git.stella-ops.org/src/Signals/StellaOps.Signals/Parsing/SimpleJsonCallgraphParser.cs

using System;
using System.Collections.Generic;
using System.IO;
using System.Text.Json;
using System.Threading;
using System.Threading.Tasks;
using StellaOps.Signals.Models;

namespace StellaOps.Signals.Parsing;

/// <summary>
/// Simple JSON-based callgraph parser used for initial language coverage.
/// </summary>
public sealed class SimpleJsonCallgraphParser : ICallgraphParser
{
    private readonly JsonSerializerOptions serializerOptions;

    public SimpleJsonCallgraphParser(string language)
    {
        ArgumentException.ThrowIfNullOrWhiteSpace(language);
        Language = language;
        serializerOptions = new JsonSerializerOptions
        {
            PropertyNameCaseInsensitive = true
        };
    }

    public string Language { get; }

    public async Task<CallgraphParseResult> ParseAsync(Stream artifactStream, CancellationToken cancellationToken)
    {
        ArgumentNullException.ThrowIfNull(artifactStream);

        using var document = await JsonDocument.ParseAsync(artifactStream, cancellationToken: cancellationToken).ConfigureAwait(false);
        var root = document.RootElement;

        if (TryParseLegacy(root, out var legacyResult))
        {
            return legacyResult;
        }

        if (TryParseSchemaV1(root, out var schemaResult))
        {
            return schemaResult;
        }

        throw new CallgraphParserValidationException("Callgraph artifact payload is empty or missing required fields.");
    }

    private static bool TryParseLegacy(JsonElement root, out CallgraphParseResult result)
    {
        result = default!;

        if (!root.TryGetProperty("graph", out var graphElement))
        {
            return false;
        }

        var nodesElement = graphElement.GetProperty("nodes");
        var edgesElement = graphElement.TryGetProperty("edges", out var edgesValue) ? edgesValue : default;

        var nodes = new List<CallgraphNode>(nodesElement.GetArrayLength());
        foreach (var nodeElement in nodesElement.EnumerateArray())
        {
            var id = nodeElement.GetProperty("id").GetString();
            if (string.IsNullOrWhiteSpace(id))
            {
                throw new CallgraphParserValidationException("Callgraph node is missing an id.");
            }

            nodes.Add(new CallgraphNode(
                Id: id.Trim(),
                Name: nodeElement.TryGetProperty("name", out var nameEl) ? nameEl.GetString() ?? id.Trim() : id.Trim(),
                Kind: nodeElement.TryGetProperty("kind", out var kindEl) ? kindEl.GetString() ?? "function" : "function",
                Namespace: nodeElement.TryGetProperty("namespace", out var nsEl) ? nsEl.GetString() : null,
                File: nodeElement.TryGetProperty("file", out var fileEl) ? fileEl.GetString() : null,
                Line: nodeElement.TryGetProperty("line", out var lineEl) && lineEl.ValueKind == JsonValueKind.Number ? lineEl.GetInt32() : null));
        }

        var edges = new List<CallgraphEdge>();
        if (edgesElement.ValueKind == JsonValueKind.Array)
        {
            foreach (var edgeElement in edgesElement.EnumerateArray())
            {
                var source = edgeElement.GetProperty("source").GetString();
                var target = edgeElement.GetProperty("target").GetString();
                if (string.IsNullOrWhiteSpace(source) || string.IsNullOrWhiteSpace(target))
                {
                    throw new CallgraphParserValidationException("Callgraph edge requires both source and target.");
                }

                var type = edgeElement.TryGetProperty("type", out var typeEl) ? typeEl.GetString() ?? "call" : "call";
                edges.Add(new CallgraphEdge(source.Trim(), target.Trim(), type));
            }
        }

        var formatVersion = root.TryGetProperty("formatVersion", out var versionEl)
            ? versionEl.GetString()
            : null;

        result = new CallgraphParseResult(nodes, edges, string.IsNullOrWhiteSpace(formatVersion) ? "1.0" : formatVersion!.Trim());
        return true;
    }

    private static bool TryParseSchemaV1(JsonElement root, out CallgraphParseResult result)
    {
        result = default!;

        if (!root.TryGetProperty("nodes", out var nodesElement) && !root.TryGetProperty("edges", out _))
        {
            return false;
        }

        var nodes = new List<CallgraphNode>();
        if (nodesElement.ValueKind == JsonValueKind.Array)
        {
            foreach (var nodeElement in nodesElement.EnumerateArray())
            {
                var id = nodeElement.TryGetProperty("sid", out var sidEl) ? sidEl.GetString() : nodeElement.GetProperty("id").GetString();
                if (string.IsNullOrWhiteSpace(id))
                {
                    throw new CallgraphParserValidationException("Callgraph node is missing an id.");
                }

                nodes.Add(new CallgraphNode(
                    Id: id.Trim(),
                    Name: nodeElement.TryGetProperty("name", out var nameEl) ? nameEl.GetString() ?? id.Trim() : id.Trim(),
                    Kind: nodeElement.TryGetProperty("kind", out var kindEl) ? kindEl.GetString() ?? "function" : "function",
                    Namespace: nodeElement.TryGetProperty("namespace", out var nsEl) ? nsEl.GetString() : null,
                    File: nodeElement.TryGetProperty("file", out var fileEl) ? fileEl.GetString() : null,
                    Line: nodeElement.TryGetProperty("line", out var lineEl) && lineEl.ValueKind == JsonValueKind.Number ? lineEl.GetInt32() : null));
            }
        }

        if (!root.TryGetProperty("edges", out var edgesElement) || edgesElement.ValueKind != JsonValueKind.Array)
        {
            edgesElement = default;
        }

        var edges = new List<CallgraphEdge>();
        if (edgesElement.ValueKind == JsonValueKind.Array)
        {
            foreach (var edgeElement in edgesElement.EnumerateArray())
            {
                var from = edgeElement.TryGetProperty("from", out var fromEl) ? fromEl.GetString() : edgeElement.GetProperty("source").GetString();
                var to = edgeElement.TryGetProperty("to", out var toEl) ? toEl.GetString() : edgeElement.GetProperty("target").GetString();
                if (string.IsNullOrWhiteSpace(from) || string.IsNullOrWhiteSpace(to))
                {
                    throw new CallgraphParserValidationException("Callgraph edge requires both source and target.");
                }

                var kind = edgeElement.TryGetProperty("kind", out var kindEl)
                    ? kindEl.GetString() ?? "call"
                    : edgeElement.TryGetProperty("type", out var typeEl)
                        ? typeEl.GetString() ?? "call"
                        : "call";

                edges.Add(new CallgraphEdge(from.Trim(), to.Trim(), kind));
            }
        }

        if (nodes.Count == 0)
        {
            // When nodes are omitted (framework overlay), derive them from the referenced edges.
            var uniqueNodeIds = new HashSet<string>(StringComparer.Ordinal);
            foreach (var edge in edges)
            {
                uniqueNodeIds.Add(edge.SourceId);
                uniqueNodeIds.Add(edge.TargetId);
            }

            foreach (var nodeId in uniqueNodeIds)
            {
                nodes.Add(new CallgraphNode(nodeId, nodeId, "function", null, null, null));
            }
        }

        var schemaVersion = root.TryGetProperty("schema_version", out var schemaEl)
            ? schemaEl.GetString()
            : "1.0";

        result = new CallgraphParseResult(nodes, edges, string.IsNullOrWhiteSpace(schemaVersion) ? "1.0" : schemaVersion!.Trim());
        return true;
    }

}