lsm/PdfPig
1
0
Fork 0
mirror of https://github.com/UglyToad/PdfPig.git synced 2025-09-19 10:47:56 +08:00
Code Issues Actions 5 Packages Projects Releases Wiki Activity

update DefaultPageSegmenter to use DlaOptions

Browse Source
This commit is contained in:
BobLd
2020-05-23 20:08:40 +01:00
committed by Eliot Jones
parent 1438fec741
commit a16f377d5a
1 changed files with 37 additions and 7 deletions
Download Patch File Download Diff File Expand all files Collapse all files

40
src/UglyToad.PdfPig.DocumentLayoutAnalysis/PageSegmenter/DefaultPageSegmenter.cs
View File

@@ -2,9 +2,11 @@
{ {
using Content; using Content;
using Core; using Core;
using System;
using System.Collections.Generic; using System.Collections.Generic;
using System.Linq; using System.Linq;
/// <inheritdoc />
/// <summary> /// <summary>
/// Default Page Segmenter. All words are included in one block. /// Default Page Segmenter. All words are included in one block.
/// </summary> /// </summary>
@@ -15,15 +17,43 @@
/// </summary> /// </summary>
public static DefaultPageSegmenter Instance { get; } = new DefaultPageSegmenter(); public static DefaultPageSegmenter Instance { get; } = new DefaultPageSegmenter();
/// <inheritdoc />
/// <summary> /// <summary>
/// Get the blocks. /// Get the blocks using default options values.
/// </summary> /// </summary>
/// <param name="pageWords">The words in the page.</param> /// <param name="words">The page's words to generate text blocks for.</param>
public IReadOnlyList<TextBlock> GetBlocks(IEnumerable<Word> pageWords) public IReadOnlyList<TextBlock> GetBlocks(IEnumerable<Word> words)
{ {
if (pageWords.Count() == 0) return EmptyArray<TextBlock>.Instance; return GetBlocks(words, new DefaultPageSegmenterOptions());
}
return new List<TextBlock>() { new TextBlock(new XYLeaf(pageWords).GetLines()) }; /// <summary>
/// Get the text blocks using options.
/// </summary>
/// <param name="words">The page's words to generate text blocks for.</param>
/// <param name="options">The <see cref="DefaultPageSegmenterOptions"/> to use.</param>
/// <returns>The <see cref="TextBlock"/>s generated by the default method.</returns>
public IReadOnlyList<TextBlock> GetBlocks(IEnumerable<Word> words, DlaOptions options)
{
if (options is DefaultPageSegmenterOptions dOptions)
{
if (words?.Any() != true)
{
return EmptyArray<TextBlock>.Instance;
}
return new List<TextBlock>() { new TextBlock(new XYLeaf(words).GetLines(dOptions.WordSeparator), dOptions.LineSeparator) };
}
else
{
throw new ArgumentException("Options provided must be of type " + nameof(DefaultPageSegmenterOptions) + ".", nameof(options));
} }
} }
/// <summary>
/// Default page segmenter options.
/// </summary>
public class DefaultPageSegmenterOptions : PageSegmenterOptions
{ }
}
} }