Files
wx-win-agent/node-agent/WxAgent.Windows/WechatMessageDbReader.cs
T

261 lines
16 KiB
C#

using System.Globalization;
using WxAgent.Core;
namespace WxAgent.Windows;
public sealed record DbMessageIncrementalPage(
IReadOnlyList<DbMessage> Messages,
IReadOnlyDictionary<string, long> NextCursors,
IReadOnlyList<string> UnavailableDatabases,
bool IsComplete,
int NewMessageCount);
/// <summary>Reads chat messages from the account's SQLCipher message databases using cached verified keys.</summary>
public static class WechatMessageDbReader
{
public static async Task<IReadOnlyList<DbMessage>> ReadAsync(
string accountRootPath, IReadOnlyList<DatabaseKeyEvidence> databases, string chatId, int limit, CancellationToken cancellationToken, long? localId = null)
{
ArgumentException.ThrowIfNullOrWhiteSpace(accountRootPath);
ArgumentException.ThrowIfNullOrWhiteSpace(chatId);
if (limit is < 1 or > 500) throw new WxAgentException(WxAgentErrorCode.InvalidArgument, "Message limit must be between 1 and 500.");
// AccountRootPath points at the account's db_storage folder; the directory above it is <wxid>_<random>.
var accountRootDirectoryName = Path.GetFileName(Path.GetDirectoryName(Path.TrimEndingDirectorySeparator(Path.GetFullPath(accountRootPath)))) ?? string.Empty;
var tableName = WechatDbMessage.TableNameFor(chatId);
if (!tableName.StartsWith("Msg_", StringComparison.Ordinal) || tableName.Length != 36 || !tableName[4..].All(Uri.IsHexDigit))
{
throw new WxAgentException(WxAgentErrorCode.InvalidArgument, "Message table name could not be derived from the chat id.");
}
var messageDatabase = await FindMessageDatabaseAsync(accountRootPath, databases, tableName, cancellationToken).ConfigureAwait(false);
if (messageDatabase is null)
{
return Array.Empty<DbMessage>();
}
var sql = "SELECT m.local_id, m.server_id, m.local_type, m.create_time, " +
"hex(m.message_content) AS hex_content, m.WCDB_CT_message_content AS is_compressed, n.user_name AS sender_wxid " +
$"FROM \"{tableName}\" m LEFT JOIN Name2Id n ON m.real_sender_id = n.rowid " +
"WHERE ($localId IS NULL OR m.local_id = $localId) " +
"ORDER BY m.create_time DESC, m.local_id DESC LIMIT $limit;";
var rows = await SqlCipherDatabaseReader.QueryRowsAsync(messageDatabase.Value.Path, messageDatabase.Value.EncKey, sql,
[new KeyValuePair<string, object?>("$limit", (long)limit), new KeyValuePair<string, object?>("$localId", localId)], cancellationToken).ConfigureAwait(false);
var senders = rows.Select(row => row.GetValueOrDefault("sender_wxid") as string)
.Where(sender => !string.IsNullOrEmpty(sender)).Cast<string>().Distinct(StringComparer.Ordinal).ToArray();
var contacts = await ReadContactsAsync(accountRootPath, databases, senders, cancellationToken).ConfigureAwait(false);
return BuildMessages(rows, chatId, accountRootDirectoryName, contacts);
}
public static async Task<DbMessageIncrementalPage> ReadIncrementalAsync(
string accountRootPath,
IReadOnlyList<DatabaseKeyEvidence> databases,
IReadOnlyList<string> chatIds,
IReadOnlyDictionary<string, long> cursors,
int perChatLimit,
CancellationToken cancellationToken,
int overlapRows = 1)
{
ArgumentException.ThrowIfNullOrWhiteSpace(accountRootPath);
if (chatIds.Count == 0) return new DbMessageIncrementalPage([], cursors, [], true, 0);
if (perChatLimit is < 1 or > 500) throw new WxAgentException(WxAgentErrorCode.InvalidArgument, "Incremental message limit must be between 1 and 500.");
var accountRootDirectoryName = Path.GetFileName(Path.GetDirectoryName(Path.TrimEndingDirectorySeparator(Path.GetFullPath(accountRootPath)))) ?? string.Empty;
var candidates = databases
.Where(database => database.RelativePath.StartsWith("message/", StringComparison.OrdinalIgnoreCase)
&& database.RelativePath.EndsWith(".db", StringComparison.OrdinalIgnoreCase))
.OrderBy(database => database.RelativePath, StringComparer.OrdinalIgnoreCase)
.ToArray();
if (candidates.Length == 0)
return new DbMessageIncrementalPage([], cursors, ["message/*.db"], false, 0);
var rowsByChat = new List<(string ChatId, string RelativePath, IReadOnlyList<IReadOnlyDictionary<string, object?>> Rows)>();
var unavailable = new HashSet<string>(StringComparer.OrdinalIgnoreCase);
foreach (var chatId in chatIds.Distinct(StringComparer.Ordinal))
{
cancellationToken.ThrowIfCancellationRequested();
var tableName = WechatDbMessage.TableNameFor(chatId);
if (!tableName.StartsWith("Msg_", StringComparison.Ordinal) || tableName.Length != 36 || !tableName[4..].All(Uri.IsHexDigit))
throw new WxAgentException(WxAgentErrorCode.InvalidArgument, "Message table name could not be derived from the chat id.");
foreach (var database in candidates)
{
cancellationToken.ThrowIfCancellationRequested();
var key = CursorKey(chatId, database.RelativePath);
var cursor = cursors.GetValueOrDefault(key);
var queryCursor = cursor > 0 ? Math.Max(0, cursor - Math.Max(0, overlapRows)) : 0;
var path = DatabasePath(accountRootPath, database.RelativePath);
try
{
var tableRows = await SqlCipherDatabaseReader.QueryRowsAsync(path, database.EncKey,
"SELECT name FROM sqlite_master WHERE type = 'table' AND name = $name;",
[new KeyValuePair<string, object?>("$name", tableName)], cancellationToken).ConfigureAwait(false);
if (tableRows.Count == 0) continue;
var sql = "SELECT m.local_id, m.server_id, m.local_type, m.create_time, " +
"hex(m.message_content) AS hex_content, m.WCDB_CT_message_content AS is_compressed, n.user_name AS sender_wxid " +
$"FROM \"{tableName}\" m LEFT JOIN Name2Id n ON m.real_sender_id = n.rowid " +
$"WHERE m.local_id {(cursor > 0 ? ">=" : ">")} $afterLocalId ORDER BY m.local_id ASC LIMIT $limit;";
var rows = await SqlCipherDatabaseReader.QueryRowsAsync(path, database.EncKey, sql,
[new KeyValuePair<string, object?>("$afterLocalId", queryCursor), new KeyValuePair<string, object?>("$limit", (long)perChatLimit)], cancellationToken).ConfigureAwait(false);
rowsByChat.Add((chatId, database.RelativePath, rows));
}
catch (OperationCanceledException) { throw; }
catch (WxAgentException)
{
unavailable.Add(database.RelativePath);
}
}
}
var allSenders = rowsByChat.SelectMany(item => item.Rows)
.Select(row => row.GetValueOrDefault("sender_wxid") as string)
.Where(sender => !string.IsNullOrWhiteSpace(sender)).Cast<string>()
.Distinct(StringComparer.Ordinal).ToArray();
IReadOnlyDictionary<string, WechatContactName> contacts;
try
{
contacts = await ReadContactsAsync(accountRootPath, databases, allSenders, cancellationToken).ConfigureAwait(false);
}
catch (WxAgentException)
{
contacts = new Dictionary<string, WechatContactName>(StringComparer.Ordinal);
unavailable.Add("contact/contact.db");
}
var nextCursors = new Dictionary<string, long>(cursors, StringComparer.Ordinal);
var messages = new List<DbMessage>();
var newMessageCount = 0;
foreach (var group in rowsByChat)
{
messages.AddRange(BuildMessages(group.Rows, group.ChatId, accountRootDirectoryName, contacts));
var cursorKey = CursorKey(group.ChatId, group.RelativePath);
var previousCursor = cursors.GetValueOrDefault(cursorKey);
newMessageCount += group.Rows.Count(row => Convert.ToInt64(row.GetValueOrDefault("local_id") ?? 0L, CultureInfo.InvariantCulture) > previousCursor);
if (group.Rows.Count > 0)
{
var next = group.Rows.Max(row => Convert.ToInt64(row.GetValueOrDefault("local_id") ?? 0L, CultureInfo.InvariantCulture));
nextCursors[CursorKey(group.ChatId, group.RelativePath)] = Math.Max(nextCursors.GetValueOrDefault(CursorKey(group.ChatId, group.RelativePath)), next);
}
}
return new DbMessageIncrementalPage(messages.OrderBy(message => message.Timestamp).ThenBy(message => message.LocalId).ToArray(), nextCursors, unavailable.OrderBy(item => item, StringComparer.OrdinalIgnoreCase).ToArray(), unavailable.Count == 0, newMessageCount);
}
public static string CursorKey(string chatId, string databaseRelativePath) => $"{chatId}\u001f{databaseRelativePath.Replace('\\', '/')}";
private static string DatabasePath(string accountRootPath, string relativePath)
{
var root = Path.GetFullPath(accountRootPath).TrimEnd(Path.DirectorySeparatorChar, Path.AltDirectorySeparatorChar) + Path.DirectorySeparatorChar;
var relative = relativePath.Replace('/', Path.DirectorySeparatorChar);
var path = Path.GetFullPath(Path.Combine(root, relative));
if (Path.IsPathRooted(relative) || !path.StartsWith(root, StringComparison.OrdinalIgnoreCase))
throw new WxAgentException(WxAgentErrorCode.InvalidArgument, "The database path must remain inside the selected account root.");
return path;
}
private static IReadOnlyList<DbMessage> BuildMessages(
IReadOnlyList<IReadOnlyDictionary<string, object?>> rows,
string chatId,
string accountRootDirectoryName,
IReadOnlyDictionary<string, WechatContactName> contacts)
{
var messages = new List<DbMessage>(rows.Count);
foreach (var row in rows)
{
var senderWxId = row.GetValueOrDefault("sender_wxid") as string;
var content = WechatDbMessage.DecodeContent(
row.GetValueOrDefault("hex_content") as string ?? string.Empty,
Convert.ToInt64(row.GetValueOrDefault("is_compressed") ?? 0L, CultureInfo.InvariantCulture) != 0);
var contact = senderWxId is null ? null : contacts.GetValueOrDefault(senderWxId);
messages.Add(new DbMessage(
Convert.ToInt64(row.GetValueOrDefault("local_id") ?? 0L, CultureInfo.InvariantCulture),
Convert.ToInt64(row.GetValueOrDefault("server_id") ?? 0L, CultureInfo.InvariantCulture),
chatId,
senderWxId,
contact?.Name,
contact?.AvatarUrl,
Convert.ToInt64(row.GetValueOrDefault("local_type") ?? 0L, CultureInfo.InvariantCulture),
content,
DateTimeOffset.FromUnixTimeSeconds(Convert.ToInt64(row.GetValueOrDefault("create_time") ?? 0L, CultureInfo.InvariantCulture)),
WechatDbMessage.IsSelf(senderWxId, accountRootDirectoryName)));
}
return messages;
}
public static async Task<DbMergedChat> ReadMergedAsync(string accountRootPath, IReadOnlyList<DatabaseKeyEvidence> databases,
string chatId, long localId, CancellationToken cancellationToken, string? databaseRelativePath = null)
{
if (localId <= 0) throw new WxAgentException(WxAgentErrorCode.InvalidArgument, "A positive local message ID is required.");
var candidates = databases.Where(database => database.RelativePath.StartsWith("message/", StringComparison.OrdinalIgnoreCase)
&& (databaseRelativePath is null || string.Equals(database.RelativePath, databaseRelativePath.Replace('\\', '/'), StringComparison.OrdinalIgnoreCase))).ToArray();
if (candidates.Length == 0) throw new WxAgentException(WxAgentErrorCode.DatabaseKeyNotFound, "No matching message database has a cached key.");
(string Path, DbMessage Message)? found = null;
foreach (var database in candidates)
{
cancellationToken.ThrowIfCancellationRequested();
var selected = databases.Where(entry => !entry.RelativePath.StartsWith("message/", StringComparison.OrdinalIgnoreCase) || entry == database).ToArray();
var rows = await ReadAsync(accountRootPath, selected, chatId, 2, cancellationToken, localId).ConfigureAwait(false);
if (rows.Count == 0) continue;
if (rows.Count != 1 || found is not null)
throw new WxAgentException(WxAgentErrorCode.InvalidOperationState, "The local ID is ambiguous across message shards; specify --database.");
found = (database.RelativePath, rows[0]);
}
if (found is null) throw new WxAgentException(WxAgentErrorCode.ControlNotFound, "The selected database message was not found.");
try { return new DbMergedChat(found.Value.Path, found.Value.Message, WechatMergedChatParser.Parse(found.Value.Message.Content)); }
catch (InvalidDataException ex) { throw new WxAgentException(WxAgentErrorCode.DatabaseOpenFailed, ex.Message, ex); }
}
private static async Task<IReadOnlyDictionary<string, WechatContactName>> ReadContactsAsync(
string accountRootPath, IReadOnlyList<DatabaseKeyEvidence> databases, IReadOnlyList<string> senders, CancellationToken cancellationToken)
{
var result = new Dictionary<string, WechatContactName>(StringComparer.Ordinal);
if (senders.Count == 0) return result;
var contactDatabase = databases.FirstOrDefault(database => string.Equals(database.RelativePath, "contact/contact.db", StringComparison.OrdinalIgnoreCase));
if (contactDatabase is null) return result;
var path = Path.GetFullPath(Path.Combine(accountRootPath, "contact", "contact.db"));
var placeholders = string.Join(", ", Enumerable.Range(0, senders.Count).Select(index => $"$p{index}"));
var parameters = senders.Select((sender, index) => new KeyValuePair<string, object?>($"$p{index}", sender)).ToArray();
var sql = $"SELECT username, nick_name, remark, small_head_url FROM contact WHERE username IN ({placeholders});";
var rows = await SqlCipherDatabaseReader.QueryRowsAsync(path, contactDatabase.EncKey, sql, parameters, cancellationToken).ConfigureAwait(false);
foreach (var row in rows)
{
var username = row.GetValueOrDefault("username") as string;
if (string.IsNullOrEmpty(username)) continue;
var remark = row.GetValueOrDefault("remark") as string;
var nickname = row.GetValueOrDefault("nick_name") as string;
var name = !string.IsNullOrEmpty(remark) ? remark : !string.IsNullOrEmpty(nickname) ? nickname : username;
result[username] = new WechatContactName(name, row.GetValueOrDefault("small_head_url") as string);
}
return result;
}
private static async Task<(string Path, string EncKey)?> FindMessageDatabaseAsync(
string accountRootPath, IReadOnlyList<DatabaseKeyEvidence> databases, string tableName, CancellationToken cancellationToken)
{
var candidates = databases
.Where(database => database.RelativePath.StartsWith("message/", StringComparison.OrdinalIgnoreCase)
&& (database.RelativePath.EndsWith(".db", StringComparison.OrdinalIgnoreCase)))
.OrderBy(database => database.RelativePath, StringComparer.OrdinalIgnoreCase)
.ToArray();
foreach (var database in candidates)
{
cancellationToken.ThrowIfCancellationRequested();
var root = Path.GetFullPath(accountRootPath).TrimEnd(Path.DirectorySeparatorChar, Path.AltDirectorySeparatorChar) + Path.DirectorySeparatorChar;
var relative = database.RelativePath.Replace('/', Path.DirectorySeparatorChar);
var path = Path.GetFullPath(Path.Combine(root, relative));
if (Path.IsPathRooted(relative) || !path.StartsWith(root, StringComparison.OrdinalIgnoreCase))
throw new WxAgentException(WxAgentErrorCode.InvalidArgument, "The database path must remain inside the selected account root.");
var sql = "SELECT name FROM sqlite_master WHERE type = 'table' AND name = $name;";
var rows = await SqlCipherDatabaseReader.QueryRowsAsync(path, database.EncKey, sql,
[new KeyValuePair<string, object?>("$name", tableName)], cancellationToken).ConfigureAwait(false);
if (rows.Count > 0) return (path, database.EncKey);
}
return null;
}
private sealed record WechatContactName(string Name, string? AvatarUrl);
}