| 1 | //! Response shaping: what a `gh` payload looks like by the time the model |
| 2 | //! sees it. |
| 3 | //! |
| 4 | //! Long bodies and diffs spill to task artifacts and come back as summaries, |
| 5 | //! and every write action returns the task metadata that records what it did. |
| 6 | |
| 7 | use std::path::{Path, PathBuf}; |
| 8 | |
| 9 | use chrono::Utc; |
| 10 | use serde_json::{Value, json}; |
| 11 | use uuid::Uuid; |
| 12 | |
| 13 | use crate::task_manager::{TaskArtifactRef, TaskGithubEvent}; |
| 14 | use crate::tools::spec::{ToolContext, ToolError}; |
| 15 | |
| 16 | pub(super) const BODY_ARTIFACT_THRESHOLD: usize = 4_000; |
| 17 | pub(super) const DIFF_ARTIFACT_THRESHOLD: usize = 8_000; |
| 18 | |
| 19 | pub(super) fn shape_large_text( |
| 20 | context: &ToolContext, |
| 21 | mut value: Value, |
| 22 | label: &str, |
| 23 | threshold: usize, |
| 24 | ) -> Result<Value, ToolError> { |
| 25 | let body = value |
| 26 | .get("body") |
| 27 | .and_then(Value::as_str) |
| 28 | .map(ToString::to_string); |
| 29 | if let Some(body) = body |
| 30 | && body.len() > threshold |
| 31 | { |
| 32 | let artifact = write_artifact_if_needed(context, label, &body, threshold)?; |
| 33 | value["body_summary"] = json!(summarize(&body, 900)); |
| 34 | value["body_artifact"] = json!(artifact); |
| 35 | value["body"] = json!(summarize(&body, 1200)); |
| 36 | } |
| 37 | Ok(value) |
| 38 | } |
| 39 | |
| 40 | pub(super) fn write_artifact_if_needed( |
| 41 | context: &ToolContext, |
| 42 | label: &str, |
| 43 | content: &str, |
| 44 | threshold: usize, |
| 45 | ) -> Result<Option<PathBuf>, ToolError> { |
| 46 | if content.len() <= threshold { |
| 47 | return Ok(None); |
| 48 | } |
| 49 | let Some(task_id) = context.runtime.active_task_id.as_deref() else { |
| 50 | return Ok(None); |
| 51 | }; |
| 52 | if let Some(manager) = context.runtime.task_manager.as_ref() { |
| 53 | return manager |
| 54 | .write_task_artifact(task_id, label, content) |
| 55 | .map(Some) |
| 56 | .map_err(|e| ToolError::execution_failed(e.to_string())); |
| 57 | } |
| 58 | let Some(data_dir) = context.runtime.task_data_dir.as_ref() else { |
| 59 | return Ok(None); |
| 60 | }; |
| 61 | let dir = data_dir.join("artifacts").join(task_id); |
| 62 | std::fs::create_dir_all(&dir) |
| 63 | .map_err(|e| ToolError::execution_failed(format!("create artifact dir: {e}")))?; |
| 64 | let absolute = dir.join(format!( |
| 65 | "{}_{}.txt", |
| 66 | Utc::now().format("%Y%m%dT%H%M%S%.3fZ"), |
| 67 | sanitize_filename(label) |
| 68 | )); |
| 69 | std::fs::write(&absolute, content) |
| 70 | .map_err(|e| ToolError::execution_failed(format!("write artifact: {e}")))?; |
| 71 | Ok(Some( |
| 72 | absolute |
| 73 | .strip_prefix(data_dir) |
| 74 | .map(Path::to_path_buf) |
| 75 | .unwrap_or(absolute), |
| 76 | )) |
| 77 | } |
| 78 | |
| 79 | pub(super) fn artifact_refs_from_context(content: &str, label: &str) -> Vec<TaskArtifactRef> { |
| 80 | let Ok(value) = serde_json::from_str::<Value>(content) else { |
| 81 | return Vec::new(); |
| 82 | }; |
| 83 | let (path_key, summary_key) = if label.ends_with("_diff") { |
| 84 | ("diff_artifact", "diff_summary") |
| 85 | } else { |
| 86 | ("body_artifact", "body_summary") |
| 87 | }; |
| 88 | let mut refs = Vec::new(); |
| 89 | collect_artifact_refs(&value, path_key, summary_key, label, &mut refs); |
| 90 | refs |
| 91 | } |
| 92 | |
| 93 | fn collect_artifact_refs( |
| 94 | value: &Value, |
| 95 | path_key: &str, |
| 96 | summary_key: &str, |
| 97 | label: &str, |
| 98 | refs: &mut Vec<TaskArtifactRef>, |
| 99 | ) { |
| 100 | match value { |
| 101 | Value::Object(map) => { |
| 102 | if let Some(path) = map.get(path_key).and_then(Value::as_str) { |
| 103 | let summary = map |
| 104 | .get(summary_key) |
| 105 | .and_then(Value::as_str) |
| 106 | .map(ToString::to_string) |
| 107 | .unwrap_or_else(|| format!("GitHub {label} artifact")); |
| 108 | refs.push(TaskArtifactRef { |
| 109 | label: label.to_string(), |
| 110 | path: PathBuf::from(path), |
| 111 | summary, |
| 112 | created_at: Utc::now(), |
| 113 | }); |
| 114 | } |
| 115 | for child in map.values() { |
| 116 | collect_artifact_refs(child, path_key, summary_key, label, refs); |
| 117 | } |
| 118 | } |
| 119 | Value::Array(items) => { |
| 120 | for child in items { |
| 121 | collect_artifact_refs(child, path_key, summary_key, label, refs); |
| 122 | } |
| 123 | } |
| 124 | _ => {} |
| 125 | } |
| 126 | } |
| 127 | |
| 128 | pub(super) fn github_event_metadata( |
| 129 | action: &str, |
| 130 | target: &str, |
| 131 | number: u64, |
| 132 | summary: String, |
| 133 | url: Option<String>, |
| 134 | artifact: Option<PathBuf>, |
| 135 | ) -> Value { |
| 136 | let artifacts = artifact |
| 137 | .map(|path| { |
| 138 | json!([TaskArtifactRef { |
| 139 | label: format!("github_{action}"), |
| 140 | path, |
| 141 | summary: summary.clone(), |
| 142 | created_at: Utc::now(), |
| 143 | }]) |
| 144 | }) |
| 145 | .unwrap_or_else(|| json!([])); |
| 146 | json!({ |
| 147 | "task_updates": { |
| 148 | "github_event": TaskGithubEvent { |
| 149 | id: format!("gh_{}", &Uuid::new_v4().to_string()[..8]), |
| 150 | action: action.to_string(), |
| 151 | target: target.to_string(), |
| 152 | number, |
| 153 | summary, |
| 154 | url, |
| 155 | recorded_at: Utc::now(), |
| 156 | }, |
| 157 | "artifacts": artifacts |
| 158 | } |
| 159 | }) |
| 160 | } |
| 161 | |
| 162 | pub(super) fn summarize(text: &str, limit: usize) -> String { |
| 163 | let mut out = String::new(); |
| 164 | for (idx, ch) in text.chars().enumerate() { |
| 165 | if idx >= limit.saturating_sub(3) { |
| 166 | out.push_str("..."); |
| 167 | return out; |
| 168 | } |
| 169 | if ch.is_control() && ch != '\n' && ch != '\t' { |
| 170 | continue; |
| 171 | } |
| 172 | out.push(ch); |
| 173 | } |
| 174 | out |
| 175 | } |
| 176 | |
| 177 | fn sanitize_filename(input: &str) -> String { |
| 178 | let mut out = String::new(); |
| 179 | for ch in input.chars() { |
| 180 | if ch.is_ascii_alphanumeric() || ch == '_' || ch == '-' { |
| 181 | out.push(ch); |
| 182 | } else { |
| 183 | out.push('_'); |
| 184 | } |
| 185 | } |
| 186 | if out.is_empty() { |
| 187 | "artifact".to_string() |
| 188 | } else { |
| 189 | out |
| 190 | } |
| 191 | } |
| 192 |