Large Scale Extraction

ChemSchematicResolver can be used for high-throughput data extraction using two methods.

Extract Images

Extract all diagrams from a directory of images with extract_images:

>>> result = csr.extract_images('<path/to/img/dir>')

This runs the extract_image method on every file in the target directory, and returns a list of the results. For example:

>>> print(result)
[[(['1a'], 'C1=CC=CC=C1')], [(['1b'], 'C1=CC=CC(=C1)C')]]    

Extract Documents

Similarly, the extract_documents method can be used to extract from a directory containing multiple documents.

>>> csr.extract_documents('<path/to/docs/dir>')  

This also returns a list, where each element describes CSR-enriched chemical records of each article. For example:

>>> print(result)
    [{'labels': ['1a'], 'roles': ['compound'], 'melting_points': [{'value': '5', 'units': '°C'}], 'diagram': { 'smiles': 'C1=CC=CC=C1', 'label': '1a' } },
     {'labels': ['1b'], 'roles': ['compound'], 'melting_points': [{'value': '-126', 'units': '°C'}], 'diagram': { 'smiles': 'C1=CC=CC(=C1)C', 'label': '1b' } }]

The user may also use the extract_all=False argument with this method: see the previous section.

ChemSchematicResolver also supports compressed directories with .zip, .tar and .tar.gz formats.

---