fpspreadsheet: Improved auto-format detection (check extension first, check file header when extension test fails). Fix ods reader crashing when extension has been renamed.

git-svn-id: https://svn.code.sf.net/p/lazarus-ccr/svn@6760 8e941d3f-bd1b-0410-a28a-d453659cc2b4
2018-12-18 00:07:16 +00:00
parent ca80c68d6b
commit 02281cd15f
3 changed files with 108 additions and 76 deletions
--- a/components/fpspreadsheet/source/common/fpsopendocument.pas
+++ b/components/fpspreadsheet/source/common/fpsopendocument.pas
@ -2613,97 +2613,103 @@ begin
      XMLStream.Free;
    end;

-    ReadFontFaces(Doc.DocumentElement.FindNode('office:font-face-decls'));
+    if Assigned(Doc) then begin
+      ReadFontFaces(Doc.DocumentElement.FindNode('office:font-face-decls'));

-    StylesNode := Doc.DocumentElement.FindNode('office:styles');
-    ReadNumFormats(StylesNode);
-    ReadStyles(StylesNode);
-    ReadAutomaticStyles(Doc.DocumentElement.FindNode('office:automatic-styles'));
-    ReadMasterStyles(Doc.DocumentElement.FindNode('office:master-styles'));
-    FreeAndNil(Doc);
+      StylesNode := Doc.DocumentElement.FindNode('office:styles');
+      ReadNumFormats(StylesNode);
+      ReadStyles(StylesNode);
+      ReadAutomaticStyles(Doc.DocumentElement.FindNode('office:automatic-styles'));
+      ReadMasterStyles(Doc.DocumentElement.FindNode('office:master-styles'));
+      FreeAndNil(Doc);
+    end;

    //process the content.xml file
    XMLStream := CreateXMLStream;
    try
      if UnzipToStream(AStream, 'content.xml', XMLStream) then
-        ReadXMLStream(Doc, XMLStream);
+        ReadXMLStream(Doc, XMLStream)
+      else
+        raise EFPSpreadsheetReader.CreateFmt(rsDefectiveInternalFileStructure, ['ods']);
    finally
      XMLStream.Free;
    end;

-    ReadFontFaces(Doc.DocumentElement.FindNode('office:font-face-decls'));
-    StylesNode := Doc.DocumentElement.FindNode('office:automatic-styles');
-    ReadNumFormats(StylesNode);
-    ReadStyles(StylesNode);
+    if Assigned(Doc) then begin
+      ReadFontFaces(Doc.DocumentElement.FindNode('office:font-face-decls'));
+      StylesNode := Doc.DocumentElement.FindNode('office:automatic-styles');
+      ReadNumFormats(StylesNode);
+      ReadStyles(StylesNode);

-    BodyNode := Doc.DocumentElement.FindNode('office:body');
-    if not Assigned(BodyNode) then
-      raise EFPSpreadsheet.Create('[TsSpreadOpenDocReader.ReadFromStream] Node "office:body" not found.');
+      BodyNode := Doc.DocumentElement.FindNode('office:body');
+      if not Assigned(BodyNode) then
+        raise EFPSpreadsheet.Create('[TsSpreadOpenDocReader.ReadFromStream] Node "office:body" not found.');

-    SpreadSheetNode := BodyNode.FindNode('office:spreadsheet');
-    if not Assigned(SpreadSheetNode) then
-      raise EFPSpreadsheet.Create('[TsSpreadOpenDocReader.ReadFromStream] Node "office:spreadsheet" not found.');
+      SpreadSheetNode := BodyNode.FindNode('office:spreadsheet');
+      if not Assigned(SpreadSheetNode) then
+        raise EFPSpreadsheet.Create('[TsSpreadOpenDocReader.ReadFromStream] Node "office:spreadsheet" not found.');

-    ReadSheets(SpreadsheetNode);
-    ReadDocumentProtection(SpreadsheetNode);
-    ReadDateMode(SpreadSheetNode);
+      ReadSheets(SpreadsheetNode);
+      ReadDocumentProtection(SpreadsheetNode);
+      ReadDateMode(SpreadSheetNode);

-    //process each table (sheet)
-    TableNode := SpreadSheetNode.FindNode('table:table');
-    while Assigned(TableNode) do
-    begin
-      nodename := TableNode.Nodename;
-      // These nodes occur due to leading spaces which are not skipped
-      // automatically any more due to PreserveWhiteSpace option applied
-      // to ReadXMLFile
-      if nodeName <> 'table:table' then
+      //process each table (sheet)
+      TableNode := SpreadSheetNode.FindNode('table:table');
+      while Assigned(TableNode) do
      begin
-        TableNode := TableNode.NextSibling;
-        continue;
-      end;
+        nodename := TableNode.Nodename;
+        // These nodes occur due to leading spaces which are not skipped
+        // automatically any more due to PreserveWhiteSpace option applied
+        // to ReadXMLFile
+        if nodeName <> 'table:table' then
+        begin
+          TableNode := TableNode.NextSibling;
+          continue;
+        end;

-      // Tables with external references contain a copy of the external table
-      // having the filename as sheet name - which is not valid for fps.
-      // Since external references are not supported ATM we skip this table.
-      if TableNode.FindNode('table:table-source') <> nil then begin
-        TableNode := TableNode.NextSibling;
-        Continue;
-      end;
+        // Tables with external references contain a copy of the external table
+        // having the filename as sheet name - which is not valid for fps.
+        // Since external references are not supported ATM we skip this table.
+        if TableNode.FindNode('table:table-source') <> nil then begin
+          TableNode := TableNode.NextSibling;
+          Continue;
+        end;

-      sheetName := GetAttrValue(TableNode, 'table:name');
-      FWorksheet := TsWorkbook(FWorkbook).GetWorksheetByName(sheetName);
+        sheetName := GetAttrValue(TableNode, 'table:name');
+        FWorksheet := TsWorkbook(FWorkbook).GetWorksheetByName(sheetName);
 //      FWorkSheet := TsWorkbook(FWorkbook).AddWorksheet(sheetName, true);
-      tablestyleName := GetAttrValue(TableNode, 'table:style-name');
-      // Read protection
-      ReadSheetProtection(TableNode, FWorksheet);
-      // Collect embedded images
-      ReadShapes(TableNode);
-      // Collect column styles used
-      ReadColumns(TableNode);
-      // Process each row inside the sheet and process each cell of the row
-      ReadRowsAndCells(TableNode);
-      // Read page layout
-      ReadPageLayout(StylesNode, GetAttrValue(TableNode, 'table:style-name'),
-        (FWorksheet as TsWorksheet).PageLayout);
-      // Repeated cols/rows already have been determined.
-      (FWorksheet as TsWorksheet).PageLayout.SetRepeatedRows(
-        FRepeatedRows.FirstIndex, FRepeatedRows.LastIndex);
-      (FWorksheet as TsWorksheet).PageLayout.SetRepeatedCols(
-        FRepeatedCols.FirstIndex, FRepeatedCols.LastIndex);
-      // Read print ranges
-      ReadPrintRanges(TableNode, FWorksheet);
-      // Apply table style
-      ApplyTableStyle(FWorksheet, tablestylename);
-      // Handle columns
-      ApplyColWidths;
-      // Page layout
-      FixCols(FWorksheet);
-      FixRows(FWorksheet);
-      // Continue with next table
-      TableNode := TableNode.NextSibling;
-    end; //while Assigned(TableNode)
+        tablestyleName := GetAttrValue(TableNode, 'table:style-name');
+        // Read protection
+        ReadSheetProtection(TableNode, FWorksheet);
+        // Collect embedded images
+        ReadShapes(TableNode);
+        // Collect column styles used
+        ReadColumns(TableNode);
+        // Process each row inside the sheet and process each cell of the row
+        ReadRowsAndCells(TableNode);
+        // Read page layout
+        ReadPageLayout(StylesNode, GetAttrValue(TableNode, 'table:style-name'),
+          (FWorksheet as TsWorksheet).PageLayout);
+        // Repeated cols/rows already have been determined.
+        (FWorksheet as TsWorksheet).PageLayout.SetRepeatedRows(
+          FRepeatedRows.FirstIndex, FRepeatedRows.LastIndex);
+        (FWorksheet as TsWorksheet).PageLayout.SetRepeatedCols(
+          FRepeatedCols.FirstIndex, FRepeatedCols.LastIndex);
+        // Read print ranges
+        ReadPrintRanges(TableNode, FWorksheet);
+        // Apply table style
+        ApplyTableStyle(FWorksheet, tablestylename);
+        // Handle columns
+        ApplyColWidths;
+        // Page layout
+        FixCols(FWorksheet);
+        FixRows(FWorksheet);
+        // Continue with next table
+        TableNode := TableNode.NextSibling;
+      end; //while Assigned(TableNode)

-    FreeAndNil(Doc);
+      FreeAndNil(Doc);
+    end;

    // process the settings.xml file (Note: it does not always exist!)
    XMLStream := CreateXMLStream;
@ -2718,9 +2724,6 @@ begin
      XMLStream.Free;
    end;

-    // Convert formulas from OpenDocument to ExcelA1 dialect
-//    FixFormulas;
-
    // Active sheet
    if FActiveSheet <> '' then
      sheet := (FWorkbook as TsWorkbook).GetWorksheetByName(FActiveSheet) else
--- a/components/fpspreadsheet/source/common/fpspreadsheet.pas
+++ b/components/fpspreadsheet/source/common/fpspreadsheet.pas
@ -8691,18 +8691,47 @@ procedure TsWorkbook.ReadFromFile(AFileName: string; APassword: String = '';
  AParams: TsStreamParams = []);
 var
  success: Boolean;
+  fmtID: TsSpreadFormatID;
  fileFormats: TsSpreadFormatIDArray;
  i: Integer;
+  found: Boolean;
 begin
  if not FileExists(AFileName) then
    raise EFPSpreadsheetReader.CreateFmt(rsFileNotFound, [AFileName]);

+  // First try to determine file format from the extension
+  if GetFormatFromFileName(AFilename, fmtID) then begin
+    try
+      ReadFromFile(AFileName, fmtID, APassword, AParams);
+      exit;
+    except
+      // format does not match. We must continue with rest of procedure
+    end;
+  end else
+    fmtID := MaxInt;
+
  // Try to get file format from file header
  GetFormatFromFileHeader(AFileName, fileformats);
  if Length(fileformats) = 0 then
    // If not successful use formats defined by extension
    fileFormats := GetSpreadFormatsFromFileName(faRead, AFileName);

+  // Remove already tested format
+  found := false;
+  i := 0;
+  while (i <= High(fileFormats)) do begin
+    if fileFormats[i] = fmtID then begin
+      found := true;
+      inc(i);
+      while (i <= High(fileFormats)) do begin
+        fileFormats[i-1] := fileFormats[i];
+        inc(i);
+      end;
+    end else
+     inc(i);
+  end;
+  if found then SetLength(fileFormats, Length(fileFormats)-1);
+
  // No file format found for this file --> error
  if Length(fileformats) = 0 then
    raise EFPSpreadsheetReader.CreateFmt(rsReaderNotFound, [AFileName]);
--- a/components/fpspreadsheet/source/common/fpsutils.pas
+++ b/components/fpspreadsheet/source/common/fpsutils.pas
@ -1370,7 +1370,7 @@ var
  fileformats: TsSpreadFormatIDArray;
 begin
  fileFormats := GetSpreadFormatsFromFileName(faRead, AFileName, ord(sfExcel8));
-  Result := (Length(fileFormats) > 0) and (fileFormats[0] <= sfidUnknown);
+  Result := (Length(fileFormats) > 0); // and (fileFormats[0] <= sfidUnknown);  // wp - removed for new format detection
  if Result then AFormatID := fileFormats[0];
 end;