Extract molecular formulas from text — extract

Finds and returns chemically valid molecular formulas from a character vector, restricted to content inside parentheses or square brackets.

Usage

extract_formulas(text_vector)

Arguments

text_vector: A character vector of text to search.

Value

A list of character vectors. Each element corresponds to one input string and contains all formulas found inside its bracketed content.

Details

Behavior:

Correctly handles parentheses, square brackets, stoichiometric numbers, and grouped substructures (e.g., "(NH3)2").
Recognizes complexes and hydrates inside brackets when they include spaces, middle dot (U+00B7), plus/minus, or periods (these are normalized before validation).
Ignores oxidation state Roman numerals in brackets, e.g., "(III)" or "( ii )".
Excludes carbon backbone ranges like "C9-12".

Examples

texts <- c(
  "Water (H2O) and ethanol (C2H5OH).",
  "Complex: [Pt(NH3)2Cl2] catalyst.",
  "Hydrate: (CuSO4 . 5H2O)",
  "Oxidation state: iron (III) chloride",  # "(III)" is ignored
  "Backbone range: C9-12 alcohols"         # "C9-12" is ignored
)
extract_formulas(texts)
#> [[1]]
#> [1] "H2O"    "C2H5OH"
#> 
#> [[2]]
#> [1] "Pt(NH3)2Cl2"
#> 
#> [[3]]
#> [1] "CuSO4 . 5H2O"
#> 
#> [[4]]
#> character(0)
#> 
#> [[5]]
#> character(0)
#>