marcbllv/regex3000

A simple regex parser written in Go

regex3000's Issues

Use slice of runes instead of string

Context

It's complex to extract runes one-by-one from a string, better convert the input regex to a slice of rune at first.

Scope and solution

Replace string usage by []rune everywhere

Handle regex with bad format properly

Context

Currently regexes with a bad format usually raise a segfault.
This must be avoided and errors should be handled in a clean way.

Scope and solution

TBD

Bug when using brackets

Bug

When runnning:

$ bin/regex3000 'I have [1-9]+ l?eg{1,2}s' 'I have 3 legs'

it outputs false, expected true.

Scope and solution

There's an issue with the brackets, we need to plug the previous states (here previous states of letter g) to the final ones.

Return matching position in string instead of cropping string

Context

Matching function currently return true/false + the rest of string that matches

It would be cleaner to return the match position (if match==true), and this would allow for further developments in an easier way (eg: matching the beginning / end of string)

Scope and solution

All Match() function in inspectors should take as input:

• a list of integer position where matching is working
• the full string to match

And return:

• return a list of matching position (incremented)

Match internal pieces of text and whole string

Context

Handle beginning and end of string with ^ & $

Scope & solution

Set values for mustMatchBeginning and mustMatchEnd in Starting & Final inspectors

Handle special backslash sets

Context

Handle special sets defined via a backslash:

\d \D
\s \S
\w \W

Rewrite State class in a better OOP way

Context

Investigate whether we could use the Go OOP style described here: https://www.toptal.com/go/golang-oop-tutorial

The State class might be a good pick for refactoring: we're using StateType as an attribute while in the aforementioned article the authors discourages such pattern in Go.

Scope and solution

Write new implementation according to specs below

Handle negative sets

Context

We cannot use "negative" sets using brackets and ^, eg: `[^a-zA-Z].

Scope and solution

Update brackets parsing function: if first char is ^, then create an OppositeSetInspector instead of a SetInspector

Better parser for braces

Context

The braces parsing is complex and not very clean to write.

Scope and solution

• Write generic parsers for ranges (braces content), string and integers
• Write a function to combine parsers
• Put this alltogether