cswatt / narratr Goto Github PK

for parser and codegen

• objects
• pocket
• how to automatically use key that works with appropriate lock
• operations: pickup/use

Corresponding codegen

• loops
• variables declared only once
• explicit delete in cleanup
• miscellaneous

Probably a collaboration between @nivvedan , @hylbyj , and myself.

in parser and codegen

• arithmetic
• if constructions
• loops

what else?

• FIRST PRIORITY: shift to argparse or getopt
• Default output file (and maintain stdout possibility)
• Verbose mode (now basically the default)

functions in parser.py

corresponding codegen

Codegen needs to be cleaned up before others also start working on it, else it could become more confusing for everyone. Specifically,

1. Comments. All logic, especially that which could be difficult for others to understand should be well commented - especially in detecting direction and moving around.
2. Fixing PEP8 style errors. nosetests should pass.
3. Removing deugging print statements. There are quite a few of them.
4. Fix and remove code that is commented out. No code should be commented out without a specific stated reason.

Comments in the commits need to addressed here. Maybe redundant from above.
899d141
869f939

• scene structure: setup/action/cleanup
• win/lose

Please also check the parse tree. Don't take the current parse tree as correct.

@nivvedan , this is your decision to make, but I figured everyone should be aware.

We never really decided what loop constructs we are going to support. I was thinking for and while, but should we support a do-while loop? I have also tentatively made the syntax as follows:

for [var] in [list]:
    [actions]

while [conditional]:
    [actions]

They're obviously based on Python's. Let me know if you have suggestions.

for parser and codegen

• numbering of scenes
• exposition
• operations: investigate, move, arithmetic

test, and_test, not_test, comparison, comparison_op, arithmetic_expression, term, power, atom

corresponding codegen

First time using issue, hope I used it in a right way. Can you guys take a look at codegen.py? I mentioned some points here: 869f939

I didn't do pep8 style check yet...

Thanks!

corresponding codegen

preprocessing, etc

functions in parser.py

In parser and codegen

The AST node should have line and column information

It is currently in action().

corresponding codegen

functions in parser.py

i.e. develop an ideal structure (and implement it). this will make the code generator much more efficient (and easier to write), since it will have to do a lot less searching for nodes.

Probably @nivvedan and myself.

functions in parser.py

All len(p) testing in the parser should be abandoned in favour of testing for p[i].type

Need to detect certain errors like scene/variable/item redeclaration while building the AST.
SymTab also make life of CodeGen easier.

functions in parser.py

• program
• newlines_optional
• newlines
• blocks
• scene_block
• item_block (this is going to be somewhat involved, I think)
• start_state
• setup_block
• action_block
• cleanup_block

It is very helpful for two reasons:

1. Syntax highlighting in Text Editors when we want to look at the target code generated,
2. Double Click to execute - Now THAT's useful.

for parser and codegen