Extract audio preparation into a utility crate (#36807)

## What changed

- Add `codex-utils-audio` as a workspace crate for canonicalizing audio inputs
  and estimating their token usage.
- Update `codex-core` to consume the new crate while preserving the existing
  audio preparation tests.

GitOrigin-RevId: d719ecc08363ef52778aa37f3df0ca14f7778324

codex-rs/Cargo.lock

@@ -2761,6 +2761,7 @@ dependencies = [
  "codex-thread-store",
  "codex-tools",
  "codex-utils-absolute-path",
+ "codex-utils-audio",
  "codex-utils-cache",
  "codex-utils-cargo-bin",
  "codex-utils-home-dir",
@@ -2802,7 +2803,6 @@ dependencies = [
  "sha1 0.10.6",
  "shlex",
  "similar",
- "symphonia",
  "tempfile",
  "test-case",
  "test-log",
@@ -4377,6 +4377,20 @@ dependencies = [
  "codex-protocol",
 ]
 
+[[package]]
+name = "codex-utils-audio"
+version = "0.0.0"
+dependencies = [
+ "base64 0.22.1",
+ "codex-protocol",
+ "codex-utils-cache",
+ "codex-utils-string",
+ "pretty_assertions",
+ "symphonia",
+ "thiserror 2.0.18",
+ "tracing",
+]
+
 [[package]]
 name = "codex-utils-cache"
 version = "0.0.0"

codex-rs/Cargo.toml

@@ -94,6 +94,7 @@ members = [
     "v8-poc",
     "websocket-client",
     "utils/absolute-path",
+    "utils/audio",
     "utils/path-uri",
     "utils/cargo-bin",
     "git-utils",
@@ -246,6 +247,7 @@ codex-tui = { path = "tui" }
 codex-uds = { path = "uds" }
 codex-utils-absolute-path = { path = "utils/absolute-path" }
 codex-utils-approval-presets = { path = "utils/approval-presets" }
+codex-utils-audio = { path = "utils/audio" }
 codex-utils-cache = { path = "utils/cache" }
 codex-utils-cargo-bin = { path = "utils/cargo-bin" }
 codex-utils-cli = { path = "utils/cli" }

codex-rs/core/Cargo.toml

@@ -69,6 +69,7 @@ codex-terminal-detection = { workspace = true }
 codex-thread-store = { workspace = true }
 codex-tools = { workspace = true }
 codex-utils-absolute-path = { workspace = true }
+codex-utils-audio = { workspace = true }
 codex-utils-cache = { workspace = true }
 codex-utils-image = { workspace = true }
 codex-utils-home-dir = { workspace = true }
@@ -103,7 +104,6 @@ serde_json = { workspace = true }
 sha1 = { workspace = true }
 shlex = { workspace = true }
 similar = { workspace = true }
-symphonia = { workspace = true }
 tempfile = { workspace = true }
 thiserror = { workspace = true }
 tokio = { workspace = true, features = [

codex-rs/core/src/context_manager/history.rs

@@ -1,4 +1,3 @@
-use crate::audio_preparation::estimate_audio_token_count;
 use crate::context::ContextualUserFragment;
 use crate::context::world_state::WorldState;
 use crate::context::world_state::WorldStateSnapshot;
@@ -23,6 +22,7 @@ use codex_protocol::protocol::TokenUsage;
 use codex_protocol::protocol::TokenUsageInfo;
 use codex_protocol::protocol::TurnContextItem;
 use codex_protocol::protocol::WorldStateItem;
+use codex_utils_audio::estimate_audio_token_count;
 use codex_utils_cache::BlockingLruCache;
 use codex_utils_cache::sha1_digest;
 use codex_utils_output_truncation::TruncationPolicy;

codex-rs/core/src/lib.rs

@@ -7,7 +7,6 @@
 
 mod apply_patch;
 mod apps;
-mod audio_preparation;
 mod client;
 mod client_common;
 mod realtime_context;

codex-rs/core/src/session/mod.rs

@@ -17,7 +17,6 @@ use crate::agent::status::is_final;
 use crate::agent_communication::AgentCommunicationContext;
 use crate::agent_communication::AgentCommunicationKind;
 use crate::attestation::AttestationProvider;
-use crate::audio_preparation::prepare_response_items as prepare_audio_response_items;
 use crate::compact;
 use crate::compact::CompactedHistoryMetadata;
 use crate::config::ManagedFeatures;
@@ -152,6 +151,7 @@ use codex_thread_store::ReadThreadParams;
 use codex_thread_store::ResumeThreadParams;
 use codex_thread_store::ThreadPersistenceMetadata;
 use codex_thread_store::ThreadStore;
+use codex_utils_audio::prepare_response_items as prepare_audio_response_items;
 use codex_utils_path_uri::PathUri;
 use futures::future::BoxFuture;
 use futures::future::Shared;

codex-rs/core/src/tools/code_mode/mod.rs

@@ -24,7 +24,6 @@ use serde_json::Value as JsonValue;
 use tokio::sync::OnceCell;
 use tokio_util::sync::CancellationToken;
 
-use crate::audio_preparation::estimate_audio_token_count;
 use crate::function_tool::FunctionCallError;
 use crate::original_image_detail::can_request_original_image_detail;
 use crate::original_image_detail::sanitize_original_image_detail as sanitize_image_detail_items;
@@ -41,6 +40,7 @@ use crate::tools::router::ToolCallSource;
 use crate::unified_exec::resolve_max_tokens;
 use codex_protocol::openai_models::ToolMode;
 use codex_tools::ToolName;
+use codex_utils_audio::estimate_audio_token_count;
 use codex_utils_output_truncation::TruncationPolicy;
 use codex_utils_output_truncation::formatted_truncate_text_content_items_with_policy;
 use codex_utils_output_truncation::truncate_function_output_items_with_policy;

codex-rs/utils/audio/BUILD.bazel

@@ -0,0 +1,6 @@
+load("//:defs.bzl", "codex_rust_crate")
+
+codex_rust_crate(
+    name = "audio",
+    crate_name = "codex_utils_audio",
+)

codex-rs/utils/audio/Cargo.toml

@@ -0,0 +1,23 @@
+[package]
+edition.workspace = true
+license.workspace = true
+name = "codex-utils-audio"
+version.workspace = true
+
+[lints]
+workspace = true
+
+[dependencies]
+base64 = { workspace = true }
+codex-protocol = { workspace = true }
+codex-utils-cache = { workspace = true }
+codex-utils-string = { workspace = true }
+symphonia = { workspace = true }
+thiserror = { workspace = true }
+tracing = { workspace = true }
+
+[dev-dependencies]
+pretty_assertions = { workspace = true }
+
+[lib]
+doctest = false

codex-rs/utils/audio/src/lib.rs

@@ -1,3 +1,5 @@
+//! Audio preparation and duration-based token estimates for model inputs.
+
 use base64::Engine;
 use base64::engine::general_purpose::STANDARD as BASE64_STANDARD;
 use codex_protocol::models::ContentItem;
@@ -5,7 +7,7 @@ use codex_protocol::models::FunctionCallOutputContentItem;
 use codex_protocol::models::ResponseItem;
 use codex_utils_cache::BlockingLruCache;
 use codex_utils_cache::sha1_digest;
-use codex_utils_output_truncation::approx_token_count;
+use codex_utils_string::approx_token_count;
 use std::io::Cursor;
 use std::num::NonZeroUsize;
 use std::sync::LazyLock;
@@ -58,7 +60,8 @@ impl AudioPreparationError {
     }
 }
 
-pub(crate) fn prepare_response_items(items: &mut [ResponseItem]) {
+/// Canonicalizes audio inputs and replaces unsupported inputs with text placeholders.
+pub fn prepare_response_items(items: &mut [ResponseItem]) {
     for item in items {
         match item {
             ResponseItem::Message { content, .. } => prepare_message_content(content),
@@ -141,7 +144,8 @@ fn canonical_audio_mime(mime: &str) -> Option<&'static str> {
     }
 }
 
-pub(crate) fn estimate_audio_token_count(audio_url: &str) -> usize {
+/// Estimates audio tokens from decoded duration, falling back to the data URL size.
+pub fn estimate_audio_token_count(audio_url: &str) -> usize {
     let key = sha1_digest(audio_url.as_bytes());
     AUDIO_TOKEN_ESTIMATE_CACHE.get_or_insert_with(key, || {
         let Some(duration_seconds) = audio_duration_seconds(audio_url) else {