List <DiffResult <string> > DiffSheetRow(ISheet sheet1, int row1, ISheet sheet2, int row2, SheetDiffStatus status)
{
var list1 = new List <string>();
var list2 = new List <string>();
if (sheet1.GetRow(row1) != null)
{
var row = sheet1.GetRow(row1);
var columnCount = books["src"].SheetValideColumn[sheet1.SheetName];
for (int i = 0; i < columnCount; ++i)
{
list1.Add(Util.GetCellValue(row.GetCell(i)));
}
}
if (sheet2.GetRow(row2) != null)
{
var row = sheet2.GetRow(row2);
var columnCount = books["dst"].SheetValideColumn[sheet2.SheetName];
for (int i = 0; i < columnCount; ++i)
{
list2.Add(Util.GetCellValue(row.GetCell(i)));
}
}
var diff = DiffUtil.Diff(list1, list2);
//var optimized = diff.ToList();// DiffUtil.OptimizeCaseDeletedFirst(diff);
var optimized = DiffUtil.OptimizeCaseDeletedFirst(diff);
optimized = DiffUtil.OptimizeCaseInsertedFirst(optimized);
return(optimized.ToList());
}
private IEnumerable <DiffResult <string> > DiffCellValue(IEnumerable <string> src, IEnumerable <string> dst)
{
var r = DiffUtil.Diff(src, dst);
r = DiffUtil.Order(r, DiffOrderType.LazyDeleteFirst);
return(DiffUtil.OptimizeCaseDeletedFirst(r));
}
public void OptimizeCaseDeleteFirst()
{
var str1 = "aaa";
var str2 = "bbb";
/*
* obj1 a a a
* obj2 b b b
* - - M + +
*/
var results = DiffUtil.Diff(str1, str2);
results = DiffUtil.Order(results, DiffOrderType.GreedyDeleteFirst);
results = DiffUtil.OptimizeCaseDeletedFirst(results);
Assert.AreEqual(DiffStatus.Deleted, results.ElementAt(0).Status);
Assert.AreEqual(DiffStatus.Deleted, results.ElementAt(1).Status);
Assert.AreEqual(DiffStatus.Modified, results.ElementAt(2).Status);
Assert.AreEqual(DiffStatus.Inserted, results.ElementAt(3).Status);
Assert.AreEqual(DiffStatus.Inserted, results.ElementAt(4).Status);
/*
* obj1 a a a
* obj2 b b b
+ + M - -
*/
results = DiffUtil.Diff(str1, str2);
results = DiffUtil.Order(results, DiffOrderType.GreedyInsertFirst);
results = DiffUtil.OptimizeCaseInsertedFirst(results);
Assert.AreEqual(DiffStatus.Inserted, results.ElementAt(0).Status);
Assert.AreEqual(DiffStatus.Inserted, results.ElementAt(1).Status);
Assert.AreEqual(DiffStatus.Modified, results.ElementAt(2).Status);
Assert.AreEqual(DiffStatus.Deleted, results.ElementAt(3).Status);
Assert.AreEqual(DiffStatus.Deleted, results.ElementAt(4).Status);
/*
* obj1 a a a
* obj2 b b b
* M M M
*/
results = DiffUtil.Diff(str1, str2);
results = DiffUtil.Order(results, DiffOrderType.LazyDeleteFirst);
results = DiffUtil.OptimizeCaseDeletedFirst(results);
Assert.AreEqual(DiffStatus.Modified, results.ElementAt(0).Status);
Assert.AreEqual(DiffStatus.Modified, results.ElementAt(1).Status);
Assert.AreEqual(DiffStatus.Modified, results.ElementAt(2).Status);
/*
* obj1 a a a
* obj2 b b b
* M M M
*/
results = DiffUtil.Diff(str1, str2);
results = DiffUtil.Order(results, DiffOrderType.LazyInsertFirst);
results = DiffUtil.OptimizeCaseInsertedFirst(results);
Assert.AreEqual(DiffStatus.Modified, results.ElementAt(0).Status);
Assert.AreEqual(DiffStatus.Modified, results.ElementAt(1).Status);
Assert.AreEqual(DiffStatus.Modified, results.ElementAt(2).Status);
}
private void UpdateValueDiff(string srcValue, string dstValue)
{
SrcValueTextBox.Document.Blocks.First().ContentStart.Paragraph.Inlines.Clear();
DstValueTextBox.Document.Blocks.First().ContentStart.Paragraph.Inlines.Clear();
var srcLines = srcValue.Split('\n').Select(s => s.TrimEnd());
var dstLines = dstValue.Split('\n').Select(s => s.TrimEnd());
var lineDiffResults = DiffCellValue(srcLines, dstLines).ToList();
var srcRange = new List <Tuple <string, Color?> >();
var dstRange = new List <Tuple <string, Color?> >();
foreach (var lineDiffResult in lineDiffResults)
{
if (lineDiffResult.Status == DiffStatus.Equal)
{
DiffEqualLine(lineDiffResult, srcRange);
DiffEqualLine(lineDiffResult, dstRange);
}
else if (lineDiffResult.Status == DiffStatus.Modified)
{
var charDiffResults = DiffUtil.Diff(lineDiffResult.Obj1, lineDiffResult.Obj2);
charDiffResults = DiffUtil.Order(charDiffResults, DiffOrderType.LazyDeleteFirst);
charDiffResults = DiffUtil.OptimizeCaseDeletedFirst(charDiffResults);
DiffModifiedLine(charDiffResults.Where(r => r.Status != DiffStatus.Inserted), srcRange, true);
DiffModifiedLine(charDiffResults.Where(r => r.Status != DiffStatus.Deleted), dstRange, false);
}
else if (lineDiffResult.Status == DiffStatus.Deleted)
{
DiffDeletedLine(lineDiffResult, srcRange, true);
DiffDeletedLine(lineDiffResult, dstRange, false);
}
else if (lineDiffResult.Status == DiffStatus.Inserted)
{
DiffInsertedLine(lineDiffResult, srcRange, true);
DiffInsertedLine(lineDiffResult, dstRange, false);
}
}
foreach (var r in srcRange)
{
var bc = r.Item2.HasValue ? new SolidColorBrush(r.Item2.Value) : new SolidColorBrush();
SrcValueTextBox.Document.Blocks.First().ContentStart.Paragraph.Inlines.Add(new Run(r.Item1)
{
Background = bc
});
}
foreach (var r in dstRange)
{
var bc = r.Item2.HasValue ? new SolidColorBrush(r.Item2.Value) : new SolidColorBrush();
DstValueTextBox.Document.Blocks.First().ContentStart.Paragraph.Inlines.Add(new Run(r.Item1)
{
Background = bc
});
}
}
private static Dictionary <int, ExcelColumnStatus> CreateColumnStatusMap(
IEnumerable <ExcelColumn> srcColumns, IEnumerable <ExcelColumn> dstColumns, ExcelSheetDiffConfig config)
{
var option = new DiffOption <ExcelColumn>();
if (config.SrcHeaderIndex >= 0)
{
option.EqualityComparer = new HeaderComparer();
foreach (var sc in srcColumns)
{
sc.HeaderIndex = config.SrcHeaderIndex;
}
}
if (config.DstHeaderIndex >= 0)
{
foreach (var dc in dstColumns)
{
dc.HeaderIndex = config.DstHeaderIndex;
}
}
var results = DiffUtil.Diff(srcColumns, dstColumns, option);
results = DiffUtil.Order(results, DiffOrderType.LazyDeleteFirst);
results = DiffUtil.OptimizeCaseDeletedFirst(results);
var ret = new Dictionary <int, ExcelColumnStatus>();
var columnIndex = 0;
foreach (var result in results)
{
var status = ExcelColumnStatus.None;
if (result.Status == DiffStatus.Deleted)
{
status = ExcelColumnStatus.Deleted;
}
else if (result.Status == DiffStatus.Inserted)
{
status = ExcelColumnStatus.Inserted;
}
ret.Add(columnIndex, status);
columnIndex++;
}
return(ret);
}
List <DiffResult <string> > DiffSheetRow(ISheet sheet1, int row1, ISheet sheet2, int row2, SheetDiffStatus status, out int maxLineCount)
{
var list1 = new List <string>();
var list2 = new List <string>();
maxLineCount = 0;
if (sheet1.GetRow(row1) != null)
{
var row = sheet1.GetRow(row1);
var columnCount = books["src"].SheetValideColumn[sheet1.SheetName];
for (int i = 0; i < columnCount; ++i)
{
var value = Util.GetCellValue(row.GetCell(i));
maxLineCount = Math.Max(maxLineCount, value.Count((c) => { return(c == '\n'); }) + 1);
list1.Add(value);
}
}
if (sheet2.GetRow(row2) != null)
{
var row = sheet2.GetRow(row2);
var columnCount = books["dst"].SheetValideColumn[sheet2.SheetName];
for (int i = 0; i < columnCount; ++i)
{
var value = Util.GetCellValue(row.GetCell(i));
maxLineCount = Math.Max(maxLineCount, value.Count((c) => { return(c == '\n'); }) + 1);
list2.Add(value);
}
}
var diff = DiffUtil.Diff(list1, list2);
//var optimized = diff.ToList();// DiffUtil.OptimizeCaseDeletedFirst(diff);
var optimized = DiffUtil.OptimizeCaseDeletedFirst(diff);
optimized = DiffUtil.OptimizeCaseInsertedFirst(optimized);
var tlist = optimized.ToList();
optimized = DiffUtil.OptimizeShift(tlist, false);
optimized = DiffUtil.OptimizeShift(optimized, true);
return(optimized.ToList());
}
// 把第一列认为是id列，检查增删, <value, 行id>
List <DiffResult <string2int> > GetIDDiffList(ISheet sheet1, ISheet sheet2, int checkCellCount, bool addRowID = false)
{
var list1 = new List <string2int>();
var list2 = new List <string2int>();
var nameHash = new HashSet <string>();
var startIdx = DiffStartIdx();
// 尝试找一个id不会重复的前几列的值作为key
for (int i = startIdx; ; i++)
{
var row = sheet1.GetRow(i);
if (row == null || !Util.CheckValideRow(row))
{
books["src"].SheetValideRow[sheet1.SheetName] = i;
break;
}
;
var val = "";
for (var j = 0; j < checkCellCount; ++j)
{
val += Util.GetCellValue(row.GetCell(j));
}
var hash_val = val;
if (addRowID)
{
hash_val = hash_val + "." + i;
}
if (nameHash.Contains(hash_val))
{
if (checkCellCount < 6)
{
return(GetIDDiffList(sheet1, sheet2, checkCellCount + 1, addRowID));
}
else
{
// 已经找不到能作为key的了。把id和行号连一块
return(GetIDDiffList(sheet1, sheet2, 1, true));
}
}
nameHash.Add(hash_val);
list1.Add(new string2int(val, i));
}
list1.Sort(delegate(string2int a, string2int b) {
var cmp = a.Key.CompareTo(b.Key);
if (cmp == 0)
{
return(a.Value.CompareTo(b.Value));
}
return(cmp);
});
nameHash.Clear();
for (int i = startIdx; ; i++)
{
var row = sheet2.GetRow(i);
if (row == null || !Util.CheckValideRow(row))
{
books["dst"].SheetValideRow[sheet2.SheetName] = i;
break;
}
var val = "";
for (var j = 0; j < checkCellCount; ++j)
{
val += Util.GetCellValue(row.GetCell(j));
}
var hash_val = val;
if (addRowID)
{
hash_val = hash_val + "." + i;
}
if (nameHash.Contains(hash_val))
{
if (checkCellCount < 6)
{
return(GetIDDiffList(sheet1, sheet2, checkCellCount + 1, addRowID));
}
else
{
// 已经找不到能作为key的了。把id和行号连一块
return(GetIDDiffList(sheet1, sheet2, 1, true));
}
}
nameHash.Add(hash_val);
list2.Add(new string2int(val, i));
}
list2.Sort(delegate(string2int a, string2int b) {
var cmp = a.Key.CompareTo(b.Key);
if (cmp == 0)
{
return(a.Value.CompareTo(b.Value));
}
return(cmp);
});
var option = new DiffOption <string2int>();
option.Optimize = false;
option.EqualityComparer = new SheetIDComparer();
var result = DiffUtil.Diff(list1, list2, option);
//var optimize = result.ToList();//
var optimize = DiffUtil.OptimizeCaseDeletedFirst(result);
return(optimize.ToList());
}
public void Diff(string file1, string file2, bool resetInitFile = true)
{
if (string.IsNullOrEmpty(file1) || string.IsNullOrEmpty(file2))
{
return;
}
if (resetInitFile)
{
SrcFile = file1;
DstFile = file2;
}
string oldsheetName = null;
if (books.ContainsKey("src"))
{
oldsheetName = books["src"].sheetname;
}
var src = InitWorkWrap(file1);
var dst = InitWorkWrap(file2);
var option = new DiffOption <SheetNameCombo>();
option.EqualityComparer = new SheetNameComboComparer();
var result = DiffUtil.Diff(src.sheetNameCombos, dst.sheetNameCombos, option);
//diffSheetName = result.ToList();//
diffSheetName = DiffUtil.OptimizeCaseDeletedFirst(result).ToList();
books["src"] = src;
books["dst"] = dst;
var srcSheetID = -1;
var dstSheetID = -1;
for (int i = 0; i < diffSheetName.Count; ++i)
{
var sheetname = diffSheetName[i];
var name = sheetname.Obj1 == null ? sheetname.Obj2.Name : sheetname.Obj1.Name;
// 只有sheet名字一样的可以diff， 先这么处理
if (sheetname.Status == DiffStatus.Equal)
{
var sheet1 = sheetname.Obj1.ID;
var sheet2 = sheetname.Obj2.ID;
sheetsDiff[name] = DiffSheet(src.book.GetSheetAt(sheet1), dst.book.GetSheetAt(sheet2));
if (sheetsDiff[name] != null)
{
oldsheetName = sheetname.Obj1.Name;
var sheetidx = 0;
if (!string.IsNullOrEmpty(oldsheetName))
{
sheetidx = src.book.GetSheetIndex(oldsheetName);
}
if (sheetsDiff[name].changed || srcSheetID == -1)
{
src.sheet = sheetidx;
srcSheetID = sheetidx;
}
if (!string.IsNullOrEmpty(oldsheetName))
{
sheetidx = dst.book.GetSheetIndex(oldsheetName);
}
if (sheetsDiff[name].changed || dstSheetID == -1)
{
dst.sheet = sheetidx;
dstSheetID = sheetidx;
}
}
}
}
// refresh ui
SrcFilePath.Content = file1;
DstFilePath.Content = file2;
SrcFileSheetsCombo.Items.Clear();
foreach (var item in src.sheetCombo)
{
int index = diffSheetName.FindIndex(a => a.Obj1 != null && a.Obj1.ID == (item.Content as SheetNameCombo).ID);
SolidColorBrush color = null;
DiffStatus status = diffSheetName[index].Status;
if (status != DiffStatus.Equal)
{
color = Util.GetColorByDiffStatus(status);
}
else
{
var name = diffSheetName[index].Obj1.Name;
color = Util.GetColorByDiffStatus(sheetsDiff.ContainsKey(name) && sheetsDiff[name] != null && sheetsDiff[name].changed ? DiffStatus.Modified : DiffStatus.Equal);
}
if (color != null)
{
item.Background = color;
}
SrcFileSheetsCombo.Items.Add(item);
}
var comboidx = src.ItemID2ComboIdx[src.sheet];
SrcFileSheetsCombo.SelectedItem = src.sheetCombo[comboidx];
DstFileSheetsCombo.Items.Clear();
foreach (var item in dst.sheetCombo)
{
int index = diffSheetName.FindIndex(a => a.Obj2 != null && a.Obj2.ID == (item.Content as SheetNameCombo).ID);
SolidColorBrush color = null;
DiffStatus status = diffSheetName[index].Status;
if (status != DiffStatus.Equal)
{
color = Util.GetColorByDiffStatus(status);
}
else
{
var name = diffSheetName[index].Obj1.Name;
color = Util.GetColorByDiffStatus(sheetsDiff.ContainsKey(name) && sheetsDiff[name] != null && sheetsDiff[name].changed ? DiffStatus.Modified : DiffStatus.Equal);
}
if (color != null)
{
item.Background = color;
}
DstFileSheetsCombo.Items.Add(item);
}
comboidx = dst.ItemID2ComboIdx[dst.sheet];
DstFileSheetsCombo.SelectedItem = dst.sheetCombo[comboidx];
DstDataGrid.RefreshData();
SrcDataGrid.RefreshData();
}
SheetDiffStatus DiffSheet(ISheet src, ISheet dst, SheetDiffStatus status = null)
{
status = status ?? new SheetDiffStatus();
bool changed = false;
var head1 = GetHeaderStrList(src);
var head2 = GetHeaderStrList(dst);
if (head1 == null || head2 == null)
{
return(null);
}
var diff = NetDiff.DiffUtil.Diff(head1, head2);
//var optimized = diff.ToList();// NetDiff.DiffUtil.OptimizeCaseDeletedFirst(diff);
var optimized = DiffUtil.OptimizeCaseDeletedFirst(diff);
changed = changed || optimized.Any(a => a.Status != DiffStatus.Equal);
status.diffHead = new SheetRowDiff()
{
diffcells = optimized.ToList()
};
status.column2diff1 = new Dictionary <int, int[]>();
status.column2diff2 = new Dictionary <int, int[]>();
status.column2diff1[0] = getColumn2Diff(status.diffHead.diffcells, true);
status.column2diff2[0] = getColumn2Diff(status.diffHead.diffcells, false);
books["src"].SheetValideColumn[src.SheetName] = head1.Count;
books["dst"].SheetValideColumn[dst.SheetName] = head2.Count;
status.diffFistColumn = GetIDDiffList(src, dst, 1);
changed = changed || status.diffFistColumn.Any(a => a.Status != DiffStatus.Equal);
status.diffSheet = new List <SheetRowDiff>();
status.rowID2DiffMap1 = new Dictionary <int, int>();
status.rowID2DiffMap2 = new Dictionary <int, int>();
status.Diff2RowID1 = new Dictionary <int, int>();
status.Diff2RowID2 = new Dictionary <int, int>();
status.RowEdited1 = status.RowEdited1 ?? new Dictionary <int, Dictionary <int, CellEditMode> >();
status.RowEdited2 = status.RowEdited2 ?? new Dictionary <int, Dictionary <int, CellEditMode> >();
foreach (var diffkv in status.diffFistColumn)
{
var rowid1 = diffkv.Obj1.Value;
var rowid2 = diffkv.Obj2.Value;
if (diffkv.Obj1.Key == null)
{
// 创建新行，方便比较
rowid1 = -1;
}
if (diffkv.Obj2.Key == null)
{
rowid2 = -1;
}
var diffrow = DiffSheetRow(src, rowid1, dst, rowid2, status);
if (diffkv.Obj1.Key == null)
{
// 创建新行，方便比较,放在后面是为了保证diff的时候是new,delete的形式，而不是modify
rowid1 = books["src"].SheetValideRow[src.SheetName];
//src.CreateRow(rowid1);
}
if (diffkv.Obj2.Key == null)
{
rowid2 = books["dst"].SheetValideRow[dst.SheetName];
//dst.CreateRow(rowid2);
}
status.column2diff1[rowid1] = getColumn2Diff(diffrow, true);
status.column2diff2[rowid2] = getColumn2Diff(diffrow, false);
int diffIdx = status.diffSheet.Count;
status.rowID2DiffMap1[rowid1] = diffIdx;
status.rowID2DiffMap2[rowid2] = diffIdx;
status.Diff2RowID1[diffIdx] = rowid1;
status.Diff2RowID2[diffIdx] = rowid2;
if (!status.RowEdited1.ContainsKey(rowid1))
{
status.RowEdited1[rowid1] = new Dictionary <int, CellEditMode>();
}
if (!status.RowEdited2.ContainsKey(rowid2))
{
status.RowEdited2[rowid2] = new Dictionary <int, CellEditMode>();
}
var rowdiff = new SheetRowDiff();
rowdiff.diffcells = diffrow;
rowdiff.changed = diffrow.Any(a => a.Status != DiffStatus.Equal);
if (rowdiff.changed)
{
rowdiff.diffcell_details = new List <List <DiffResult <char> > >();
foreach (var cell in diffrow)
{
if (cell.Status == DiffStatus.Modified)
{
var cell_diff = NetDiff.DiffUtil.Diff(cell.Obj1, cell.Obj2);
//var optimized = diff.ToList();// NetDiff.DiffUtil.OptimizeCaseDeletedFirst(diff);
var opt_cell_diff = DiffUtil.OptimizeCaseDeletedFirst(cell_diff);
rowdiff.diffcell_details.Add(opt_cell_diff.ToList());
}
else
{
rowdiff.diffcell_details.Add(null);
}
}
}
status.diffSheet.Add(rowdiff);
changed = changed || rowdiff.changed;
if (changed)
{
changed = true;
}
}
status.changed = changed;
return(status);
}
public static ExcelSheetDiff Diff(ExcelSheet src, ExcelSheet dst, ExcelSheetDiffConfig config)
{
var srcColumns = src.CreateColumns();
var dstColumns = dst.CreateColumns();
var columnStatusMap = CreateColumnStatusMap(srcColumns, dstColumns, config);
var option = new DiffOption <ExcelRow>();
option.EqualityComparer =
new RowComparer(new HashSet <int>(columnStatusMap.Where(i => i.Value != ExcelColumnStatus.None).Select(i => i.Key)));
// 这里实际上计算的是有没有插入新列
foreach (var row in src.Rows.Values)
{
var shifted = new List <ExcelCell>();
var index = 0;
var queue = new Queue <ExcelCell>(row.Cells);
while (queue.Any())
{
if (columnStatusMap[index] == ExcelColumnStatus.Inserted)
{
shifted.Add(new ExcelCell(string.Empty, 0, 0));
}
else
{
shifted.Add(queue.Dequeue());
}
index++;
}
row.UpdateCells(shifted);
}
foreach (var row in dst.Rows.Values)
{
var shifted = new List <ExcelCell>();
var index = 0;
var queue = new Queue <ExcelCell>(row.Cells);
while (queue.Any())
{
if (columnStatusMap[index] == ExcelColumnStatus.Deleted)
{
shifted.Add(new ExcelCell(string.Empty, 0, 0));
}
else
{
shifted.Add(queue.Dequeue());
}
index++;
}
row.UpdateCells(shifted);
}
var r = DiffUtil.Diff(src.Rows.Values, dst.Rows.Values, option);
r = DiffUtil.Order(r, DiffOrderType.LazyDeleteFirst);
var resultArray = DiffUtil.OptimizeCaseDeletedFirst(r).ToArray();
if (resultArray.Length > 10000)
{
var count = 0;
var indices = Enumerable.Range(0, 100).ToList();
foreach (var result in resultArray)
{
if (result.Status != DiffStatus.Equal)
{
indices.AddRange(Enumerable.Range(Math.Max(0, count - 100), 200));
}
count++;
}
indices = indices.Distinct().ToList();
resultArray = indices.Where(i => i < resultArray.Length).Select(i => resultArray[i]).ToArray();
}
var sheetDiff = new ExcelSheetDiff(src, dst);
DiffCells(resultArray, sheetDiff, columnStatusMap);
return(sheetDiff);
}