/// <summary>
/// 提取word中的图片
/// </summary>
/// <param name="filePath">word文件路径</param>
/// <param name="savePath">保存文件路径</param>
/// <returns></returns>
public static IEnumerable <string> ExportImageFromWordFile(string filePath, string savePath = "")
{
if (!File.Exists(filePath))
{
yield return(string.Empty);
}
if (string.IsNullOrEmpty(savePath))
{
savePath = $"{AppDomain.CurrentDomain.BaseDirectory}\\Temp\\";
}
//加载word
Document doc = new Document(filePath);
NodeCollection shapes = doc.GetChildNodes(NodeType.Shape, true);
int imageIndex = 0;
Shape shape;
for (int i = 0; i < shapes.Count; i++)
{
shape = shapes[i] as Shape;
if (shape.HasImage)
{
//扩展名
string ex = FileFormatUtil.ImageTypeToExtension(shape.ImageData.ImageType);
//文件名
string fileName = $"{imageIndex + 1}{ex}";
shape.ImageData.Save(savePath + fileName);
yield return(fileName);
imageIndex++;
}
}
}
public void ExtractImagesToFiles()
{
//ExStart
//ExFor:Shape
//ExFor:Shape.ImageData
//ExFor:Shape.HasImage
//ExFor:ImageData
//ExFor:FileFormatUtil.ImageTypeToExtension(ImageType)
//ExFor:ImageData.ImageType
//ExFor:ImageData.Save(String)
//ExFor:CompositeNode.GetChildNodes(NodeType, bool)
//ExSummary:Shows how to extract images from a document and save them as files.
Document doc = new Document(MyDir + "Images.docx");
NodeCollection shapes = doc.GetChildNodes(NodeType.Shape, true);
Assert.AreEqual(9, shapes.Count(s => ((Shape)s).HasImage));
int imageIndex = 0;
foreach (Shape shape in shapes.OfType <Shape>())
{
if (shape.HasImage)
{
string imageFileName =
$"File.ExtractImagesToFiles.{imageIndex}{FileFormatUtil.ImageTypeToExtension(shape.ImageData.ImageType)}";
shape.ImageData.Save(ArtifactsDir + imageFileName);
imageIndex++;
}
}
//ExEnd
Assert.AreEqual(9, Directory.GetFiles(ArtifactsDir).
Count(s => Regex.IsMatch(s, @"^.+\.(jpeg|png|emf|wmf)$") && s.StartsWith(ArtifactsDir + "File.ExtractImagesToFiles")));
}
public void ExtractImageFromWordDocumentFeature()
{
Document doc = new Document(MyDir + "Extract image.docx");
// Save the document to memory and reload it.
using (MemoryStream stream = new MemoryStream())
{
doc.Save(stream, SaveFormat.Doc);
Document doc2 = new Document(stream);
// "Shape" nodes that have the "HasImage" flag set contain and display images.
IEnumerable <Shape> shapes = doc2.GetChildNodes(NodeType.Shape, true)
.OfType <Shape>().Where(s => s.HasImage);
int imageIndex = 0;
foreach (Shape shape in shapes)
{
string imageFileName =
$"Image.ExportImages.{imageIndex}_Aspose.Words_{FileFormatUtil.ImageTypeToExtension(shape.ImageData.ImageType)}";
shape.ImageData.Save(ArtifactsDir + imageFileName);
imageIndex++;
}
}
}
static void Main(string[] args)
{
string FilePath = @"..\..\..\..\Sample Files\";
string File = FilePath + "Extract Image - Aspose.docx";
Document doc = new Document(File);
// Save document as DOC in memory
MemoryStream stream = new MemoryStream();
doc.Save(stream, SaveFormat.Doc);
// Reload document as DOC to extract images.
Document doc2 = new Document(stream);
NodeCollection shapes = doc2.GetChildNodes(NodeType.Shape, true);
int imageIndex = 0;
foreach (Shape shape in shapes)
{
if (shape.HasImage)
{
string imageFileName = string.Format(
"Image.ExportImages.{0}_out_{1}", imageIndex, FileFormatUtil.ImageTypeToExtension(shape.ImageData.ImageType));
shape.ImageData.Save(FilePath + imageFileName);
imageIndex++;
}
}
}
static void Main(string[] args)
{
// Check for license and apply if exists
string licenseFile = AppDomain.CurrentDomain.BaseDirectory + "Aspose.Words.lic";
if (File.Exists(licenseFile))
{
// Apply Aspose.Words API License
Aspose.Words.License license = new Aspose.Words.License();
// Place license file in Bin/Debug/ Folder
license.SetLicense("Aspose.Words.lic");
}
string filePath = Path.GetDirectoryName(Path.GetDirectoryName(Directory.GetCurrentDirectory())) + @"\data\" + "Extract Images from Word Document.doc";
Document wordDocument = new Document(filePath);
NodeCollection pictures = wordDocument.GetChildNodes(NodeType.Shape, true);
int imageindex = 0;
foreach (Shape shape in pictures)
{
string imageType = FileFormatUtil.ImageTypeToExtension(shape.ImageData.ImageType);
if (shape.HasImage)
{
string imageFileName = "Aspose_" + (imageindex++).ToString() + "_" + shape.Name + imageType;
shape.ImageData.Save(imageFileName);
}
}
}
[Test] //ExSkip
public void ExtractImagesToFiles()
{
Aspose.Words.Document doc = new Aspose.Words.Document(MyDir + "Image.SampleImages.doc");
NodeCollection shapes = doc.GetChildNodes(NodeType.Shape, true);
int imageIndex = 0;
foreach (Shape shape in shapes)
{
if (shape.HasImage)
{
string imageFileName = string.Format(
"Image.ExportImages.{0} Out{1}", imageIndex, FileFormatUtil.ImageTypeToExtension(shape.ImageData.ImageType));
shape.ImageData.Save(MyDir + imageFileName);
imageIndex++;
}
}
}
[Test] //ExSkip
public void ExtractImagesToFiles()
{
Document doc = new Document(MyDir + "Image.SampleImages.doc");
NodeCollection shapes = doc.GetChildNodes(NodeType.Shape, true);
int imageIndex = 0;
foreach (Shape shape in shapes.OfType <Shape>())
{
if (shape.HasImage)
{
string imageFileName =
$"Image.ExportImages.{imageIndex}{FileFormatUtil.ImageTypeToExtension(shape.ImageData.ImageType)}";
shape.ImageData.Save(ArtifactsDir + imageFileName);
imageIndex++;
}
}
}
public void ExtractImages()
{
//ExStart
//ExFor:Shape
//ExFor:Shape.ImageData
//ExFor:Shape.HasImage
//ExFor:ImageData
//ExFor:FileFormatUtil.ImageTypeToExtension(ImageType)
//ExFor:ImageData.ImageType
//ExFor:ImageData.Save(String)
//ExFor:CompositeNode.GetChildNodes(NodeType, bool)
//ExSummary:Shows how to extract images from a document, and save them to the local file system as individual files.
Document doc = new Document(MyDir + "Images.docx");
// Get the collection of shapes from the document,
// and save the image data of every shape with an image as a file to the local file system.
NodeCollection shapes = doc.GetChildNodes(NodeType.Shape, true);
Assert.AreEqual(9, shapes.Count(s => ((Shape)s).HasImage));
int imageIndex = 0;
foreach (Shape shape in shapes.OfType <Shape>())
{
if (shape.HasImage)
{
// The image data of shapes may contain images of many possible image formats.
// We can determine a file extension for each image automatically, based on its format.
string imageFileName =
$"File.ExtractImages.{imageIndex}{FileFormatUtil.ImageTypeToExtension(shape.ImageData.ImageType)}";
shape.ImageData.Save(ArtifactsDir + imageFileName);
imageIndex++;
}
}
//ExEnd
Assert.AreEqual(9, Directory.GetFiles(ArtifactsDir).
Count(s => Regex.IsMatch(s, @"^.+\.(jpeg|png|emf|wmf)$") && s.StartsWith(ArtifactsDir + "File.ExtractImages")));
}
public static void Run()
{
//ExStart:ExtractImagesToFiles
// The path to the documents directory.
string dataDir = RunExamples.GetDataDir_WorkingWithImages();
Document doc = new Document(dataDir + "Image.SampleImages.doc");
NodeCollection shapes = doc.GetChildNodes(NodeType.Shape, true);
int imageIndex = 0;
foreach (Shape shape in shapes)
{
if (shape.HasImage)
{
string imageFileName = string.Format(
"Image.ExportImages.{0}_out_{1}", imageIndex, FileFormatUtil.ImageTypeToExtension(shape.ImageData.ImageType));
shape.ImageData.Save(dataDir + imageFileName);
imageIndex++;
}
}
//ExEnd:ExtractImagesToFiles
Console.WriteLine("\nAll images extracted from document.");
}
public void test()
{
string fileName = Path.Combine(System.AppDomain.CurrentDomain.SetupInformation.ApplicationBase, "二衬检测.docx");
Document doc = new Document(fileName);
var marksList = doc.Range.Bookmarks;
var pageHeader = marksList["Image1"];
Node next = pageHeader.BookmarkStart.NextSibling;
Console.WriteLine(next.GetType());
Shape shape = next as Shape;
var savePath = AppDomain.CurrentDomain.BaseDirectory;
string time = DateTime.Now.ToString("HHmmssfff");
//扩展名
string ex = FileFormatUtil.ImageTypeToExtension(shape.ImageData.ImageType);
//文件名
string filename = savePath + "\\images";
if (!Directory.Exists(filename)) //如果该文件夹不存在就建立这个新文件夹
{
Directory.CreateDirectory(filename);
}
else
{
Directory.Delete(filename, true);
Directory.CreateDirectory(filename);
}
string file_Name = string.Format("{0}_{1}{2}", time, 0, ex);
shape.ImageData.Save(Path.Combine(filename + "\\", file_Name));
//Console.WriteLine(pageHeader.GetType());
//var text = pageHeader.Text;
//Console.WriteLine(pageHeader.Text);
var table0 = doc.GetChildNodes(NodeType.Table, true)[5] as Aspose.Words.Tables.Table;
Cell cell = table0.Rows[0].Cells[2];
Console.WriteLine(cell.GetText());
Console.WriteLine(table0.FirstRow.Cells[2].GetText());
//List<string> list = new List<string>();
//var savePath = AppDomain.CurrentDomain.BaseDirectory;
//NodeCollection shapes = doc.GetChildNodes(NodeType.Shape, true);
//int imageIndex = 0;
//foreach (Shape shape in shapes)
//{
// if (shape.HasImage)
// {
// string time = DateTime.Now.ToString("HHmmssfff");
// //扩展名
// string ex = FileFormatUtil.ImageTypeToExtension(shape.ImageData.ImageType);
// //文件名
// string filename = savePath + "\\images";
// if (!Directory.Exists(filename)) //如果该文件夹不存在就建立这个新文件夹
// {
// Directory.CreateDirectory(filename);
// }else
// {
// Directory.Delete(filename, true);
// Directory.CreateDirectory(filename);
// }
// string file_Name = string.Format("{0}_{1}{2}", time, imageIndex, ex);
// shape.ImageData.Save(Path.Combine(filename + "\\", file_Name));
// //添加文件到集合
// list.Add(file_Name);
// imageIndex++;
// }
//}
//NodeCollection shapes = doc.GetChildNodes(NodeType.Shape, true);
//foreach (Shape shape in shapes)
//{
// Console.WriteLine(shape.Name);
// if(shape.Name == "Image1")
// {
// Console.WriteLine(shape.Name);
// }
//}
//odc.Save(fileName, SaveFormat.Docx);
}