Add script for generating tol JSON from txt file. Significantly increase dataset size.

1 files changed, 243 insertions, 0 deletions

diff --git a/src/tol.txt b/src/tol.txt
new file mode 100644
index 0000000..a881d55
--- /dev/null
+++ b/src/tol.txt
@@ -0,0 +1,243 @@
+Root
+	Viruses
+		Coronaviridae
+		Enterovirus
+		Rabies lyssavirus
+	Archaea
+	Bacteria
+		Cyanobacteria
+		Proteobacteria
+	Eukaryota
+		Amoebozoa
+		Plantae
+			Rhodopyhta
+			Viridiplantae
+				Prasinophytes
+				Ulvophyceae
+				Streptophyta
+					Charales
+					Embryophytes
+						Marchantiomorpha
+						Anthocerotophyta
+						Bryophyta
+						Lycopodiopsida
+							Lycopodiidae
+							Selaginellales
+						Polypodiopsida
+							Polypodiidae
+								Polypodiales
+							Equisetidae
+						Spermatopsida
+							Cycads
+							Conifers
+								Taxaceae
+								Cupressaceae
+								Pinaceae
+									Pinus
+									Picea
+									Larix
+									Cedrus
+									Abies
+							Ginkgos
+							Angiosperms
+								Illicium
+								magnoliids
+									Piperales
+										Piperaceae
+									Magnoliales
+										Annonaceae
+										Myristicaceae
+									Laurales
+										Lauraceae
+								Monocotyledons
+									Alismatanae
+									Aranae
+									Liliaceae
+									Asparagales
+										Amaryllidaceae
+										Asparagaceae
+										Asphodelaceae
+										Iridaceae
+										Orchidaceae
+									Dioscoreaceae
+									Arecanae
+										Cocoeae
+										Phoeniceae
+									Zingiberanae
+										Musaceae
+										Strelitziaceae
+										Zingiberaceae
+									Commelinanae
+										Bromeliaceae
+										Cyperaceae
+										Typhaceae
+										Poaceae
+								eudicots
+									Ranunculales
+										Papaveraceae
+										Ranunculaceae
+									Proteales
+										Proteaceae
+										Nelumbo
+									Core Eudicots
+										Saxifragales
+										Rosids
+											Fabaceae
+												Mimosoideae
+												IRLC (Inverted Repat-lacking clade)
+													Trifolieae
+													Fabeae
+											Rosales
+												Rosaceae
+												Ulmaceae
+												Urticaceae
+												Moraceae
+												Cannabaceae
+											Fagales
+												Fagaceae
+												Betulaceae
+												Juglandaceae
+											Cucurbitales
+												Cucurbitaceae
+											Malpighiales
+												Salicaceae
+												Violaceae
+												Passifloraceae
+												Erythroxylaceae
+												Rhizophoraceae
+												Euphorbiaceae
+												Linaceae
+												Rafflesiaceae
+											Myrtales
+												Myrtaceae
+												Onagraceae
+												Lythraceae
+											Brassicales
+												Caricaceae
+												Brassicaceae
+											Malvales
+												Core Malvales
+													Malvoideae
+													Bombacoideae
+													Sterculioideae
+													Helicteroideae
+													Byttnerioideae
+											Sapindales
+												Anacardiaceae
+												Burseraceae
+												Meliaceae
+												Rutaceae
+												Sapindaceae
+										Vitaceae
+										Caryophyllales
+											Polygonaceae
+											Droseraceae
+											Nepenthaceae
+											core Caryophyllales
+												Cactaceae
+												Amaranthaceae
+										Asterids
+											Ericales
+												Actinidiaceae
+												Ericaceae
+												Lecythidaceae
+												Sapotaceae
+												Ebenaceae
+												Theaceae
+											Solanales
+												Solanaceae
+												Convolvulaceae
+											Lamiales
+												Oleaceae
+													Fraxinus
+												Bignoniaceae
+												Pedaliaceae
+												Lentibulariaceae
+												Lamiaceae
+											Gentianales
+												Rubiaceae
+											Asterales
+												Campanulaceae
+												Asteraceae
+													Carduoideae
+														Cardueae
+													Cichorioideae
+														Cichorieae
+													Asteroideae
+														Asterodae
+														Helianthodae
+											Apiales
+												Apiaceae
+												Araliaceae
+											Aquifoliaceae
+		Fungi
+			Fungi 1
+				Dikarya
+					Basidiomycota
+						Agaricomycotina
+							Agaricomycetes
+								Agaricomycetes 1
+									Agaricomycetidae
+										Agaricales
+											Strophariaceae strict-sense
+											Psathyrellaceae
+											Agaricaceae
+											Nidulariaceae
+											Marasmiaceae
+											Physalacriaceae
+											Pleurotaceae
+											Amanitaceae
+											Podoserpula
+										Boletales
+											Serpulaceae
+											Sclerodermataceae
+											Boletaceae
+									Russulales
+									Hymenochaetales
+								Phallomycetidae
+									Geastrales
+									Gomphales
+									Phallales
+								Cantharellales
+								Auriculariales
+							Tremellomycetes
+						Ustilaginomycotina
+						Pucciniomycotina
+							Pucciniomycetes
+								Septobasidiales
+								Pucciniales
+							Mixiomycetes
+							Tritirachiomycetes
+						Entorrhizomycetes
+						Wallemiomycetes
+					Ascomycota
+						Pezizomycotina
+							Pezizomycetes
+							'Leotiomyceta'
+								Eurotiomycetes
+								Geoglossaceae
+								Sordariomycetes
+									Hypocreomycetidae
+									Sordariomycetidae
+								Laboulbeniomycetes
+								Pleosporomycetidae
+						Saccharomycotina
+						Taphrinomycotina
+							Schizosaccharomycetes
+							Pneumocystidiomycetes
+							Taphrinomycetes
+				Glomeromycota
+			Zygomycota
+				Endogonales
+				Mucorales
+			Blastocladiomycota
+			Chytridiomycota
+			Neocallimastigomycota
+			Microsporidia
+		Metazoa
+			Cnidaria
+			Arthropoda
+			Vertebrata
+				Amphibia
+				Reptilia
+				Synapsida