Files
bim-dogma-poc/ExcelKv.Core/ExcelLoader.cs

234 lines
9.7 KiB
C#

using System;
using System.Collections.Generic;
using System.Linq;
using System.Threading.Tasks;
using MiniExcelLibs;
namespace ExcelKv.Core;
public class RegionConfig
{
public int TopHeaderStartRow { get; set; } = 0;
public int TopHeaderDepth { get; set; } = 3;
public int LeftHeaderStartCol { get; set; } = 0;
public int LeftHeaderWidth { get; set; } = 4;
// Optional Parsing Limits (inclusive indices)
public int? DataEndRow { get; set; }
public int? DataEndCol { get; set; }
}
public interface IStorageWrapper
{
Task SetAsync(string key, string value);
// New: Metadata overload for Traceability
Task SetAsync(string key, string value, int row, int col);
Task IncrementAsync(string key, double value);
}
// Live Aggregator Helper (Moved here for shared access)
public class LiveAggregator
{
private static readonly System.Text.RegularExpressions.Regex MetricPattern =
new System.Text.RegularExpressions.Regex(@"(수량|합계|Volume|Weight|Total|Usage)", System.Text.RegularExpressions.RegexOptions.IgnoreCase | System.Text.RegularExpressions.RegexOptions.Compiled);
public static bool IsMetric(string colName) => MetricPattern.IsMatch(colName);
public static string GenerateGroupKey(string sheetName, string category, string spec, string metricName)
=> $"Stats:{Use_(category)}:{Use_(spec)}:{Use_(metricName)}";
private static string Use_(string s) => string.IsNullOrWhiteSpace(s) ? "Unknown" : s.Replace(" ", "_");
}
public class ExcelLoader
{
public static async Task<List<string>> GetSheetNamesAsync(string filePath)
{
return await Task.Run(() => MiniExcel.GetSheetNames(filePath).ToList());
}
public static async Task<List<string[]>> GetPreviewRowsAsync(string filePath, string sheetName, int count = 50)
{
return await Task.Run(() =>
{
var rows = MiniExcel.Query(filePath, sheetName: sheetName, useHeaderRow: false).Take(count).ToList();
List<string[]> result = new();
foreach(IDictionary<string, object> r in rows)
{
result.Add(FlattenDictionaryRow(r));
}
return result;
});
}
public static async Task<int> ProcessFileAsync(
string filePath,
string sheetName,
RegionConfig config,
IStorageWrapper storage,
SchemaRegistry registry)
{
Console.WriteLine($"[Core.Loader] Processing {filePath} Sheet: {sheetName}...");
var rows = MiniExcel.Query(filePath, sheetName: sheetName, useHeaderRow: false).ToList();
// Validation
int dataStartRow = config.TopHeaderStartRow + config.TopHeaderDepth;
if (rows.Count <= dataStartRow) return 0;
// 1. Analyze Top Headers (Data Columns)
var topHeaderRows = new List<string[]>();
for (int i = config.TopHeaderStartRow; i < dataStartRow; i++)
{
topHeaderRows.Add(FlattenDictionaryRow((IDictionary<string, object>)rows[i]));
}
// Calculate Global Max Col from Data Rows to ensure we don't truncate data
// Optimization: check a sample or assume logical limit. For correctness, check valid rows.
int globalMaxCol = 0;
int limitRow = config.DataEndRow.HasValue ? Math.Min(rows.Count, config.DataEndRow.Value + 1) : rows.Count;
for(int i = dataStartRow; i < limitRow; i++)
{
var d = (IDictionary<string, object>)rows[i];
if(d.Count > globalMaxCol) globalMaxCol = d.Count; // Approximate
// FlattenDictionaryRow is cleaner but expensive to call just for count.
// d.Keys.Count is effectively the column count for that row.
}
// Flatten Data Headers (Right Side)
var topAxisKeys = FlattenTopHeaders(topHeaderRows, config.LeftHeaderStartCol + config.LeftHeaderWidth, globalMaxCol);
// ... (Left Axis Header logic omitted for brevity, unchanged) ...
// ** New: Extract Left Axis Headers (Corner Region) **
// These are the headers *above* the Left Key columns.
var leftAxisHeaders = new List<string>();
var bottomHeaderRow = FlattenDictionaryRow((IDictionary<string, object>)rows[dataStartRow - 1]);
for (int c = 0; c < config.LeftHeaderWidth; c++)
{
int absCol = config.LeftHeaderStartCol + c;
string headerVal = (absCol < bottomHeaderRow.Length) ? bottomHeaderRow[absCol] : $"Col{c}";
leftAxisHeaders.Add(headerVal);
}
string ns = System.IO.Path.GetFileName(filePath) + "_" + sheetName;
registry.RegisterSchema(ns, topAxisKeys);
// 2. Process Data Rows
string[] lastLeftValues = new string[config.LeftHeaderWidth];
int processedCount = 0;
for (int i = dataStartRow; i < limitRow; i++)
{
var rowDict = (IDictionary<string, object>)rows[i];
var rowVals = FlattenDictionaryRow(rowDict);
// ... (Left Key Logic Unchanged) ...
var currentLeftParts = new List<string>();
bool rowHasContent = false;
for (int c = 0; c < config.LeftHeaderWidth; c++)
{
int absCol = config.LeftHeaderStartCol + c;
string val = (absCol < rowVals.Length) ? rowVals[absCol] : "";
if (string.IsNullOrWhiteSpace(val)) val = lastLeftValues[c];
else lastLeftValues[c] = val;
if(!string.IsNullOrWhiteSpace(val)) {
currentLeftParts.Add(val);
rowHasContent = true;
}
}
if (!rowHasContent) continue;
string leftKey = string.Join("----", currentLeftParts);
// B. Map Values
int limitCol = config.DataEndCol.HasValue ? Math.Min(rowVals.Length, config.DataEndCol.Value + 1) : rowVals.Length;
for (int c = config.LeftHeaderStartCol + config.LeftHeaderWidth; c < limitCol; c++)
{
int topIndex = c - (config.LeftHeaderStartCol + config.LeftHeaderWidth);
if (topIndex < 0 || topIndex >= topAxisKeys.Count) continue;
string topKey = topAxisKeys[topIndex];
if (string.IsNullOrWhiteSpace(topKey)) continue; // Skip if no header key
string val = rowVals[c];
if (string.IsNullOrWhiteSpace(val)) continue;
// ** Feature: Rounding to 6 Decimal Places **
// "All data should be formatted to 6 decimal places by default"
if (double.TryParse(val, out double dVal))
{
val = Math.Round(dVal, 6).ToString(); // Simple G-format or fixed? User said "Decimal 6 places".
// Let's use standard string representation of the rounded 6-digit number to avoid trailing zeros "1.100000".
// Math.Round(1.1, 6) -> 1.1.
// If user meant "Fixed 6 places" (1.100000), use F6. Usually "representation" means significant digits.
// Given SI context, significant precision matters. Let's use normalization (remove trailing zeros)
// Update: User said "소수점 6자리까지 표현하는걸로 통일". Could mean truncate or round. Safe bet is Round.
}
string fullKey = $"{sheetName}:{leftKey}*{topKey}";
await storage.SetAsync(fullKey, val, i, c);
if (LiveAggregator.IsMetric(topKey) && double.TryParse(val, out double statVal))
{
await storage.IncrementAsync($"Stats:Global:{topKey}", statVal);
}
}
processedCount++;
}
Console.WriteLine($"[Core.Loader] Finished. Processed: {processedCount} items.");
return processedCount;
}
private static string[] FlattenDictionaryRow(IDictionary<string, object> rowDict)
{
var sortedKeys = rowDict.Keys.OrderBy(k => k.Length).ThenBy(k => k).ToList();
var result = new string[sortedKeys.Count];
for (int i = 0; i < sortedKeys.Count; i++)
{
var val = rowDict[sortedKeys[i]];
result[i] = val?.ToString() ?? "";
}
return result;
}
private static List<string> FlattenTopHeaders(List<string[]> headerRows, int startCol, int globalMaxCol)
{
if (headerRows.Count == 0) return new List<string>();
// Ensure we cover all data columns, even if headers are short
int maxCol = Math.Max(headerRows.Max(r => r.Length), globalMaxCol);
var flatHeaders = new List<string>();
// Iterate Columns
var lastValues = new string[headerRows.Count];
for (int c = startCol; c < maxCol; c++)
{
var parts = new List<string>();
// Iterate Rows (Depth)
for (int r = 0; r < headerRows.Count; r++)
{
string val = (c < headerRows[r].Length) ? headerRows[r][c] : "";
// Horizontal Forward Fill (Re-enabled for Merged Headers)
if (string.IsNullOrWhiteSpace(val)) val = lastValues[r];
else lastValues[r] = val;
if (!string.IsNullOrWhiteSpace(val)) parts.Add(val);
}
// If empty, use "Col_Index" fallback or keep empty?
// User schema usually requires keys. If empty, it's skipped in mapping.
// Let's keep it empty, but if data exists, it won't map unless we have a key.
// If parts is empty, let's leave valid empty string so mapping can decide.
flatHeaders.Add(string.Join("----", parts));
}
return flatHeaders;
}
}