Example:

int main() {
    char* test = "This is a really long string and I am just going to keep 
        typing random stuff to make it more than 80 chars";
}

Outputs:

>>> check_all()
test.c:
   3: Indentation error (expected 4, got 8)

Possible fix:

• Add a check for ';' in the line and if not on that line then the next lines tab needs to be 8.

Oh yeah and while I remember, my script currently fails to correctly identify variable declarations where the type of the variable is not declared in the current file (you'll see this if you run it over a multiple file assignment 3 code).

I know why - it's because the variable naming code requires a list of types, which it builds though an initial parse of the file - but it's a bit annoying to fix. I think we'd have to switch to a proper tokeniser in this respect.

Sean Purdon

When checking the line lengths, comments that go past 79 characters are ignored.

To get vera to output basic tokens you run 'vera++ -rule DUMP source.c' and
this will spit out something like the tokens below. To use in the main python script
we need a wrapper around either this output, or a replication of the original tcl
library (see http://www.inspirel.com/vera/ce/doc/tclapi.html), or something in
between.

There are a variety of ways to merge C++ code with python code, some which
involve python changes only, others which involve modifying the C++ (using
Boost or otherwise) so that C++ objects become 'native' inside python.

Note: we probably only have binaries for the vera++ Windows install, unless we
create a CMake file to cross-compile the tokenizer than any C++ changes would
restrict the platform (which isn't the end of the world but..)

We need to:

1. decide on what functionality we need from the tokenizer in the python script
2. choose which method of integration is appropriate
3. actually code that up.

Example vera++ output:

Tokens in file test.c:
1/0 ccomment /* hello world */
1/17 newline

2/0 newline

3/0 pp_hheader #include <stdio.h>
3/18 newline

4/0 newline

5/0 identifier main
5/4 leftparen (
5/5 rightparen )
5/6 space
5/7 leftbrace {
5/8 newline

6/0 space
6/4 identifier printf
6/10 leftparen (
6/11 stringlit "hello, world\n"
6/27 rightparen )
6/28 semicolon ;
6/29 newline

7/0 space
7/4 return return
7/10 space
7/11 intlit 0
7/12 semicolon ;
7/13 newline

8/0 rightbrace }
8/1 newline

9/0 eof

Some bugs in checking variable names and struct names:

• Globally declared variables appear to be not checked
• Function parameters are not checked
• Multiple variables declared on the same line are not checked.
• Function pointers are not checked.
• struct names are not checked (e.g. typedef struct bobStruct { } or struct bobStruct { })

We really need some test cases to get this project started..

To start with we don't need to a really formal collection using python's 'unittest' framework,
can start with some ad hoc tests as a basis and formalise later..

Basically just need to check the tokenizer (vera++) compiles on other peoples machines (Linux/Mac).
It works on my Mac but I might have forgotten about setup steps I've performed in the past..