Parse Molecular Formula

[ctapobep]

The goal is to parse molecular formulas (MF) like H2O. The result is a structure that tells us how many and which atoms are present in MF. The resulting algorithm should be the fastest possible, but it's interesting to look at the elegant ones as well.

Use cases to handle (Java Tests):

Input	Result	Comment
H2O	H2O
CH3CH3	C2H6
Na.Cl	NaCl	Multiple molecules or counterions are combined
C(CH3CH3)2	C5H12	Group coefficient 2 applied to (xxx)
C(CH3CH3)	C3H6	If no group coefficient, then it's 1
(C(OH)2)2	C2H4O4	Nested parentheses
2NH3	N2H6	Group coefficient at the start
2NH3.4CH3	N2C4H20	Each component can have its own coefficient
[HSO4]-	HSO4	The charge is simply ignored
[HSO4]2-	HSO4	Double charge is ignored too
[(2H2O.NaCl)3S.N]2-	H12O6Na3Cl3SN	A hairy example
		1.3 million dataset exported from Meve*

Not needed:

• The order of elements in the resulting MF doesn't have to be conventional - any order is fine. Here we're concentrated on the parsing only.
• The whole topic of isotopes can be ignored

*This dataset doesn't contain anything with parentheses. So in the benchmarks we generate modifications with the parentheses.