public static String Extract(String url, String saveTo = null)
{
try
{
WebClient wc = new WebClient();
byte[] data = wc.DownloadData(url);
// 如果设置了保存到文件，则写入文件
if (!String.IsNullOrWhiteSpace(saveTo) && data != null && data.Length > 0)
{
if (!Directory.Exists(saveTo))
{
Directory.CreateDirectory(saveTo);
}
String file = GeneratePath(saveTo, url);
try
{
using (FileStream fs = new FileStream(file, FileMode.CreateNew))
{
fs.Write(data, 0, data.Length);
}
}
catch (IOException ioe)
{
}
}
using (PdfReader reader = new PdfReader(data))
{
try
{
StringBuilder sb = new StringBuilder();
ITextExtractionStrategy extract = new SimpleTextExtractionStrategy();
for (int i = 1; i <= reader.NumberOfPages; i++)
{
sb.Append(PdfTextExtractor.GetTextFromPage(reader, i, extract));
reader.ReleasePage(i);
}
return sb.ToString();
}
finally
{
if (reader != null)
{
reader.Close();
}
}
}
}
catch (Exception e)
{
return String.Empty;
}
}
/**
* Gets a <CODE>List</CODE> with the bookmarks that are children of <CODE>outline</CODE>. It returns <CODE>null</CODE> if
* the document doesn't have any bookmarks.
* @param reader the document
* @param outline the outline dictionary to get bookmarks from
* @param includeRoot indicates if to include <CODE>outline</CODE> parameter itself into returned list of bookmarks
* @return a <CODE>List</CODE> with the bookmarks or <CODE>null</CODE> if the
* document doesn't have any
*/
public static IList <Dictionary <String, Object> > GetBookmark(PdfReader reader, PdfDictionary outline, bool includeRoot)
{
if (outline == null)
{
return(null);
}
IntHashtable pages = new IntHashtable();
int numPages = reader.NumberOfPages;
for (int k = 1; k <= numPages; ++k)
{
pages[reader.GetPageOrigRef(k).Number] = k;
reader.ReleasePage(k);
}
if (includeRoot)
{
return(BookmarkDepth(reader, outline, pages, true));
}
else
{
return(BookmarkDepth(reader, (PdfDictionary)PdfReader.GetPdfObjectRelease(outline.Get(PdfName.FIRST)), pages, false));
}
}
/**
* Gets a <CODE>List</CODE> with the bookmarks that are children of <CODE>outline</CODE>. It returns <CODE>null</CODE> if
* the document doesn't have any bookmarks.
* @param reader the document
* @param outline the outline dictionary to get bookmarks from
* @param includeRoot indicates if to include <CODE>outline</CODE> parameter itself into returned list of bookmarks
* @return a <CODE>List</CODE> with the bookmarks or <CODE>null</CODE> if the
* document doesn't have any
*/
public static IList<Dictionary<String, Object>> GetBookmark(PdfReader reader, PdfDictionary outline, bool includeRoot) {
if (outline == null)
return null;
IntHashtable pages = new IntHashtable();
int numPages = reader.NumberOfPages;
for (int k = 1; k <= numPages; ++k) {
pages[reader.GetPageOrigRef(k).Number] = k;
reader.ReleasePage(k);
}
if (includeRoot)
return BookmarkDepth(reader, outline, pages, true);
else
return BookmarkDepth(reader, (PdfDictionary)PdfReader.GetPdfObjectRelease(outline.Get(PdfName.FIRST)), pages, false);
}
/**
* Gets a <CODE>List</CODE> with the bookmarks. It returns <CODE>null</CODE> if
* the document doesn't have any bookmarks.
* @param reader the document
* @return a <CODE>List</CODE> with the bookmarks or <CODE>null</CODE> if the
* document doesn't have any
*/
public static IList<Dictionary<String, Object>> GetBookmark(PdfReader reader) {
PdfDictionary catalog = reader.Catalog;
PdfObject obj = PdfReader.GetPdfObjectRelease(catalog.Get(PdfName.OUTLINES));
if (obj == null || !obj.IsDictionary())
return null;
PdfDictionary outlines = (PdfDictionary)obj;
IntHashtable pages = new IntHashtable();
int numPages = reader.NumberOfPages;
for (int k = 1; k <= numPages; ++k) {
pages[reader.GetPageOrigRef(k).Number] = k;
reader.ReleasePage(k);
}
return BookmarkDepth(reader, (PdfDictionary)PdfReader.GetPdfObjectRelease(outlines.Get(PdfName.FIRST)), pages);
}
