Translator – Developmental Release

Another change is needed in the Token class - the “code” field needs to be changed to “index” because it is really an index of a Table entry and not the code enumeration. It was originally named code because it was going to be the code within the program in memory, which will make the program easy to execute (the index is still going to be used in the program memory). The table entry will have a function pointer to the routine that will execute the command or operator. If the code from the Code enumeration was used, then the run-time module would have to first index through the code-to-index conversion array to get the index of the table and then index to the table entries array.

Changes have be made to the Parser (special EOL  token and changing the Token member code to index), Table (added unary_code field and entries for Null, Neg, and EOL codes) and the test code (renamed test_parser.cpp to test_ibcp.cpp because it's going to contain all of the test code; separated code that prints token contents into its own function so it can be called from other function like from the test_translator() function to be written; modified to work with the new EOL token; and was setup to allow for additional test functions where the command line arguments have been changed in that now “-p” activates the Parser test code).

The ibcp_0.1.3-src.zip file has been uploaded at Sourceforge IBCP Project along with the binary for the program. The code essentially does what it did before, but contains the recent changes mentioned. The reason for the release was to follow the spirit of open source code (release early, release often) and I wanted to verify that the changes so far compiled and worked. Also, I discovered the source for Release 0.1.2 was missing the test_parser.cpp source file.

Translator Class – Table/Parser Changes

The new unary operator Neg_Code for Sub_Code needs to be added, which will represent negation, along with it's corresponding table entry. To other two codes also need to be added, Null_Code and EOL_Code.

The Null_Code will be used for two things. One being a very low (lowest) precedence operator that will be pushed onto the operator stack before processing tokens. This will simplify the Translator when it is checking and popping off high precedence operators as it won't need to check if the stack is empty first. The second will be for the new unary_code field that needs to be added to the Table entries to indicate an operator's associated unary operator with Null_Code indicating that there is no unary operator.

The Parser needs to be modified to return a token with an EOL_Code at the end of the line instead of returning a NULL pointer. The reason for this changes is so that this EOL token can be passed into the Translator. The EOL will be treated as an operator and it's precedence will also be very low, but higher than the Null operator. The EOL operator will force the Translator to empty the operator stack of any operators except for the Null operator. Also by using the EOL operator, the Translator can make sure an end of line is allowed. In other words, if the Translator is expecting an operand (as after it has just seen an operator), but sees EOL (an operator), an error will be reported (“operand expected”).