| 1 | //! Durable context reference records and media attachment parsing. |
| 2 | //! |
| 3 | //! Separated from composer completion and terminal-only UI so session persistence, |
| 4 | //! image attachment, and engine history can reference context and attachment items |
| 5 | //! without depending on `codewhale-tui`. |
| 6 | |
| 7 | use serde::{Deserialize, Serialize}; |
| 8 | |
| 9 | /// The transcript keeps the user's compact text (`@path` or `[Attached ...]`) |
| 10 | /// readable. This record preserves the exact target and inclusion state for |
| 11 | /// the context inspector and for session resume without leaking raw metadata |
| 12 | /// into the visible history cell. |
| 13 | #[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize)] |
| 14 | pub struct ContextReference { |
| 15 | pub kind: ContextReferenceKind, |
| 16 | pub source: ContextReferenceSource, |
| 17 | /// Short badge for terminal display, e.g. `file`, `dir`, `image`. |
| 18 | pub badge: String, |
| 19 | /// Compact display label from the transcript, without the leading `@`. |
| 20 | pub label: String, |
| 21 | /// Resolved target path or URI-equivalent string. |
| 22 | pub target: String, |
| 23 | pub included: bool, |
| 24 | pub expanded: bool, |
| 25 | #[serde(default, skip_serializing_if = "Option::is_none")] |
| 26 | pub detail: Option<String>, |
| 27 | } |
| 28 | |
| 29 | #[derive(Debug, Clone, Copy, PartialEq, Eq, Serialize, Deserialize)] |
| 30 | #[serde(rename_all = "snake_case")] |
| 31 | pub enum ContextReferenceKind { |
| 32 | File, |
| 33 | Directory, |
| 34 | Missing, |
| 35 | Unsupported, |
| 36 | MediaMention, |
| 37 | MediaAttachment, |
| 38 | /// `@git` / `@diff` — curated git context rather than a path (#4067). |
| 39 | GitContext, |
| 40 | } |
| 41 | |
| 42 | #[derive(Debug, Clone, Copy, PartialEq, Eq, Serialize, Deserialize)] |
| 43 | #[serde(rename_all = "snake_case")] |
| 44 | pub enum ContextReferenceSource { |
| 45 | AtMention, |
| 46 | Attachment, |
| 47 | } |
| 48 | |
| 49 | #[derive(Debug, Clone, PartialEq, Eq)] |
| 50 | pub struct MediaAttachmentReference { |
| 51 | pub kind: String, |
| 52 | pub path: String, |
| 53 | pub start_byte: usize, |
| 54 | pub end_byte: usize, |
| 55 | } |
| 56 | |
| 57 | /// Extract media attachment references from text formatted as `[Attached <kind>: <path>]`. |
| 58 | #[must_use] |
| 59 | pub fn media_attachment_references(input: &str) -> Vec<MediaAttachmentReference> { |
| 60 | let mut out = Vec::new(); |
| 61 | let mut offset = 0usize; |
| 62 | for line in input.split_inclusive('\n') { |
| 63 | let start_byte = offset; |
| 64 | let end_byte = offset + line.len(); |
| 65 | offset = end_byte; |
| 66 | let trimmed = line.trim(); |
| 67 | let Some(body) = trimmed |
| 68 | .strip_prefix("[Attached ") |
| 69 | .and_then(|value| value.strip_suffix(']')) |
| 70 | else { |
| 71 | continue; |
| 72 | }; |
| 73 | let Some((kind, rest)) = body.split_once(": ") else { |
| 74 | continue; |
| 75 | }; |
| 76 | let path = attachment_path(rest).trim(); |
| 77 | if !path.is_empty() { |
| 78 | out.push(MediaAttachmentReference { |
| 79 | kind: kind.trim().to_string(), |
| 80 | path: path.to_string(), |
| 81 | start_byte, |
| 82 | end_byte, |
| 83 | }); |
| 84 | } |
| 85 | } |
| 86 | out |
| 87 | } |
| 88 | |
| 89 | /// The path in `<description> at <path>` or a bare `<path>`. Attachment |
| 90 | /// producers write absolute paths, and a description never starts with one, |
| 91 | /// so the separator is the first ` at ` followed by an absolute path. |
| 92 | /// Splitting at the last ` at ` broke paths that contain one, such as macOS |
| 93 | /// screenshot names (`Screenshot 2026-09-28 at 9.18.11 AM.png`). Text with no |
| 94 | /// absolute candidate (hand-written or older sessions) keeps the last-` at ` |
| 95 | /// split it always had. |
| 96 | fn attachment_path(rest: &str) -> &str { |
| 97 | if looks_absolute(rest) { |
| 98 | return rest; |
| 99 | } |
| 100 | rest.match_indices(" at ") |
| 101 | .map(|(idx, sep)| &rest[idx + sep.len()..]) |
| 102 | .find(|candidate| looks_absolute(candidate)) |
| 103 | .or_else(|| rest.rsplit_once(" at ").map(|(_, path)| path)) |
| 104 | .unwrap_or(rest) |
| 105 | } |
| 106 | |
| 107 | fn looks_absolute(path: &str) -> bool { |
| 108 | let bytes = path.as_bytes(); |
| 109 | path.starts_with(['/', '\\', '~']) |
| 110 | || (bytes.len() >= 3 |
| 111 | && bytes[0].is_ascii_alphabetic() |
| 112 | && bytes[1] == b':' |
| 113 | && matches!(bytes[2], b'\\' | b'/')) |
| 114 | } |
| 115 | |
| 116 | #[cfg(test)] |
| 117 | mod tests { |
| 118 | use super::*; |
| 119 | |
| 120 | #[test] |
| 121 | fn serialization_roundtrip() { |
| 122 | let reference = ContextReference { |
| 123 | kind: ContextReferenceKind::File, |
| 124 | source: ContextReferenceSource::AtMention, |
| 125 | badge: "file".to_string(), |
| 126 | label: "test.rs".to_string(), |
| 127 | target: "/path/to/test.rs".to_string(), |
| 128 | included: true, |
| 129 | expanded: false, |
| 130 | detail: Some("included".to_string()), |
| 131 | }; |
| 132 | let json = serde_json::to_string(&reference).unwrap(); |
| 133 | let deserialized: ContextReference = serde_json::from_str(&json).unwrap(); |
| 134 | assert_eq!(reference, deserialized); |
| 135 | } |
| 136 | |
| 137 | #[test] |
| 138 | fn parses_media_attachments() { |
| 139 | let input = "Here is the screenshot:\n[Attached image: 100x100 at /tmp/shot.png]\nPlease analyze it."; |
| 140 | let refs = media_attachment_references(input); |
| 141 | assert_eq!(refs.len(), 1); |
| 142 | assert_eq!(refs[0].kind, "image"); |
| 143 | assert_eq!(refs[0].path, "/tmp/shot.png"); |
| 144 | } |
| 145 | |
| 146 | #[test] |
| 147 | fn attachment_paths_may_contain_at() { |
| 148 | let shot = "/Users/x/Desktop/Screenshot 2026-09-28 at 9.18.11 AM.png"; |
| 149 | let bare = format!("[Attached image: {shot}]"); |
| 150 | assert_eq!(media_attachment_references(&bare)[0].path, shot); |
| 151 | let described = format!("[Attached image: 8x4 PNG (2KB) at {shot}]"); |
| 152 | assert_eq!(media_attachment_references(&described)[0].path, shot); |
| 153 | let windows = r"[Attached image: 8x4 PNG at C:\Users\x\Shot at noon.png]"; |
| 154 | assert_eq!( |
| 155 | media_attachment_references(windows)[0].path, |
| 156 | r"C:\Users\x\Shot at noon.png" |
| 157 | ); |
| 158 | let relative = "[Attached image: 8x4 PNG at shots/a.png]"; |
| 159 | assert_eq!(media_attachment_references(relative)[0].path, "shots/a.png"); |
| 160 | } |
| 161 | } |
| 162 |