use serde_json::Value;
use sha2::{Digest, Sha256};
use std::sync::LazyLock;
use regex::Regex;
use crate::error::{Result, SparError};
static FENCE: LazyLock<Regex> = LazyLock::new(|| {
Regex::new(r"(?s)```(?:json)?\s*(\{.*?\}|\[.*?\])\s*```").expect("fence pattern")
});
pub fn extract_json(text: &str) -> Result<Value> {
if text.trim().is_empty() {
return Err(SparError::new("empty response, expected JSON"));
}
candidates(text)
.into_iter()
.next()
.ok_or_else(|| SparError::new(format!("no JSON found in response:\n{}", head(text, 800))))
}
pub fn candidates(text: &str) -> Vec<Value> {
let mut out = Vec::new();
let mut push = |value: Value| {
if !out.contains(&value) {
out.push(value);
}
};
let fenced: Vec<&str> = FENCE
.captures_iter(text)
.filter_map(|c| c.get(1).map(|m| m.as_str()))
.collect();
for blob in fenced.iter().rev() {
if let Ok(value) = serde_json::from_str::<Value>(blob) {
push(value);
}
}
let bytes = text.as_bytes();
for (opener, closer) in [(b'{', b'}'), (b'[', b']')] {
let mut end = rfind_byte(bytes, closer, bytes.len());
while let Some(e) = end {
let mut depth = 0i32;
let mut start = None;
for i in (0..=e).rev() {
if bytes[i] == closer {
depth += 1;
} else if bytes[i] == opener {
depth -= 1;
if depth == 0 {
start = Some(i);
break;
}
}
}
if let Some(s) = start {
if let Ok(value) = serde_json::from_str::<Value>(&text[s..=e]) {
push(value);
}
}
end = rfind_byte(bytes, closer, e);
}
}
out
}
pub fn looks_truncated(text: &str) -> bool {
let mut opened = 0i64;
let mut in_string = false;
let mut escaped = false;
for c in text.chars() {
if escaped {
escaped = false;
continue;
}
match c {
'\\' if in_string => escaped = true,
'"' => in_string = !in_string,
'{' if !in_string => opened += 1,
'}' if !in_string => opened -= 1,
_ => {}
}
}
opened > 0
}
pub fn extract_into<T: serde::de::DeserializeOwned>(text: &str) -> Result<T> {
let found = candidates(text);
if found.is_empty() {
return Err(SparError::new(if looks_truncated(text) {
format!(
"the response was cut off before any complete JSON:\n{}",
head(text, 400)
)
} else {
format!("no JSON found in response:\n{}", head(text, 800))
}));
}
let mut last_error = None;
for value in &found {
match serde_json::from_value::<T>(value.clone()) {
Ok(parsed) => return Ok(parsed),
Err(e) => last_error = Some((e, value)),
}
}
let (error, value) = last_error.expect("non-empty");
if looks_truncated(text) {
return Err(SparError::new(format!(
"the response was cut off before the answer was complete, so only fragments of it \
parsed ({error}). Ask for less in one go, or give this agent a CLI flag for native \
structured output."
)));
}
Err(SparError::new(format!(
"response did not match the expected shape ({error}).\nGot: {}",
head(&value.to_string(), 600)
)))
}
fn rfind_byte(haystack: &[u8], needle: u8, before: usize) -> Option<usize> {
haystack[..before.min(haystack.len())]
.iter()
.rposition(|b| *b == needle)
}
fn head(text: &str, max: usize) -> String {
text.chars().take(max).collect()
}
pub fn finding_key(title: &str, file: &str) -> String {
let basis: String = format!("{} {}", title.trim(), file.trim())
.to_lowercase()
.chars()
.filter(|c| c.is_ascii_alphanumeric() || matches!(c, ' ' | '/' | '.' | '_' | '-'))
.collect();
let basis = basis.split_whitespace().collect::<Vec<_>>().join(" ");
let digest = Sha256::digest(basis.as_bytes());
let hex: String = digest.iter().map(|b| format!("{b:02x}")).collect();
hex[..12].to_string()
}
#[cfg(test)]
mod tests {
use super::*;
#[test]
fn bare_object() {
assert_eq!(
serde_json::json!({"a": 1}),
extract_json(r#"{"a": 1}"#).unwrap()
);
}
#[test]
fn fenced_block() {
let out = extract_json("here you go:\n```json\n{\"a\": 1}\n```\n").unwrap();
assert_eq!(serde_json::json!({"a": 1}), out);
}
#[test]
fn trailing_prose() {
let out = extract_json("Thoughts...\n{\"verdict\": \"approve\"}\nDone.").unwrap();
assert_eq!(serde_json::json!({"verdict": "approve"}), out);
}
#[test]
fn picks_the_last_fenced_block() {
let text = "```json\n{\"n\": 1}\n```\nrevised:\n```json\n{\"n\": 2}\n```";
assert_eq!(serde_json::json!({"n": 2}), extract_json(text).unwrap());
}
#[test]
fn nested_braces() {
let payload = r#"{"findings": [{"severity": "nit", "d": {"x": [1, 2]}}]}"#;
let out = extract_json(&format!("blah {payload} blah")).unwrap();
assert_eq!(1, out["findings"].as_array().unwrap().len());
}
#[test]
fn top_level_array() {
let out = extract_json("result: [1, 2, 3]").unwrap();
assert_eq!(3, out.as_array().unwrap().len());
}
#[test]
fn multibyte_prose_around_the_payload_does_not_panic() {
let out = extract_json("\u{1f600}\u{1f600} {\"a\": 1} \u{1f600}").unwrap();
assert_eq!(serde_json::json!({"a": 1}), out);
}
#[test]
fn raises_when_there_is_none() {
assert!(extract_json("no json here at all").is_err());
}
#[test]
fn raises_on_empty() {
assert!(extract_json(" ").is_err());
}
#[test]
fn malformed_trailing_object_falls_back_to_an_earlier_one() {
let text = "{\"good\": true}\nthen: {\"bad\": ,}";
assert_eq!(
serde_json::json!({"good": true}),
extract_json(text).unwrap()
);
}
#[test]
fn key_is_stable_across_wording_noise() {
assert_eq!(
finding_key("Unbounded loop!", "src/x.rs"),
finding_key("unbounded loop", "src/x.rs")
);
}
#[test]
fn key_differs_by_file() {
assert_ne!(finding_key("t", "a.rs"), finding_key("t", "b.rs"));
}
#[test]
fn key_is_case_insensitive_in_the_path_too() {
assert_eq!(
finding_key("t", "src/Main.rs"),
finding_key("t", "src/main.rs")
);
}
#[test]
fn key_is_stable_across_whitespace() {
assert_eq!(finding_key("a b", "x.rs"), finding_key(" a b ", "x.rs"));
}
#[test]
fn key_is_twelve_hex_characters() {
let key = finding_key("anything", "file.rs");
assert_eq!(12, key.len());
assert!(key.chars().all(|c| c.is_ascii_hexdigit()));
}
}
#[cfg(test)]
mod truncation_tests {
use super::*;
use serde::Deserialize;
#[derive(Debug, Deserialize)]
struct Review {
verdict: String,
findings: Vec<Finding>,
}
#[derive(Debug, Deserialize)]
struct Finding {
title: String,
}
const TRUNCATED: &str = r#"Here is my review.
{"verdict":"changes_requested","next_action":"hand_back","summary":"Two problems.",
"findings":[
{"severity":"blocking","title":"First","detail":"one","file":"a.ts","in_scope":true},
{"severity":"non-blocking","title":"Second","detail":"numbers unnamed keys by position"#;
#[test]
fn a_truncated_review_is_reported_as_truncated_not_as_the_wrong_shape() {
let err = extract_into::<Review>(TRUNCATED).unwrap_err().to_string();
assert!(err.contains("cut off"), "{err}");
assert!(!err.contains("did not match the expected shape"), "{err}");
}
#[test]
fn truncation_is_detected_from_the_unclosed_braces() {
assert!(looks_truncated(TRUNCATED));
assert!(!looks_truncated(r#"{"a":1}"#));
assert!(!looks_truncated(r#"{"a":"a { in a string"}"#));
assert!(!looks_truncated(r#"{"a":"an escaped \" quote { here"}"#));
}
#[test]
fn the_review_is_found_even_with_nested_objects_after_it() {
let text = r#"Thinking out loud first.
{"verdict":"approve","next_action":"merge","summary":"Fine.","findings":[{"severity":"nit","title":"Wording","detail":"d","file":"a.ts","in_scope":true}]}
And here is a stray object afterwards: {"title":"not the review"}"#;
let review: Review = extract_into(text).unwrap();
assert_eq!("approve", review.verdict);
assert_eq!(1, review.findings.len());
assert_eq!("Wording", review.findings[0].title);
}
#[test]
fn candidates_are_offered_most_likely_first() {
let text = "```json\n{\"verdict\":\"approve\",\"findings\":[]}\n```\ntrailing {\"x\":1}";
let review: Review = extract_into(text).unwrap();
assert_eq!("approve", review.verdict);
}
#[test]
fn a_genuinely_wrong_shape_still_says_so() {
let err = extract_into::<Review>(r#"{"colour":"blue"}"#)
.unwrap_err()
.to_string();
assert!(err.contains("did not match the expected shape"), "{err}");
assert!(!err.contains("cut off"), "{err}");
}
#[test]
fn nothing_parseable_is_still_reported_plainly() {
let err = extract_into::<Review>("no json at all")
.unwrap_err()
.to_string();
assert!(err.contains("no JSON found"), "{err}");
}
}