feat(server): split runtime knobs out of .env into settings.toml

.env shrinks to bootstrap values only (port, paths, log, auth, deploy flags). The 14 admin-tunable values (retention, whisper, ollama, llm) move to a new settings.toml with serde(default)-based per-section defaults and an empty-prompt fallback to the code default. settings.toml is gitignored alongside .env; settings.toml.example is the template. This commit fixes the file layout. The server-side wiring (Config shrink, AppState integration, consumer migration) follows separately.
2026-04-27 11:21:23 +02:00
parent 0205301018
commit 6bbbf10f77
4 changed files with 298 additions and 33 deletions
@@ -1,6 +1,7 @@
 target/
 .env
 users.toml
+settings.toml
 *.snap.new
 repomix-output.xml
 tmpdata/
@@ -1,42 +1,22 @@
-# === Required (Phase 1) ===
+# Doctate operator config — bootstrap, logging, auth, deploy topology.
+# Admin-tunable values (retention, providers, prompts) live in settings.toml
+# (see settings.toml.example for the template).

+# Server bootstrap
 SERVER_PORT=3000
 DATA_PATH=/data
 USERS_FILE=users.toml
+
+# Logging
 LOG_LEVEL=info
 LOG_PATH=/var/log/recorder
 LOG_MAX_DAYS=90

-# === Optional (Phase 2+) ===
-
-# Retention
-RETENTION_AUDIO_DAYS=30
-RETENTION_TRANSCRIPT_DAYS=30
-RETENTION_DOCUMENT_DAYS=0
-
-# faster-whisper
-WHISPER_URL=http://localhost:10300
-WHISPER_TIMEOUT_SECONDS=120
-
-# Ollama
-OLLAMA_URL=http://localhost:11434
-OLLAMA_MODEL=gemma3:4b
-OLLAMA_KEEP_ALIVE=0
-
-# LLM provider (Ionos)
-LLM_URL=https://openai.inference.de-txl.ionos.com
-LLM_API_KEY=
-LLM_MODEL=
-LLM_TEMPERATURE=0
-LLM_TIMEOUT_SECONDS=180
-# Optional override of the consolidation system prompt. If unset or empty, a default is used.
-# Single line — escape newlines with \n if your shell preserves them, or put the prompt on one logical line.
-LLM_SYSTEM_PROMPT="Du bist ein medizinischer Assistent. Fasse die folgenden diktierten Abschnitte zu einem zusammenhängenden Text zusammen. Bereinigen und strukturieren, nichts hinzufügen, nichts interpretieren, keine Diagnose, keine Arztbrief-Struktur. Entferne Redundanzen. Spätere Aufnahmen haben Vorrang — Korrekturen, Nachträge und Widersprüche zugunsten der chronologisch letzten Aussage auflösen. Antworte auf Deutsch. Nur der zusammengefasste Text, keine Einleitung, kein Abschlusssatz."
-
-# Gazetteer (pre-LLM proper-name correction)
-# Directory with *.txt files, one entry per line. Blank and #-comment lines
-# are skipped. Missing directory is non-fatal — analysis runs without hints.
-VOCAB_DIR=./vocab
-
-# Session
+# Auth
 SESSION_TIMEOUT_HOURS=8
+# Set to false for plain-HTTP local dev; browsers refuse Secure cookies on http://
+COOKIE_SECURE=true
+
+# Optional filesystem resources (uncomment if used)
+# VOCAB_DIR=./vocab
+# HUNSPELL_DICT=/usr/share/hunspell/de_DE.dic
@@ -0,0 +1,46 @@
+# Doctate runtime settings — admin-editable via WebUI.
+# Bootstrap values (port, data paths, log paths, auth) live in .env.
+# Copy to settings.toml and adjust per deployment.
+
+[retention]
+audio_days = 30
+transcript_days = 30
+document_days = 0
+
+[whisper]
+url = "http://localhost:9000"
+timeout_seconds = 120
+
+[ollama]
+url = "http://localhost:11434"
+model = "gemma3:4b"
+keep_alive = 0
+
+# LLM provider: swap url/api_key/model between blocks to switch providers.
+# Only one [llm] block may be active at a time — comment the others.
+
+# Example: Ionos hosted
+[llm]
+url = "https://openai.inference.de-txl.ionos.com"
+api_key = ""
+model = ""
+temperature = 0
+timeout_seconds = 180
+# Single line or TOML triple-string for multi-line prompts.
+system_prompt = """
+Du bist ein medizinischer Assistent. Fasse die folgenden diktierten Abschnitte
+zu einem zusammenhängenden Text zusammen. Bereinigen und strukturieren, nichts
+hinzufügen, nichts interpretieren, keine Diagnose, keine Arztbrief-Struktur.
+Entferne Redundanzen. Spätere Aufnahmen haben Vorrang — Korrekturen, Nachträge
+und Widersprüche zugunsten der chronologisch letzten Aussage auflösen.
+Antworte auf Deutsch. Nur der zusammengefasste Text, keine Einleitung, kein
+Abschlusssatz.
+"""
+
+# Example: Ollama local
+# [llm]
+# url = "http://localhost:11434"
+# api_key = ""
+# model = "SimonPu/gpt-oss:20b_Q4_K_M"
+# temperature = 0
+# timeout_seconds = 300
@@ -0,0 +1,238 @@
+//! Runtime-tunable settings loaded from `settings.toml`.
+//!
+//! Bootstrap config (port, data paths, log paths, auth) lives in [`crate::config::Config`].
+//! Edits to `settings.toml` take effect on the next server start — hot
+//! reload is a follow-up plan.
+
+use serde::Deserialize;
+
+#[derive(Debug, Clone, Default, Deserialize)]
+pub struct Settings {
+    #[serde(default)]
+    pub retention: RetentionSettings,
+    #[serde(default)]
+    pub whisper: WhisperSettings,
+    #[serde(default)]
+    pub ollama: OllamaSettings,
+    #[serde(default)]
+    pub llm: LlmSettings,
+}
+
+#[derive(Debug, Clone, Deserialize)]
+pub struct RetentionSettings {
+    #[serde(default = "default_retention_audio_days")]
+    pub audio_days: u32,
+    #[serde(default = "default_retention_transcript_days")]
+    pub transcript_days: u32,
+    #[serde(default = "default_retention_document_days")]
+    pub document_days: u32,
+}
+
+impl Default for RetentionSettings {
+    fn default() -> Self {
+        Self {
+            audio_days: default_retention_audio_days(),
+            transcript_days: default_retention_transcript_days(),
+            document_days: default_retention_document_days(),
+        }
+    }
+}
+
+fn default_retention_audio_days() -> u32 {
+    30
+}
+fn default_retention_transcript_days() -> u32 {
+    30
+}
+fn default_retention_document_days() -> u32 {
+    0
+}
+
+#[derive(Debug, Clone, Deserialize)]
+pub struct WhisperSettings {
+    #[serde(default = "default_whisper_url")]
+    pub url: String,
+    #[serde(default = "default_whisper_timeout_seconds")]
+    pub timeout_seconds: u64,
+}
+
+impl Default for WhisperSettings {
+    fn default() -> Self {
+        Self {
+            url: default_whisper_url(),
+            timeout_seconds: default_whisper_timeout_seconds(),
+        }
+    }
+}
+
+fn default_whisper_url() -> String {
+    "http://localhost:10300".into()
+}
+fn default_whisper_timeout_seconds() -> u64 {
+    120
+}
+
+#[derive(Debug, Clone, Deserialize)]
+pub struct OllamaSettings {
+    #[serde(default = "default_ollama_url")]
+    pub url: String,
+    #[serde(default = "default_ollama_model")]
+    pub model: String,
+    #[serde(default = "default_ollama_keep_alive")]
+    pub keep_alive: u32,
+}
+
+impl Default for OllamaSettings {
+    fn default() -> Self {
+        Self {
+            url: default_ollama_url(),
+            model: default_ollama_model(),
+            keep_alive: default_ollama_keep_alive(),
+        }
+    }
+}
+
+fn default_ollama_url() -> String {
+    "http://localhost:11434".into()
+}
+fn default_ollama_model() -> String {
+    "gemma3:4b".into()
+}
+fn default_ollama_keep_alive() -> u32 {
+    0
+}
+
+#[derive(Debug, Clone, Deserialize)]
+pub struct LlmSettings {
+    #[serde(default)]
+    pub url: String,
+    #[serde(default)]
+    pub api_key: String,
+    #[serde(default)]
+    pub model: String,
+    #[serde(default)]
+    pub temperature: f32,
+    #[serde(default = "default_llm_timeout_seconds")]
+    pub timeout_seconds: u64,
+    /// Empty values fall back to the code default in [`Settings::load_or_default`]
+    /// — symmetric to the previous env-var behaviour.
+    #[serde(default = "default_llm_system_prompt")]
+    pub system_prompt: String,
+}
+
+impl Default for LlmSettings {
+    fn default() -> Self {
+        Self {
+            url: String::new(),
+            api_key: String::new(),
+            model: String::new(),
+            temperature: 0.0,
+            timeout_seconds: default_llm_timeout_seconds(),
+            system_prompt: default_llm_system_prompt(),
+        }
+    }
+}
+
+fn default_llm_timeout_seconds() -> u64 {
+    180
+}
+fn default_llm_system_prompt() -> String {
+    crate::analyze::prompt::SYSTEM_PROMPT.to_string()
+}
+
+impl Settings {
+    /// Missing file → defaults + warn. Parse error → panic (startup-fail-fast).
+    /// An empty `llm.system_prompt` field falls back to the code default.
+    pub fn load_or_default(path: &str) -> Self {
+        match std::fs::read_to_string(path) {
+            Ok(content) => {
+                let mut parsed: Settings = toml::from_str(&content)
+                    .unwrap_or_else(|e| panic!("Failed to parse {path}: {e}"));
+                if parsed.llm.system_prompt.is_empty() {
+                    parsed.llm.system_prompt = default_llm_system_prompt();
+                }
+                tracing::info!(path, "settings loaded");
+                parsed
+            }
+            Err(_) => {
+                tracing::warn!(path, "settings file not found — using code defaults");
+                Self::default()
+            }
+        }
+    }
+
+    /// `api_key` is intentionally not required — Ollama's OpenAI-compatible
+    /// endpoint accepts unauthenticated requests.
+    pub fn llm_configured(&self) -> bool {
+        !self.llm.url.is_empty() && !self.llm.model.is_empty()
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[test]
+    fn parses_full_toml() {
+        let s: Settings = toml::from_str(
+            r#"
+[whisper]
+url = "http://h:1"
+timeout_seconds = 60
+
+[ollama]
+url = "http://h:2"
+model = "x"
+keep_alive = 5
+
+[llm]
+url = "http://h:3"
+api_key = "k"
+model = "m"
+temperature = 0.5
+timeout_seconds = 30
+system_prompt = "p"
+"#,
+        )
+        .unwrap();
+        assert_eq!(s.whisper.url, "http://h:1");
+        assert_eq!(s.whisper.timeout_seconds, 60);
+        assert_eq!(s.ollama.keep_alive, 5);
+        assert_eq!(s.llm.api_key, "k");
+        assert_eq!(s.llm.system_prompt, "p");
+    }
+
+    #[test]
+    fn empty_toml_yields_defaults() {
+        let s: Settings = toml::from_str("").unwrap();
+        assert_eq!(s.whisper.url, "http://localhost:10300");
+        assert_eq!(s.whisper.timeout_seconds, 120);
+        assert_eq!(s.ollama.url, "http://localhost:11434");
+        assert_eq!(s.ollama.model, "gemma3:4b");
+        assert_eq!(s.llm.timeout_seconds, 180);
+    }
+
+    #[test]
+    fn partial_toml_fills_in_defaults() {
+        let s: Settings = toml::from_str(
+            r#"
+[whisper]
+url = "http://only-whisper"
+"#,
+        )
+        .unwrap();
+        assert_eq!(s.whisper.url, "http://only-whisper");
+        assert_eq!(s.whisper.timeout_seconds, 120);
+        assert_eq!(s.ollama.url, "http://localhost:11434");
+    }
+
+    #[test]
+    fn llm_configured_requires_url_and_model() {
+        let mut s = Settings::default();
+        assert!(!s.llm_configured());
+        s.llm.url = "http://x".into();
+        assert!(!s.llm_configured());
+        s.llm.model = "m".into();
+        assert!(s.llm_configured());
+    }
+}