react-cheminfo 0.46.0 → 0.48.0

package/lib/taxonomy/core/index.js

@@ -0,0 +1,5 @@
+export { genusOrBelow, isPlaceholderTaxon, lineageFromNodes, lineageFromRanks, principalLineage, } from "./lineage.js";
+export { ncbiTaxonomyUrl } from "./ncbi.js";
+export { splitOrganismName } from "./organismName.js";
+export { PRINCIPAL_RANKS, TAXON_RANKS, isAboveGenus, isGenusOrBelow, isPrincipalRank, rankOrder, } from "./ranks.js";
+//# sourceMappingURL=index.js.map

package/lib/taxonomy/core/index.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"index.js","sourceRoot":"","sources":["../../../src/taxonomy/core/index.ts"],"names":[],"mappings":"AACA,OAAO,EACL,YAAY,EACZ,kBAAkB,EAClB,gBAAgB,EAChB,gBAAgB,EAChB,gBAAgB,GACjB,MAAM,cAAc,CAAC;AACtB,OAAO,EAAE,eAAe,EAAE,MAAM,WAAW,CAAC;AAE5C,OAAO,EAAE,iBAAiB,EAAE,MAAM,mBAAmB,CAAC;AAEtD,OAAO,EACL,eAAe,EACf,WAAW,EACX,YAAY,EACZ,cAAc,EACd,eAAe,EACf,SAAS,GACV,MAAM,YAAY,CAAC"}

package/lib/taxonomy/core/lineage.d.ts

@@ -0,0 +1,83 @@
+/**
+ * A lineage: the taxa from the top of the tree of life down to one of them.
+ *
+ * It is read three ways, because the family meets it in three shapes: a map of
+ * NCBI nodes each pointing at its parent (what a database built on the NCBI
+ * dump answers), a record of names by rank (what octochemdb stores), or a list
+ * already in order.
+ */
+/** One taxon of a lineage. */
+export interface Taxon {
+    /** The rank, as NCBI writes it: `genus`, `no rank`, `clade`… */
+    rank: string;
+    /** The scientific name. */
+    name: string;
+    /**
+     * The NCBI Taxonomy id.
+     * @default undefined — the taxon is known by its name alone
+     */
+    taxId?: number;
+}
+/** The taxa from the top of the tree down to one of them, in that order. */
+export type TaxonLineage = readonly Taxon[];
+/** A taxon of the NCBI tree, as a map of nodes holds it. */
+export interface TaxonNode {
+    /** The scientific name. */
+    name: string;
+    /** The rank, as NCBI writes it. */
+    rank: string;
+    /** The id of the taxon above; the root is its own parent. */
+    parentId: number;
+}
+/**
+ * The taxa a lineage is summarised by: those of a principal rank (domain,
+ * kingdom, phylum, class, order, family, genus, species), and the last one
+ * whatever its rank, since it is the taxon the lineage leads to.
+ *
+ * An NCBI placeholder for an unidentified species ('Streptomyces sp.') says
+ * no more than its genus, so it is left out and the genus is the leaf. The
+ * root, 'cellular organisms', the clades and the steps of no rank above the
+ * leaf never show.
+ * @param lineage - The lineage, from the top down.
+ * @returns The taxa kept, in the same order.
+ */
+export declare function principalLineage(lineage: TaxonLineage): Taxon[];
+/**
+ * Whether a taxon is an NCBI placeholder for an unidentified species of a
+ * genus: a species named the genus and 'sp.', such as 'Streptomyces sp.'.
+ * One named for a strain ('Streptomyces sp. CNQ-509') is a taxon of its own.
+ * @param taxon - The taxon.
+ * @returns Whether it is one.
+ */
+export declare function isPlaceholderTaxon(taxon: Taxon): boolean;
+/**
+ * The lineage of a taxon, walked up a map of NCBI nodes to the root.
+ * @param nodes - The nodes, by id: a record keyed by the id written as a
+ * string, as JSON delivers it, or a map keyed by the number.
+ * @param taxId - The taxon to start from.
+ * @returns Every taxon from the root down to it, each with its id; empty when
+ * the map does not hold the taxon.
+ */
+export declare function lineageFromNodes(nodes: Readonly<Record<string, TaxonNode>> | ReadonlyMap<number, TaxonNode>, taxId: number): Taxon[];
+/**
+ * The lineage written by a record of names by rank, the shape octochemdb
+ * stores (`{ kingdom: 'Fungi', genus: 'Penicillium', … }`).
+ *
+ * A key is matched whatever its case, so `superKingdom` reads as
+ * `superkingdom`; a key that names no rank, and an empty name, are left out.
+ * @param ranks - The names, by rank.
+ * @returns The taxa, from the top down.
+ */
+export declare function lineageFromRanks(ranks: Readonly<Record<string, string | null | undefined>>): Taxon[];
+/**
+ * Which taxa of a lineage sit at the genus or under it — the ones whose names
+ * nomenclature sets in italics.
+ *
+ * A taxon with no rank (a clade, a strain NCBI left unranked) takes the place
+ * of the taxa above it: under a genus it is one of the genus's, above it it is
+ * a clade like Embryophyta, written upright.
+ * @param lineage - The lineage, from the top down.
+ * @returns One flag per taxon.
+ */
+export declare function genusOrBelow(lineage: TaxonLineage): boolean[];
+//# sourceMappingURL=lineage.d.ts.map

package/lib/taxonomy/core/lineage.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"lineage.d.ts","sourceRoot":"","sources":["../../../src/taxonomy/core/lineage.ts"],"names":[],"mappings":"AAAA;;;;;;;GAOG;AASH,8BAA8B;AAC9B,MAAM,WAAW,KAAK;IACpB,gEAAgE;IAChE,IAAI,EAAE,MAAM,CAAC;IACb,2BAA2B;IAC3B,IAAI,EAAE,MAAM,CAAC;IACb;;;OAGG;IACH,KAAK,CAAC,EAAE,MAAM,CAAC;CAChB;AAED,4EAA4E;AAC5E,MAAM,MAAM,YAAY,GAAG,SAAS,KAAK,EAAE,CAAC;AAE5C,4DAA4D;AAC5D,MAAM,WAAW,SAAS;IACxB,2BAA2B;IAC3B,IAAI,EAAE,MAAM,CAAC;IACb,mCAAmC;IACnC,IAAI,EAAE,MAAM,CAAC;IACb,6DAA6D;IAC7D,QAAQ,EAAE,MAAM,CAAC;CAClB;AAED;;;;;;;;;;;GAWG;AACH,wBAAgB,gBAAgB,CAAC,OAAO,EAAE,YAAY,GAAG,KAAK,EAAE,CAc/D;AAED;;;;;;GAMG;AACH,wBAAgB,kBAAkB,CAAC,KAAK,EAAE,KAAK,GAAG,OAAO,CAExD;AAED;;;;;;;GAOG;AACH,wBAAgB,gBAAgB,CAC9B,KAAK,EAAE,QAAQ,CAAC,MAAM,CAAC,MAAM,EAAE,SAAS,CAAC,CAAC,GAAG,WAAW,CAAC,MAAM,EAAE,SAAS,CAAC,EAC3E,KAAK,EAAE,MAAM,GACZ,KAAK,EAAE,CAiBT;AAED;;;;;;;;GAQG;AACH,wBAAgB,gBAAgB,CAC9B,KAAK,EAAE,QAAQ,CAAC,MAAM,CAAC,MAAM,EAAE,MAAM,GAAG,IAAI,GAAG,SAAS,CAAC,CAAC,GACzD,KAAK,EAAE,CAcT;AAED;;;;;;;;;GASG;AACH,wBAAgB,YAAY,CAAC,OAAO,EAAE,YAAY,GAAG,OAAO,EAAE,CAY7D"}

package/lib/taxonomy/core/lineage.js

@@ -0,0 +1,126 @@
+/**
+ * A lineage: the taxa from the top of the tree of life down to one of them.
+ *
+ * It is read three ways, because the family meets it in three shapes: a map of
+ * NCBI nodes each pointing at its parent (what a database built on the NCBI
+ * dump answers), a record of names by rank (what octochemdb stores), or a list
+ * already in order.
+ */
+import { TAXON_RANKS, isGenusOrBelow, isPrincipalRank, rankOrder, } from "./ranks.js";
+/**
+ * The taxa a lineage is summarised by: those of a principal rank (domain,
+ * kingdom, phylum, class, order, family, genus, species), and the last one
+ * whatever its rank, since it is the taxon the lineage leads to.
+ *
+ * An NCBI placeholder for an unidentified species ('Streptomyces sp.') says
+ * no more than its genus, so it is left out and the genus is the leaf. The
+ * root, 'cellular organisms', the clades and the steps of no rank above the
+ * leaf never show.
+ * @param lineage - The lineage, from the top down.
+ * @returns The taxa kept, in the same order.
+ */
+export function principalLineage(lineage) {
+    const kept = [];
+    let last = lineage.length - 1;
+    if (last > 0 && isPlaceholderTaxon(lineage[last])) {
+        // The genus becomes the leaf, past 'unclassified Streptomyces'.
+        last--;
+        while (last > 0 && !isPrincipalRank(lineage[last].rank))
+            last--;
+    }
+    for (let index = 0; index <= last; index++) {
+        const taxon = lineage[index];
+        if (index < last && isPlaceholderTaxon(taxon))
+            continue;
+        if (index === last || isPrincipalRank(taxon.rank))
+            kept.push(taxon);
+    }
+    return kept;
+}
+/**
+ * Whether a taxon is an NCBI placeholder for an unidentified species of a
+ * genus: a species named the genus and 'sp.', such as 'Streptomyces sp.'.
+ * One named for a strain ('Streptomyces sp. CNQ-509') is a taxon of its own.
+ * @param taxon - The taxon.
+ * @returns Whether it is one.
+ */
+export function isPlaceholderTaxon(taxon) {
+    return taxon.rank === 'species' && /\ssp\.$/.test(taxon.name);
+}
+/**
+ * The lineage of a taxon, walked up a map of NCBI nodes to the root.
+ * @param nodes - The nodes, by id: a record keyed by the id written as a
+ * string, as JSON delivers it, or a map keyed by the number.
+ * @param taxId - The taxon to start from.
+ * @returns Every taxon from the root down to it, each with its id; empty when
+ * the map does not hold the taxon.
+ */
+export function lineageFromNodes(nodes, taxId) {
+    const read = nodes instanceof Map
+        ? (id) => nodes.get(id)
+        : (id) => nodes[String(id)];
+    const upward = [];
+    const seen = new Set();
+    let id = taxId;
+    while (!seen.has(id)) {
+        const node = read(id);
+        if (node === undefined)
+            break;
+        seen.add(id);
+        upward.push({ rank: node.rank, name: node.name, taxId: id });
+        id = node.parentId;
+    }
+    return upward.toReversed();
+}
+/**
+ * The lineage written by a record of names by rank, the shape octochemdb
+ * stores (`{ kingdom: 'Fungi', genus: 'Penicillium', … }`).
+ *
+ * A key is matched whatever its case, so `superKingdom` reads as
+ * `superkingdom`; a key that names no rank, and an empty name, are left out.
+ * @param ranks - The names, by rank.
+ * @returns The taxa, from the top down.
+ */
+export function lineageFromRanks(ranks) {
+    const byRank = new Map();
+    for (const key of Object.keys(ranks)) {
+        const name = ranks[key]?.trim();
+        if (name === undefined || name === '')
+            continue;
+        const rank = key.toLowerCase();
+        if (rankOrder(rank) !== undefined)
+            byRank.set(rank, name);
+    }
+    const lineage = [];
+    for (const rank of TAXON_RANKS) {
+        const name = byRank.get(rank);
+        if (name !== undefined)
+            lineage.push({ rank, name });
+    }
+    return lineage;
+}
+/**
+ * Which taxa of a lineage sit at the genus or under it — the ones whose names
+ * nomenclature sets in italics.
+ *
+ * A taxon with no rank (a clade, a strain NCBI left unranked) takes the place
+ * of the taxa above it: under a genus it is one of the genus's, above it it is
+ * a clade like Embryophyta, written upright.
+ * @param lineage - The lineage, from the top down.
+ * @returns One flag per taxon.
+ */
+export function genusOrBelow(lineage) {
+    const flags = [];
+    let below = false;
+    for (const taxon of lineage) {
+        if (isGenusOrBelow(taxon.rank)) {
+            below = true;
+        }
+        else if (rankOrder(taxon.rank) !== undefined) {
+            below = false;
+        }
+        flags.push(below);
+    }
+    return flags;
+}
+//# sourceMappingURL=lineage.js.map

package/lib/taxonomy/core/lineage.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"lineage.js","sourceRoot":"","sources":["../../../src/taxonomy/core/lineage.ts"],"names":[],"mappings":"AAAA;;;;;;;GAOG;AAEH,OAAO,EACL,WAAW,EACX,cAAc,EACd,eAAe,EACf,SAAS,GACV,MAAM,YAAY,CAAC;AA4BpB;;;;;;;;;;;GAWG;AACH,MAAM,UAAU,gBAAgB,CAAC,OAAqB;IACpD,MAAM,IAAI,GAAY,EAAE,CAAC;IACzB,IAAI,IAAI,GAAG,OAAO,CAAC,MAAM,GAAG,CAAC,CAAC;IAC9B,IAAI,IAAI,GAAG,CAAC,IAAI,kBAAkB,CAAC,OAAO,CAAC,IAAI,CAAU,CAAC,EAAE,CAAC;QAC3D,gEAAgE;QAChE,IAAI,EAAE,CAAC;QACP,OAAO,IAAI,GAAG,CAAC,IAAI,CAAC,eAAe,CAAE,OAAO,CAAC,IAAI,CAAW,CAAC,IAAI,CAAC;YAAE,IAAI,EAAE,CAAC;IAC7E,CAAC;IACD,KAAK,IAAI,KAAK,GAAG,CAAC,EAAE,KAAK,IAAI,IAAI,EAAE,KAAK,EAAE,EAAE,CAAC;QAC3C,MAAM,KAAK,GAAG,OAAO,CAAC,KAAK,CAAU,CAAC;QACtC,IAAI,KAAK,GAAG,IAAI,IAAI,kBAAkB,CAAC,KAAK,CAAC;YAAE,SAAS;QACxD,IAAI,KAAK,KAAK,IAAI,IAAI,eAAe,CAAC,KAAK,CAAC,IAAI,CAAC;YAAE,IAAI,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC;IACtE,CAAC;IACD,OAAO,IAAI,CAAC;AACd,CAAC;AAED;;;;;;GAMG;AACH,MAAM,UAAU,kBAAkB,CAAC,KAAY;IAC7C,OAAO,KAAK,CAAC,IAAI,KAAK,SAAS,IAAI,SAAS,CAAC,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC;AAChE,CAAC;AAED;;;;;;;GAOG;AACH,MAAM,UAAU,gBAAgB,CAC9B,KAA2E,EAC3E,KAAa;IAEb,MAAM,IAAI,GACR,KAAK,YAAY,GAAG;QAClB,CAAC,CAAC,CAAC,EAAU,EAAE,EAAE,CAAC,KAAK,CAAC,GAAG,CAAC,EAAE,CAAC;QAC/B,CAAC,CAAC,CAAC,EAAU,EAAE,EAAE,CACZ,KAA6C,CAAC,MAAM,CAAC,EAAE,CAAC,CAAC,CAAC;IACnE,MAAM,MAAM,GAAY,EAAE,CAAC;IAC3B,MAAM,IAAI,GAAG,IAAI,GAAG,EAAU,CAAC;IAC/B,IAAI,EAAE,GAAG,KAAK,CAAC;IACf,OAAO,CAAC,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,EAAE,CAAC;QACrB,MAAM,IAAI,GAAG,IAAI,CAAC,EAAE,CAAC,CAAC;QACtB,IAAI,IAAI,KAAK,SAAS;YAAE,MAAM;QAC9B,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,CAAC;QACb,MAAM,CAAC,IAAI,CAAC,EAAE,IAAI,EAAE,IAAI,CAAC,IAAI,EAAE,IAAI,EAAE,IAAI,CAAC,IAAI,EAAE,KAAK,EAAE,EAAE,EAAE,CAAC,CAAC;QAC7D,EAAE,GAAG,IAAI,CAAC,QAAQ,CAAC;IACrB,CAAC;IACD,OAAO,MAAM,CAAC,UAAU,EAAE,CAAC;AAC7B,CAAC;AAED;;;;;;;;GAQG;AACH,MAAM,UAAU,gBAAgB,CAC9B,KAA0D;IAE1D,MAAM,MAAM,GAAG,IAAI,GAAG,EAAkB,CAAC;IACzC,KAAK,MAAM,GAAG,IAAI,MAAM,CAAC,IAAI,CAAC,KAAK,CAAC,EAAE,CAAC;QACrC,MAAM,IAAI,GAAG,KAAK,CAAC,GAAG,CAAC,EAAE,IAAI,EAAE,CAAC;QAChC,IAAI,IAAI,KAAK,SAAS,IAAI,IAAI,KAAK,EAAE;YAAE,SAAS;QAChD,MAAM,IAAI,GAAG,GAAG,CAAC,WAAW,EAAE,CAAC;QAC/B,IAAI,SAAS,CAAC,IAAI,CAAC,KAAK,SAAS;YAAE,MAAM,CAAC,GAAG,CAAC,IAAI,EAAE,IAAI,CAAC,CAAC;IAC5D,CAAC;IACD,MAAM,OAAO,GAAY,EAAE,CAAC;IAC5B,KAAK,MAAM,IAAI,IAAI,WAAW,EAAE,CAAC;QAC/B,MAAM,IAAI,GAAG,MAAM,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC;QAC9B,IAAI,IAAI,KAAK,SAAS;YAAE,OAAO,CAAC,IAAI,CAAC,EAAE,IAAI,EAAE,IAAI,EAAE,CAAC,CAAC;IACvD,CAAC;IACD,OAAO,OAAO,CAAC;AACjB,CAAC;AAED;;;;;;;;;GASG;AACH,MAAM,UAAU,YAAY,CAAC,OAAqB;IAChD,MAAM,KAAK,GAAc,EAAE,CAAC;IAC5B,IAAI,KAAK,GAAG,KAAK,CAAC;IAClB,KAAK,MAAM,KAAK,IAAI,OAAO,EAAE,CAAC;QAC5B,IAAI,cAAc,CAAC,KAAK,CAAC,IAAI,CAAC,EAAE,CAAC;YAC/B,KAAK,GAAG,IAAI,CAAC;QACf,CAAC;aAAM,IAAI,SAAS,CAAC,KAAK,CAAC,IAAI,CAAC,KAAK,SAAS,EAAE,CAAC;YAC/C,KAAK,GAAG,KAAK,CAAC;QAChB,CAAC;QACD,KAAK,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC;IACpB,CAAC;IACD,OAAO,KAAK,CAAC;AACf,CAAC"}

package/lib/taxonomy/core/ncbi.d.ts

@@ -0,0 +1,8 @@
+/**
+ * The page of a taxon in the NCBI Taxonomy Browser.
+ * @param taxId - The NCBI Taxonomy id.
+ * @returns Its address, e.g.
+ * `https://www.ncbi.nlm.nih.gov/Taxonomy/Browser/wwwtax.cgi?id=5076`.
+ */
+export declare function ncbiTaxonomyUrl(taxId: number): string;
+//# sourceMappingURL=ncbi.d.ts.map

package/lib/taxonomy/core/ncbi.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"ncbi.d.ts","sourceRoot":"","sources":["../../../src/taxonomy/core/ncbi.ts"],"names":[],"mappings":"AAAA;;;;;GAKG;AACH,wBAAgB,eAAe,CAAC,KAAK,EAAE,MAAM,GAAG,MAAM,CAErD"}

package/lib/taxonomy/core/ncbi.js

@@ -0,0 +1,10 @@
+/**
+ * The page of a taxon in the NCBI Taxonomy Browser.
+ * @param taxId - The NCBI Taxonomy id.
+ * @returns Its address, e.g.
+ * `https://www.ncbi.nlm.nih.gov/Taxonomy/Browser/wwwtax.cgi?id=5076`.
+ */
+export function ncbiTaxonomyUrl(taxId) {
+    return `https://www.ncbi.nlm.nih.gov/Taxonomy/Browser/wwwtax.cgi?id=${encodeURIComponent(String(taxId))}`;
+}
+//# sourceMappingURL=ncbi.js.map

package/lib/taxonomy/core/ncbi.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"ncbi.js","sourceRoot":"","sources":["../../../src/taxonomy/core/ncbi.ts"],"names":[],"mappings":"AAAA;;;;;GAKG;AACH,MAAM,UAAU,eAAe,CAAC,KAAa;IAC3C,OAAO,+DAA+D,kBAAkB,CAAC,MAAM,CAAC,KAAK,CAAC,CAAC,EAAE,CAAC;AAC5G,CAAC"}

package/lib/taxonomy/core/organismName.d.ts

@@ -0,0 +1,28 @@
+/**
+ * Which parts of an organism's name nomenclature sets in italics.
+ *
+ * The genus, the species epithet and an infraspecific epithet are italic; the
+ * authority, a strain, `sp.`, a cultivar and the words announcing a rank
+ * (`subsp.`, `var.`, `f.`) are upright. Names arrive the way their source wrote
+ * them — "Catharanthus roseus (L.) G. Don", "Streptomyces sp. CHQ-64",
+ * "Isodon Shikokiana Var. Occidentalis" — so this reads them by their shape,
+ * and a name it cannot place is left upright rather than guessed at.
+ */
+/** A run of an organism's name, set one way. */
+export interface OrganismNamePart {
+    /** The text, exactly as the name carries it. */
+    text: string;
+    /** Whether nomenclature sets it in italics. */
+    italic: boolean;
+}
+/**
+ * The runs of an organism's name, each set upright or in italics.
+ *
+ * Joined together, the runs give back the name exactly. A consortium written
+ * `A + B` is read as two names; a virus, and anything that does not open with
+ * a genus, is one upright run.
+ * @param name - The name, as its source writes it.
+ * @returns The runs, consecutive runs never set the same way.
+ */
+export declare function splitOrganismName(name: string): OrganismNamePart[];
+//# sourceMappingURL=organismName.d.ts.map

package/lib/taxonomy/core/organismName.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"organismName.d.ts","sourceRoot":"","sources":["../../../src/taxonomy/core/organismName.ts"],"names":[],"mappings":"AAAA;;;;;;;;;GASG;AAYH,gDAAgD;AAChD,MAAM,WAAW,gBAAgB;IAC/B,gDAAgD;IAChD,IAAI,EAAE,MAAM,CAAC;IACb,+CAA+C;IAC/C,MAAM,EAAE,OAAO,CAAC;CACjB;AAyBD;;;;;;;;GAQG;AACH,wBAAgB,iBAAiB,CAAC,IAAI,EAAE,MAAM,GAAG,gBAAgB,EAAE,CAYlE"}

package/lib/taxonomy/core/organismName.js

@@ -0,0 +1,188 @@
+/**
+ * Which parts of an organism's name nomenclature sets in italics.
+ *
+ * The genus, the species epithet and an infraspecific epithet are italic; the
+ * authority, a strain, `sp.`, a cultivar and the words announcing a rank
+ * (`subsp.`, `var.`, `f.`) are upright. Names arrive the way their source wrote
+ * them — "Catharanthus roseus (L.) G. Don", "Streptomyces sp. CHQ-64",
+ * "Isodon Shikokiana Var. Occidentalis" — so this reads them by their shape,
+ * and a name it cannot place is left upright rather than guessed at.
+ */
+import { BEFORE_EPITHET, CANDIDATUS, HYBRID, NOT_EPITHET, NOT_GENUS, RANK_MARKERS, UNNAMED, } from "./organismWords.js";
+const CONSORTIUM = /(?<plus>\s+\+\s+)/;
+const VIRUS = /\b(?:virus|viruses|phage|bacteriophage|viroid|satellite)\b/i;
+const SEPARATOR = /^[\s_]+$/;
+const TRAILING_MARK = /^(?<word>.+?)(?<mark>[,;]+)$/;
+const GENUS = /^[A-Z][a-z]+$/;
+const ABBREVIATED_GENUS = /^[A-Z]\.$/;
+// A family, a subfamily or a tribe written where a genus would stand:
+// -aceae, -oideae and -eae in botany, -idae in zoology. No genus ends so.
+const FAMILY = /(?:eae|idae)$/;
+const BRACKETED_GENUS = /^\[(?<genus>[A-Z][a-z]+)\]$/;
+const JOINED_MARKER = /^(?<marker>subsp\.|ssp\.|var\.|f\.|pv\.)(?<epithet>[a-z][a-z-]+)$/i;
+const EPITHET = /^[a-zß-öø-ÿ][a-zß-öø-ÿ-]+$/;
+const CAPITALISED_EPITHET = /^[A-Z][a-zß-öø-ÿ-]+$/;
+const STARTS_WITH_LETTER = /^\p{L}/u;
+const STARTS_UPPERCASE = /^\p{Lu}/u;
+/**
+ * The runs of an organism's name, each set upright or in italics.
+ *
+ * Joined together, the runs give back the name exactly. A consortium written
+ * `A + B` is read as two names; a virus, and anything that does not open with
+ * a genus, is one upright run.
+ * @param name - The name, as its source writes it.
+ * @returns The runs, consecutive runs never set the same way.
+ */
+export function splitOrganismName(name) {
+    if (name === '')
+        return [];
+    if (VIRUS.test(name))
+        return [{ text: name, italic: false }];
+    const pieces = [];
+    for (const segment of name.split(CONSORTIUM)) {
+        if (CONSORTIUM.test(segment)) {
+            pieces.push({ text: segment, word: false, italic: false });
+        }
+        else {
+            pieces.push(...styledPieces(segment));
+        }
+    }
+    return mergedRuns(pieces);
+}
+function styledPieces(text) {
+    const pieces = tokenize(text);
+    const words = pieces.filter((piece) => piece.word);
+    styleWords(words);
+    for (let index = 0; index < pieces.length; index++) {
+        const piece = pieces[index];
+        if (piece.word)
+            continue;
+        piece.italic =
+            pieces[index - 1]?.italic === true && pieces[index + 1]?.italic === true;
+    }
+    return pieces;
+}
+function tokenize(text) {
+    const pieces = [];
+    for (const chunk of text.split(/(?<gap>[\s_]+)/)) {
+        if (chunk === '')
+            continue;
+        if (SEPARATOR.test(chunk)) {
+            pieces.push({ text: chunk, word: false, italic: false });
+            continue;
+        }
+        // A comma or a semicolon closing a word is punctuation, not part of it.
+        const { word: bare = chunk, mark } = TRAILING_MARK.exec(chunk)?.groups ?? {};
+        const joined = JOINED_MARKER.exec(bare)?.groups;
+        const bracketed = BRACKETED_GENUS.exec(bare)?.groups;
+        const words = joined !== undefined
+            ? [joined.marker, joined.epithet]
+            : bracketed !== undefined
+                ? ['[', bracketed.genus, ']']
+                : [bare];
+        for (const word of words) {
+            pieces.push({ text: word ?? '', word: true, italic: false });
+        }
+        if (mark !== undefined) {
+            pieces.push({ text: mark, word: false, italic: false });
+        }
+    }
+    return pieces;
+}
+function styleWords(words) {
+    let index = words[0]?.text === '×' ? 1 : 0;
+    const first = words[index];
+    if (first === undefined)
+        return;
+    if (CANDIDATUS.has(first.text)) {
+        first.italic = true;
+        return;
+    }
+    let genus;
+    if (first.text === '[' && words[index + 2]?.text === ']') {
+        genus = words[index + 1];
+        index += 3;
+    }
+    else if (GENUS.test(first.text) || ABBREVIATED_GENUS.test(first.text)) {
+        genus = first;
+        index += 1;
+    }
+    if (genus === undefined ||
+        NOT_GENUS.has(genus.text.toLowerCase()) ||
+        FAMILY.test(genus.text)) {
+        return;
+    }
+    genus.italic = true;
+    const titleCase = isTitleCase(words);
+    const next = words[index]?.text;
+    if (next === undefined || UNNAMED.has(next.toLowerCase()))
+        return;
+    let announced = false;
+    if (HYBRID.has(next) || BEFORE_EPITHET.has(next.toLowerCase())) {
+        index += 1;
+        announced = true;
+    }
+    if (isEpithet(words[index], titleCase || announced)) {
+        words[index].italic = true;
+        index += 1;
+        // A zoological trinomial names its subspecies with no word announcing it.
+        if (isEpithet(words[index], false)) {
+            words[index].italic = true;
+            index += 1;
+        }
+    }
+    for (; index < words.length; index++) {
+        const word = words[index].text;
+        const marks = RANK_MARKERS.has(word.toLowerCase()) || HYBRID.has(word);
+        if (marks && isEpithet(words[index + 1], true)) {
+            words[index + 1].italic = true;
+            index += 1;
+        }
+    }
+}
+function isEpithet(word, capitalised) {
+    if (word === undefined)
+        return false;
+    const lower = word.text.toLowerCase();
+    if (NOT_EPITHET.has(lower) ||
+        UNNAMED.has(lower) ||
+        RANK_MARKERS.has(lower) ||
+        HYBRID.has(word.text)) {
+        return false;
+    }
+    return (EPITHET.test(word.text) ||
+        (capitalised && CAPITALISED_EPITHET.test(word.text)));
+}
+// A source that capitalises every word ("Agave Americana") still means a
+// binomial; the words announcing a rank are left out of the count.
+function isTitleCase(words) {
+    let lettered = 0;
+    for (const { text } of words) {
+        const lower = text.toLowerCase();
+        if (RANK_MARKERS.has(lower) || UNNAMED.has(lower) || HYBRID.has(text)) {
+            continue;
+        }
+        if (!STARTS_WITH_LETTER.test(text))
+            continue;
+        if (!STARTS_UPPERCASE.test(text))
+            return false;
+        lettered += 1;
+    }
+    return lettered >= 2;
+}
+function mergedRuns(pieces) {
+    const runs = [];
+    for (const piece of pieces) {
+        if (piece.text === '')
+            continue;
+        const last = runs.at(-1);
+        if (last?.italic === piece.italic) {
+            last.text += piece.text;
+        }
+        else {
+            runs.push({ text: piece.text, italic: piece.italic });
+        }
+    }
+    return runs;
+}
+//# sourceMappingURL=organismName.js.map

package/lib/taxonomy/core/organismName.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"organismName.js","sourceRoot":"","sources":["../../../src/taxonomy/core/organismName.ts"],"names":[],"mappings":"AAAA;;;;;;;;;GASG;AAEH,OAAO,EACL,cAAc,EACd,UAAU,EACV,MAAM,EACN,WAAW,EACX,SAAS,EACT,YAAY,EACZ,OAAO,GACR,MAAM,oBAAoB,CAAC;AAgB5B,MAAM,UAAU,GAAG,mBAAmB,CAAC;AACvC,MAAM,KAAK,GAAG,6DAA6D,CAAC;AAC5E,MAAM,SAAS,GAAG,UAAU,CAAC;AAC7B,MAAM,aAAa,GAAG,8BAA8B,CAAC;AACrD,MAAM,KAAK,GAAG,eAAe,CAAC;AAC9B,MAAM,iBAAiB,GAAG,WAAW,CAAC;AACtC,sEAAsE;AACtE,0EAA0E;AAC1E,MAAM,MAAM,GAAG,eAAe,CAAC;AAC/B,MAAM,eAAe,GAAG,6BAA6B,CAAC;AACtD,MAAM,aAAa,GACjB,oEAAoE,CAAC;AACvE,MAAM,OAAO,GAAG,4BAA4B,CAAC;AAC7C,MAAM,mBAAmB,GAAG,sBAAsB,CAAC;AACnD,MAAM,kBAAkB,GAAG,SAAS,CAAC;AACrC,MAAM,gBAAgB,GAAG,UAAU,CAAC;AAEpC;;;;;;;;GAQG;AACH,MAAM,UAAU,iBAAiB,CAAC,IAAY;IAC5C,IAAI,IAAI,KAAK,EAAE;QAAE,OAAO,EAAE,CAAC;IAC3B,IAAI,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC;QAAE,OAAO,CAAC,EAAE,IAAI,EAAE,IAAI,EAAE,MAAM,EAAE,KAAK,EAAE,CAAC,CAAC;IAC7D,MAAM,MAAM,GAAY,EAAE,CAAC;IAC3B,KAAK,MAAM,OAAO,IAAI,IAAI,CAAC,KAAK,CAAC,UAAU,CAAC,EAAE,CAAC;QAC7C,IAAI,UAAU,CAAC,IAAI,CAAC,OAAO,CAAC,EAAE,CAAC;YAC7B,MAAM,CAAC,IAAI,CAAC,EAAE,IAAI,EAAE,OAAO,EAAE,IAAI,EAAE,KAAK,EAAE,MAAM,EAAE,KAAK,EAAE,CAAC,CAAC;QAC7D,CAAC;aAAM,CAAC;YACN,MAAM,CAAC,IAAI,CAAC,GAAG,YAAY,CAAC,OAAO,CAAC,CAAC,CAAC;QACxC,CAAC;IACH,CAAC;IACD,OAAO,UAAU,CAAC,MAAM,CAAC,CAAC;AAC5B,CAAC;AAED,SAAS,YAAY,CAAC,IAAY;IAChC,MAAM,MAAM,GAAG,QAAQ,CAAC,IAAI,CAAC,CAAC;IAC9B,MAAM,KAAK,GAAG,MAAM,CAAC,MAAM,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC;IACnD,UAAU,CAAC,KAAK,CAAC,CAAC;IAClB,KAAK,IAAI,KAAK,GAAG,CAAC,EAAE,KAAK,GAAG,MAAM,CAAC,MAAM,EAAE,KAAK,EAAE,EAAE,CAAC;QACnD,MAAM,KAAK,GAAG,MAAM,CAAC,KAAK,CAAU,CAAC;QACrC,IAAI,KAAK,CAAC,IAAI;YAAE,SAAS;QACzB,KAAK,CAAC,MAAM;YACV,MAAM,CAAC,KAAK,GAAG,CAAC,CAAC,EAAE,MAAM,KAAK,IAAI,IAAI,MAAM,CAAC,KAAK,GAAG,CAAC,CAAC,EAAE,MAAM,KAAK,IAAI,CAAC;IAC7E,CAAC;IACD,OAAO,MAAM,CAAC;AAChB,CAAC;AAED,SAAS,QAAQ,CAAC,IAAY;IAC5B,MAAM,MAAM,GAAY,EAAE,CAAC;IAC3B,KAAK,MAAM,KAAK,IAAI,IAAI,CAAC,KAAK,CAAC,gBAAgB,CAAC,EAAE,CAAC;QACjD,IAAI,KAAK,KAAK,EAAE;YAAE,SAAS;QAC3B,IAAI,SAAS,CAAC,IAAI,CAAC,KAAK,CAAC,EAAE,CAAC;YAC1B,MAAM,CAAC,IAAI,CAAC,EAAE,IAAI,EAAE,KAAK,EAAE,IAAI,EAAE,KAAK,EAAE,MAAM,EAAE,KAAK,EAAE,CAAC,CAAC;YACzD,SAAS;QACX,CAAC;QACD,wEAAwE;QACxE,MAAM,EAAE,IAAI,EAAE,IAAI,GAAG,KAAK,EAAE,IAAI,EAAE,GAChC,aAAa,CAAC,IAAI,CAAC,KAAK,CAAC,EAAE,MAAM,IAAI,EAAE,CAAC;QAC1C,MAAM,MAAM,GAAG,aAAa,CAAC,IAAI,CAAC,IAAI,CAAC,EAAE,MAAM,CAAC;QAChD,MAAM,SAAS,GAAG,eAAe,CAAC,IAAI,CAAC,IAAI,CAAC,EAAE,MAAM,CAAC;QACrD,MAAM,KAAK,GACT,MAAM,KAAK,SAAS;YAClB,CAAC,CAAC,CAAC,MAAM,CAAC,MAAM,EAAE,MAAM,CAAC,OAAO,CAAC;YACjC,CAAC,CAAC,SAAS,KAAK,SAAS;gBACvB,CAAC,CAAC,CAAC,GAAG,EAAE,SAAS,CAAC,KAAK,EAAE,GAAG,CAAC;gBAC7B,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC;QACf,KAAK,MAAM,IAAI,IAAI,KAAK,EAAE,CAAC;YACzB,MAAM,CAAC,IAAI,CAAC,EAAE,IAAI,EAAE,IAAI,IAAI,EAAE,EAAE,IAAI,EAAE,IAAI,EAAE,MAAM,EAAE,KAAK,EAAE,CAAC,CAAC;QAC/D,CAAC;QACD,IAAI,IAAI,KAAK,SAAS,EAAE,CAAC;YACvB,MAAM,CAAC,IAAI,CAAC,EAAE,IAAI,EAAE,IAAI,EAAE,IAAI,EAAE,KAAK,EAAE,MAAM,EAAE,KAAK,EAAE,CAAC,CAAC;QAC1D,CAAC;IACH,CAAC;IACD,OAAO,MAAM,CAAC;AAChB,CAAC;AAED,SAAS,UAAU,CAAC,KAAc;IAChC,IAAI,KAAK,GAAG,KAAK,CAAC,CAAC,CAAC,EAAE,IAAI,KAAK,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;IAC3C,MAAM,KAAK,GAAG,KAAK,CAAC,KAAK,CAAC,CAAC;IAC3B,IAAI,KAAK,KAAK,SAAS;QAAE,OAAO;IAChC,IAAI,UAAU,CAAC,GAAG,CAAC,KAAK,CAAC,IAAI,CAAC,EAAE,CAAC;QAC/B,KAAK,CAAC,MAAM,GAAG,IAAI,CAAC;QACpB,OAAO;IACT,CAAC;IAED,IAAI,KAAwB,CAAC;IAC7B,IAAI,KAAK,CAAC,IAAI,KAAK,GAAG,IAAI,KAAK,CAAC,KAAK,GAAG,CAAC,CAAC,EAAE,IAAI,KAAK,GAAG,EAAE,CAAC;QACzD,KAAK,GAAG,KAAK,CAAC,KAAK,GAAG,CAAC,CAAC,CAAC;QACzB,KAAK,IAAI,CAAC,CAAC;IACb,CAAC;SAAM,IAAI,KAAK,CAAC,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,IAAI,iBAAiB,CAAC,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,EAAE,CAAC;QACxE,KAAK,GAAG,KAAK,CAAC;QACd,KAAK,IAAI,CAAC,CAAC;IACb,CAAC;IACD,IACE,KAAK,KAAK,SAAS;QACnB,SAAS,CAAC,GAAG,CAAC,KAAK,CAAC,IAAI,CAAC,WAAW,EAAE,CAAC;QACvC,MAAM,CAAC,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,EACvB,CAAC;QACD,OAAO;IACT,CAAC;IACD,KAAK,CAAC,MAAM,GAAG,IAAI,CAAC;IAEpB,MAAM,SAAS,GAAG,WAAW,CAAC,KAAK,CAAC,CAAC;IACrC,MAAM,IAAI,GAAG,KAAK,CAAC,KAAK,CAAC,EAAE,IAAI,CAAC;IAChC,IAAI,IAAI,KAAK,SAAS,IAAI,OAAO,CAAC,GAAG,CAAC,IAAI,CAAC,WAAW,EAAE,CAAC;QAAE,OAAO;IAClE,IAAI,SAAS,GAAG,KAAK,CAAC;IACtB,IAAI,MAAM,CAAC,GAAG,CAAC,IAAI,CAAC,IAAI,cAAc,CAAC,GAAG,CAAC,IAAI,CAAC,WAAW,EAAE,CAAC,EAAE,CAAC;QAC/D,KAAK,IAAI,CAAC,CAAC;QACX,SAAS,GAAG,IAAI,CAAC;IACnB,CAAC;IACD,IAAI,SAAS,CAAC,KAAK,CAAC,KAAK,CAAC,EAAE,SAAS,IAAI,SAAS,CAAC,EAAE,CAAC;QACnD,KAAK,CAAC,KAAK,CAAW,CAAC,MAAM,GAAG,IAAI,CAAC;QACtC,KAAK,IAAI,CAAC,CAAC;QACX,0EAA0E;QAC1E,IAAI,SAAS,CAAC,KAAK,CAAC,KAAK,CAAC,EAAE,KAAK,CAAC,EAAE,CAAC;YAClC,KAAK,CAAC,KAAK,CAAW,CAAC,MAAM,GAAG,IAAI,CAAC;YACtC,KAAK,IAAI,CAAC,CAAC;QACb,CAAC;IACH,CAAC;IAED,OAAO,KAAK,GAAG,KAAK,CAAC,MAAM,EAAE,KAAK,EAAE,EAAE,CAAC;QACrC,MAAM,IAAI,GAAI,KAAK,CAAC,KAAK,CAAW,CAAC,IAAI,CAAC;QAC1C,MAAM,KAAK,GAAG,YAAY,CAAC,GAAG,CAAC,IAAI,CAAC,WAAW,EAAE,CAAC,IAAI,MAAM,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC;QACvE,IAAI,KAAK,IAAI,SAAS,CAAC,KAAK,CAAC,KAAK,GAAG,CAAC,CAAC,EAAE,IAAI,CAAC,EAAE,CAAC;YAC9C,KAAK,CAAC,KAAK,GAAG,CAAC,CAAW,CAAC,MAAM,GAAG,IAAI,CAAC;YAC1C,KAAK,IAAI,CAAC,CAAC;QACb,CAAC;IACH,CAAC;AACH,CAAC;AAED,SAAS,SAAS,CAAC,IAAuB,EAAE,WAAoB;IAC9D,IAAI,IAAI,KAAK,SAAS;QAAE,OAAO,KAAK,CAAC;IACrC,MAAM,KAAK,GAAG,IAAI,CAAC,IAAI,CAAC,WAAW,EAAE,CAAC;IACtC,IACE,WAAW,CAAC,GAAG,CAAC,KAAK,CAAC;QACtB,OAAO,CAAC,GAAG,CAAC,KAAK,CAAC;QAClB,YAAY,CAAC,GAAG,CAAC,KAAK,CAAC;QACvB,MAAM,CAAC,GAAG,CAAC,IAAI,CAAC,IAAI,CAAC,EACrB,CAAC;QACD,OAAO,KAAK,CAAC;IACf,CAAC;IACD,OAAO,CACL,OAAO,CAAC,IAAI,CAAC,IAAI,CAAC,IAAI,CAAC;QACvB,CAAC,WAAW,IAAI,mBAAmB,CAAC,IAAI,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC,CACrD,CAAC;AACJ,CAAC;AAED,yEAAyE;AACzE,mEAAmE;AACnE,SAAS,WAAW,CAAC,KAAuB;IAC1C,IAAI,QAAQ,GAAG,CAAC,CAAC;IACjB,KAAK,MAAM,EAAE,IAAI,EAAE,IAAI,KAAK,EAAE,CAAC;QAC7B,MAAM,KAAK,GAAG,IAAI,CAAC,WAAW,EAAE,CAAC;QACjC,IAAI,YAAY,CAAC,GAAG,CAAC,KAAK,CAAC,IAAI,OAAO,CAAC,GAAG,CAAC,KAAK,CAAC,IAAI,MAAM,CAAC,GAAG,CAAC,IAAI,CAAC,EAAE,CAAC;YACtE,SAAS;QACX,CAAC;QACD,IAAI,CAAC,kBAAkB,CAAC,IAAI,CAAC,IAAI,CAAC;YAAE,SAAS;QAC7C,IAAI,CAAC,gBAAgB,CAAC,IAAI,CAAC,IAAI,CAAC;YAAE,OAAO,KAAK,CAAC;QAC/C,QAAQ,IAAI,CAAC,CAAC;IAChB,CAAC;IACD,OAAO,QAAQ,IAAI,CAAC,CAAC;AACvB,CAAC;AAED,SAAS,UAAU,CAAC,MAAwB;IAC1C,MAAM,IAAI,GAAuB,EAAE,CAAC;IACpC,KAAK,MAAM,KAAK,IAAI,MAAM,EAAE,CAAC;QAC3B,IAAI,KAAK,CAAC,IAAI,KAAK,EAAE;YAAE,SAAS;QAChC,MAAM,IAAI,GAAG,IAAI,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC,CAAC;QACzB,IAAI,IAAI,EAAE,MAAM,KAAK,KAAK,CAAC,MAAM,EAAE,CAAC;YAClC,IAAI,CAAC,IAAI,IAAI,KAAK,CAAC,IAAI,CAAC;QAC1B,CAAC;aAAM,CAAC;YACN,IAAI,CAAC,IAAI,CAAC,EAAE,IAAI,EAAE,KAAK,CAAC,IAAI,EAAE,MAAM,EAAE,KAAK,CAAC,MAAM,EAAE,CAAC,CAAC;QACxD,CAAC;IACH,CAAC;IACD,OAAO,IAAI,CAAC;AACd,CAAC"}

package/lib/taxonomy/core/organismWords.d.ts

@@ -0,0 +1,20 @@
+/**
+ * The words an organism's name is read by: the signs and abbreviations that
+ * announce a rank or qualify a name, and the ordinary words a source writes
+ * where a genus or an epithet would stand.
+ */
+/** The hybrid sign, and the letters written for it. */
+export declare const HYBRID: Set<string>;
+/** The word that opens a provisional bacterial name; it alone is italic. */
+export declare const CANDIDATUS: Set<string>;
+/** Qualifiers set before the epithet they qualify, in lower case. */
+export declare const BEFORE_EPITHET: Set<string>;
+/** An unnamed species of the genus, in lower case. */
+export declare const UNNAMED: Set<string>;
+/** The abbreviations announcing a lower-rank epithet, in lower case. */
+export declare const RANK_MARKERS: Set<string>;
+/** Words that open a description rather than a binomial, in lower case. */
+export declare const NOT_GENUS: Set<string>;
+/** Lower-case words that follow a binomial without being an epithet. */
+export declare const NOT_EPITHET: Set<string>;
+//# sourceMappingURL=organismWords.d.ts.map

package/lib/taxonomy/core/organismWords.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"organismWords.d.ts","sourceRoot":"","sources":["../../../src/taxonomy/core/organismWords.ts"],"names":[],"mappings":"AAAA;;;;GAIG;AAEH,uDAAuD;AACvD,eAAO,MAAM,MAAM,aAA2B,CAAC;AAC/C,4EAA4E;AAC5E,eAAO,MAAM,UAAU,aAAiC,CAAC;AACzD,qEAAqE;AACrE,eAAO,MAAM,cAAc,aAA+C,CAAC;AAC3E,sDAAsD;AACtD,eAAO,MAAM,OAAO,aAAwC,CAAC;AAC7D,wEAAwE;AACxE,eAAO,MAAM,YAAY,aAiBvB,CAAC;AACH,2EAA2E;AAC3E,eAAO,MAAM,SAAS,aAmBpB,CAAC;AACH,wEAAwE;AACxE,eAAO,MAAM,WAAW,aA8BtB,CAAC"}

package/lib/taxonomy/core/organismWords.js

@@ -0,0 +1,86 @@
+/**
+ * The words an organism's name is read by: the signs and abbreviations that
+ * announce a rank or qualify a name, and the ordinary words a source writes
+ * where a genus or an epithet would stand.
+ */
+/** The hybrid sign, and the letters written for it. */
+export const HYBRID = new Set(['×', 'x', 'X']);
+/** The word that opens a provisional bacterial name; it alone is italic. */
+export const CANDIDATUS = new Set(['Candidatus', 'Ca.']);
+/** Qualifiers set before the epithet they qualify, in lower case. */
+export const BEFORE_EPITHET = new Set(['cf.', 'cf', 'aff.', 'aff', 'nr.']);
+/** An unnamed species of the genus, in lower case. */
+export const UNNAMED = new Set(['sp.', 'sp', 'spp.', 'spp']);
+/** The abbreviations announcing a lower-rank epithet, in lower case. */
+export const RANK_MARKERS = new Set([
+    'subsp.',
+    'ssp.',
+    'nothosubsp.',
+    'var.',
+    'nothovar.',
+    'subvar.',
+    'f.',
+    'fo.',
+    'forma',
+    'subf.',
+    'pv.',
+    'subgen.',
+    'sect.',
+    'subsect.',
+    'ser.',
+    'subser.',
+]);
+/** Words that open a description rather than a binomial, in lower case. */
+export const NOT_GENUS = new Set([
+    'unknown',
+    'unidentified',
+    'uncultured',
+    'unclassified',
+    'environmental',
+    'fungal',
+    'bacterial',
+    'marine',
+    'endophytic',
+    'mixed',
+    'fungus',
+    'bacterium',
+    'endophyte',
+    'strain',
+    'isolate',
+    'plant',
+    'soil',
+    'sponge',
+]);
+/** Lower-case words that follow a binomial without being an epithet. */
+export const NOT_EPITHET = new Set([
+    'strain',
+    'isolate',
+    'symbiont',
+    'endophyte',
+    'clone',
+    'group',
+    'complex',
+    'culture',
+    'cultivar',
+    'serovar',
+    'biovar',
+    'pathovar',
+    'species',
+    'incertae',
+    'sedis',
+    'and',
+    'et',
+    'ex',
+    'in',
+    'non',
+    'sensu',
+    'auct',
+    'de',
+    'du',
+    'la',
+    'le',
+    'van',
+    'von',
+    'der',
+]);
+//# sourceMappingURL=organismWords.js.map

package/lib/taxonomy/core/organismWords.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"organismWords.js","sourceRoot":"","sources":["../../../src/taxonomy/core/organismWords.ts"],"names":[],"mappings":"AAAA;;;;GAIG;AAEH,uDAAuD;AACvD,MAAM,CAAC,MAAM,MAAM,GAAG,IAAI,GAAG,CAAC,CAAC,GAAG,EAAE,GAAG,EAAE,GAAG,CAAC,CAAC,CAAC;AAC/C,4EAA4E;AAC5E,MAAM,CAAC,MAAM,UAAU,GAAG,IAAI,GAAG,CAAC,CAAC,YAAY,EAAE,KAAK,CAAC,CAAC,CAAC;AACzD,qEAAqE;AACrE,MAAM,CAAC,MAAM,cAAc,GAAG,IAAI,GAAG,CAAC,CAAC,KAAK,EAAE,IAAI,EAAE,MAAM,EAAE,KAAK,EAAE,KAAK,CAAC,CAAC,CAAC;AAC3E,sDAAsD;AACtD,MAAM,CAAC,MAAM,OAAO,GAAG,IAAI,GAAG,CAAC,CAAC,KAAK,EAAE,IAAI,EAAE,MAAM,EAAE,KAAK,CAAC,CAAC,CAAC;AAC7D,wEAAwE;AACxE,MAAM,CAAC,MAAM,YAAY,GAAG,IAAI,GAAG,CAAC;IAClC,QAAQ;IACR,MAAM;IACN,aAAa;IACb,MAAM;IACN,WAAW;IACX,SAAS;IACT,IAAI;IACJ,KAAK;IACL,OAAO;IACP,OAAO;IACP,KAAK;IACL,SAAS;IACT,OAAO;IACP,UAAU;IACV,MAAM;IACN,SAAS;CACV,CAAC,CAAC;AACH,2EAA2E;AAC3E,MAAM,CAAC,MAAM,SAAS,GAAG,IAAI,GAAG,CAAC;IAC/B,SAAS;IACT,cAAc;IACd,YAAY;IACZ,cAAc;IACd,eAAe;IACf,QAAQ;IACR,WAAW;IACX,QAAQ;IACR,YAAY;IACZ,OAAO;IACP,QAAQ;IACR,WAAW;IACX,WAAW;IACX,QAAQ;IACR,SAAS;IACT,OAAO;IACP,MAAM;IACN,QAAQ;CACT,CAAC,CAAC;AACH,wEAAwE;AACxE,MAAM,CAAC,MAAM,WAAW,GAAG,IAAI,GAAG,CAAC;IACjC,QAAQ;IACR,SAAS;IACT,UAAU;IACV,WAAW;IACX,OAAO;IACP,OAAO;IACP,SAAS;IACT,SAAS;IACT,UAAU;IACV,SAAS;IACT,QAAQ;IACR,UAAU;IACV,SAAS;IACT,UAAU;IACV,OAAO;IACP,KAAK;IACL,IAAI;IACJ,IAAI;IACJ,IAAI;IACJ,KAAK;IACL,OAAO;IACP,MAAM;IACN,IAAI;IACJ,IAAI;IACJ,IAAI;IACJ,IAAI;IACJ,KAAK;IACL,KAAK;IACL,KAAK;CACN,CAAC,CAAC"}