feat(cli): restructure langcodec config

Author: WendellXY

README.md

@@ -127,16 +127,20 @@ langcodec sync --source source.xcstrings --target target.xcstrings --match-lang
 Create a `langcodec.toml` in your project:
 
 ```toml
-[ai]
-provider = "openai"
-model = "gpt-4.1-mini"
+[openai]
+model = "gpt-5.4"
 
 [translate]
+concurrency = 4
+
+[translate.input]
 source = "locales/Localizable.xcstrings"
-source_lang = "en"
-target_lang = "fr,de"
+lang = "en"
 status = ["new", "stale"]
-concurrency = 4
+
+[translate.output]
+lang = ["fr", "de"]
+status = "translated"
 
 [annotate]
 input = "locales/Localizable.xcstrings"
@@ -151,7 +155,7 @@ langcodec translate
 langcodec annotate
 ```
 
-`translate` still accepts legacy `translate.provider` and `translate.model` if you have older config files. For larger projects, `translate.sources = [...]` can fan out parallel runs from config.
+When exactly one provider section is configured, `translate` and `annotate` use it automatically. If you configure multiple providers, choose one with `--provider` or `translate.provider`. For larger projects, `translate.input.sources = [...]` can fan out parallel runs from config.
 
 `annotate` also supports `annotate.inputs = [...]` for config-driven in-place runs across multiple xcstrings files.
 

langcodec-cli/README.md

@@ -99,6 +99,7 @@ langcodec translate \
 - config defaults from `langcodec.toml`
 - multiple target languages for multi-language outputs
 - live progress updates
+- `--ui auto|plain|tui` for dashboard or plain terminal output
 - preflight validation before model requests
 - translation result summaries at the end
 
@@ -120,21 +121,26 @@ langcodec annotate \
 - preserving manual comments
 - config defaults from `langcodec.toml`
 - source shortlisting before agent lookup
+- `--ui auto|plain|tui` for dashboard or plain terminal output
 - `--dry-run` and `--check` for CI-friendly runs
 
 ## Example Config
 
 ```toml
-[ai]
-provider = "openai"
-model = "gpt-4.1-mini"
+[openai]
+model = "gpt-5.4"
 
 [translate]
+concurrency = 4
+
+[translate.input]
 source = "locales/Localizable.xcstrings"
-source_lang = "en"
-target_lang = "fr,de"
+lang = "en"
 status = ["new", "stale"]
-concurrency = 4
+
+[translate.output]
+lang = ["fr", "de"]
+status = "translated"
 
 [annotate]
 input = "locales/Localizable.xcstrings"
@@ -149,7 +155,7 @@ langcodec translate
 langcodec annotate
 ```
 
-Legacy configs using `translate.provider` and `translate.model` still work. For larger repos, `translate.sources = [...]` can fan out parallel runs from config.
+When exactly one provider section is configured, `translate` and `annotate` use it automatically. If you configure multiple providers, choose one with `--provider` or `translate.provider`. For larger repos, `translate.input.sources = [...]` can fan out parallel runs from config.
 
 For annotate fan-out runs, use `annotate.inputs = [...]` and omit `annotate.output` so each catalog is updated in place.
 

langcodec-cli/src/ai.rs

@@ -1,5 +1,6 @@
 use std::sync::Arc;
 
+use crate::config::CliConfig;
 use mentra::{
     BuiltinProvider,
     provider::{self, Provider},
@@ -58,17 +59,27 @@ pub(crate) struct ProviderSetup {
 
 pub(crate) fn resolve_provider(
     cli: Option<&str>,
-    shared_cfg: Option<&str>,
-    legacy_cfg: Option<&str>,
+    config: Option<&CliConfig>,
+    translate_cfg: Option<&str>,
 ) -> Result<ProviderKind, String> {
     if let Some(value) = cli {
         return ProviderKind::parse(value);
     }
-    if let Some(value) = shared_cfg {
+    if let Some(value) = translate_cfg {
         return ProviderKind::parse(value);
     }
-    if let Some(value) = legacy_cfg {
-        return ProviderKind::parse(value);
+    if let Some(config) = config {
+        let configured = config.configured_provider_names();
+        match configured.len() {
+            1 => return ProviderKind::parse(configured[0]),
+            0 => {}
+            _ => {
+                return Err(
+                    "Multiple provider sections are configured; specify --provider or set translate.provider in langcodec.toml"
+                        .to_string(),
+                );
+            }
+        }
     }
 
     let mut available = Vec::new();
@@ -85,28 +96,36 @@ pub(crate) fn resolve_provider(
     match available.len() {
         1 => Ok(available.remove(0)),
         0 => Err(
-            "--provider is required (or set ai.provider in langcodec.toml, or use legacy translate.provider, or configure exactly one provider API key)"
+            "--provider is required (or configure exactly one provider section like [openai] in langcodec.toml, set translate.provider, or configure exactly one provider API key)"
                 .to_string(),
         ),
         _ => Err(
-            "Multiple provider API keys are configured; specify --provider or set ai.provider in langcodec.toml"
+            "Multiple provider API keys are configured; specify --provider or configure a single provider section in langcodec.toml"
                 .to_string(),
         ),
     }
 }
 
 pub(crate) fn resolve_model(
     cli: Option<&str>,
-    shared_cfg: Option<&str>,
-    legacy_cfg: Option<&str>,
+    config: Option<&CliConfig>,
+    provider: &ProviderKind,
+    translate_cfg: Option<&str>,
 ) -> Result<String, String> {
     cli.map(ToOwned::to_owned)
-        .or_else(|| shared_cfg.map(ToOwned::to_owned))
-        .or_else(|| legacy_cfg.map(ToOwned::to_owned))
+        .or_else(|| {
+            config.and_then(|cfg| {
+                cfg.provider_model(provider.display_name())
+                    .map(ToOwned::to_owned)
+            })
+        })
+        .or_else(|| translate_cfg.map(ToOwned::to_owned))
         .or_else(|| std::env::var("MENTRA_MODEL").ok())
         .ok_or_else(|| {
-            "--model is required (or set ai.model in langcodec.toml, or use legacy translate.model, or set MENTRA_MODEL)"
-                .to_string()
+            format!(
+                "--model is required (or set [{}].model in langcodec.toml, set translate.model, or set MENTRA_MODEL)",
+                provider.display_name()
+            )
         })
 }
 
@@ -134,3 +153,57 @@ pub(crate) fn build_provider(kind: &ProviderKind) -> Result<ProviderSetup, Strin
         provider,
     })
 }
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[test]
+    fn resolve_provider_uses_single_configured_provider_section() {
+        let config: CliConfig = toml::from_str(
+            r#"
+[openai]
+model = "gpt-5.4"
+"#,
+        )
+        .expect("parse config");
+
+        let provider = resolve_provider(None, Some(&config), None).expect("resolve provider");
+        assert_eq!(provider, ProviderKind::OpenAI);
+    }
+
+    #[test]
+    fn resolve_provider_rejects_multiple_configured_provider_sections() {
+        let config: CliConfig = toml::from_str(
+            r#"
+[openai]
+model = "gpt-5.4"
+
+[anthropic]
+model = "claude-sonnet"
+"#,
+        )
+        .expect("parse config");
+
+        let err = resolve_provider(None, Some(&config), None).unwrap_err();
+        assert!(err.contains("Multiple provider sections are configured"));
+    }
+
+    #[test]
+    fn resolve_model_prefers_selected_provider_section() {
+        let config: CliConfig = toml::from_str(
+            r#"
+[openai]
+model = "gpt-5.4"
+
+[anthropic]
+model = "claude-sonnet"
+"#,
+        )
+        .expect("parse config");
+
+        let model = resolve_model(None, Some(&config), &ProviderKind::Anthropic, None)
+            .expect("resolve model");
+        assert_eq!(model, "claude-sonnet");
+    }
+}