CodeSnip Old Code (SVN)

Changed v4 file format again. Checksums no longer have preceding size - size is known and fixed.

* Added new TBackupFileInfo helper class used by reader and writer classes.
* Stream being read initialised by skipping over any watermark.
* Refactored file reader classes, pulling out parent classes for text and binary based reader classes.

Modified v4 file format: file id and file count are now written in binary rather than as formatted text.

Substantially revised way that TBackupFileLoaderFactory class operates. Now much more general and more complex. Uses watermarks to identify required loader classes.

Added support for new v4 backup file format, implemented in TV4BackupFileLoader. Also added new TBackupFileWriter class to do the actual backup file creation.

Added new TBinDataStreamReader and TBinDataStreamWriter classes for reading and writing formatted binary data using a wrapped stream.

* Changed UUtils unit's ListFiles routine to return full file specification in list of files rather than just the base name.
* Changed all the using code accordingly.

Changed TSourceGen's IncFileAsString and UnitAsString methods to use TStringBuilder to build output string instead of a writer object that wraps a string stream.

Changed TParsedHiliter to use TStringBuilder to build output string instead of a writer object that wraps a string stream.

* Renamed UTextStreamReader unit as UStringReader and TTextStreamReader class as TStringReader. Changed TStringReader to read its data from a string instead of a stream.
* Updated using Hiliter.UPasLexer unit to use revised unit and class names.
* Updated chain of Hiliter classes that use Hiliter.UPasLexer to pass source code as a string instead of a stream. When streams were used they had to contain Unicode text, so it is safer and easier just to use strings instead.