magi-code 0.100.0

//! Provider-only evidence derived from structured local reads, never from disk.

use serde::{Deserialize, Serialize};
use serde_json::Value;

use crate::tools::skill_provenance::digest_hex;

#[derive(Debug, Clone, Serialize, Deserialize, PartialEq, Eq)]
pub(crate) struct FileReadProvenance {
    pub(crate) identity: String,
    pub(crate) tag: String,
    pub(crate) start: usize,
    pub(crate) end: usize,
    pub(crate) sha256: String,
}

impl FileReadProvenance {
    pub(crate) fn validates(&self, output: &str) -> bool {
        self.tag.len() == 4
            && self.tag.bytes().all(|byte| byte.is_ascii_hexdigit())
            && self.start < self.end
            && output
                .get(self.start..self.end)
                .is_some_and(|body| digest_hex(body.as_bytes()) == self.sha256)
    }
}

pub(crate) fn validated_file_reads(
    tool_name: &str,
    success: bool,
    output: &str,
    metadata: &Value,
) -> Vec<FileReadProvenance> {
    if tool_name != "read" || !success {
        return Vec::new();
    }
    let Some(files) = metadata
        .pointer("/structured_result/files")
        .and_then(Value::as_array)
    else {
        return Vec::new();
    };
    if files.len() > 8 {
        return Vec::new();
    }
    let summaries = metadata.get("results").and_then(Value::as_array);
    let mut cursor = 0;
    let mut reads = Vec::new();
    for (index, file) in files.iter().enumerate() {
        let summary = match summaries {
            Some(summaries) => summaries.get(index).unwrap_or(&Value::Null),
            None => metadata,
        };
        let Some(text) = file.get("text").and_then(Value::as_str) else {
            continue;
        };
        // Match the actual section boundary, never a substring in another file's body.
        if summaries.is_some() {
            let Some(path) = file.get("path").and_then(Value::as_str) else {
                break;
            };
            let header = format!(
                "{}--- FILE: {path} ---\n",
                if index == 0 { "" } else { "\n\n" }
            );
            if !output
                .get(cursor..)
                .is_some_and(|tail| tail.starts_with(&header))
            {
                break;
            }
            cursor += header.len();
        }
        if !output
            .get(cursor..)
            .is_some_and(|tail| tail.starts_with(text))
        {
            break;
        }
        let start = cursor;
        cursor += text.len();
        if file.get("success") != Some(&Value::Bool(true))
            || file.get("complete") != Some(&Value::Bool(true))
            || file.get("request_complete") != Some(&Value::Bool(true))
            || summary.get("kind").and_then(Value::as_str) != Some("file")
            || summary.get("raw") != Some(&Value::Bool(false))
        {
            continue;
        }
        let Some(path) = summary.get("path").and_then(Value::as_str) else {
            continue;
        };
        let Some(tag) = summary.get("hashline_tag").and_then(Value::as_str) else {
            continue;
        };
        if path.len() > 4096 || path.chars().any(char::is_control) {
            continue;
        }
        let header = format!("[{path}#{tag}]\n");
        let Some(body) = text.strip_prefix(&header) else {
            continue;
        };
        let Some(selector) = summary.get("selector") else {
            continue;
        };
        let identity = serde_json::json!([path, selector, tag]).to_string();
        if identity.len() > 8192 {
            continue;
        }
        let read = FileReadProvenance {
            identity: digest_hex(identity.as_bytes()),
            tag: tag.to_owned(),
            start: start + header.len(),
            end: cursor,
            sha256: digest_hex(body.as_bytes()),
        };
        if read.validates(output) {
            reads.push(read);
        }
    }
    reads
}