edbeard/chemdataextractor-pv

Fork of ChemDataExtractor 2.0, tailored to extract information about dye-sensitized photovoltaic devices

chemdataextractor-pv's Issues

Bracketed errors not extracted from tables

Example:

def test_errors_brackets_table(self):
    input = [['Dye', 'FF (%)'], ['T-1', '71.43(±0.09)']]
    expected = []

    self.do_table_cell(input, expected, FillFactor)

Returns:

[{'FillFactor': {'raw_value': '71.43', 'specifier': 'FF', 'value': [71.43]}}]

NB: Issue likely to do with core CDE. Will submit a fix to the dev version when done.

Add [sigfig, tabledataextractor] Library to requirement.txt please

Contextual SentenceDye

The SentenceDye parser is too permissive. Currently idenifying sentences like:
"Although, this HOMO level is little lower than those of Spiro-OMeTAD, X51 and P3HT, based on the HOMO energy level of the CH3NH3PbI3 perovskite sensitizer (−5.46 eV (ref. 34)), the 0.17 eV over potential is large enough driving force for the complete hole transportation (Fig. S3†), and thereby a complete regeneration of the oxidized dye can be ensured."

This is picking up the word sensitizer, and then looking for any instance of an alphanumeric in the same sentence.

Could restrict the raw_value parser for this, or try a different approach (such as the dictionary of popular dyes...)

PhotovoltaicCell parser specifier is too lenient

The current parse_expression for the specifier is parse_expression=Any().hide() . This is lenient and slows down CDE, as every element is determine to be a PhotovoltaicCell, and then thrown away at later stage. Example:

{'PhotovoltaicCell': {'semiconductor': {'Semiconductor': {'specifier': 'photoanode', 'raw_value': 'TiO2', 'thickness': {'SemiconductorThickness': {'raw_value': '20', 'raw_units': 'nm', 'value': [20.0], 'units': '(10^-9.0) * Meter^(1.0)', 'specifier': 'photoanode'}}}}}}

{'PhotovoltaicCell': {}}

{'PhotovoltaicCell': {}}

{'PhotovoltaicCell': {}}

{'PhotovoltaicCell': {}}

Might be solved by adding an option for automatic parser that doesn't need a specifier