using HtmlAgilityPack;
namespace Service.EmailServices.EmailParse;
public class EmailParseFormCcsvc(
ILogger logger,
IOpenAiService openAiService
) : EmailParseServicesBase(logger, openAiService)
{
public override bool CanParse(string from, string subject, string body)
{
if (!from.Contains("ccsvc@message.cmbchina.com"))
{
return false;
}
if (!subject.Contains("每日信用管家"))
{
return false;
}
// 必须包含HTML标签
if (!Regex.IsMatch(body, "<.*?>"))
{
return false;
}
return true;
}
public override async Task<(
string card,
string reason,
decimal amount,
decimal balance,
TransactionType type,
DateTime? occurredAt
)[]> ParseEmailContentAsync(string emailContent)
{
var doc = new HtmlDocument();
doc.LoadHtml(emailContent);
var result = new List<(string, string, decimal, decimal, TransactionType, DateTime?)>();
// 1. Get Date
var dateNode = doc.DocumentNode.SelectSingleNode("//font[contains(text(), '您的消费明细如下')]");
var dateText = dateNode.InnerText.Trim();
// "2025/12/21 您的消费明细如下:"
var dateMatch = Regex.Match(dateText, @"\d{4}/\d{1,2}/\d{1,2}");
if (!dateMatch.Success || !DateTime.TryParse(dateMatch.Value, out var date))
{
logger.LogWarning("Failed to parse date from: {DateText}", dateText);
return [];
}
// 2. Get Balance (Available Limit)
decimal balance = 0;
// Find "可用额度" label
var limitLabelNode = doc.DocumentNode.SelectSingleNode("//font[contains(text(), '可用额度')]");
{
// Go up to TR
var tr = limitLabelNode.Ancestors("tr").FirstOrDefault();
if (tr != null)
{
var prevTr = tr.PreviousSibling;
while (prevTr.Name != "tr") prevTr = prevTr.PreviousSibling;
var balanceNode = prevTr.SelectSingleNode(".//font[contains(text(), '¥')]");
var balanceStr = balanceNode.InnerText.Replace("¥", "").Replace(",", "").Trim();
decimal.TryParse(balanceStr, out balance);
}
}
// 3. Get Transactions
var transactionNodes = doc.DocumentNode.SelectNodes("//span[@id='fixBand4']");
foreach (var node in transactionNodes)
{
try
{
// Time
var timeNode = node.SelectSingleNode(".//span[@id='fixBand5']//font");
var timeText = timeNode.InnerText.Trim(); // "10:13:43"
DateTime? occurredAt = date;
if (!string.IsNullOrEmpty(timeText) && DateTime.TryParse($"{date:yyyy-MM-dd} {timeText}", out var dt))
{
occurredAt = dt;
}
// Info Block
var infoNode = node.SelectSingleNode(".//span[@id='fixBand12']");
// Amount
var amountNode = infoNode.SelectSingleNode(".//font[contains(text(), 'CNY')]");
var amountText = amountNode.InnerText.Replace("CNY", "").Replace(" ", "").Trim();
if (!decimal.TryParse(amountText, out var amount))
{
continue;
}
// Description
var descNode = infoNode.SelectSingleNode(".//tr[2]//font");
var descText = descNode.InnerText;
// Replace and non-breaking space (\u00A0) with normal space
descText = descText.Replace(" ", " ");
descText = HtmlEntity.DeEntitize(descText).Replace((char)160, ' ').Trim();
// Parse Description: "尾号4390 消费 财付通-luckincoffee瑞幸咖啡"
var parts = descText.Split([' '], StringSplitOptions.RemoveEmptyEntries);
string card = "";
string reason = descText;
TransactionType type;
if (parts.Length > 0 && parts[0].StartsWith("尾号"))
{
card = parts[0].Replace("尾号", "");
}
if (parts.Length > 2)
{
reason = string.Join(" ", parts.Skip(2));
}
// 招商信用卡特殊,消费金额为正数,退款为负数
if(amount > 0)
{
type = TransactionType.Expense;
}
else
{
type = TransactionType.Income;
amount = Math.Abs(amount);
}
result.Add((card, reason, amount, balance, type, occurredAt));
}
catch (Exception ex)
{
logger.LogError(ex, "Error parsing transaction node");
}
}
return await Task.FromResult(result.ToArray());
}
}