1use crate::PredictionProvider;
2use crate::paths::WORKTREES_DIR;
3use crate::qa::QaResult;
4use anyhow::{Context as _, Result};
5use collections::HashMap;
6use edit_prediction::example_spec::ExampleSpec;
7use edit_prediction::udiff::OpenedBuffers;
8use gpui::Entity;
9use http_client::Url;
10use language::{Anchor, Buffer};
11use project::Project;
12use serde::{Deserialize, Serialize};
13use std::{
14 borrow::Cow,
15 collections::VecDeque,
16 io::Read,
17 path::{Path, PathBuf},
18 sync::Arc,
19};
20use zeta_prompt::RelatedFile;
21
22#[derive(Clone, Debug, Serialize, Deserialize)]
23pub struct Example {
24 #[serde(flatten)]
25 pub spec: ExampleSpec,
26
27 /// The full content of the file where an edit is being predicted, and the
28 /// actual cursor offset.
29 #[serde(skip_serializing_if = "Option::is_none")]
30 pub prompt_inputs: Option<ExamplePromptInputs>,
31
32 /// The input and expected output from the edit prediction model.
33 #[serde(skip_serializing_if = "Option::is_none")]
34 pub prompt: Option<ExamplePrompt>,
35
36 /// The actual predictions from the model.
37 #[serde(default, skip_serializing_if = "Vec::is_empty")]
38 pub predictions: Vec<ExamplePrediction>,
39
40 /// The scores, for how well the actual predictions match the expected
41 /// predictions.
42 #[serde(default, skip_serializing_if = "Vec::is_empty")]
43 pub score: Vec<ExampleScore>,
44
45 /// QA evaluation results for each prediction (indexed parallel to `predictions`).
46 #[serde(default, skip_serializing_if = "Vec::is_empty")]
47 pub qa: Vec<Option<QaResult>>,
48
49 /// The application state used to process this example.
50 #[serde(skip)]
51 pub state: Option<ExampleState>,
52}
53
54#[derive(Clone, Debug)]
55pub struct ExampleState {
56 pub project: Entity<Project>,
57 pub buffer: Entity<Buffer>,
58 pub cursor_position: Anchor,
59 pub _open_buffers: OpenedBuffers,
60}
61
62#[derive(Clone, Debug, Serialize, Deserialize)]
63pub struct ExamplePromptInputs {
64 pub content: String,
65 pub cursor_row: u32,
66 pub cursor_column: u32,
67 pub cursor_offset: usize,
68 pub edit_history: Vec<Arc<zeta_prompt::Event>>,
69 pub related_files: Option<Vec<RelatedFile>>,
70}
71
72#[derive(Clone, Debug, Serialize, Deserialize)]
73pub struct ExamplePrompt {
74 pub input: String,
75 pub expected_output: String,
76 pub provider: PredictionProvider,
77}
78
79#[derive(Clone, Debug, Serialize, Deserialize)]
80pub struct ExamplePrediction {
81 #[serde(default, skip_serializing_if = "Option::is_none")]
82 pub actual_patch: Option<String>,
83 #[serde(deserialize_with = "deserialize_null_as_empty_string")]
84 pub actual_output: String,
85 #[serde(default, skip_serializing_if = "Option::is_none")]
86 pub error: Option<String>,
87 pub provider: PredictionProvider,
88}
89
90fn deserialize_null_as_empty_string<'de, D>(deserializer: D) -> Result<String, D::Error>
91where
92 D: serde::Deserializer<'de>,
93{
94 let opt = Option::<String>::deserialize(deserializer)?;
95 Ok(opt.unwrap_or_default())
96}
97
98#[derive(Clone, Debug, Serialize, Deserialize)]
99pub struct ExampleScore {
100 pub delta_chr_f: f32,
101 pub braces_disbalance: usize,
102 #[serde(default)]
103 pub exact_lines_tp: usize,
104 #[serde(default)]
105 pub exact_lines_fp: usize,
106 #[serde(default)]
107 pub exact_lines_fn: usize,
108}
109
110impl Example {
111 pub fn repo_name(&self) -> Result<RepoName<'_>> {
112 // git@github.com:owner/repo.git
113 if self.spec.repository_url.contains('@') {
114 let (owner, repo) = self
115 .spec
116 .repository_url
117 .split_once(':')
118 .context("expected : in git url")?
119 .1
120 .split_once('/')
121 .context("expected / in git url")?;
122 Ok(RepoName {
123 owner: Cow::Borrowed(owner),
124 name: Cow::Borrowed(repo.trim_end_matches(".git")),
125 })
126 // http://github.com/owner/repo.git
127 } else {
128 let url = Url::parse(&self.spec.repository_url)?;
129 let mut segments = url.path_segments().context("empty http url")?;
130 let owner = segments
131 .next()
132 .context("expected owner path segment")?
133 .to_string();
134 let repo = segments
135 .next()
136 .context("expected repo path segment")?
137 .trim_end_matches(".git")
138 .to_string();
139 assert!(segments.next().is_none());
140
141 Ok(RepoName {
142 owner: Cow::Owned(owner),
143 name: Cow::Owned(repo),
144 })
145 }
146 }
147}
148
149pub struct RepoName<'a> {
150 pub owner: Cow<'a, str>,
151 pub name: Cow<'a, str>,
152}
153
154impl RepoName<'_> {
155 pub fn worktree_path(&self) -> PathBuf {
156 WORKTREES_DIR
157 .join(self.owner.as_ref())
158 .join(self.name.as_ref())
159 }
160}
161
162pub fn read_example_files(inputs: &[PathBuf]) -> Vec<Example> {
163 let mut examples = Vec::new();
164
165 for path in inputs {
166 let is_stdin = path.as_path() == Path::new("-");
167 let content = if is_stdin {
168 let mut buffer = String::new();
169 std::io::stdin()
170 .read_to_string(&mut buffer)
171 .expect("Failed to read from stdin");
172 buffer
173 } else {
174 std::fs::read_to_string(path)
175 .unwrap_or_else(|_| panic!("Failed to read path: {:?}", &path))
176 };
177 let filename = path.file_stem().unwrap().to_string_lossy().to_string();
178 let ext = if !is_stdin {
179 path.extension()
180 .map(|ext| ext.to_string_lossy().to_string())
181 .unwrap_or_else(|| panic!("{} should have an extension", path.display()))
182 } else {
183 "jsonl".to_string()
184 };
185
186 match ext.as_ref() {
187 "json" => {
188 let mut example =
189 serde_json::from_str::<Example>(&content).unwrap_or_else(|error| {
190 panic!("Failed to parse example file: {}\n{error}", path.display())
191 });
192 if example.spec.name.is_empty() {
193 example.spec.name = filename;
194 }
195 examples.push(example);
196 }
197 "jsonl" => examples.extend(
198 content
199 .lines()
200 .enumerate()
201 .map(|(line_ix, line)| {
202 let mut example =
203 serde_json::from_str::<Example>(line).unwrap_or_else(|error| {
204 panic!(
205 "Failed to parse example on {}:{}\n{error}",
206 path.display(),
207 line_ix + 1
208 )
209 });
210 if example.spec.name.is_empty() {
211 example.spec.name = format!("{filename}-{line_ix}")
212 }
213 example
214 })
215 .collect::<Vec<Example>>(),
216 ),
217 "md" => {
218 let mut example = parse_markdown_example(&content).unwrap();
219 if example.spec.name.is_empty() {
220 example.spec.name = filename;
221 }
222 examples.push(example);
223 }
224 ext => {
225 panic!("{} has invalid example extension `{ext}`", path.display())
226 }
227 }
228 }
229
230 examples
231}
232
233pub fn sort_examples_by_repo_and_rev(examples: &mut [Example]) {
234 examples.sort_by(|a, b| {
235 a.spec
236 .repository_url
237 .cmp(&b.spec.repository_url)
238 .then(b.spec.revision.cmp(&a.spec.revision))
239 });
240}
241
242pub fn group_examples_by_repo(examples: Vec<Example>) -> VecDeque<Vec<Example>> {
243 let mut examples_by_repo = HashMap::default();
244 for example in examples {
245 examples_by_repo
246 .entry(example.spec.repository_url.clone())
247 .or_insert_with(Vec::new)
248 .push(example);
249 }
250 examples_by_repo.into_values().collect()
251}
252
253fn parse_markdown_example(input: &str) -> Result<Example> {
254 let spec = ExampleSpec::from_markdown(input)?;
255 Ok(Example {
256 spec,
257 prompt_inputs: None,
258 prompt: None,
259 predictions: Vec::new(),
260 score: Vec::new(),
261 qa: Vec::new(),
262 state: None,
263 })
264}