gitdataai/libs/service/file_tools/json.rs

//! read_json — parse, validate, and query JSON / JSONC files.

use crate::file_tools::MAX_FILE_SIZE;
use crate::git_tools::ctx::GitToolCtx;
use agent::{ToolDefinition, ToolHandler, ToolParam, ToolRegistry, ToolSchema};
use serde_json::Value as JsonValue;
use std::collections::HashMap;

/// Remove comments from JSONC (lines starting with // or /* */) for parsing.
fn strip_jsonc_comments(input: &str) -> String {
    let mut result = String::with_capacity(input.len());
    let mut chars = input.chars().peekable();
    let mut in_string = false;
    let mut escaped = false;

    while let Some(c) = chars.next() {
        if escaped {
            result.push(c);
            escaped = false;
            continue;
        }

        if c == '\\' && in_string {
            result.push(c);
            escaped = true;
            continue;
        }

        if c == '"' {
            result.push(c);
            in_string = !in_string;
            continue;
        }

        if !in_string {
            if c == '/' {
                if let Some(&next) = chars.peek() {
                    if next == '/' {
                        // Line comment — skip to end of line
                        chars.next();
                        while let Some(nc) = chars.next() {
                            if nc == '\n' {
                                result.push(nc);
                                break;
                            }
                        }
                        continue;
                    } else if next == '*' {
                        // Block comment — skip until */
                        chars.next();
                        while let Some(nc) = chars.next() {
                            if nc == '*' {
                                if let Some(&'/') = chars.peek() {
                                    chars.next();
                                    break;
                                }
                            }
                        }
                        continue;
                    }
                }
            }
        }

        result.push(c);
    }

    result
}

fn infer_schema(value: &JsonValue, max_depth: usize) -> JsonValue {
    if max_depth == 0 {
        return serde_json::json!({ "type": "MAX_DEPTH" });
    }

    match value {
        JsonValue::Null => serde_json::json!({ "type": "null" }),
        JsonValue::Bool(_) => serde_json::json!({ "type": "boolean" }),
        JsonValue::Number(_) => serde_json::json!({ "type": "number" }),
        JsonValue::String(_) => serde_json::json!({ "type": "string" }),
        JsonValue::Array(arr) => {
            if arr.is_empty() {
                serde_json::json!({ "type": "array", "items": null })
            } else {
                serde_json::json!({
                    "type": "array",
                    "length": arr.len(),
                    "items": infer_schema(&arr[0], max_depth - 1)
                })
            }
        }
        JsonValue::Object(obj) => {
            let mut schema = serde_json::Map::new();
            schema.insert("type".into(), serde_json::Value::String("object".into()));
            let mut properties = serde_json::Map::new();
            for (k, v) in obj {
                properties.insert(k.clone(), infer_schema(v, max_depth - 1));
            }
            schema.insert("properties".into(), serde_json::Value::Object(properties));
            schema.insert("keyCount".into(), serde_json::json!(obj.len()));
            serde_json::Value::Object(schema)
        }
    }
}

async fn read_json_exec(
    ctx: GitToolCtx,
    args: serde_json::Value,
) -> Result<serde_json::Value, String> {
    let p: serde_json::Map<String, serde_json::Value> =
        serde_json::from_value(args).map_err(|e| e.to_string())?;

    let project_name = p
        .get("project_name")
        .and_then(|v| v.as_str())
        .ok_or("missing project_name")?;
    let repo_name = p
        .get("repo_name")
        .and_then(|v| v.as_str())
        .ok_or("missing repo_name")?;
    let path = p.get("path").and_then(|v| v.as_str()).ok_or("missing path")?;
    let rev = p
        .get("rev")
        .and_then(|v| v.as_str())
        .map(String::from)
        .unwrap_or_else(|| "HEAD".to_string());
    let query = p.get("query").and_then(|v| v.as_str()).map(String::from);
    let max_depth = p.get("schema_depth").and_then(|v| v.as_u64()).unwrap_or(4) as usize;
    let pretty = p.get("pretty").and_then(|v| v.as_bool()).unwrap_or(false);

    let domain = ctx.open_repo(project_name, repo_name).await?;

    let commit_oid = if rev.len() >= 40 {
        git::commit::types::CommitOid::new(&rev)
    } else {
        domain
            .commit_get_prefix(&rev)
            .map_err(|e| e.to_string())?
            .oid
    };

    let entry = domain
        .tree_entry_by_path_from_commit(&commit_oid, path)
        .map_err(|e| e.to_string())?;
    let content = domain.blob_content(&entry.oid).map_err(|e| e.to_string())?;

    let data = &content.content;
    if data.len() > MAX_FILE_SIZE {
        return Err(format!(
            "file too large ({} bytes), max {} bytes",
            data.len(),
            MAX_FILE_SIZE
        ));
    }

    let text = String::from_utf8_lossy(data);
    // Only treat as JSONC if the extension indicates it, or if we can
    // confirm a comment-like pattern outside of a string context.
    let is_jsonc = path.ends_with(".jsonc");

    let json_text = if is_jsonc {
        strip_jsonc_comments(&text)
    } else {
        text.to_string()
    };

    let parsed: JsonValue = serde_json::from_str(&json_text)
        .map_err(|e| format!("JSON parse error at {}: {}", e.line(), e))?;

    // Apply JSONPath-like query
    let result = if let Some(ref q) = query {
        query_json(&parsed, q)?
    } else {
        parsed
    };

    let schema = infer_schema(&result, max_depth);

    let display = if pretty {
        serde_json::to_string_pretty(&result).unwrap_or_default()
    } else {
        serde_json::to_string(&result).unwrap_or_default()
    };

    Ok(serde_json::json!({
        "path": path,
        "rev": rev,
        "format": if is_jsonc { "jsonc" } else { "json" },
        "size_bytes": data.len(),
        "schema": schema,
        "data": if display.chars().count() > 5000 {
            let truncated: String = display.chars().take(5000).collect();
            format!("{}... (truncated, {} chars total)", truncated, display.chars().count())
        } else { display },
    }))
}

/// Simple JSONPath-like query support.
/// Supports: $.key, $[0], $.key.nested, $.arr[0].field
/// Bracket notation ["key.with.dots"] allows accessing keys containing dots.
fn query_json(value: &JsonValue, query: &str) -> Result<JsonValue, String> {
    let query = query.trim();
    let query = if query.starts_with("$.") {
        &query[2..]
    } else if query.starts_with('$') && query.len() > 1 {
        &query[1..]
    } else {
        query
    };

    let mut current = value.clone();

    // Parse into access segments: Key("name"), Index(0), BracketKey("key.with.dots")
    enum Segment { Key(String), Index(usize), BracketKey(String) }
    let mut segments: Vec<Segment> = Vec::new();
    let mut i = 0;
    let q_chars: Vec<char> = query.chars().collect();
    while i < q_chars.len() {
        if q_chars[i] == '[' {
            // Find matching ]
            let mut j = i + 1;
            let mut bracket_content = String::new();
            while j < q_chars.len() && q_chars[j] != ']' {
                bracket_content.push(q_chars[j]);
                j += 1;
            }
            if j < q_chars.len() && q_chars[j] == ']' {
                let content = bracket_content.trim();
                // Check if it's a quoted string key or a numeric index
                if content.starts_with('"') && content.ends_with('"') {
                    let key = content[1..content.len()-1].to_string();
                    segments.push(Segment::BracketKey(key));
                } else if content.starts_with("'") && content.ends_with("'") {
                    let key = content[1..content.len()-1].to_string();
                    segments.push(Segment::BracketKey(key));
                } else if let Ok(idx) = content.parse::<usize>() {
                    segments.push(Segment::Index(idx));
                } else {
                    return Err(format!("Invalid bracket notation: [{}]", content));
                }
                i = j + 1;
                // Skip dot after bracket if present
                if i < q_chars.len() && q_chars[i] == '.' {
                    i += 1;
                }
            } else {
                return Err("Unmatched [ in query".into());
            }
        } else {
            // Read key until . or [
            let mut key = String::new();
            while i < q_chars.len() && q_chars[i] != '.' && q_chars[i] != '[' {
                key.push(q_chars[i]);
                i += 1;
            }
            if !key.is_empty() {
                // Check if key contains a numeric-only segment (array index shorthand)
                segments.push(Segment::Key(key));
            }
            if i < q_chars.len() && q_chars[i] == '.' {
                i += 1;
            }
        }
    }

    for seg in &segments {
        match seg {
            Segment::Key(key) | Segment::BracketKey(key) => {
                if let JsonValue::Object(obj) = &current {
                    current = obj.get(key).cloned().unwrap_or(JsonValue::Null);
                } else {
                    return Err(format!("cannot access property '{}' on non-object", key));
                }
            }
            Segment::Index(idx) => {
                if let JsonValue::Array(arr) = &current {
                    current = arr.get(*idx).cloned().unwrap_or(JsonValue::Null);
                } else {
                    return Err(format!("index {} on non-array", idx));
                }
            }
        }
    }

    Ok(current)
}

pub fn register_json_tools(registry: &mut ToolRegistry) {
    let p = HashMap::from([
        ("project_name".into(), ToolParam { name: "project_name".into(), param_type: "string".into(), description: Some("Project name (slug)".into()), required: true, properties: None, items: None }),
        ("repo_name".into(), ToolParam { name: "repo_name".into(), param_type: "string".into(), description: Some("Repository name".into()), required: true, properties: None, items: None }),
        ("path".into(), ToolParam { name: "path".into(), param_type: "string".into(), description: Some("File path to the JSON or JSONC file".into()), required: true, properties: None, items: None }),
        ("rev".into(), ToolParam { name: "rev".into(), param_type: "string".into(), description: Some("Git revision (default: HEAD)".into()), required: false, properties: None, items: None }),
        ("query".into(), ToolParam { name: "query".into(), param_type: "string".into(), description: Some("JSONPath-like query (e.g. $.config.items[0].name) to extract a subset of the document".into()), required: false, properties: None, items: None }),
        ("schema_depth".into(), ToolParam { name: "schema_depth".into(), param_type: "integer".into(), description: Some("How deep to infer the JSON schema (default: 4)".into()), required: false, properties: None, items: None }),
        ("pretty".into(), ToolParam { name: "pretty".into(), param_type: "boolean".into(), description: Some("Pretty-print the output (default: false)".into()), required: false, properties: None, items: None }),
    ]);
    let schema = ToolSchema { schema_type: "object".into(), properties: Some(p), required: Some(vec!["project_name".into(), "repo_name".into(), "path".into()]) };
    registry.register(
        ToolDefinition::new("read_json")
            .description("Parse, validate, and query JSON and JSONC files. Supports JSONPath-like queries ($.key, $.arr[0]), schema inference, and pretty-printing. Automatically detects JSONC (with // comments).")
            .parameters(schema),
        ToolHandler::new(|ctx, args| {
            let gctx = GitToolCtx::new(ctx);
            Box::pin(async move {
                read_json_exec(gctx, args).await.map_err(agent::ToolError::ExecutionError)
            })
        }),
    );
}