mirror of
https://github.com/UglyToad/PdfPig.git
synced 2026-01-09 14:04:35 +08:00
correct Word bounding box
This commit is contained in:
@@ -5,7 +5,6 @@
|
||||
using System.Collections.Generic;
|
||||
using System.Linq;
|
||||
using System.Text;
|
||||
using UglyToad.PdfPig.Geometry;
|
||||
|
||||
/// <summary>
|
||||
/// A word.
|
||||
@@ -40,7 +39,7 @@
|
||||
/// <summary>
|
||||
/// Create a new <see cref="Word"/>.
|
||||
/// </summary>
|
||||
/// <param name="letters">The letters contained in the word.</param>
|
||||
/// <param name="letters">The letters contained in the word, in the correct order.</param>
|
||||
public Word(IReadOnlyList<Letter> letters)
|
||||
{
|
||||
if (letters == null)
|
||||
@@ -69,7 +68,6 @@
|
||||
}
|
||||
|
||||
Tuple<string, PdfRectangle> data;
|
||||
|
||||
switch (tempTextOrientation)
|
||||
{
|
||||
case TextOrientation.Horizontal:
|
||||
@@ -106,162 +104,160 @@
|
||||
{
|
||||
var builder = new StringBuilder();
|
||||
|
||||
var minX = double.MaxValue;
|
||||
var maxX = double.MinValue;
|
||||
var minY = double.MaxValue;
|
||||
var maxY = double.MinValue;
|
||||
var blX = double.MaxValue;
|
||||
var trX = double.MinValue;
|
||||
var blY = double.MaxValue;
|
||||
var trY = double.MinValue;
|
||||
|
||||
for (var i = 0; i < letters.Count; i++)
|
||||
{
|
||||
var letter = letters[i];
|
||||
builder.Append(letter.Value);
|
||||
|
||||
if (letter.StartBaseLine.X < minX)
|
||||
if (letter.StartBaseLine.X < blX)
|
||||
{
|
||||
minX = letter.StartBaseLine.X;
|
||||
blX = letter.StartBaseLine.X;
|
||||
}
|
||||
|
||||
if (letter.StartBaseLine.Y < minY)
|
||||
if (letter.StartBaseLine.Y < blY)
|
||||
{
|
||||
minY = letter.StartBaseLine.Y;
|
||||
blY = letter.StartBaseLine.Y;
|
||||
}
|
||||
|
||||
var right = letter.StartBaseLine.X + Math.Max(letter.Width, letter.GlyphRectangle.Width);
|
||||
if (right > maxX)
|
||||
if (right > trX)
|
||||
{
|
||||
maxX = right;
|
||||
trX = right;
|
||||
}
|
||||
|
||||
if (letter.GlyphRectangle.Top > maxY)
|
||||
if (letter.GlyphRectangle.TopLeft.Y > trY)
|
||||
{
|
||||
maxY = letter.GlyphRectangle.Top;
|
||||
trY = letter.GlyphRectangle.TopLeft.Y;
|
||||
}
|
||||
}
|
||||
|
||||
return new Tuple<string, PdfRectangle>(builder.ToString(), new PdfRectangle(minX, minY, maxX, maxY));
|
||||
return new Tuple<string, PdfRectangle>(builder.ToString(), new PdfRectangle(blX, blY, trX, trY));
|
||||
}
|
||||
|
||||
private Tuple<string, PdfRectangle> GetBoundingBox180(IReadOnlyList<Letter> letters)
|
||||
{
|
||||
var builder = new StringBuilder();
|
||||
|
||||
var minX = double.MaxValue;
|
||||
var maxX = double.MinValue;
|
||||
var maxY = double.MinValue;
|
||||
var minY = double.MaxValue;
|
||||
var blX = double.MinValue;
|
||||
var blY = double.MinValue;
|
||||
var trX = double.MaxValue;
|
||||
var trY = double.MaxValue;
|
||||
|
||||
for (var i = 0; i < letters.Count; i++)
|
||||
{
|
||||
var letter = letters[i];
|
||||
builder.Append(letter.Value);
|
||||
|
||||
if (letter.StartBaseLine.X > maxX)
|
||||
if (letter.StartBaseLine.X > blX)
|
||||
{
|
||||
maxX = letter.StartBaseLine.X;
|
||||
blX = letter.StartBaseLine.X;
|
||||
}
|
||||
|
||||
if (letter.StartBaseLine.Y > maxY)
|
||||
if (letter.StartBaseLine.Y > blY)
|
||||
{
|
||||
maxY = letter.StartBaseLine.Y;
|
||||
blY = letter.StartBaseLine.Y;
|
||||
}
|
||||
|
||||
var right = letter.StartBaseLine.X + Math.Min(letter.Width, letter.GlyphRectangle.Width);
|
||||
if (right < minX)
|
||||
var right = letter.StartBaseLine.X - Math.Max(letter.Width, letter.GlyphRectangle.Width);
|
||||
if (right < trX)
|
||||
{
|
||||
minX = right;
|
||||
trX = right;
|
||||
}
|
||||
|
||||
if (letter.GlyphRectangle.Top < minY)
|
||||
if (letter.GlyphRectangle.TopRight.Y < trY)
|
||||
{
|
||||
minY = letter.GlyphRectangle.Top;
|
||||
trY = letter.GlyphRectangle.TopRight.Y;
|
||||
}
|
||||
}
|
||||
|
||||
return new Tuple<string, PdfRectangle>(builder.ToString(), new PdfRectangle(maxX, maxY, minX, minY));
|
||||
return new Tuple<string, PdfRectangle>(builder.ToString(), new PdfRectangle(blX, blY, trX, trY));
|
||||
}
|
||||
|
||||
private Tuple<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;
|
||||
var b = double.MaxValue;
|
||||
var r = double.MaxValue;
|
||||
var t = double.MinValue;
|
||||
var l = double.MinValue;
|
||||
|
||||
for (var i = 0; i < letters.Count; i++)
|
||||
{
|
||||
var letter = letters[i];
|
||||
builder.Append(letter.Value);
|
||||
|
||||
if (letter.StartBaseLine.X < minX)
|
||||
if (letter.StartBaseLine.X < b)
|
||||
{
|
||||
minX = letter.StartBaseLine.X;
|
||||
b = letter.StartBaseLine.X;
|
||||
}
|
||||
|
||||
if (letter.EndBaseLine.Y < minY)
|
||||
if (letter.EndBaseLine.Y < r)
|
||||
{
|
||||
minY = letter.EndBaseLine.Y;
|
||||
r = letter.EndBaseLine.Y;
|
||||
}
|
||||
|
||||
var right = letter.StartBaseLine.X - letter.GlyphRectangle.Height;
|
||||
if (right > maxX)
|
||||
var right = letter.StartBaseLine.X + letter.GlyphRectangle.Height;
|
||||
if (right > t)
|
||||
{
|
||||
maxX = right;
|
||||
t = right;
|
||||
}
|
||||
|
||||
if (letter.GlyphRectangle.Top > maxY)
|
||||
if (letter.GlyphRectangle.BottomLeft.Y > l)
|
||||
{
|
||||
maxY = letter.GlyphRectangle.Top;
|
||||
l = letter.GlyphRectangle.BottomLeft.Y;
|
||||
}
|
||||
}
|
||||
|
||||
return new Tuple<string, PdfRectangle>(builder.ToString(), new PdfRectangle(new PdfPoint(maxX, maxY),
|
||||
new PdfPoint(maxX, minY),
|
||||
new PdfPoint(minX, maxY),
|
||||
new PdfPoint(minX, minY)));
|
||||
return new Tuple<string, PdfRectangle>(builder.ToString(), new PdfRectangle(
|
||||
new PdfPoint(b, l), new PdfPoint(t, l),
|
||||
new PdfPoint(t, r), new PdfPoint(b, r)));
|
||||
}
|
||||
|
||||
private Tuple<string, PdfRectangle> GetBoundingBox270(IReadOnlyList<Letter> letters)
|
||||
{
|
||||
var builder = new StringBuilder();
|
||||
|
||||
var minX = double.MaxValue;
|
||||
var maxX = double.MinValue;
|
||||
var minY = double.MaxValue;
|
||||
var maxY = double.MinValue;
|
||||
var t = double.MaxValue;
|
||||
var b = double.MinValue;
|
||||
var l = double.MaxValue;
|
||||
var r = double.MinValue;
|
||||
|
||||
for (var i = 0; i < letters.Count; i++)
|
||||
{
|
||||
var letter = letters[i];
|
||||
builder.Append(letter.Value);
|
||||
|
||||
if (letter.StartBaseLine.X > maxX)
|
||||
if (letter.StartBaseLine.X > b)
|
||||
{
|
||||
maxX = letter.StartBaseLine.X;
|
||||
b = letter.StartBaseLine.X;
|
||||
}
|
||||
|
||||
if (letter.StartBaseLine.Y < minY)
|
||||
if (letter.StartBaseLine.Y < l)
|
||||
{
|
||||
minY = letter.StartBaseLine.Y;
|
||||
l = letter.StartBaseLine.Y;
|
||||
}
|
||||
|
||||
var right = letter.StartBaseLine.X - letter.GlyphRectangle.Height;
|
||||
if (right < minX)
|
||||
if (right < t)
|
||||
{
|
||||
minX = right;
|
||||
t = right;
|
||||
}
|
||||
|
||||
if (letter.GlyphRectangle.Bottom > maxY)
|
||||
if (letter.GlyphRectangle.BottomRight.Y > r)
|
||||
{
|
||||
maxY = letter.GlyphRectangle.Bottom;
|
||||
r = letter.GlyphRectangle.BottomRight.Y;
|
||||
}
|
||||
}
|
||||
|
||||
return new Tuple<string, PdfRectangle>(builder.ToString(), new PdfRectangle(new PdfPoint(minX, minY),
|
||||
new PdfPoint(minX, maxY),
|
||||
new PdfPoint(maxX, minY),
|
||||
new PdfPoint(maxX, maxY)));
|
||||
return new Tuple<string, PdfRectangle>(builder.ToString(), new PdfRectangle(
|
||||
new PdfPoint(b, l), new PdfPoint(t, l),
|
||||
new PdfPoint(t, r), new PdfPoint(b, r)));
|
||||
}
|
||||
|
||||
private Tuple<string, PdfRectangle> GetBoundingBoxOther(IReadOnlyList<Letter> letters)
|
||||
@@ -272,97 +268,121 @@
|
||||
builder.Append(letters[i].Value);
|
||||
}
|
||||
|
||||
var baseLinePoints = letters.SelectMany(r => new[]
|
||||
if (letters.Count == 1)
|
||||
{
|
||||
r.StartBaseLine,
|
||||
r.EndBaseLine,
|
||||
}).ToList();
|
||||
|
||||
// Fitting a line through the base lines points
|
||||
// to find the orientation (slope)
|
||||
double x0 = baseLinePoints.Average(p => p.X);
|
||||
double y0 = baseLinePoints.Average(p => p.Y);
|
||||
double sumProduct = 0;
|
||||
double sumDiffSquaredX = 0;
|
||||
|
||||
for (int i = 0; i < baseLinePoints.Count; i++)
|
||||
{
|
||||
var point = baseLinePoints[i];
|
||||
var x_diff = point.X - x0;
|
||||
var y_diff = point.Y - y0;
|
||||
sumProduct += x_diff * y_diff;
|
||||
sumDiffSquaredX += x_diff * x_diff;
|
||||
return new Tuple<string, PdfRectangle>(builder.ToString(), letters[0].GlyphRectangle);
|
||||
}
|
||||
|
||||
var slope = sumProduct / sumDiffSquaredX;
|
||||
|
||||
// Rotate the points to build the axis-aligned bounding box (AABB)
|
||||
var angleRad = Math.Atan(slope);
|
||||
var cos = Math.Cos(angleRad);
|
||||
var sin = Math.Sin(angleRad);
|
||||
|
||||
var inverseRotation = new TransformationMatrix(
|
||||
cos, -sin, 0,
|
||||
sin, cos, 0,
|
||||
0, 0, 1);
|
||||
|
||||
var transformedPoints = letters.SelectMany(r => new[]
|
||||
else
|
||||
{
|
||||
r.StartBaseLine,
|
||||
r.EndBaseLine,
|
||||
r.GlyphRectangle.TopLeft,
|
||||
r.GlyphRectangle.TopRight
|
||||
}).Distinct().Select(p => inverseRotation.Transform(p));
|
||||
var aabb = new PdfRectangle(transformedPoints.Min(p => p.X),
|
||||
transformedPoints.Min(p => p.Y),
|
||||
transformedPoints.Max(p => p.X),
|
||||
transformedPoints.Max(p => p.Y));
|
||||
var baseLinePoints = letters.SelectMany(r => new[]
|
||||
{
|
||||
r.StartBaseLine,
|
||||
r.EndBaseLine,
|
||||
}).ToList();
|
||||
|
||||
// Rotate back the AABB to obtain to oriented bounding box (OBB)
|
||||
var rotateBack = new TransformationMatrix(
|
||||
cos, sin, 0,
|
||||
-sin, cos, 0,
|
||||
0, 0, 1);
|
||||
// Fitting a line through the base lines points
|
||||
// to find the orientation (slope)
|
||||
double x0 = baseLinePoints.Average(p => p.X);
|
||||
double y0 = baseLinePoints.Average(p => p.Y);
|
||||
double sumProduct = 0;
|
||||
double sumDiffSquaredX = 0;
|
||||
|
||||
// Candidates bounding boxes
|
||||
var obb = rotateBack.Transform(aabb);
|
||||
var obb1 = new PdfRectangle(obb.BottomLeft, obb.TopLeft, obb.BottomRight, obb.TopRight);
|
||||
var obb2 = new PdfRectangle(obb.TopRight, obb.BottomRight, obb.TopLeft, obb.BottomLeft);
|
||||
var obb3 = new PdfRectangle(obb.BottomRight, obb.BottomLeft, obb.TopRight, obb.TopLeft);
|
||||
for (int i = 0; i < baseLinePoints.Count; i++)
|
||||
{
|
||||
var point = baseLinePoints[i];
|
||||
var x_diff = point.X - x0;
|
||||
var y_diff = point.Y - y0;
|
||||
sumProduct += x_diff * y_diff;
|
||||
sumDiffSquaredX += x_diff * x_diff;
|
||||
}
|
||||
|
||||
// Find the orientation of the OBB, using the baseline angle
|
||||
var firstLetter = letters[0];
|
||||
var lastLetter = letters[letters.Count - 1];
|
||||
double cos = 0;
|
||||
double sin = 1;
|
||||
if (sumDiffSquaredX > 1e-3)
|
||||
{
|
||||
// not vertical line
|
||||
double angleRad = Math.Atan(sumProduct / sumDiffSquaredX); // -π/2 ≤ θ ≤ π/2
|
||||
cos = Math.Cos(angleRad);
|
||||
sin = Math.Sin(angleRad);
|
||||
}
|
||||
|
||||
var baseLineAngle = Math.Atan2(
|
||||
lastLetter.EndBaseLine.Y - firstLetter.StartBaseLine.Y,
|
||||
lastLetter.EndBaseLine.X - firstLetter.StartBaseLine.X) * 180 / Math.PI;
|
||||
// Rotate the points to build the axis-aligned bounding box (AABB)
|
||||
var inverseRotation = new TransformationMatrix(
|
||||
cos, -sin, 0,
|
||||
sin, cos, 0,
|
||||
0, 0, 1);
|
||||
|
||||
double deltaAngle = Math.Abs(baseLineAngle - obb.Rotation);
|
||||
double deltaAngle1 = Math.Abs(baseLineAngle - obb1.Rotation);
|
||||
if (deltaAngle1 < deltaAngle)
|
||||
{
|
||||
deltaAngle = deltaAngle1;
|
||||
obb = obb1;
|
||||
var transformedPoints = letters.SelectMany(r => new[]
|
||||
{
|
||||
r.StartBaseLine,
|
||||
r.EndBaseLine,
|
||||
r.GlyphRectangle.TopLeft,
|
||||
r.GlyphRectangle.TopRight
|
||||
}).Distinct().Select(p => inverseRotation.Transform(p));
|
||||
|
||||
var aabb = new PdfRectangle(transformedPoints.Min(p => p.X),
|
||||
transformedPoints.Min(p => p.Y),
|
||||
transformedPoints.Max(p => p.X),
|
||||
transformedPoints.Max(p => p.Y));
|
||||
|
||||
// Rotate back the AABB to obtain to oriented bounding box (OBB)
|
||||
var rotateBack = new TransformationMatrix(
|
||||
cos, sin, 0,
|
||||
-sin, cos, 0,
|
||||
0, 0, 1);
|
||||
|
||||
// Candidates bounding boxes
|
||||
var obb = rotateBack.Transform(aabb);
|
||||
var obb1 = new PdfRectangle(obb.BottomRight, obb.BottomLeft, obb.TopLeft, obb.TopRight);
|
||||
var obb2 = new PdfRectangle(obb.TopRight, obb.BottomRight, obb.BottomLeft, obb.TopLeft);
|
||||
var obb3 = new PdfRectangle(obb.TopLeft, obb.TopRight, obb.BottomRight, obb.BottomLeft);
|
||||
|
||||
// Find the orientation of the OBB, using the baseline angle
|
||||
// Assumes word order is correct
|
||||
var firstLetter = letters[0];
|
||||
var lastLetter = letters[letters.Count - 1];
|
||||
|
||||
var baseLineAngle = Math.Atan2(
|
||||
lastLetter.EndBaseLine.Y - firstLetter.StartBaseLine.Y,
|
||||
lastLetter.EndBaseLine.X - firstLetter.StartBaseLine.X) * 180 / Math.PI;
|
||||
|
||||
double deltaAngle = Math.Abs(BoundAngle180(obb.Rotation - baseLineAngle));
|
||||
double deltaAngle1 = Math.Abs(BoundAngle180(obb1.Rotation - baseLineAngle));
|
||||
if (deltaAngle1 < deltaAngle)
|
||||
{
|
||||
deltaAngle = deltaAngle1;
|
||||
obb = obb1;
|
||||
}
|
||||
|
||||
double deltaAngle2 = Math.Abs(BoundAngle180(obb2.Rotation - baseLineAngle));
|
||||
if (deltaAngle2 < deltaAngle)
|
||||
{
|
||||
deltaAngle = deltaAngle2;
|
||||
obb = obb2;
|
||||
}
|
||||
|
||||
double deltaAngle3 = Math.Abs(BoundAngle180(obb3.Rotation - baseLineAngle));
|
||||
if (deltaAngle3 < deltaAngle)
|
||||
{
|
||||
obb = obb3;
|
||||
}
|
||||
|
||||
return new Tuple<string, PdfRectangle>(builder.ToString(), obb);
|
||||
}
|
||||
|
||||
double deltaAngle2 = Math.Abs(baseLineAngle - obb2.Rotation);
|
||||
if (deltaAngle2 < deltaAngle)
|
||||
{
|
||||
deltaAngle = deltaAngle2;
|
||||
obb = obb2;
|
||||
}
|
||||
|
||||
double deltaAngle3 = Math.Abs(baseLineAngle - obb3.Rotation);
|
||||
if (deltaAngle3 < deltaAngle)
|
||||
{
|
||||
obb = obb3;
|
||||
}
|
||||
|
||||
return new Tuple<string, PdfRectangle>(builder.ToString(), obb);
|
||||
}
|
||||
#endregion
|
||||
|
||||
/// <summary>
|
||||
/// Bound angle so that -180 ≤ θ ≤ 180.
|
||||
/// </summary>
|
||||
/// <param name="angle">The angle to bound.</param>
|
||||
private static double BoundAngle180(double angle)
|
||||
{
|
||||
angle = (angle + 180) % 360;
|
||||
if (angle < 0) angle += 360;
|
||||
return angle - 180;
|
||||
}
|
||||
|
||||
/// <inheritdoc />
|
||||
public override string ToString()
|
||||
{
|
||||
|
||||
Reference in New Issue
Block a user