feat(worker): ClaudeHistoryReader distills session logs
This commit is contained in:
mika kuns
131
src/ClaudeDo.Worker/Report/ClaudeHistoryReader.cs
Normal file
131
src/ClaudeDo.Worker/Report/ClaudeHistoryReader.cs
Normal file
@@ -0,0 +1,131 @@
|
||||
using System.Text.Json;
|
||||
using ClaudeDo.Worker.Report.Interfaces;
|
||||
|
||||
namespace ClaudeDo.Worker.Report;
|
||||
|
||||
public sealed class ClaudeHistoryReader : IClaudeHistoryReader
|
||||
{
|
||||
private readonly string _projectsRoot;
|
||||
|
||||
public ClaudeHistoryReader(string projectsRoot) => _projectsRoot = projectsRoot;
|
||||
|
||||
public Task<IReadOnlyList<RepoActivity>> ReadAsync(
|
||||
DateOnly start, DateOnly end, IReadOnlyList<string> excludedPrefixes, CancellationToken ct = default)
|
||||
{
|
||||
var buckets = new Dictionary<(string Repo, DateOnly Date), DayActivity>();
|
||||
var normalizedExcludes = excludedPrefixes
|
||||
.Select(NormalizePath).Where(p => p.Length > 0).ToArray();
|
||||
|
||||
if (Directory.Exists(_projectsRoot))
|
||||
{
|
||||
foreach (var file in Directory.EnumerateFiles(_projectsRoot, "*.jsonl", SearchOption.AllDirectories))
|
||||
{
|
||||
ct.ThrowIfCancellationRequested();
|
||||
ReadFile(file, start, end, normalizedExcludes, buckets);
|
||||
}
|
||||
}
|
||||
|
||||
var repos = buckets
|
||||
.GroupBy(kv => kv.Key.Repo)
|
||||
.Select(g =>
|
||||
{
|
||||
var ra = new RepoActivity { RepoPath = g.Key };
|
||||
foreach (var day in g.OrderBy(kv => kv.Key.Date).Select(kv => kv.Value))
|
||||
ra.Days.Add(day);
|
||||
return ra;
|
||||
})
|
||||
.OrderBy(r => r.RepoPath)
|
||||
.ToList();
|
||||
|
||||
return Task.FromResult<IReadOnlyList<RepoActivity>>(repos);
|
||||
}
|
||||
|
||||
private static void ReadFile(
|
||||
string file, DateOnly start, DateOnly end, string[] excludes,
|
||||
Dictionary<(string, DateOnly), DayActivity> buckets)
|
||||
{
|
||||
string? lastAssistantText = null;
|
||||
string? lastAssistantRepo = null;
|
||||
DateOnly lastAssistantDate = default;
|
||||
|
||||
foreach (var line in File.ReadLines(file))
|
||||
{
|
||||
if (string.IsNullOrWhiteSpace(line)) continue;
|
||||
JsonDocument doc;
|
||||
try { doc = JsonDocument.Parse(line); }
|
||||
catch (JsonException) { continue; }
|
||||
using (doc)
|
||||
{
|
||||
var root = doc.RootElement;
|
||||
if (root.ValueKind != JsonValueKind.Object) continue;
|
||||
if (!root.TryGetProperty("type", out var typeEl)) continue;
|
||||
var type = typeEl.GetString();
|
||||
if (type is not ("user" or "assistant")) continue;
|
||||
if (!root.TryGetProperty("cwd", out var cwdEl) || cwdEl.ValueKind != JsonValueKind.String) continue;
|
||||
var cwd = cwdEl.GetString()!;
|
||||
if (IsExcluded(cwd, excludes)) continue;
|
||||
if (!root.TryGetProperty("timestamp", out var tsEl) ||
|
||||
!DateTimeOffset.TryParse(tsEl.GetString(), out var ts)) continue;
|
||||
var date = DateOnly.FromDateTime(ts.LocalDateTime);
|
||||
if (date < start || date > end) continue;
|
||||
|
||||
var text = ExtractText(root);
|
||||
if (string.IsNullOrWhiteSpace(text)) continue;
|
||||
|
||||
if (type == "user")
|
||||
{
|
||||
if (text.Contains("<system-reminder>", StringComparison.OrdinalIgnoreCase)) continue;
|
||||
Bucket(buckets, cwd, date).Prompts.Add(text.Trim());
|
||||
}
|
||||
else
|
||||
{
|
||||
lastAssistantText = text.Trim();
|
||||
lastAssistantRepo = cwd;
|
||||
lastAssistantDate = date;
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
if (lastAssistantText is not null && lastAssistantRepo is not null)
|
||||
Bucket(buckets, lastAssistantRepo, lastAssistantDate).Summaries.Add(lastAssistantText);
|
||||
}
|
||||
|
||||
private static DayActivity Bucket(
|
||||
Dictionary<(string, DateOnly), DayActivity> buckets, string repo, DateOnly date)
|
||||
{
|
||||
var key = (repo, date);
|
||||
if (!buckets.TryGetValue(key, out var day))
|
||||
{
|
||||
day = new DayActivity { Date = date };
|
||||
buckets[key] = day;
|
||||
}
|
||||
return day;
|
||||
}
|
||||
|
||||
private static string ExtractText(JsonElement root)
|
||||
{
|
||||
if (!root.TryGetProperty("message", out var msg) ||
|
||||
!msg.TryGetProperty("content", out var content)) return "";
|
||||
if (content.ValueKind == JsonValueKind.String) return content.GetString() ?? "";
|
||||
if (content.ValueKind != JsonValueKind.Array) return "";
|
||||
|
||||
var parts = new List<string>();
|
||||
foreach (var item in content.EnumerateArray())
|
||||
{
|
||||
if (item.ValueKind != JsonValueKind.Object) continue;
|
||||
if (item.TryGetProperty("type", out var t) && t.GetString() == "text" &&
|
||||
item.TryGetProperty("text", out var txt) && txt.ValueKind == JsonValueKind.String)
|
||||
parts.Add(txt.GetString() ?? "");
|
||||
}
|
||||
return string.Join("\n", parts);
|
||||
}
|
||||
|
||||
private static bool IsExcluded(string cwd, string[] excludes)
|
||||
{
|
||||
var norm = NormalizePath(cwd);
|
||||
return excludes.Any(p => norm.StartsWith(p, StringComparison.Ordinal));
|
||||
}
|
||||
|
||||
private static string NormalizePath(string p) =>
|
||||
(p ?? "").Replace('/', '\\').TrimEnd('\\').ToLowerInvariant();
|
||||
}
|
||||
@@ -0,0 +1,78 @@
|
||||
using ClaudeDo.Worker.Report;
|
||||
|
||||
namespace ClaudeDo.Worker.Tests.Report;
|
||||
|
||||
public class ClaudeHistoryReaderTests : IDisposable
|
||||
{
|
||||
private readonly string _root;
|
||||
public ClaudeHistoryReaderTests()
|
||||
{
|
||||
_root = Path.Combine(Path.GetTempPath(), $"cdh_{Guid.NewGuid():N}");
|
||||
Directory.CreateDirectory(_root);
|
||||
}
|
||||
public void Dispose() { try { Directory.Delete(_root, true); } catch { } }
|
||||
|
||||
private void WriteSession(string projectDir, string file, params string[] lines)
|
||||
{
|
||||
var dir = Path.Combine(_root, projectDir);
|
||||
Directory.CreateDirectory(dir);
|
||||
File.WriteAllLines(Path.Combine(dir, file), lines);
|
||||
}
|
||||
|
||||
private static string UserLine(string cwd, string ts, string text) =>
|
||||
$$$"""{"type":"user","cwd":{{{Json(cwd)}}},"timestamp":"{{{ts}}}","message":{"role":"user","content":[{"type":"text","text":{{{Json(text)}}}}]}}""";
|
||||
|
||||
private static string AssistantLine(string cwd, string ts, string text) =>
|
||||
$$$"""{"type":"assistant","cwd":{{{Json(cwd)}}},"timestamp":"{{{ts}}}","message":{"role":"assistant","content":[{"type":"text","text":{{{Json(text)}}}}]}}""";
|
||||
|
||||
private static string Json(string s) => System.Text.Json.JsonSerializer.Serialize(s);
|
||||
|
||||
[Fact]
|
||||
public async Task Extracts_Prompts_And_Last_Assistant_Summary_GroupedByRepoAndDay()
|
||||
{
|
||||
WriteSession("proj", "s1.jsonl",
|
||||
UserLine(@"C:\Dev\Repos\App", "2026-06-01T08:00:00Z", "Add login"),
|
||||
AssistantLine(@"C:\Dev\Repos\App", "2026-06-01T08:05:00Z", "first summary"),
|
||||
AssistantLine(@"C:\Dev\Repos\App", "2026-06-01T08:30:00Z", "final summary"));
|
||||
|
||||
var reader = new ClaudeHistoryReader(_root);
|
||||
var result = await reader.ReadAsync(new DateOnly(2026, 6, 1), new DateOnly(2026, 6, 3),
|
||||
Array.Empty<string>());
|
||||
|
||||
var repo = Assert.Single(result);
|
||||
Assert.Equal(@"C:\Dev\Repos\App", repo.RepoPath);
|
||||
var day = Assert.Single(repo.Days);
|
||||
Assert.Equal(new[] { "Add login" }, day.Prompts);
|
||||
Assert.Equal(new[] { "final summary" }, day.Summaries);
|
||||
}
|
||||
|
||||
[Fact]
|
||||
public async Task Drops_Sessions_Under_Excluded_Prefix_CaseInsensitive()
|
||||
{
|
||||
WriteSession("priv", "s.jsonl",
|
||||
UserLine(@"C:\Private\Secret", "2026-06-01T08:00:00Z", "private work"));
|
||||
|
||||
var reader = new ClaudeHistoryReader(_root);
|
||||
var result = await reader.ReadAsync(new DateOnly(2026, 6, 1), new DateOnly(2026, 6, 3),
|
||||
new[] { @"c:\private" });
|
||||
|
||||
Assert.Empty(result);
|
||||
}
|
||||
|
||||
[Fact]
|
||||
public async Task Filters_By_Date_Window_And_Skips_Noise_And_Malformed()
|
||||
{
|
||||
WriteSession("proj", "s.jsonl",
|
||||
"this is not json",
|
||||
UserLine(@"C:\Dev\App", "2026-05-01T08:00:00Z", "too old"),
|
||||
UserLine(@"C:\Dev\App", "2026-06-02T08:00:00Z", "in range"),
|
||||
UserLine(@"C:\Dev\App", "2026-06-02T09:00:00Z", "noise <system-reminder> blah"));
|
||||
|
||||
var reader = new ClaudeHistoryReader(_root);
|
||||
var result = await reader.ReadAsync(new DateOnly(2026, 6, 1), new DateOnly(2026, 6, 3),
|
||||
Array.Empty<string>());
|
||||
|
||||
var day = Assert.Single(Assert.Single(result).Days);
|
||||
Assert.Equal(new[] { "in range" }, day.Prompts);
|
||||
}
|
||||
}
|
||||
Reference in New Issue
Block a user