This repository contains a trivial piece of demonstration code, which is to be used in my conjunction with my brief talk on parsing, lexing and CI-linting
There is a trivial lexer which will read a string which is expected to contain PHP code, and turn that into a list of entries:
- Multi-line comments (i.e arbitrary text bound between
/*and*/). - Class definitions (i.e. "
class foo").
The lexer is contained in lexer.py and was written with the intention it would be easy to understand, rather than efficient.
There's a driver in main.py which will let you see output:
frodo ~/Repos/github.com/skx/php-class-comments $ ./main.py
{'type': 'comment', 'value': ' Class Bar is known for serving alcohol. '}
{'type': 'class', 'name': 'Bar'}
{'type': 'comment', 'value': ' Class Foo reminds us that soft-drinks are also fine. '}
{'type': 'class', 'name': 'Foo'}
{'type': 'comment', 'value': ' PHP code goes here '}
{'type': 'comment', 'value': ' Class Foo reminds us that soft-drinks are also fine. '}
{'type': 'class', 'name': 'Baz'}
{'type': 'comment', 'value': " Trailing comment - I'm not even closed!\n"}
Pretend we're linting now
Found class with name Bar
The previous token was a comment. Yay!
The comment refers to a class ..
OK : The comment refers to the expected class!
Found class with name Foo
The previous token was a comment. Yay!
The comment refers to a class ..
OK : The comment refers to the expected class!
Found class with name Baz
The previous token was a comment. Yay!
The comment refers to a class ..
FAIL: Comment does not refer to the expected class: Baz
I maintain a small collection of sysadmin-tools called sysbox, packaged as a single binary.
One of the subcommands allows extraction of comments from multiple languages, as described in this blog-post:
As a result of using this tool I submitted a couple of spelling-mistakes to comments in the git tool: