Expressor
Source Code Compression and Language Translator
Expressor is the application which is the heart of SSDS. Expressor is a program that compresses a source file from any language. Through its compression algorithm, Expressor translates source code from one language into its own language, Express.
Compression is achieved in seven steps:
- All unnecessary whitespace (tabs, spaces, returns) is removed from the file.
- Comments are removed.
- Each occurrence of a variable name, object name, argument name, and procedure name (all words which are not reserved language objects) is replaced by minimal sequence of letters and numbers. For example, if the variables, gCounter and bIsNeeded, were the most frequently used variables within a source file, respectively; each occurrence of gCounter would be replaced by the character, a, and each occurrence of bIsNeeded would be replaced by the letter, b.
- Reserved language words are replaced by a minimal sequence of characters. For example, the Java method, paint(), can be replaced by @A.
- Repetitive elements are replaced by a minimal set of characters. For example, in a C source file, a deeply nested 'if' statement may end with four '}'. The four '}' would be replaced by $4. '$' symbol would represent '}', and the second number represents the number of repetitions.
- The use of special symbols in steps 4 and 5 would allow for the elimination of even more whitespace since variable names and reserved keywords would no longer need to be separated by whitespace. For example, a Java variable declaration would be "boolean bIsFlag;". A space is needed between the two words so that the compiler can know where the declaration ends and the variable name begins. The space is not need in Express because Expressor would convert the statement to "@#c;". The compiler would know that the "@" is beginning of declaration which has a two character format. Any symbols which comes after the two characters must be variable names.
- The source file generated by Express itself can be compressed further with industry standard Ziv-Lempel compression. The file would be decompressed before compilation begins.
Expressor does not have to perform all of the compression steps in order to be effective. Performing steps 1-3 alone yields significant compression ratios, while keeping the source files compilable by today's compilers. Performing steps 4-6 would require that compilers and libraries (zbo) would have to be redesigned in order to compile an Express file.
I converted a java graphical program to an ebo file. The original Java source file was 132 KB. The uncompressed jar file created from the java program was 47 KB. After performing steps 1-3, the Ebo file was 42 KB. This Ebo file is still compilable by the standard java compiler. After performing steps 1-6, the Ebo file was 27 KB. Therefore, Ebo files are compact enough to be sent over the net..