Skip to content

Commit

Permalink
allocation fighting and performance boosting
Browse files Browse the repository at this point in the history
  • Loading branch information
@petriashev
petriashev committed Feb 26, 2023
1 parent 7790d48 commit a0b77c8
Show file tree
Hide file tree
Showing 39 changed files with 1,166 additions and 222 deletions.
19 changes: 5 additions & 14 deletions src/MicroElements.Metadata.OpenXml/OpenXml/Excel/Parsing/ExcelElement.cs
View file
Original file line number Diff line number Diff line change
Expand Up @@ -21,8 +21,7 @@ public interface IExcelElement<TOpenXmlElement> : IMetadataProvider
/// <summary>
/// Gets OpenXml element.
/// </summary>
[MaybeNull]
TOpenXmlElement Data { get; }
TOpenXmlElement? Data { get; }

/// <summary>
/// Returns true if <see cref="Data"/> is not null.
Expand All @@ -45,21 +44,14 @@ public class ExcelElement<TOpenXmlElement> : IExcelElement<TOpenXmlElement>
/// <summary>
/// Gets OpenXml element.
/// </summary>
[MaybeNull]
public TOpenXmlElement Data { get; }

///// <summary>
///// Gets OpenXml element as <see cref="Option{A}"/>.
///// </summary>
///// <returns>Optional OpenXml element.</returns>
//public Option<TOpenXmlElement> AsOption() => Data!;
public TOpenXmlElement? Data { get; }

/// <summary>
/// Initializes a new instance of the <see cref="ExcelElement{TOpenXmlElement}"/> class.
/// </summary>
/// <param name="doc">OpenXml document that contains this element.</param>
/// <param name="data">OpenXml element.</param>
public ExcelElement(SpreadsheetDocument doc, [MaybeNull] TOpenXmlElement data)
public ExcelElement(SpreadsheetDocument doc, TOpenXmlElement? data)
{
Doc = doc.AssertArgumentNotNull(nameof(doc));
Data = data;
Expand Down Expand Up @@ -95,15 +87,14 @@ public ExcelElement(SpreadsheetDocument doc, [MaybeNull] TOpenXmlElement data)
/// <summary>
/// Gets OpenXml element.
/// </summary>
[MaybeNull]
public TOpenXmlElement Data { get; }
public TOpenXmlElement? Data { get; }

/// <summary>
/// Initializes a new instance of the <see cref="ExcelElementLight{TOpenXmlElement}"/> struct.
/// </summary>
/// <param name="doc">OpenXml document that contains this element.</param>
/// <param name="data">OpenXml element.</param>
public ExcelElementLight(SpreadsheetDocument doc, [MaybeNull] TOpenXmlElement data)
public ExcelElementLight(SpreadsheetDocument doc, TOpenXmlElement? data)
{
Doc = doc.AssertArgumentNotNull(nameof(doc));
Data = data;
Expand Down
75 changes: 75 additions & 0 deletions src/MicroElements.Metadata.OpenXml/OpenXml/Excel/Parsing/ExcelParsing.CellReference.cs
View file
Original file line number Diff line number Diff line change
@@ -0,0 +1,75 @@
// Copyright (c) MicroElements. All rights reserved.
// Licensed under the MIT license. See LICENSE file in the project root for full license information.

using System.Linq;
using DocumentFormat.OpenXml;
using MicroElements.CodeContracts;
using MicroElements.Collections.Cache;

namespace MicroElements.Metadata.OpenXml.Excel.Parsing
{
public static partial class ExcelParsingExtensions
{
/// <summary>
/// Gets cell reference by column row index.
/// Example: (0,0)->A1.
/// </summary>
/// <param name="column">Column index.</param>
/// <param name="row">Row index.</param>
/// <param name="zeroBased">Is column and row zero based.</param>
/// <returns>Cell reference.</returns>
public static StringValue GetCellReference(int column, int row, bool zeroBased = true)
{
int columnIndex = zeroBased ? column : column - 1;
string columnName = GetColumnName(columnIndex);
int rowName = zeroBased ? row + 1 : row;
return new StringValue(string.Concat(columnName, rowName.ToString()));
}

/// <summary>
/// Gets column index (cached).
/// </summary>
public static string GetColumnName(int columnIndex = 0)
{
return Cache
.Instance<int, string>("ColumnName")
.GetOrAdd(columnIndex, i => GetColumnName(string.Empty, i));
}

private static string GetColumnName(string prefix, int columnIndex = 0)
{
return columnIndex < 26
? $"{prefix}{(char)(65 + columnIndex)}"
: GetColumnName(GetColumnName(prefix, ((columnIndex - (columnIndex % 26)) / 26) - 1), columnIndex % 26);
}

/// <summary>
/// Gets column reference from cell reference.
/// For example: A1->A, CD22->CD.
/// </summary>
/// <param name="cellReference">Cell reference.</param>
/// <returns>Column reference.</returns>
public static string GetColumnReference(this StringValue cellReference)
{
cellReference.AssertArgumentNotNull(nameof(cellReference));

return cellReference.Value.GetColumnReference();
}

/// <summary>
/// Gets column reference from cell reference.
/// For example: A1->A, CD22->CD.
/// </summary>
/// <param name="cellReference">Cell reference.</param>
/// <returns>Column reference.</returns>
public static string GetColumnReference(this string cellReference)
{
cellReference.AssertArgumentNotNull(nameof(cellReference));

if (cellReference.Length == 2)
return cellReference.Substring(0, 1);

return new string(cellReference.TakeWhile(char.IsLetter).ToArray());
}
}
}
110 changes: 30 additions & 80 deletions src/MicroElements.Metadata.OpenXml/OpenXml/Excel/Parsing/ExcelParsingExtensions.cs
View file
Original file line number Diff line number Diff line change
Expand Up @@ -10,7 +10,9 @@
using DocumentFormat.OpenXml.Packaging;
using DocumentFormat.OpenXml.Spreadsheet;
using MicroElements.CodeContracts;
using MicroElements.Collections.Cache;
using MicroElements.Collections.Extensions.Iterate;
using MicroElements.Collections.TwoLayerCache;
using MicroElements.Diagnostics;
using MicroElements.Metadata.Parsing;
using MicroElements.Validation;
Expand Down Expand Up @@ -64,17 +66,6 @@ public static IEnumerable<ExcelElement<Column>> GetColumns(this ExcelElement<She
return Array.Empty<ExcelElement<Column>>();
}

/// <summary>
/// Gets rows from sheet.
/// </summary>
/// <param name="sheet">Source sheet.</param>
/// <returns>Sheet rows.</returns>
public static IEnumerable<ExcelElement<Row>> GetRows(this ExcelElement<Sheet> sheet)
{
return GetOpenXmlRows(sheet)
.Zip(Enumerable.Repeat(sheet, int.MaxValue), (row, sh) => new ExcelElement<Row>(sh.Doc, row));
}

/// <summary>
/// Gets rows from sheet.
/// </summary>
Expand All @@ -96,65 +87,13 @@ public static IEnumerable<Row> GetOpenXmlRows(this ExcelElement<Sheet> sheet)
}

/// <summary>
/// Gets cell reference by column row index.
/// Example: (0,0)->A1.
/// </summary>
/// <param name="column">Column index.</param>
/// <param name="row">Row index.</param>
/// <param name="zeroBased">Is column and row zero based.</param>
/// <returns>Cell reference.</returns>
public static StringValue GetCellReference(int column, int row, bool zeroBased = true)
{
int columnIndex = zeroBased ? column : column - 1;
string columnName = GetColumnName(columnIndex);
int rowName = zeroBased ? row + 1 : row;
return new StringValue(string.Concat(columnName, rowName.ToString()));
}

private static readonly ConcurrentDictionary<int, string> _columnIndexes = new ConcurrentDictionary<int, string>();

/// <summary>
/// Gets column index (cached).
/// </summary>
public static string GetColumnName(int columnIndex = 0)
{
return _columnIndexes.GetOrAdd(columnIndex, i => GetColumnName(string.Empty, i));
}

private static string GetColumnName(string prefix, int columnIndex = 0)
{
return columnIndex < 26
? $"{prefix}{(char)(65 + columnIndex)}"
: GetColumnName(GetColumnName(prefix, ((columnIndex - (columnIndex % 26)) / 26) - 1), columnIndex % 26);
}

/// <summary>
/// Gets column reference from cell reference.
/// For example: A1->A, CD22->CD.
/// </summary>
/// <param name="cellReference">Cell reference.</param>
/// <returns>Column reference.</returns>
public static string GetColumnReference(this StringValue cellReference)
{
cellReference.AssertArgumentNotNull(nameof(cellReference));

return cellReference.Value.GetColumnReference();
}

/// <summary>
/// Gets column reference from cell reference.
/// For example: A1->A, CD22->CD.
/// Gets rows from sheet.
/// </summary>
/// <param name="cellReference">Cell reference.</param>
/// <returns>Column reference.</returns>
public static string GetColumnReference(this string cellReference)
/// <param name="sheet">Source sheet.</param>
/// <returns>Sheet rows.</returns>
public static IEnumerable<ExcelElement<Row>> GetRows(this ExcelElement<Sheet> sheet)
{
cellReference.AssertArgumentNotNull(nameof(cellReference));

if (cellReference.Length == 2)
return cellReference.Substring(0, 1);

return new string(cellReference.TakeWhile(char.IsLetter).ToArray());
return GetOpenXmlRows(sheet).Select(row => new ExcelElement<Row>(sheet.Doc, row));
}

/// <summary>
Expand Down Expand Up @@ -253,7 +192,7 @@ public static ExcelElement<HeaderCell>[] GetHeaders(this ExcelElement<Row> row)
if (cell != null)
{
IPropertyParser? propertyParser = header.GetMetadata<IPropertyParser>();
rowValues[i] = cell.GetCellValue(nullValue, propertyParser);
rowValues[i] = cell.GetCellValue(nullValue, propertyParser?.TargetType);

//StringValue cellReference = cell.Data.CellReference;
}
Expand Down Expand Up @@ -309,28 +248,37 @@ private static string GetFormattedValue(this ExcelElement<Cell> cell)
/// Uses SharedStringTable if needed.
/// For DateTime, LocalDate and LocalTime tries to convert double excel value to ISO format.
/// </summary>
public static string? GetCellValue(this ExcelElement<Cell> cell, string? nullValue = null, IPropertyParser? propertyParser = null)
public static string? GetCellValue<T>(
this T cellElement,
string? nullValue = null,
Type? targetType = null)
where T : IExcelElement<Cell>
{
Cell? cellData = cell.Data;
string? cellValue = cellData?.CellValue?.InnerText ?? nullValue;
Cell? cell = cellElement.Data;
SharedStringTable sharedStringTable = cellElement.Doc.WorkbookPart.SharedStringTablePart.SharedStringTable;

string? cellValue = cell?.CellValue?.InnerText ?? nullValue;
string? cellTextValue = null;

if (cellData == null)
if (cell == null)
return cellValue;

CellValues? dataTypeValue = cellData.DataType?.Value;
CellValues? dataTypeValue = cell.DataType?.Value;

if (cellValue != null && dataTypeValue == CellValues.SharedString)
{
cellTextValue = cell.Doc.WorkbookPart.SharedStringTablePart.SharedStringTable.ChildElements.GetItem(int.Parse(cellValue)).InnerText;
// cellTextValue = sharedStringTable.ChildElements.GetItem(int.Parse(cellValue)).InnerText;
cellTextValue = sharedStringTable
.GetWeakCache<string, string>()
.GetOrAdd(cellValue, static (cellValue, sharedStringTable) => sharedStringTable.ChildElements.GetItem(int.Parse(cellValue)).InnerText, sharedStringTable);
}

if (cellTextValue == null && cellValue != null)
{
propertyParser ??= cell.GetMetadata<IPropertyParser>();

if (propertyParser != null)
cellTextValue = TryParseAsDateType(cellValue, propertyParser.TargetType);
if (targetType != null)
{
cellTextValue = TryParseAsDateType(cellValue, targetType);
}
}

return cellTextValue ?? cellValue;
Expand Down Expand Up @@ -578,11 +526,13 @@ public static partial class ExcelParsingExtensions
// Use first row as headers
headers = row.GetHeaders();

var propertyParsers = parserProvider.GetParsers().ToArray();

// Associate parser for each header
foreach (var header in headers)
{
// TODO: RIGID SEARCH. Use predicate?
var propertyParser = parserProvider.GetParsers().FirstOrDefault(parser => parser.SourceName == header.Data.Name);
var propertyParser = propertyParsers.FirstOrDefault(parser => parser.SourceName == header.Data.Name);
if (propertyParser != null)
{
header.SetMetadata(propertyParser);
Expand Down
Loading

0 comments on commit a0b77c8

Please sign in to comment.