This repository will be a collection of documents and scripts that help reverse engineering binary save formats of various Microsoft BASIC versions.
The folder qb contains the current work for QuickBASIC versions 4.0, 4.5, 7.0, 7.1 as well as Visual Basic for MS-DOS.
It contains doc.html, documenting the file format as it is known, and a script qb.py that takes as argument the file name of a BASIC file, saved in binary format, and outputs a textual representation.
The output should be as close to the text format saved by QuickBASIC.
Unlike the MS-DOS versions, QuickBASIC for Macintosh uses a tokenized file format, similar to GW-BASIC.
Like QuickBASIC, line numbers are optional, and variable names are stored in a table.
The script mac.py in the folder mac converts a Macintosh binary file into text output.
The first byte is F1 (hexadecimal) for tokenized files (F0 for protected files, format unknown).
It is then followed by a series of lines.
Each line starts with a 16-bit big endian word, encoding the length of the line in bytes (including this 16-bit word).
The most significant bit is set if the line starts with a line number.
Then a single byte encodes the number of spaces (one less if it starts with a line number) preceding the first non-space character.
If the line has a line number (that is, the most significant bit of the initial 16-bit word is set), then a 16-bit big endian word follows, storing the number.
The line body then follows as a token stream, with bytes 20 to 7E (hexadecimal) encoding their ASCII values.
Each line is then terminated by a null byte.
The last line is followed by a 16-bit null word, an impossible byte length for a line (it is always at least 2 bytes long). Then the section containing identifier names follows. It always starts on an even address, with a single null byte, followed by the identifiers. Each identifier begins with a single byte, encoding the length of the identifier, followed by the characters of the identifier.
Identifiers are indexed by their number in the sequence: the first identifier is accessed as identifier 0, the second one as identifier 1, and so on.
For the various token types, see mac.py under the mac folder.
This work is done without any access to disassembly or internal documentation to the BASIC executables. All information is provided as-is, with no warranty as to its correctness or usability. However, the author believes the code produces generally correct output.
GW-BASIC, the predecessor to QuickBASIC, uses its own binary format, already documented in other places: