lsm/PdfPig
1
0
Fork 0
mirror of https://github.com/UglyToad/PdfPig.git synced 2025-09-20 11:37:57 +08:00
Code Issues Actions 1 Packages Projects Releases Wiki Activity

Fix the bug that happens when all the words in the current leaf for VerticalCut/HorizontalCut are all white spaces.

Browse Source
This commit is contained in:
Zhiguan Hu
2019-12-10 15:03:05 -06:00
parent 677d2b5e8f
commit 30247ba774
1 changed files with 35 additions and 8 deletions
Download Patch File Download Diff File Expand all files Collapse all files

43
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. XYLeaf root = new XYLeaf(pageWords); // Create a root node.
XYNode node = VerticalCut(root, minimumWidth, dominantFontWidthFunc, dominantFontHeightFunc); XYNode node = VerticalCut(root, minimumWidth, dominantFontWidthFunc, dominantFontHeightFunc);
var leafs = node.GetLeafs(); if (node.IsLeaf)
if (leafs.Count > 0)
{ {
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>(); return new List<TextBlock>();
@@ -84,6 +91,18 @@ namespace UglyToad.PdfPig.DocumentLayoutAnalysis
Func<IEnumerable<decimal>, decimal> dominantFontWidthFunc, Func<IEnumerable<decimal>, decimal> dominantFontWidthFunc,
Func<IEnumerable<decimal>, decimal> dominantFontHeightFunc, int level = 0) 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) if (leaf.CountWords() <= 1 || leaf.BoundingBox.Width <= minimumWidth)
{ {
// we stop cutting if // we stop cutting if
@@ -92,9 +111,6 @@ namespace UglyToad.PdfPig.DocumentLayoutAnalysis
return leaf; 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 // determine dominantFontWidth and dominantFontHeight
decimal domFontWidth = dominantFontWidthFunc(words.SelectMany(x => x.Letters) decimal domFontWidth = dominantFontWidthFunc(words.SelectMany(x => x.Letters)
.Select(x => Math.Abs(x.GlyphRectangle.Width))); .Select(x => Math.Abs(x.GlyphRectangle.Width)));
@@ -177,6 +193,18 @@ namespace UglyToad.PdfPig.DocumentLayoutAnalysis
Func<IEnumerable<decimal>, decimal> dominantFontWidthFunc, Func<IEnumerable<decimal>, decimal> dominantFontWidthFunc,
Func<IEnumerable<decimal>, decimal> dominantFontHeightFunc, int level = 0) 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) if (leaf.CountWords() <= 1)
{ {
// we stop cutting if // we stop cutting if
@@ -184,7 +212,6 @@ namespace UglyToad.PdfPig.DocumentLayoutAnalysis
return leaf; 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 // determine dominantFontWidth and dominantFontHeight
decimal domFontWidth = dominantFontWidthFunc(words.SelectMany(x => x.Letters) decimal domFontWidth = dominantFontWidthFunc(words.SelectMany(x => x.Letters)