Categorise figure caption style to aid parsing

[nickynicolson]

Issue #12 is about parsing the figure caption. Aside from formatting errors (some examples shown in the comments on issue 12) different articles may use different styles to compose the figure caption. These include things like:

• Specimen reference only (as used in https://doi.org/10.3897/phytokeys.22.4041). Example:
  • Solanum angustifidum Bitter. (A drawn from Tressens et al. 560 B–E drawn from Pedersen 7150 F drawn from Cristóbal et al. 1618). Illustration by Bobbi Angell.
• Description and specimen reference (as used in https://doi.org/10.3897/phytokeys.84.12695). Example:
  • Solanum houstonii Martyn. A Whole plant B Leaf with prickly petiole and stellate trichomes C Lobed leaf D Prickly stem with stellate trichomes E Bud F Staminate flower G Ovary and style of staminate flower H Anthers of staminate flower I Hermaphroditic flower spread J Gynoecium of hermaphroditic flower K Anther of hermaphroditic flower L Fruits M Seed. Drawn by Bobbi Angell A, E from Fryxell 3706 (NY00751643) B from Durán et al. 4737 (NY00751647) C from Reina G. 98-2082 (NY00751617) D from Palmer 1243 (NY00751622) F–H from Reina G. 98-2123 (NY00751616) I–K from Cabrera 10755 (NY00751572) L from Cabrera 11000 (NY00751567).
• Description, specimen reference and scale (as used in https://doi.org/10.3897/phytokeys.66.8457). Example:
  • Solanum betroka D’Arcy & Rakot. A Flowering stem B Flowering stem with lobed leaf C Flower D Immature berry E Flower just before anthesis, showing elongate calyx lobes F Simple trichome G Dendritic trichome (Based on: A–D, F, GBosser 17370; ELeroy 3). Scale bar: A, B = 2 cm; C, E = 7 mm; D = 1 cm; F, G = 0.3 mm. Drawn by Lucy T. Smith.

Possible strategies to categorise these:

• After discarding the first italic tag (which contains the species name) count the number of bold and italic tags. If the bold and italic counts are equal, then we have specimen reference only. If not we have one of the other caption styles
• Extract all the bold tags and see how many include the letter A:
  • Bolded A count == 1 we have specimen references only
  • Bolded A count == 2 we have description and specimen reference
  • Bolded A count == 3 (and presence of dimension indicators like mm, cm etc) we have a 3 part caption (description, specimen reference and scale).

This categorisation can then be used to direct the caption parsing.