COBOL to bytecode translation
View Patent ↗A system that translates source code of a compiled high level language into bytecode. Compiled languages are so named because their implementations are typically compilers that generate machine code. The described system takes source code of the compiled high level language and translates it into bytecode. The bytecode can be optimized with control flow analysis and method splitting.
1. A system comprising:
a processor; and
a memory, wherein the memory stores a compiler that translates source code of a compiled high level language into bytecode, and wherein the compiler comprises:
a front end that analyzes the source code to build an internal representation of the source code; and
a bytecode generator that translates the internal representation into bytecode, wherein the bytecode is not machine code and requires further processing in order to be executed, wherein the bytecode generator performs control flow analysis by detecting and optimizing flow-affected code, wherein the compiled high level language is COBOL, wherein the flow-affected code includes overlapping PERFORM ranges, wherein the optimizing is accomplished through resolving the overlapping PERFORM ranges, and wherein the overlapping PERFORM ranges are detected by:
identifying one or more regions having an entry point;
selecting a PERFORM instruction associated with the entry point to analyze;
determining a PERFORM range for the selected PERFORM instruction;
identifying an instruction that changes control flow subsequent to the selected PERFORM instruction; and
determining flow-affected code resulting from the instruction, wherein the overlapping PERFORM ranges are resolved by determining reachable regions and forming an adjusted PERFORM range for the selected PERFORM instruction based on the determined reachable regions.
2. The system of claim 1 , wherein the further processing is performed by a virtual machine.
3. The system of claim 1 , wherein the internal representation is a syntax tree.
4. The system of claim 1 , wherein the bytecode generator performs method splitting by converting a section of code into multiple smaller sections of code.
5. The system of claim 4 , wherein the conversion is accomplished by:
selecting a sequence of instructions from code, wherein the sequence of instructions have a jump instruction being configured to alter an order of execution of the sequence of instructions;
identifying a portion of the code from the sequence of instructions based upon the jump instruction;
generating a code block based upon the portion of the code, wherein the code block stores a start point for the portion of the code and an end point for the portion of the code;
creating another method based upon the code block; and
replacing the portion of the code with a call associated with the another method.
6. A method comprising:
receiving source code from a compiled high level language;
translating the source code into an internal representation;
optimizing the internal representation by performing control flow analysis on the internal representation by detecting and optimizing flow-affected code, wherein the compiled high level language is COBOL, wherein the flow-affected code includes overlapping PERFORM ranges, wherein the optimizing is accomplished through resolving the overlapping PERFORM ranges, and wherein the overlapping PERFORM ranges are detected by:
identifying one or more regions having an entry point;
selecting a PERFORM instruction associated with the entry point to analyze;
determining a PERFORM range for the selected PERFORM instruction;
identifying an instruction that changes control flow subsequent to the selected PERFORM instruction; and
determining flow-affected code resulting from the instruction, wherein the overlapping PERFORM ranges are resolved by determining reachable regions and forming an adjusted PERFORM range for the selected PERFORM instruction based on the determined reachable regions;
splitting the optimized internal representation by converting a section of the optimized internal representation into multiple smaller sections; and
generating bytecode from the multiple smaller sections.
7. A computer program product embodied in a non-transitory computer readable medium and comprising computer instructions configured to:
receive source code from a compiled high level language;
translate the source code into an internal representation;
optimize the internal representation by performing control flow analysis on the internal representation by detecting and optimizing flow-affected code, wherein the compiled high level language is COBOL, wherein the flow-affected code includes overlapping PERFORM ranges, wherein the optimizing is accomplished through resolving the overlapping PERFORM ranges, and wherein the overlapping PERFORM ranges are detected by:
identifying one or more regions having an entry point;
selecting a PERFORM instruction associated with the entry point to analyze;
determining a PERFORM range for the selected PERFORM instruction;
identifying an instruction that changes control flow subsequent to the selected PERFORM instruction; and
determining flow-affected code resulting from the instruction, wherein the overlapping PERFORM ranges are resolved by determining reachable regions and forming an adjusted PERFORM range for the selected PERFORM instruction based on the determined reachable regions;
split the optimized internal representation by converting a section of the optimized internal representation into multiple smaller sections; and
generate bytecode from the multiple smaller sections.