Add ability to select path segments using Python re regexprs

[AlainLich]

Hi,
In case there is an interest, this allows to use python re regular expressions, like in following examples:

import dpath
dpath.options.DPATH_ACCEPT_RE_REGEXP = True

js = loadJson()
selPath = 'Config/{(Env|Cmd)}'
x = dpath.util.search(js.lod, selPath)
print(x)

selPath = [ re.compile('(Config|Graph)') , re.compile('(Env|Cmd|Data)') ]
x = dpath.util.search(js.lod, selPath)
print(x)

selPath = '{(Config|Graph)}/{(Env|Cmd|Data)}'
x = dpath.util.search(js.lod, selPath)
print(x)

I found this useful navigating JSON.

Remarks:

. This runs under Python3 as shown on https://github.com/AlainLich/dpath-python/actions/runs/896380770 ;
. the trickwith dpath.options.DPATH_ACCEPT_RE_REGEXP, means the option has to be enabled make this compatible
with earlier versions and will not break if users have path segments in curly brackets.

WARNING: this will not work under Python2.7, since it requires Python3 features like 'f' strings; however I might
do a compatible version if there is interest.

Changes (most in random tests):

1. quite straightforward changes in dpath/segments.py
2. added DPATH_ACCEPT_RE_REGEXP in options as explained above
3. in dpath/util.py re.compile extended regexprs once when they are found
4. added `test/test_path_ext.py ̀ to test this extension
5. made all tests using hypothesis run under unittest, in particular this permits to select and reproduce a single test function
6. debugged a rare situation in the random test generation, reproduced in file issues/err_walk.py where a path segment
   gets generated which is interpreted by the bash globbing function resulting in a test failure (e.g. ̀seg[0]) which is globbed into something accepting seg0` but not its literal version with the brackets

[moomoohk]

@AlainLich Hey, are you still interested in working on this PR?

[AlainLich]

@moomoohk Hello, I have not touched this for a year and a half, could be interested to work on this PR, likely at a slow pace, since this would facilitate my own use (loading via pip ).

However, only on Python3... not interested in Python2 compatibility and such.

[moomoohk]

That's great to hear :)
Slow pace is alright by me, as well as Python 3+. On that note, I've decided to drop Python 2 support entirely a while ago!
Minimum supported version is 3.7 as of now.
I've refactored the codebase quite a bit in the last year and a half. Mainly modernizing, upgrading to Python 3 and blowing some dust off...
If you have any questions feel free to reach out here or on Gitter.

[AlainLich]

How do you suggest to proceed. I suppose the simpler is for me to pull the latest from branch master or any other branch you find more relevant and look at the changes. Then I will try to assess what needs to be changed to integrate the PR 'as is'.

Please advise.

[moomoohk]

Yes indeed. Please start by merging master into your branch and take it from there.

[AlainLich]

How do you suggest to proceed. I suppose the simpler is for me to pull the latest from branch master or any other branch you find more relevant and look at the changes. Then I will try to assess what needs to be changed to integrate the PR 'as is'.

Please advise.

[AlainLich]

Hi, I have entered a new PR #177 with revised code integrated in the current commit for branch master.

This PR should be closed.