refactor(arr): share the LLM prompt helpers across backends

crates/arr-subs/src/openai.rs

@@ -25,7 +25,10 @@ use serde::{Deserialize, Serialize};
 
 use arr_core::Language;
 
-use crate::translate::{Backend, BackendId, Batch, Error, Result, TranslateFuture, TranslatedCue};
+use crate::translate::{
+    strip_code_fence, system_prompt, Backend, BackendId, Batch, Error, Result, TranslateFuture,
+    TranslatedCue,
+};
 
 const DEFAULT_BASE_URL: &str = "https://api.openai.com/v1/";
 const BACKEND_NAME: &str = "openai";
@@ -277,69 +280,6 @@ fn truncate(body: &str) -> String {
     body.chars().take(MAX_ERROR_BODY).collect()
 }
 
-/// Strip a ```` ```json ```` / ```` ``` ```` fence a model wrapped its reply
-/// in despite being told not to. Cheaper than retrying the whole batch over a
-/// habit some models never break.
-fn strip_code_fence(content: &str) -> &str {
-    let trimmed = content.trim();
-    let trimmed = trimmed
-        .strip_prefix("```json")
-        .or_else(|| trimmed.strip_prefix("```"))
-        .unwrap_or(trimmed);
-    trimmed.strip_suffix("```").unwrap_or(trimmed).trim()
-}
-
-/// How a language reads in a prompt meant for an LLM, not a settings row.
-fn language_name(language: &Language) -> String {
-    match language {
-        Language::PortuguesePortugal => "European Portuguese (pt-PT)".to_owned(),
-        Language::PortugueseBrazil => "Brazilian Portuguese (pt-BR)".to_owned(),
-        Language::PortugueseUnverified => "Portuguese".to_owned(),
-        Language::Other(tag) => tag.clone(),
-    }
-}
-
-/// A concrete nudge against the one mistake DESIGN.md §15 calls out by name:
-/// answering the wrong Portuguese. Generic advice ("respect the variant")
-/// is easy for a model to nod at and ignore; naming the words it must not
-/// reach for is not.
-fn variant_note(target: &Language) -> Option<&'static str> {
-    match target {
-        Language::PortuguesePortugal => Some(
-            "This is continental European Portuguese, not Brazilian Portuguese: write \"tu\", \
-             \"comboio\", \"casa de banho\" and European spelling and verb forms, never \
-             Brazilian ones.",
-        ),
-        Language::PortugueseBrazil => Some(
-            "This is Brazilian Portuguese, not European Portuguese: write \"você\", \"trem\", \
-             \"banheiro\" and Brazilian spelling and verb forms, never European ones.",
-        ),
-        _ => None,
-    }
-}
-
-fn system_prompt(source: &Language, target: &Language) -> String {
-    let mut prompt = format!(
-        "You translate subtitle cues for a film or TV episode from {} into {}. Write natural, \
-         spoken subtitle language a native speaker would actually use, not a stiff literal \
-         translation.",
-        language_name(source),
-        language_name(target)
-    );
-    if let Some(note) = variant_note(target) {
-        prompt.push(' ');
-        prompt.push_str(note);
-    }
-    prompt.push_str(
-        " Reply with nothing but a JSON array, one object per cue, each shaped \
-         {\"number\": <the same number you were given>, \"text\": \"<translation>\"}, in the \
-         order you received them. Every cue you were given must appear exactly once, under its \
-         original number: never add, drop, merge or renumber a cue. No prose, no markdown code \
-         fences, no explanation — JSON only.",
-    );
-    prompt
-}
-
 #[derive(Serialize)]
 struct RequestCue<'a> {
     number: usize,
@@ -386,8 +326,9 @@ struct ChatReplyMessage {
 mod tests {
     use arr_core::Language;
 
-    use super::{language_name, strip_code_fence, system_prompt, OpenAi};
+    use super::OpenAi;
     use crate::translate::Backend;
+    use crate::translate::{language_name, strip_code_fence, system_prompt};
 
     fn backend() -> OpenAi {
         OpenAi::new("gpt-4o-mini", super::OpenAiConfig { api_key: None }).expect("client builds")

crates/arr-subs/src/translate.rs

@@ -153,6 +153,72 @@ pub trait Backend: fmt::Debug + Send + Sync {
 /// Result alias for translation.
 pub type Result<T> = std::result::Result<T, Error>;
 
+/// Strip a ```` ```json ```` / ```` ``` ```` fence a model wrapped its reply
+/// in despite being told not to. Cheaper than retrying the whole batch over a
+/// habit some models never break.
+pub(crate) fn strip_code_fence(content: &str) -> &str {
+    let trimmed = content.trim();
+    let trimmed = trimmed
+        .strip_prefix("```json")
+        .or_else(|| trimmed.strip_prefix("```"))
+        .unwrap_or(trimmed);
+    trimmed.strip_suffix("```").unwrap_or(trimmed).trim()
+}
+
+/// How a language reads in a prompt meant for an LLM, not a settings row.
+pub(crate) fn language_name(language: &Language) -> String {
+    match language {
+        Language::PortuguesePortugal => "European Portuguese (pt-PT)".to_owned(),
+        Language::PortugueseBrazil => "Brazilian Portuguese (pt-BR)".to_owned(),
+        Language::PortugueseUnverified => "Portuguese".to_owned(),
+        Language::Other(tag) => tag.clone(),
+    }
+}
+
+/// A concrete nudge against the one mistake DESIGN.md §15 calls out by name:
+/// answering the wrong Portuguese. Generic advice ("respect the variant")
+/// is easy for a model to nod at and ignore; naming the words it must not
+/// reach for is not.
+fn variant_note(target: &Language) -> Option<&'static str> {
+    match target {
+        Language::PortuguesePortugal => Some(
+            "This is continental European Portuguese, not Brazilian Portuguese: write \"tu\", \
+             \"comboio\", \"casa de banho\" and European spelling and verb forms, never \
+             Brazilian ones.",
+        ),
+        Language::PortugueseBrazil => Some(
+            "This is Brazilian Portuguese, not European Portuguese: write \"você\", \"trem\", \
+             \"banheiro\" and Brazilian spelling and verb forms, never European ones.",
+        ),
+        _ => None,
+    }
+}
+
+/// The instruction every LLM-shaped backend sends ahead of the batch. Shared
+/// rather than per-backend so the reply contract — JSON only, every cue back
+/// under its original number — cannot drift between them.
+pub(crate) fn system_prompt(source: &Language, target: &Language) -> String {
+    let mut prompt = format!(
+        "You translate subtitle cues for a film or TV episode from {} into {}. Write natural, \
+         spoken subtitle language a native speaker would actually use, not a stiff literal \
+         translation.",
+        language_name(source),
+        language_name(target)
+    );
+    if let Some(note) = variant_note(target) {
+        prompt.push(' ');
+        prompt.push_str(note);
+    }
+    prompt.push_str(
+        " Reply with nothing but a JSON array, one object per cue, each shaped \
+         {\"number\": <the same number you were given>, \"text\": \"<translation>\"}, in the \
+         order you received them. Every cue you were given must appear exactly once, under its \
+         original number: never add, drop, merge or renumber a cue. No prose, no markdown code \
+         fences, no explanation — JSON only.",
+    );
+    prompt
+}
+
 /// Everything that can go wrong translating a subtitle.
 ///
 /// Kept apart from the provider [`Error`](crate::Error): the two report to