lsm/PdfPig
1
0
Fork 0
mirror of https://github.com/UglyToad/PdfPig.git synced 2025-10-15 11:44:51 +08:00
Code Issues Actions 5 Packages Projects Releases Wiki Activity

Merge pull request #100 from huzhiguan/master

Browse Source 
fix issue with leaf/nodes in RecursiveXYCut that only have whitespaces words.
This commit is contained in:
Eliot Jones
2019-12-14 16:44:00 +00:00
committed by GitHub
parent af1217f910 9baa8c3ca0
commit 7a6b8ce6d6
1 changed files with 36 additions and 9 deletions
Download Patch File Download Diff File Expand all files Collapse all files

45
src/UglyToad.PdfPig/DocumentLayoutAnalysis/RecursiveXYCut.cs
View File

@@ -70,11 +70,18 @@ namespace UglyToad.PdfPig.DocumentLayoutAnalysis
XYLeaf root = new XYLeaf(pageWords); // Create a root node.
XYNode node = VerticalCut(root, minimumWidth, dominantFontWidthFunc, dominantFontHeightFunc);
var leafs = node.GetLeafs();
if (leafs.Count > 0)
if (node.IsLeaf)
{
return leafs.Select(l => new TextBlock(l.GetLines())).ToList();
return new List<TextBlock>{ new TextBlock((node as XYLeaf).GetLines())};
}
else
{
var leafs = node.GetLeafs();
if (leafs.Count > 0)
{
return leafs.Select(l => new TextBlock(l.GetLines())).ToList();
}
}
return new List<TextBlock>();
@@ -84,6 +91,19 @@ namespace UglyToad.PdfPig.DocumentLayoutAnalysis
Func<IEnumerable<decimal>, decimal> dominantFontWidthFunc,
Func<IEnumerable<decimal>, decimal> dominantFontHeightFunc, int level = 0)
{
// order words left to right
var words = leaf.Words.Where(w => !string.IsNullOrWhiteSpace(w.Text)).OrderBy(w => w.BoundingBox.Left).ToArray();
if(!words.Any())
{
return new XYNode(null);
}
else
{
//Create new leaf with non-whitespace words.
leaf = new XYLeaf(words);
}
if (leaf.CountWords() <= 1 || leaf.BoundingBox.Width <= minimumWidth)
{
// we stop cutting if
@@ -92,9 +112,6 @@ namespace UglyToad.PdfPig.DocumentLayoutAnalysis
return leaf;
}
// order words left to right
var words = leaf.Words.Where(w => !string.IsNullOrWhiteSpace(w.Text)).OrderBy(w => w.BoundingBox.Left).ToArray();
// determine dominantFontWidth and dominantFontHeight
decimal domFontWidth = dominantFontWidthFunc(words.SelectMany(x => x.Letters)
.Select(x => Math.Abs(x.GlyphRectangle.Width)));
@@ -177,6 +194,18 @@ namespace UglyToad.PdfPig.DocumentLayoutAnalysis
Func<IEnumerable<decimal>, decimal> dominantFontWidthFunc,
Func<IEnumerable<decimal>, decimal> dominantFontHeightFunc, int level = 0)
{
var words = leaf.Words.Where(w => !string.IsNullOrWhiteSpace(w.Text)).OrderBy(w => w.BoundingBox.Bottom).ToArray(); // order bottom to top
if (!words.Any())
{
return new XYNode(null);
}
else
{
//Create new leaf with non-whitespace words.
leaf = new XYLeaf(words);
}
if (leaf.CountWords() <= 1)
{
// we stop cutting if
@@ -184,8 +213,6 @@ namespace UglyToad.PdfPig.DocumentLayoutAnalysis
return leaf;
}
var words = leaf.Words.Where(w => !string.IsNullOrWhiteSpace(w.Text)).OrderBy(w => w.BoundingBox.Bottom).ToArray(); // order bottom to top
// determine dominantFontWidth and dominantFontHeight
decimal domFontWidth = dominantFontWidthFunc(words.SelectMany(x => x.Letters)
.Select(x => Math.Abs(x.GlyphRectangle.Width)));