lsm/PdfPig
1
0
Fork 0
mirror of https://github.com/UglyToad/PdfPig.git synced 2025-10-15 03:34:52 +08:00
Code Issues Actions 5 Packages Projects Releases Wiki Activity

Fix NearestNeighbourWordExtractor for rotated text

Browse Source
This commit is contained in:
BobLd
2020-01-16 14:43:06 +00:00
committed by Eliot Jones
parent 8ab2838063
commit 75821919a7
4 changed files with 216 additions and 62 deletions
Download Patch File Download Diff File Expand all files Collapse all files

52
src/UglyToad.PdfPig.Core/PdfRectangle.cs
View File

@@ -73,51 +73,31 @@
/// <summary>
/// Create a new <see cref="PdfRectangle"/>.
/// </summary>
public PdfRectangle(PdfPoint point1, PdfPoint point2) : this(point1.X, point1.Y, point2.X, point2.Y) { }
/// <param name="bottomLeft">Bottom left point of the rectangle.</param>
/// <param name="topRight">Top right point of the rectangle.</param>
public PdfRectangle(PdfPoint bottomLeft, PdfPoint topRight) :
this(bottomLeft.X, bottomLeft.Y, topRight.X, topRight.Y)
{ }
/// <summary>
/// Create a new <see cref="PdfRectangle"/>.
/// </summary>
/// <param name="x1">Bottom left point's x coordinate of the rectangle.</param>
/// <param name="y1">Bottom left point's y coordinate of the rectangle.</param>
/// <param name="x2">Top right point's x coordinate of the rectangle.</param>
/// <param name="y2">Top right point's y coordinate of the rectangle.</param>
public PdfRectangle(short x1, short y1, short x2, short y2) : this((double)x1, y1, x2, y2) { }
/// <summary>
/// Create a new <see cref="PdfRectangle"/>.
/// </summary>
public PdfRectangle(double x1, double y1, double x2, double y2)
{
double bottom;
double top;
if (y1 <= y2)
{
bottom = y1;
top = y2;
}
else
{
bottom = y2;
top = y1;
}
double left;
double right;
if (x1 <= x2)
{
left = x1;
right = x2;
}
else
{
left = x2;
right = x1;
}
TopLeft = new PdfPoint(left, top);
TopRight = new PdfPoint(right, top);
BottomLeft = new PdfPoint(left, bottom);
BottomRight = new PdfPoint(right, bottom);
}
/// <param name="x1">Bottom left point's x coordinate of the rectangle.</param>
/// <param name="y1">Bottom left point's y coordinate of the rectangle.</param>
/// <param name="x2">Top right point's x coordinate of the rectangle.</param>
/// <param name="y2">Top right point's y coordinate of the rectangle.</param>
public PdfRectangle(double x1, double y1, double x2, double y2) :
this(new PdfPoint(x1, y2), new PdfPoint(x2, y2), new PdfPoint(x1, y1), new PdfPoint(x2, y1))
{ }
/// <summary>
/// Create a new <see cref="PdfRectangle"/>.

18
src/UglyToad.PdfPig.DocumentLayoutAnalysis/TextLine.cs
View File

@@ -51,10 +51,12 @@
Text = string.Join(" ", words.Where(s => !string.IsNullOrWhiteSpace(s.Text)).Select(x => x.Text));
var minX = words.Min(x => x.BoundingBox.Left);
var minY = words.Min(x => x.BoundingBox.Bottom);
var maxX = words.Max(x => x.BoundingBox.Right);
var maxY = words.Max(x => x.BoundingBox.Top);
var normalisedBoundingBoxes = words.Select(x => NormaliseRectangle(x.BoundingBox)).ToList();
var minX = normalisedBoundingBoxes.Min(x => x.Left);
var minY = normalisedBoundingBoxes.Min(x => x.Bottom);
var maxX = normalisedBoundingBoxes.Max(x => x.Right);
var maxY = normalisedBoundingBoxes.Max(x => x.Top);
BoundingBox = new PdfRectangle(minX, minY, maxX, maxY);
if (words.All(x => x.TextDirection == words[0].TextDirection))
@@ -67,6 +69,14 @@
}
}
private PdfRectangle NormaliseRectangle(PdfRectangle rectangle)
{
return new PdfRectangle(Math.Min(rectangle.Left, rectangle.Right),
Math.Min(rectangle.Bottom, rectangle.Top),
Math.Max(rectangle.Left, rectangle.Right),
Math.Max(rectangle.Bottom, rectangle.Top));
}
/// <inheritdoc />
public override string ToString()
{

23
src/UglyToad.PdfPig.DocumentLayoutAnalysis/WordExtractor/NearestNeighbourWordExtractor.cs
View File

@@ -38,9 +38,17 @@
.OrderByDescending(x => x.BoundingBox.Bottom)
.ThenBy(x => x.BoundingBox.Left).ToList();
List<Word> words270 = GetWords(
letters.Where(l => l.TextDirection == TextDirection.Rotate270),
(l1, l2) => Math.Max(l1.GlyphRectangle.Width, l2.GlyphRectangle.Width) * 0.2,
Distances.Manhattan, MaxDegreeOfParallelism)
.OrderBy(x => x.BoundingBox.Right)
.ThenByDescending(x => x.BoundingBox.Bottom).ToList();
wordsH.AddRange(words270);
List<Word> words180 = GetWords(
letters.Where(l => l.TextDirection == TextDirection.Rotate180),
(l1, l2) => Math.Max(l1.GlyphRectangle.Width, l2.GlyphRectangle.Width) * 0.2,
(l1, l2) => Math.Max(Math.Abs(l1.GlyphRectangle.Width), Math.Abs(l2.GlyphRectangle.Width)) * 0.2,
Distances.Manhattan, MaxDegreeOfParallelism)
.OrderBy(x => x.BoundingBox.Top)
.ThenByDescending(x => x.BoundingBox.Right).ToList();
@@ -48,20 +56,12 @@
List<Word> words90 = GetWords(
letters.Where(l => l.TextDirection == TextDirection.Rotate90),
(l1, l2) => Math.Max(l1.GlyphRectangle.Height, l2.GlyphRectangle.Height) * 0.2,
(l1, l2) => Math.Max(Math.Abs(l1.GlyphRectangle.Width), Math.Abs(l2.GlyphRectangle.Width)) * 0.2,
Distances.Manhattan, MaxDegreeOfParallelism)
.OrderByDescending(x => x.BoundingBox.Left)
.ThenBy(x => x.BoundingBox.Top).ToList();
wordsH.AddRange(words90);
List<Word> words270 = GetWords(
letters.Where(l => l.TextDirection == TextDirection.Rotate270),
(l1, l2) => Math.Max(l1.GlyphRectangle.Height, l2.GlyphRectangle.Height) * 0.2,
Distances.Manhattan, MaxDegreeOfParallelism)
.OrderBy(x => x.BoundingBox.Right)
.ThenByDescending(x => x.BoundingBox.Bottom).ToList();
wordsH.AddRange(words270);
List<Word> wordsU = GetWords(
letters.Where(l => l.TextDirection == TextDirection.Unknown),
(l1, l2) => Math.Max(l1.GlyphRectangle.Width, l2.GlyphRectangle.Width) * 0.2,
@@ -85,7 +85,7 @@
/// <param name="maxDegreeOfParallelism">Sets the maximum number of concurrent tasks enabled.
/// <para>A positive property value limits the number of concurrent operations to the set value.
/// If it is -1, there is no limit on the number of concurrently running operations.</para></param>
private List<Word> GetWords(IEnumerable<Letter> pageLetters,
public List<Word> GetWords(IEnumerable<Letter> pageLetters,
Func<Letter, Letter, double> maxDistanceFunction, Func<PdfPoint, PdfPoint, double> distMeasure,
int maxDegreeOfParallelism)
{
@@ -97,6 +97,7 @@
throw new ArgumentException("NearestNeighbourWordExtractor.GetWords(): Mixed Text Direction.");
}
// TO DO: orderFunc should also take in account the edge relationships found by 'ClusterNearestNeighbours'
Func<IEnumerable<Letter>, IReadOnlyList<Letter>> orderFunc = l => l.OrderBy(x => x.GlyphRectangle.Left).ToList();
if (textDirection == TextDirection.Rotate180)
{

185
src/UglyToad.PdfPig/Content/Word.cs
View File

@@ -2,6 +2,7 @@
{
using System;
using System.Collections.Generic;
using System.Linq;
using System.Text;
using Core;
@@ -53,11 +54,53 @@
Letters = letters;
var tempTextDirection = letters[0].TextDirection;
if (letters.Any(l => l.TextDirection != tempTextDirection))
{
tempTextDirection = TextDirection.Unknown;
}
(string, PdfRectangle) data;
switch (tempTextDirection)
{
case TextDirection.Horizontal:
data = GetBoundingBoxH(letters);
break;
case TextDirection.Rotate180:
data = GetBoundingBox180(letters);
break;
case TextDirection.Rotate90:
data = GetBoundingBox90(letters);
break;
case TextDirection.Rotate270:
data = GetBoundingBox270(letters);
break;
case TextDirection.Unknown:
default:
data = GetBoundingBoxH(letters);
break;
}
Text = data.Item1;
BoundingBox = data.Item2;
FontName = letters[0].FontName;
TextDirection = tempTextDirection;
}
#region Bounding box
private (string, PdfRectangle) GetBoundingBoxH(IReadOnlyList<Letter> letters)
{
var builder = new StringBuilder();
var minX = double.MaxValue;
var minY = double.MaxValue;
var maxX = double.MinValue;
var minY = double.MaxValue;
var maxY = double.MinValue;
for (var i = 0; i < letters.Count; i++)
@@ -65,17 +108,17 @@
var letter = letters[i];
builder.Append(letter.Value);
if (letter.Location.X < minX)
if (letter.StartBaseLine.X < minX)
{
minX = letter.Location.X;
minX = letter.StartBaseLine.X;
}
if (letter.Location.Y < minY)
if (letter.StartBaseLine.Y < minY)
{
minY = letter.Location.Y;
minY = letter.StartBaseLine.Y;
}
var right = letter.Location.X + letter.Width;
var right = letter.StartBaseLine.X + letter.GlyphRectangle.Width;
if (right > maxX)
{
maxX = right;
@@ -87,13 +130,133 @@
}
}
Text = builder.ToString();
BoundingBox = new PdfRectangle(minX, minY, maxX, maxY);
FontName = letters[0].FontName;
TextDirection = letters[0].TextDirection;
return (builder.ToString(), new PdfRectangle(minX, minY, maxX, maxY));
}
private (string, PdfRectangle) GetBoundingBox180(IReadOnlyList<Letter> letters)
{
var builder = new StringBuilder();
var maxX = double.MinValue;
var minX = double.MaxValue;
var maxY = double.MinValue;
var minY = double.MaxValue;
for (var i = 0; i < letters.Count; i++)
{
var letter = letters[i];
builder.Append(letter.Value);
if (letter.StartBaseLine.X > maxX)
{
maxX = letter.StartBaseLine.X;
}
if (letter.StartBaseLine.Y > maxY)
{
maxY = letter.StartBaseLine.Y;
}
var right = letter.StartBaseLine.X + letter.GlyphRectangle.Width;
if (right < minX)
{
minX = right;
}
if (letter.GlyphRectangle.Top < minY)
{
minY = letter.GlyphRectangle.Top;
}
}
return (builder.ToString(), new PdfRectangle(maxX, maxY, minX, minY));
}
private (string, PdfRectangle) GetBoundingBox90(IReadOnlyList<Letter> letters)
{
var builder = new StringBuilder();
var minX = double.MaxValue;
var maxX = double.MinValue;
var minY = double.MaxValue;
var maxY = double.MinValue;
for (var i = 0; i < letters.Count; i++)
{
var letter = letters[i];
builder.Append(letter.Value);
if (letter.StartBaseLine.X < minX)
{
minX = letter.StartBaseLine.X;
}
if (letter.EndBaseLine.Y < minY)
{
minY = letter.EndBaseLine.Y;
}
var right = letter.StartBaseLine.X - letter.GlyphRectangle.Width;
if (right > maxX)
{
maxX = right;
}
if (letter.GlyphRectangle.Top > maxY)
{
maxY = letter.GlyphRectangle.Top;
}
}
return (builder.ToString(), new PdfRectangle(new PdfPoint(maxX, maxY),
new PdfPoint(maxX, minY),
new PdfPoint(minX, maxY),
new PdfPoint(minX, minY)));
}
private (string, PdfRectangle) GetBoundingBox270(IReadOnlyList<Letter> letters)
{
var builder = new StringBuilder();
var maxX = double.MinValue;
var minX = double.MaxValue;
var minY = double.MaxValue;
var maxY = double.MinValue;
for (var i = 0; i < letters.Count; i++)
{
var letter = letters[i];
builder.Append(letter.Value);
if (letter.StartBaseLine.X > maxX)
{
maxX = letter.StartBaseLine.X;
}
if (letter.StartBaseLine.Y < minY)
{
minY = letter.StartBaseLine.Y;
}
var right = letter.StartBaseLine.X - letter.GlyphRectangle.Width;
if (right < minX)
{
minX = right;
}
if (letter.GlyphRectangle.Bottom > maxY)
{
maxY = letter.GlyphRectangle.Bottom;
}
}
return (builder.ToString(), new PdfRectangle(new PdfPoint(minX, minY),
new PdfPoint(minX, maxY),
new PdfPoint(maxX, minY),
new PdfPoint(maxX, maxY)));
}
#endregion
/// <inheritdoc />
public override string ToString()
{