Ticket #1958: html.2.patch

src/main/java/org/sophie2/extra/func/html/HtmlTextImportManager.java

@@ -1 +1 @@
 package org.sophie2.extra.func.html;
 
+import java.io.BufferedReader;
+import java.io.ByteArrayInputStream;
 import java.io.File;
+import java.io.FileNotFoundException;
 import java.io.FileReader;
 import java.io.IOException;
-import java.io.Reader;
+import java.io.InputStream;
+import java.util.regex.Matcher;
+import java.util.regex.Pattern;
 
 import javax.swing.JTextPane;
 import javax.swing.filechooser.FileFilter;
@@ -28 +33 @@
         /**
          * The file filter of the html resources.
          */
-        public static final FileFilter FILE_FILTER = 
-                new FileExtensionFilter("HTML (*.html, *.htm)", "html", "htm");
-        
+        public static final FileFilter FILE_FILTER = new FileExtensionFilter(
+                        "HTML (*.html, *.htm)", "html", "htm");
+
         /**
          * The import role of the <code>HtmlTextImportManager</code>.
          */
         public static final String IMPORT_ROLE = "import html text";
-        
+
         @Override
         public String getImportRole() {
                 return IMPORT_ROLE;
         }
-
-        @Override
+   private static final String splitBadTagStr = "<meta\\s*http-equiv" +
+                "=\\s*\"Content-Type\"\\s*content=\"text/html;";
+        private static final String splitEndTagStr = ">";
+   @Override
         public ImmHotText getResourceData(File resFile, ResourceH parentResource,
                         Object additionalData) {
-                if (resFile == null || !(resFile.getName().endsWith("html")
-                                || resFile.getName().endsWith("htm"))) {
+           
+           BufferedReader input = null;
+                if (resFile == null
+                                || !(resFile.getName().endsWith("html") || resFile.getName()
+                                                .endsWith("htm"))) {
                         return null;
                 }
                 
+                StringBuilder fileContent = new StringBuilder();
+                try {
+                         input =  new BufferedReader(new FileReader(resFile));
+                        String readLine;
+                        while ((readLine = input.readLine()) != null) {
+                                        fileContent.append(readLine);
+                        }
+
+                } catch (FileNotFoundException e) {
+                        throw new RuntimeException("Could not import HTML file", e);
+                } catch (IOException e) {
+                        throw new RuntimeException("Could not import HTML file", e);
+                }
+                Pattern badTagPattern = Pattern.compile(splitBadTagStr,Pattern.CASE_INSENSITIVE);
+                Pattern endTagPattern = Pattern.compile(splitEndTagStr);
+                Matcher badTagMatcher = badTagPattern.matcher(fileContent);
+                Matcher endTagMatcher = endTagPattern.matcher(fileContent);
+                if(badTagMatcher.find()) {
+                        int start = badTagMatcher.start();
+                        int end = badTagMatcher.end();
+                        if(endTagMatcher.find(end)) {
+                        end = endTagMatcher.end();
+                        }
+                        fileContent.replace(start, end, "");
+                }
+                
                 HTMLEditorKit kit = new HTMLEditorKit();
                 HTMLDocument doc = (HTMLDocument) kit.createDefaultDocument();
-
                 JTextPane pane = new JTextPane();
                 pane.setEditorKit(kit);
                 pane.setDocument(doc);
                 View view = pane.getUI().getRootView(pane);
-
-                Reader reader = null;
-                String htmlPlain;
                 try {
-                        reader = new FileReader(resFile);
-                        kit.read(reader, doc, 0);
-                        reader.close();
+                        InputStream is = new ByteArrayInputStream(fileContent.toString().getBytes());
+                        kit.read(is, doc, 0);
+                        is.close();
 
-                        htmlPlain = doc.getText(0, doc.getLength());
                 } catch (BadLocationException e) {
                         throw new RuntimeException("Could not import HTML file", e);
                 } catch (IOException e) {
                         throw new RuntimeException("Could not import HTML file", e);
                 }
 
-                ImmHotText text = ImmHotText.createPlain(htmlPlain);
-
-                ApplyHtmlStylesUtility.applyStyles(text, view);
+      ImmHotText text = null;
                 
+                try {
+                        ApplyHtmlStylesUtility utility = new ApplyHtmlStylesUtility();
+                        text = utility.createStyled(view);
+                } catch (BadLocationException e) {
+                        throw new RuntimeException(e);
+                }
+                
                 return text;
+                
         }
 
 }