| 1 | //! The closed set of message roles Codewhale can put on a transcript. |
| 2 | //! |
| 3 | //! Roles used to be free-form `String`s on [`crate::request::Message`], and |
| 4 | //! four wire adapters each decided independently what an unfamiliar role |
| 5 | //! meant: two dropped it silently, one forwarded it verbatim for the provider |
| 6 | //! to reject, one failed closed. [`Role`] closes the set so that decision can |
| 7 | //! be made once, in one table, instead of four times by accident. |
| 8 | //! |
| 9 | //! Two properties are load-bearing and are covered by tests: |
| 10 | //! |
| 11 | //! * **Byte-identical serialization.** A `Role` serializes as exactly the |
| 12 | //! string it replaced and deserializes from any string. Saved transcripts |
| 13 | //! therefore need no schema bump and no migration ladder, and a session |
| 14 | //! written by a newer build still loads here — an unfamiliar role lands in |
| 15 | //! [`Role::Unrecognized`] rather than failing the whole session load. |
| 16 | //! * **`assistant_interrupted` stays distinct.** The interrupted-assistant |
| 17 | //! sentinel is its own variant, not a flavour of [`Role::Assistant`], so it |
| 18 | //! keeps round-tripping as a separate session item. |
| 19 | |
| 20 | use std::fmt; |
| 21 | |
| 22 | use serde::{Deserialize, Deserializer, Serialize, Serializer}; |
| 23 | |
| 24 | use crate::request::INTERRUPTED_ASSISTANT_ROLE; |
| 25 | |
| 26 | /// Who authored a message in a transcript. |
| 27 | /// |
| 28 | /// `Unrecognized` is deliberately part of the type: it is what lets a |
| 29 | /// transcript written by a future build round-trip through this one. It |
| 30 | /// carries no trust and grants no placement of its own — see the wire |
| 31 | /// adapters' placement table for what each dialect does with it. |
| 32 | #[derive(Debug, Clone, PartialEq, Eq, Hash)] |
| 33 | pub enum Role { |
| 34 | /// Input authored by the human operator or by the harness on their behalf. |
| 35 | User, |
| 36 | /// Output authored by the model. |
| 37 | Assistant, |
| 38 | /// Harness-authored context injected into the transcript body — compaction |
| 39 | /// summaries, branch summaries, sub-agent framing. |
| 40 | System, |
| 41 | /// Provider-supported instruction content embedded at a transcript |
| 42 | /// position. Unlike the top-level system prompt, this role is load-bearing |
| 43 | /// history and must retain its position on wires that support it. |
| 44 | Developer, |
| 45 | /// Assistant text that was visible before the turn was interrupted. Kept |
| 46 | /// distinct from [`Role::Assistant`] so replay can mark it as incomplete. |
| 47 | InterruptedAssistant, |
| 48 | /// A role string this build does not know. Preserved verbatim so loading |
| 49 | /// and re-saving a transcript is lossless. |
| 50 | Unrecognized(String), |
| 51 | } |
| 52 | |
| 53 | impl Role { |
| 54 | /// The exact wire/persisted string for this role. |
| 55 | /// |
| 56 | /// This is the serialization; do not let it drift from [`Serialize`]. |
| 57 | #[must_use] |
| 58 | pub fn as_str(&self) -> &str { |
| 59 | match self { |
| 60 | Self::User => "user", |
| 61 | Self::Assistant => "assistant", |
| 62 | Self::System => "system", |
| 63 | Self::Developer => "developer", |
| 64 | Self::InterruptedAssistant => INTERRUPTED_ASSISTANT_ROLE, |
| 65 | Self::Unrecognized(raw) => raw.as_str(), |
| 66 | } |
| 67 | } |
| 68 | |
| 69 | /// True for roles the model itself authored, interrupted output included. |
| 70 | #[must_use] |
| 71 | pub fn is_assistant_like(&self) -> bool { |
| 72 | matches!(self, Self::Assistant | Self::InterruptedAssistant) |
| 73 | } |
| 74 | } |
| 75 | |
| 76 | impl From<&str> for Role { |
| 77 | fn from(value: &str) -> Self { |
| 78 | match value { |
| 79 | "user" => Self::User, |
| 80 | "assistant" => Self::Assistant, |
| 81 | "system" => Self::System, |
| 82 | "developer" => Self::Developer, |
| 83 | INTERRUPTED_ASSISTANT_ROLE => Self::InterruptedAssistant, |
| 84 | other => Self::Unrecognized(other.to_string()), |
| 85 | } |
| 86 | } |
| 87 | } |
| 88 | |
| 89 | impl From<String> for Role { |
| 90 | fn from(value: String) -> Self { |
| 91 | match value.as_str() { |
| 92 | "user" => Self::User, |
| 93 | "assistant" => Self::Assistant, |
| 94 | "system" => Self::System, |
| 95 | "developer" => Self::Developer, |
| 96 | INTERRUPTED_ASSISTANT_ROLE => Self::InterruptedAssistant, |
| 97 | _ => Self::Unrecognized(value), |
| 98 | } |
| 99 | } |
| 100 | } |
| 101 | |
| 102 | impl fmt::Display for Role { |
| 103 | fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { |
| 104 | f.write_str(self.as_str()) |
| 105 | } |
| 106 | } |
| 107 | |
| 108 | impl PartialEq<str> for Role { |
| 109 | fn eq(&self, other: &str) -> bool { |
| 110 | self.as_str() == other |
| 111 | } |
| 112 | } |
| 113 | |
| 114 | impl PartialEq<&str> for Role { |
| 115 | fn eq(&self, other: &&str) -> bool { |
| 116 | self.as_str() == *other |
| 117 | } |
| 118 | } |
| 119 | |
| 120 | impl PartialEq<String> for Role { |
| 121 | fn eq(&self, other: &String) -> bool { |
| 122 | self.as_str() == other.as_str() |
| 123 | } |
| 124 | } |
| 125 | |
| 126 | impl PartialEq<Role> for str { |
| 127 | fn eq(&self, other: &Role) -> bool { |
| 128 | self == other.as_str() |
| 129 | } |
| 130 | } |
| 131 | |
| 132 | impl PartialEq<Role> for &str { |
| 133 | fn eq(&self, other: &Role) -> bool { |
| 134 | *self == other.as_str() |
| 135 | } |
| 136 | } |
| 137 | |
| 138 | impl PartialEq<Role> for String { |
| 139 | fn eq(&self, other: &Role) -> bool { |
| 140 | self.as_str() == other.as_str() |
| 141 | } |
| 142 | } |
| 143 | |
| 144 | impl Serialize for Role { |
| 145 | fn serialize<S: Serializer>(&self, serializer: S) -> Result<S::Ok, S::Error> { |
| 146 | serializer.serialize_str(self.as_str()) |
| 147 | } |
| 148 | } |
| 149 | |
| 150 | impl<'de> Deserialize<'de> for Role { |
| 151 | fn deserialize<D: Deserializer<'de>>(deserializer: D) -> Result<Self, D::Error> { |
| 152 | Ok(Self::from(String::deserialize(deserializer)?)) |
| 153 | } |
| 154 | } |
| 155 | |
| 156 | #[cfg(test)] |
| 157 | mod tests { |
| 158 | use super::Role; |
| 159 | use crate::request::INTERRUPTED_ASSISTANT_ROLE; |
| 160 | |
| 161 | #[test] |
| 162 | fn known_roles_serialize_as_the_strings_they_replaced() { |
| 163 | for (role, expected) in [ |
| 164 | (Role::User, "\"user\""), |
| 165 | (Role::Assistant, "\"assistant\""), |
| 166 | (Role::System, "\"system\""), |
| 167 | (Role::Developer, "\"developer\""), |
| 168 | (Role::InterruptedAssistant, "\"assistant_interrupted\""), |
| 169 | ] { |
| 170 | assert_eq!(serde_json::to_string(&role).expect("serialize"), expected); |
| 171 | } |
| 172 | } |
| 173 | |
| 174 | #[test] |
| 175 | fn serialized_bytes_match_the_raw_string_encoding() { |
| 176 | // The persisted format must not shift: a `Role` has to produce the |
| 177 | // same bytes the `String` field produced, or every saved session |
| 178 | // would need a schema bump and a migration ladder. |
| 179 | for raw in [ |
| 180 | "user", |
| 181 | "assistant", |
| 182 | "system", |
| 183 | "developer", |
| 184 | INTERRUPTED_ASSISTANT_ROLE, |
| 185 | ] { |
| 186 | assert_eq!( |
| 187 | serde_json::to_vec(&Role::from(raw)).expect("serialize role"), |
| 188 | serde_json::to_vec(raw).expect("serialize string"), |
| 189 | "role {raw} must serialize like its string", |
| 190 | ); |
| 191 | } |
| 192 | } |
| 193 | |
| 194 | #[test] |
| 195 | fn unknown_roles_round_trip_verbatim() { |
| 196 | let role = Role::from("future_role"); |
| 197 | assert_eq!(role, Role::Unrecognized("future_role".to_string())); |
| 198 | let encoded = serde_json::to_string(&role).expect("serialize"); |
| 199 | assert_eq!(encoded, "\"future_role\""); |
| 200 | let decoded: Role = serde_json::from_str(&encoded).expect("deserialize"); |
| 201 | assert_eq!(decoded, role); |
| 202 | } |
| 203 | |
| 204 | #[test] |
| 205 | fn interrupted_assistant_is_not_assistant() { |
| 206 | let decoded: Role = |
| 207 | serde_json::from_str("\"assistant_interrupted\"").expect("deserialize sentinel"); |
| 208 | assert_eq!(decoded, Role::InterruptedAssistant); |
| 209 | assert_ne!(decoded, Role::Assistant); |
| 210 | assert_ne!(decoded, "assistant"); |
| 211 | assert!(decoded.is_assistant_like()); |
| 212 | } |
| 213 | |
| 214 | #[test] |
| 215 | fn string_comparisons_work_in_both_directions() { |
| 216 | assert_eq!(Role::User, "user"); |
| 217 | assert_eq!("user", Role::User); |
| 218 | assert_eq!(Role::User, "user".to_string()); |
| 219 | assert_ne!(Role::System, "user"); |
| 220 | } |
| 221 | } |
| 222 |