1use serde_json::Value;
4use sha2::{Digest, Sha256};
5use std::sync::LazyLock;
6
7use regex::Regex;
8
9use crate::error::{Result, SparError};
10
11static FENCE: LazyLock<Regex> = LazyLock::new(|| {
12 Regex::new(r"(?s)```(?:json)?\s*(\{.*?\}|\[.*?\])\s*```").expect("fence pattern")
13});
14
15pub fn extract_json(text: &str) -> Result<Value> {
21 if text.trim().is_empty() {
22 return Err(SparError::new("empty response, expected JSON"));
23 }
24 candidates(text)
25 .into_iter()
26 .next()
27 .ok_or_else(|| SparError::new(format!("no JSON found in response:\n{}", head(text, 800))))
28}
29
30pub fn candidates(text: &str) -> Vec<Value> {
35 let mut out = Vec::new();
36 let mut push = |value: Value| {
37 if !out.contains(&value) {
38 out.push(value);
39 }
40 };
41
42 let fenced: Vec<&str> = FENCE
43 .captures_iter(text)
44 .filter_map(|c| c.get(1).map(|m| m.as_str()))
45 .collect();
46 for blob in fenced.iter().rev() {
47 if let Ok(value) = serde_json::from_str::<Value>(blob) {
48 push(value);
49 }
50 }
51
52 let bytes = text.as_bytes();
53 for (opener, closer) in [(b'{', b'}'), (b'[', b']')] {
54 let mut end = rfind_byte(bytes, closer, bytes.len());
55 while let Some(e) = end {
56 let mut depth = 0i32;
57 let mut start = None;
58 for i in (0..=e).rev() {
59 if bytes[i] == closer {
60 depth += 1;
61 } else if bytes[i] == opener {
62 depth -= 1;
63 if depth == 0 {
64 start = Some(i);
65 break;
66 }
67 }
68 }
69 if let Some(s) = start {
70 if let Ok(value) = serde_json::from_str::<Value>(&text[s..=e]) {
71 push(value);
72 }
73 }
74 end = rfind_byte(bytes, closer, e);
75 }
76 }
77 out
78}
79
80pub fn looks_truncated(text: &str) -> bool {
86 let mut opened = 0i64;
87 let mut in_string = false;
88 let mut escaped = false;
89 for c in text.chars() {
90 if escaped {
91 escaped = false;
92 continue;
93 }
94 match c {
95 '\\' if in_string => escaped = true,
96 '"' => in_string = !in_string,
97 '{' if !in_string => opened += 1,
98 '}' if !in_string => opened -= 1,
99 _ => {}
100 }
101 }
102 opened > 0
103}
104
105pub fn extract_into<T: serde::de::DeserializeOwned>(text: &str) -> Result<T> {
112 let found = candidates(text);
113 if found.is_empty() {
114 return Err(SparError::new(if looks_truncated(text) {
115 format!(
116 "the response was cut off before any complete JSON:\n{}",
117 head(text, 400)
118 )
119 } else {
120 format!("no JSON found in response:\n{}", head(text, 800))
121 }));
122 }
123
124 let mut last_error = None;
125 for value in &found {
126 match serde_json::from_value::<T>(value.clone()) {
127 Ok(parsed) => return Ok(parsed),
128 Err(e) => last_error = Some((e, value)),
129 }
130 }
131
132 let (error, value) = last_error.expect("non-empty");
133 if looks_truncated(text) {
134 return Err(SparError::new(format!(
135 "the response was cut off before the answer was complete, so only fragments of it \
136 parsed ({error}). Ask for less in one go, or give this agent a CLI flag for native \
137 structured output."
138 )));
139 }
140 Err(SparError::new(format!(
141 "response did not match the expected shape ({error}).\nGot: {}",
142 head(&value.to_string(), 600)
143 )))
144}
145
146fn rfind_byte(haystack: &[u8], needle: u8, before: usize) -> Option<usize> {
147 haystack[..before.min(haystack.len())]
148 .iter()
149 .rposition(|b| *b == needle)
150}
151
152fn head(text: &str, max: usize) -> String {
153 text.chars().take(max).collect()
154}
155
156pub fn finding_key(title: &str, file: &str) -> String {
162 let basis: String = format!("{} {}", title.trim(), file.trim())
163 .to_lowercase()
164 .chars()
165 .filter(|c| c.is_ascii_alphanumeric() || matches!(c, ' ' | '/' | '.' | '_' | '-'))
166 .collect();
167 let basis = basis.split_whitespace().collect::<Vec<_>>().join(" ");
168
169 let digest = Sha256::digest(basis.as_bytes());
170 let hex: String = digest.iter().map(|b| format!("{b:02x}")).collect();
171 hex[..12].to_string()
172}
173
174#[cfg(test)]
175mod tests {
176 use super::*;
177
178 #[test]
179 fn bare_object() {
180 assert_eq!(
181 serde_json::json!({"a": 1}),
182 extract_json(r#"{"a": 1}"#).unwrap()
183 );
184 }
185
186 #[test]
187 fn fenced_block() {
188 let out = extract_json("here you go:\n```json\n{\"a\": 1}\n```\n").unwrap();
189 assert_eq!(serde_json::json!({"a": 1}), out);
190 }
191
192 #[test]
193 fn trailing_prose() {
194 let out = extract_json("Thoughts...\n{\"verdict\": \"approve\"}\nDone.").unwrap();
195 assert_eq!(serde_json::json!({"verdict": "approve"}), out);
196 }
197
198 #[test]
199 fn picks_the_last_fenced_block() {
200 let text = "```json\n{\"n\": 1}\n```\nrevised:\n```json\n{\"n\": 2}\n```";
201 assert_eq!(serde_json::json!({"n": 2}), extract_json(text).unwrap());
202 }
203
204 #[test]
205 fn nested_braces() {
206 let payload = r#"{"findings": [{"severity": "nit", "d": {"x": [1, 2]}}]}"#;
207 let out = extract_json(&format!("blah {payload} blah")).unwrap();
208 assert_eq!(1, out["findings"].as_array().unwrap().len());
209 }
210
211 #[test]
212 fn top_level_array() {
213 let out = extract_json("result: [1, 2, 3]").unwrap();
214 assert_eq!(3, out.as_array().unwrap().len());
215 }
216
217 #[test]
218 fn multibyte_prose_around_the_payload_does_not_panic() {
219 let out = extract_json("\u{1f600}\u{1f600} {\"a\": 1} \u{1f600}").unwrap();
220 assert_eq!(serde_json::json!({"a": 1}), out);
221 }
222
223 #[test]
224 fn raises_when_there_is_none() {
225 assert!(extract_json("no json here at all").is_err());
226 }
227
228 #[test]
229 fn raises_on_empty() {
230 assert!(extract_json(" ").is_err());
231 }
232
233 #[test]
234 fn malformed_trailing_object_falls_back_to_an_earlier_one() {
235 let text = "{\"good\": true}\nthen: {\"bad\": ,}";
236 assert_eq!(
237 serde_json::json!({"good": true}),
238 extract_json(text).unwrap()
239 );
240 }
241
242 #[test]
245 fn key_is_stable_across_wording_noise() {
246 assert_eq!(
247 finding_key("Unbounded loop!", "src/x.rs"),
248 finding_key("unbounded loop", "src/x.rs")
249 );
250 }
251
252 #[test]
253 fn key_differs_by_file() {
254 assert_ne!(finding_key("t", "a.rs"), finding_key("t", "b.rs"));
255 }
256
257 #[test]
258 fn key_is_case_insensitive_in_the_path_too() {
259 assert_eq!(
260 finding_key("t", "src/Main.rs"),
261 finding_key("t", "src/main.rs")
262 );
263 }
264
265 #[test]
266 fn key_is_stable_across_whitespace() {
267 assert_eq!(finding_key("a b", "x.rs"), finding_key(" a b ", "x.rs"));
268 }
269
270 #[test]
271 fn key_is_twelve_hex_characters() {
272 let key = finding_key("anything", "file.rs");
273 assert_eq!(12, key.len());
274 assert!(key.chars().all(|c| c.is_ascii_hexdigit()));
275 }
276}
277
278#[cfg(test)]
279mod truncation_tests {
280 use super::*;
281 use serde::Deserialize;
282
283 #[derive(Debug, Deserialize)]
284 struct Review {
285 verdict: String,
286 findings: Vec<Finding>,
287 }
288 #[derive(Debug, Deserialize)]
289 struct Finding {
290 title: String,
291 }
292
293 const TRUNCATED: &str = r#"Here is my review.
298{"verdict":"changes_requested","next_action":"hand_back","summary":"Two problems.",
299 "findings":[
300 {"severity":"blocking","title":"First","detail":"one","file":"a.ts","in_scope":true},
301 {"severity":"non-blocking","title":"Second","detail":"numbers unnamed keys by position"#;
302
303 #[test]
304 fn a_truncated_review_is_reported_as_truncated_not_as_the_wrong_shape() {
305 let err = extract_into::<Review>(TRUNCATED).unwrap_err().to_string();
306 assert!(err.contains("cut off"), "{err}");
307 assert!(!err.contains("did not match the expected shape"), "{err}");
308 }
309
310 #[test]
311 fn truncation_is_detected_from_the_unclosed_braces() {
312 assert!(looks_truncated(TRUNCATED));
313 assert!(!looks_truncated(r#"{"a":1}"#));
314 assert!(!looks_truncated(r#"{"a":"a { in a string"}"#));
316 assert!(!looks_truncated(r#"{"a":"an escaped \" quote { here"}"#));
317 }
318
319 #[test]
322 fn the_review_is_found_even_with_nested_objects_after_it() {
323 let text = r#"Thinking out loud first.
324{"verdict":"approve","next_action":"merge","summary":"Fine.","findings":[{"severity":"nit","title":"Wording","detail":"d","file":"a.ts","in_scope":true}]}
325And here is a stray object afterwards: {"title":"not the review"}"#;
326 let review: Review = extract_into(text).unwrap();
327 assert_eq!("approve", review.verdict);
328 assert_eq!(1, review.findings.len());
329 assert_eq!("Wording", review.findings[0].title);
330 }
331
332 #[test]
333 fn candidates_are_offered_most_likely_first() {
334 let text = "```json\n{\"verdict\":\"approve\",\"findings\":[]}\n```\ntrailing {\"x\":1}";
335 let review: Review = extract_into(text).unwrap();
336 assert_eq!("approve", review.verdict);
337 }
338
339 #[test]
340 fn a_genuinely_wrong_shape_still_says_so() {
341 let err = extract_into::<Review>(r#"{"colour":"blue"}"#)
342 .unwrap_err()
343 .to_string();
344 assert!(err.contains("did not match the expected shape"), "{err}");
345 assert!(!err.contains("cut off"), "{err}");
346 }
347
348 #[test]
349 fn nothing_parseable_is_still_reported_plainly() {
350 let err = extract_into::<Review>("no json at all")
351 .unwrap_err()
352 .to_string();
353 assert!(err.contains("no JSON found"), "{err}");
354 }
355}