
+    return mod
+
+
+_stub_if_missing("tiktoken", _build_tiktoken)
+_stub_if_missing("other_functions", _build_other_functions)
+_stub_if_missing("openai", _build_openai)
+
+import ai_functions  # noqa: E402  (import after stubbing)
+import openai  # noqa: E402  (real or stub, same object ai_functions uses)
+
+
+def _reset_active(name="gpt"):
+    ai_functions._ACTIVE_CONFIG_NAME = name
+
+
+def test_split_extracts_system_and_starts_with_user():
+    system, convo = ai_functions._to_anthropic_messages(
+        [
+            {"role": "system", "content": "SYS1"},
+            {"role": "system", "content": "SYS2"},
+            {"role": "assistant", "content": "leading-assistant-dropped"},
+            {"role": "user", "content": "u1"},
+            {"role": "assistant", "content": "a1"},
+            {"role": "user", "content": "hi"},
+        ]
+    )
+    assert system == "SYS1\n\nSYS2"
+    assert convo[0] == {"role": "user", "content": "u1"}
+    assert convo == [
+        {"role": "user", "content": "u1"},
+        {"role": "assistant", "content": "a1"},
+        {"role": "user", "content": "hi"},
+    ]
+
+
+def test_split_synthesises_user_turn_when_only_system():
+    _system, convo = ai_functions._to_anthropic_messages(
+        [{"role": "system", "content": "only"}]
+    )
+    assert convo == [{"role": "user", "content": " "}]
+
+
+def test_select_model_gpt_active():
+    _reset_active("gpt")
+    # legacy default auto-selects; MUSIC is cheap; explicit ids are honoured
+    assert ai_functions.select_model("GENERAL", "gpt-4o") == "gpt-4o"
+    assert ai_functions.select_model("MUSIC", "gpt-4o") == "gpt-4o-mini"
+    assert ai_functions.select_model("MUSIC", "auto") == "gpt-4o-mini"
+    assert ai_functions.select_model("GENERAL", "o1-preview") == "o1-preview"
+
+
+def test_select_model_claude_active_maps_legacy_default():
+    _reset_active("claude")
+    try:
+        # the old gpt-4o default must not leak to Claude - it auto-maps
+        assert ai_functions.select_model("GENERAL", "gpt-4o") == "claude-opus-4-8"
+        assert ai_functions.select_model("MUSIC", "gpt-4o") == "claude-haiku-4-5"
+        # a real, deliberate model id is still honoured verbatim
+        assert ai_functions.select_model("GENERAL", "claude-sonnet-5") == "claude-sonnet-5"
+    finally:
+        _reset_active("gpt")
+
+
+def test_list_ai_configs_hides_templates():
+    names = ai_functions.list_ai_configs()
+    assert "_template" not in names
+    assert {"gpt", "claude"}.issubset(set(names))
+
+
+def _bare(cls):
+    # Build an instance without invoking __init__ - the real openai SDK
+    # exceptions require response/body kwargs, the CI stubs don't. isinstance
+    # (all _map_openai_error cares about) works on __new__-created instances.
+    return cls.__new__(cls)
+
+
+def test_map_openai_error_categories():
+    assert ai_functions._map_openai_error(_bare(openai.AuthenticationError)).category == "auth"
+    assert ai_functions._map_openai_error(_bare(openai.RateLimitError)).category == "rate_limit"
+    assert ai_functions._map_openai_error(_bare(openai.APITimeoutError)).category == "timeout"
+    assert ai_functions._map_openai_error(ValueError("x")).category == "api"

tests/unit/test_constants.py

@@ -33,3 +33,18 @@ def test_conan_defaults_present():
     # Feature toggles default to "off" so the bridge stays dormant.
     assert constants.CONAN_JOIN_CHANNEL_ID == 0
     assert constants.CONAN_RCON_HOST == ""
+
+
+def test_ai_configs_expose_gpt_and_claude_backends():
+    cfgs = constants.AI_CONFIGS
+    assert cfgs["gpt"]["provider"] == "openai"
+    assert cfgs["claude"]["provider"] == "anthropic"
+    # Claude carries a max_tokens (Anthropic requires it) and omits temperature
+    # (Opus 4.8 / Sonnet 5 reject sampling params).
+    assert "max_tokens" in cfgs["claude"]
+    assert "temperature" not in cfgs["claude"]
+
+
+def test_default_ai_config_is_a_known_config():
+    # The single startup switch must always resolve to a real, selectable config.
+    assert constants.DEFAULT_AI_CONFIG in constants.AI_CONFIGS

voice_recognition_commands.py

@@ -1,6 +1,5 @@
 import asyncio
 import logging
-import netrc
 import time
 import wave
 
@@ -9,11 +8,17 @@ import discord
 from discord.ext import commands, tasks, voice_recv
 from discord.opus import Decoder as OpusDecoder
 
-# Replace with your API key
-NETRC_FILE = "/home/pi/.netrc"
-netrc_mod = netrc.netrc(NETRC_FILE)
-authTokens = netrc_mod.authenticators("assemblyai")
-aai.settings.api_key = authTokens[2]
+from constants import ASSEMBLYAI_API_KEY, TRANSCRIPTS_PATH
+
+# Credentials come from constants (env ASSEMBLYAI_API_KEY, or the 'assemblyai'
+# machine in the netrc at CONJURER_NETRC_FILE). Raising here means the guarded
+# extension loader logs the reason and disables ONLY this cog.
+if not ASSEMBLYAI_API_KEY:
+    raise RuntimeError(
+        "AssemblyAI API key not configured (netrc machine 'assemblyai' or "
+        "ASSEMBLYAI_API_KEY env) - voice recognition stays disabled"
+    )
+aai.settings.api_key = ASSEMBLYAI_API_KEY
 
 discord.opus._load_default()
 CHANNELS = OpusDecoder.CHANNELS
@@ -23,7 +28,7 @@ SAMPLING_RATE = OpusDecoder.SAMPLING_RATE
 # rotate file after there is 0.5s between last received pcm for user.
 # delete messsages after user disconnect
 
-LOCATION = "/home/pi/Conjurer/transcripts/"
+LOCATION = TRANSCRIPTS_PATH
 
 
 class CommunicationObject: