Whitespace skipping breaks matches unrelated to whitespaces

[Hi-Angel]

Given the default behavior of whitespace skipping, it means that LineStart() + Regexp('word') should match both of: 1. 'word' and 2. ' word' (note the leading spaces) strings. However 2 doesn't match. One might think, perhaps it is a corner case where whitespace is singificant — but no, LineStart() + Regexp('\s+') will not match 2 either.

Steps to reproduce (in terms of terminal commands)

 λ cat test2.py
from pyparsing import LineStart, Regex

r = LineStart() + Regex('(\s+|word)')
s = '  word'
print([r for r in r.scanString(s)])
 λ python test2.py
[]

Expected

The list of matches is not empty

Actual

The list of matches is empty

Known workarounds

Disabling whitespace skipping as in the following code makes it work.

from pyparsing import LineStart, Regex, ParserElement

ParserElement.setDefaultWhitespaceChars('')
r = LineStart() + Regex('(\s+|word)')
s = '  word'
print([r for r in r.scanString(s)])

Version

2.4.7

[Hi-Angel]

Oh, I figured, the problem is more severe, it's not even related to Regexp class. The following code fails to find a match too:

from pyparsing import LineStart

r = LineStart() + 'word'
s = '  word'
print([r for r in r.scanString(s)])

[Hi-Angel]

In case someone asks: LineStart per see seems not to be affected by whitespace skipping, because the testcases do work if you remove leading whitespace from the string. That is, LineStart does match despite that whitespace skipping is enabled.

[ptmcg]

Taking a look at this. LineStart needs some special handling to treat StringStart like LineStart. Also, LineStart and whitespace-skipping have always not gotten along together.

[ptmcg]

Released in 3.0.0rc2