@larsgw/formica 0.8.4 → 0.8.5

package/src/resources/parse-text.ts

@@ -1,38 +1,7 @@
 import * as yaml from 'js-yaml'
 import { WorkResource } from './resource'
 import { createDiff, ResourceDiffType } from './diff-resource'
-
-const RANKS: Rank[] = [
-    'phylum',
-    'subphylum',
-    'class',
-    'infraclass',
-    'superorder',
-    'order',
-    'suborder',
-    'infraorder',
-    'superfamily',
-    'family',
-    'subfamily',
-    'tribe',
-    'subtribe',
-    'genus',
-    'subgenus',
-    'section', // not ICZN
-    'subsection', // not ICZN
-    'series', // not ICZN
-    'group',
-    'subgroup', // ...
-    'aggregate', // not ICZN
-    'complex', // not ICZN
-    'species',
-    'subspecies',
-    'variety',
-    'form',
-    'aberration', // not ICZN
-    'race', // not ICZN
-    'stirps' // not ICZN
-]
+import { parseName, RANKS } from './parse-name'
 
 const MAIN_RANKS: Rank[] = [
     'kingdom',
@@ -62,313 +31,15 @@ const FLAGS: ResourceFlag[] = [
     'MISSING_AUTHORSHIP'
 ]
 
-const TAXONOMIC_STATUS: Record<string, TaxonStatus> = {
-    '>': 'incorrect',
-    '+': 'heterotypic synonym',
-    '=': 'synonym'
-}
-
-const RANK_LABELS: Record<Rank, string> = {
-    'subspecies': 'subsp.',
-    'variety': 'var.',
-    'form': 'f.',
-    'aberration': 'ab.',
-    'race': 'r.',
-    'stirps': 'st.'
-}
-
-const RANK_LABELS_REVERSE: Record<string, Rank> = {
-    'st': 'stirps',
-    'r': 'race',
-    'ab': 'aberration',
-    'f': 'form',
-    'var': 'variety',
-    'ssp': 'subspecies',
-    'subsp': 'subspecies'
-}
-
-const HYBRID_SIGN = '\u00D7'
-
-/**
- *   1. Any number of
- *      - capitalized words
- *      - "&"
- *      - " in "
- *      - " ex "
- *      - lowercase name particles
- *   2. Followed by a capitalized word
- *   3. Optionally, followed by "et al."
- */
-const LOWERCASE_NAME_PARTICLES = ['y', 'der', 'den', 'de', 'van', 'von'].join('|')
-const SIMPLE_AUTHOR_PATTERN = '(?:(?:\\p{Lu}\\S*|&|in|ex|' + LOWERCASE_NAME_PARTICLES + ')\\s*)*\\p{Lu}\\S+(?:\\s+et\\s+al\\.)?'
-
-const NAME_PATTERN = new RegExp(
-    '^' +
-        // $1 main name part
-        '(\\S+)' +
-        // $2 optional author citation
-        '(?: ' +
-            // but not auct(t)., etc.
-            '(?!auctt?\\.|(?:syn|comb|sp|spec|nom|gen|subgen)\\. n(?:ov)?\\.|s(?:ens[.u]|\\.)|in part|partim)' +
-        '(' +
-            // $2.1 anything in parentheses, followed by optional revising author(s)
-            '\\(.+?\\)(?:\\s+' + SIMPLE_AUTHOR_PATTERN + ')?' +
-        '|' +
-            // $2.2 anything followed by a year
-            '.+?\\d{4}\\)?' +
-        '|' +
-            // $2.3 author(s)
-            SIMPLE_AUTHOR_PATTERN +
-        '))?' +
-        // $3 optional notes
-        '(?:,? (.+))?' +
-    '$',
-    'u'
-)
-
-/**
- * Structure
- *   $1 genus+subgenus (+ trailing space): (?:([A-Z]\S+) (?:\(([A-Z]\S+?)\) )?)?
- *     $1.1 genus: ((?:x )?[A-Z]\S+)
- *     $1.2 subgenus: (?:\(([A-Z]\S+?)\) )?
- *   $2 species: (x [a-z-]+|[a-z-][^\s.]+(?: x [a-z-]+)?|[A-Z][a-z]+_[a-z-]+ x [A-Z][a-z]+_[a-z-]+)
- *     $2a: x [a-z-]+
- *     $2b hybrid: [a-z-][^\s.]+(?: x [a-z-]+)?
- *     $2c intergeneric hybrid: [A-Z][a-z]+_[a-z-]+ x [A-Z][a-z]+_[a-z-]+
- */
-const BINAME_PATTERN = /^(?:((?:x )?[A-Z]\S+) (?:\(([A-Z]\S+?)\) )?)?(x [a-z-]+|[a-z-][^\s.]+(?: x [a-z-]+)?|[A-Z][a-z]+_[a-z-]+ x [A-Z][a-z]+_[a-z-]+)(?= |$)/
-
-function compareRanks (a: Rank, b: Rank): number {
-    return RANKS.indexOf(a) - RANKS.indexOf(b)
-}
-
-function capitalize (name: string): string {
-    return name[0].toUpperCase() + name.slice(1).toLowerCase()
-}
-
-function capitalizeGenericName (name: string): string {
-    if (name[0] === HYBRID_SIGN) {
-        return HYBRID_SIGN + capitalize(name.slice(1))
-    }
+const RESOURCE_DELIMITER = '\n\n===\n\n'
+const INDENT = 2
 
-    return capitalize(name)
+function makeParseError (message: string, line: number, column: number = 1): SyntaxError {
+    return new SyntaxError(`[${line}:${column}] ${message}`)
 }
 
-function isUpperCase (name: string): boolean {
-    return name === name.toUpperCase()
-}
-
-function getSynonymRank (name: string, rank: Rank): Rank {
-    const rest = name.replace(BINAME_PATTERN, '')
-    const rankPrefix = rest.match(/^(?: |^)(st|r|ab|f|var|ssp|subsp)\. /)
-    if (rankPrefix) {
-        return RANK_LABELS_REVERSE[rankPrefix[1]] as string
-    } else if (!BINAME_PATTERN.test(name)) {
-        return rank
-    } else if (/^ (?!sensu)[a-z0-9-]+($| )/.test(rest)) {
-        return 'subspecies'
-    } else {
-        return 'species'
-    }
-}
-
-function capitalizeAuthors (authors: string): string {
-    return authors
-        .replace(
-            /[^\x00-\x40\x5B-\x60\x7B-\x7F]+/g, // eslint-disable-line no-control-regex
-            name => isUpperCase(name) ? capitalize(name) : name
-        )
-        .replace(/ Y /g, ' y ')
-}
-
-function parseName (name: string, rank: Rank, parent: WorkingTaxon): WorkingTaxon {
-    const item = {} as WorkingTaxon
-
-    // Synonyms have the accepted name usage as 'parent'.
-    const isSynonym = /^[+=>] /.test(name)
-    if (isSynonym) {
-        item.taxonomicStatus = TAXONOMIC_STATUS[name[0]]
-        name = name.replace(/^[+=>] (\? ?)?/, '')
-        rank = getSynonymRank(name, parent.taxonRank as Rank)
-    } else {
-        item.taxonomicStatus = 'accepted'
-    }
-
-    // Clusters
-    if (/^\[(_|\d+)\] /.test(name)) {
-        name = name.replace(/^\[(_|\d+)\] /, '')
-    }
-
-    // Set verbatim identification after subsequent syntax is removed.
-    item.verbatimIdentification = name.replace(/(?<=^| )x(?=$| )/g, HYBRID_SIGN).replace(/_/g, ' ')
-
-    // Parent context is used for parsing and formatting binomial names.
-    // For formatting, it needs to match external databases (i.e. be correct).
-    // For parsing, it needs to match the current file. If relevant parents
-    // (i.e. genus, species) had mistakes that were corrected, the uncorrected
-    // genus and species names need to be used.
-    const parentContext = {
-        genus: parent.genus,
-        subgenus: parent.subgenus,
-        specificEpithet: parent.specificEpithet,
-        incorrect: {
-            genus: parent.incorrect && parent.incorrect.genus,
-            specificEpithet: parent.incorrect && parent.incorrect.specificEpithet
-        }
-    }
-
-    // Both contexts should be amended in the two cases where binomial names
-    // are fully used: (1) synonyms and (2) multinomial taxa without parents to
-    // provide parts of the name (e.g. bare species without a genus parent, or
-    // even subspecies without a species or genus parent).
-    if (isSynonym || !parentContext.genus || (compareRanks('species', rank) < 0 && !parentContext.specificEpithet)) {
-        const [, genus, subgenus, species] = name.match(BINAME_PATTERN) || []
-        if (genus) {
-            parentContext.incorrect.genus = genus
-            parentContext.genus = capitalizeGenericName(genus.replace(/(^| )x /, HYBRID_SIGN))
-        }
-        if (subgenus) {
-            parentContext.subgenus = capitalize(subgenus)
-        } else if (genus) {
-            // If a genus is given but no subgenus, remove any existing subgenus
-            // from the parent context.
-            delete parentContext.subgenus
-        }
-        if (species && compareRanks('species', rank) < 0) {
-            parentContext.incorrect.specificEpithet = species
-            parentContext.specificEpithet = species.replace(/(^| )x /, HYBRID_SIGN)
-        }
-    }
-
-    // In taxa of group, species or lower, the name should just contain the
-    // (infra)specific epithet and the author information & remarks when processing
-    // further.
-    if (compareRanks('group', rank) <= 0) {
-        // Remove genus
-        const genus = parentContext.incorrect.genus || parentContext.genus || ''
-        if (name[0] === genus[0] && name.toLowerCase().startsWith(genus.toLowerCase() + ' ')) {
-            name = name.slice(genus.length + 1)
-        }
-
-        // Remove subgenus
-        name = name.replace(/^\(.*?\) /, '')
-
-        // Infraspecific taxa
-        if (compareRanks('species', rank) < 0) {
-            // Remove specific epithet
-            const species = parentContext.incorrect.specificEpithet || parentContext.specificEpithet || ''
-            if (name.startsWith(species + ' ')) {
-                name = name.slice(species.length + 1)
-            }
-
-            // Remove rank abbreviations
-            name = name.replace(/^(st|r|ab|f|var|ssp|subsp)\. /, '')
-        }
-    }
-
-    // Hybrids
-    if (rank === 'genus' && name.startsWith('x ')) {
-        name = HYBRID_SIGN + name.slice(2)
-    }
-
-    if (rank === 'species' && /(^| )x /.test(name)) {
-        name = name.replace(/(^| )x /, HYBRID_SIGN)
-    }
-
-    // Divide the name into the main scientific name (only the epithet for taxa
-    // lower than genus), the authorship information, and optionally remarks
-    const nameParts = name.match(NAME_PATTERN)
-    if (!nameParts) {
-        throw new Error(`Taxon "${name}" could not be parsed`)
-    }
-
-    // To encode old names with spaces (e.g. "Orsillus pini canariensis Lindberg, 1953")
-    // underscores are used, which are replaced here. This is also used for undescribed
-    // species (e.g. "Leiobunum species A") and intergeneric hybrids (e.g. "×Festulpia
-    // Festuca rubra × Vulpia bromoides")
-    if (nameParts[1].includes('_')) {
-        nameParts[1] = nameParts[1].replace(/_/g, ' ')
-    }
-
-    const [_, taxon, citation = '', notes] = nameParts
-    item.scientificNameAuthorship = capitalizeAuthors(citation)
-    item.taxonRemarks = notes
-    item.taxonRank = rank
-
-    // @ts-expect-error TS1501: This regular expression flag is only available when targeting 'es6' or later.
-    if (/[^\p{L}0-9\u{00D7}\- ]/u.test(taxon)) {
-        throw new Error(`Taxon name contains unexpected characters: "${taxon}"`)
-    }
-
-    // Validate names and recompose binomial and trinomial names
-    if (rank === 'genus') {
-        item.scientificName = capitalizeGenericName(taxon)
-        if (taxon[0].toUpperCase() !== taxon[0] || (taxon[0] === HYBRID_SIGN && taxon[1].toUpperCase() !== taxon[1])) {
-            throw new Error(`Generic epithet should be capitalized: "${taxon}"`)
-        }
-    } else if (compareRanks('group', rank) > 0) {
-        item.scientificName = capitalize(taxon)
-        if (taxon[0].toUpperCase() !== taxon[0]) {
-            throw new Error(`Taxon name (${rank}) should be capitalized: "${taxon}"`)
-        }
-    } else if (rank === 'group') {
-        item.genericName = parentContext.genus
-        item.infragenericEpithet = parentContext.subgenus
-        const specificEpithet = taxon.toLowerCase().replace(/(-group)?$/, '')
-        item.scientificName = `${item.genericName} ${specificEpithet}-group`
-        if (taxon.toLowerCase() !== taxon) {
-            console.log(item, taxon)
-            throw new Error(`Group name should be lowercase: "${taxon}"`)
-        }
-    } else if (rank === 'subgroup') {
-        item.genericName = parentContext.genus
-        item.infragenericEpithet = parentContext.subgenus
-        const specificEpithet = taxon.toLowerCase().replace(/(-subgroup)?$/, '')
-        item.scientificName = `${item.genericName} ${specificEpithet}-subgroup`
-        if (taxon.toLowerCase() !== taxon) {
-            console.log(item, taxon)
-            throw new Error(`Subgroup name should be lowercase: "${taxon}"`)
-        }
-    } else if (rank === 'species') {
-        item.genericName = parentContext.genus
-        item.infragenericEpithet = parentContext.subgenus
-        if (taxon.toLowerCase() !== taxon && !/^[A-Z][a-z]+ [a-z]+\xD7[A-Z][a-z]+ [a-z]+$/.test(taxon)) {
-            console.log(item, taxon)
-            throw new Error(`Specific epithet should be lowercase: "${taxon}"`)
-        }
-        item.specificEpithet = taxon
-        item.scientificName = `${item.genericName} ${item.specificEpithet}`
-    } else if (compareRanks('species', rank) < 0) {
-        item.genericName = parentContext.genus
-        item.infragenericEpithet = parentContext.subgenus
-        item.specificEpithet = parentContext.specificEpithet
-        item.infraspecificEpithet = taxon.toLowerCase()
-
-        // If possible, names below species should have abbreviations for ranks,
-        // like "subsp."
-        const nameParts = [
-            item.genericName,
-            item.specificEpithet,
-            item.infraspecificEpithet
-        ]
-        if (item.taxonRank in RANK_LABELS) {
-            nameParts.splice(2, 0, RANK_LABELS[item.taxonRank])
-        }
-        item.scientificName = nameParts.join(' ')
-
-        if (item.infraspecificEpithet !== taxon) {
-            console.log(item, taxon)
-            throw new Error(`Infraspecific epithet should be lowercase: "${taxon}"`)
-        }
-    }
-
-    // Re-add authorship information
-    item.scientificNameOnly = item.scientificName
-    if (item.scientificNameAuthorship) {
-        item.scientificName += ` ${item.scientificNameAuthorship}`
-    }
-
-    return item
+function mergeParserErrors (errors: SyntaxError[]): SyntaxError {
+    return new SyntaxError(errors.map(error => error.message).join('\n'))
 }
 
 function parseHeader (header: string): ResourceMetadata {
@@ -447,113 +118,101 @@ function parseHeader (header: string): ResourceMetadata {
     return metadata
 }
 
-function validateResource (config: ResourceMetadata, content: string) {
-    // Check for too much indentation
-    const longerIndent = new RegExp(`^(  ){${config.levels.length - 1}}(?!  [+=>] |    > ) `, 'm')
-    const longerIndentMatch = content.match(longerIndent)
-    if (longerIndentMatch !== null) {
-        const offset = longerIndentMatch.index
-        const line = (content.slice(0, offset).match(/\n/g) || []).length + 1
-        throw new SyntaxError(`Too much indentation at ${line}:0
-${content.slice(offset).split('\n', 1)}
-^`)
-    }
+function parseResource (resource: FilePart): [ResourceMetadata, FilePart] {
+    const [header, _, ...rest] = resource.content.split(/(\n---\n+)/)
+    let config
 
-    // Check for missing leaf taxa
-    const leafTaxonRank = config.levels.filter(rank => MAIN_RANKS.includes(rank)).pop() as string
-    const leafTaxonParentIndent = config.levels.indexOf(leafTaxonRank) - 1
-    if (leafTaxonRank && leafTaxonParentIndent >= 0) {
-        const missingLeafTaxa = new RegExp(`^((?:  ){0,${leafTaxonParentIndent}})(?![+=> ] ).*\\n(\\1(  )+[+=>].*\\n)*(?!\\1  )`, 'm')
-        const missingLeafTaxaMatch = content.match(missingLeafTaxa)
-        if (missingLeafTaxaMatch !== null) {
-            const offset = missingLeafTaxaMatch.index
-            const line = (content.slice(0, offset).match(/\n/g) || []).length + 1
-            throw new SyntaxError(`Missing leaf taxon at ${line}:0
-${content.slice(offset).split('\n', 1)}
-^`)
-        }
+    try {
+        config = parseHeader(header)
+    } catch (error) {
+        throw makeParseError(error.message, resource.offsetLine + 1)
     }
-}
 
-function parseResource (resource: string): [ResourceMetadata, string] {
-    const [header, _, ...rest] = resource.split(/(\n---\n+)/)
-    const config = parseHeader(header)
     const content = rest.join('')
+    const offsetLine = resource.offsetLine + (header + _).split('\n').length - 1
 
-    return [config, content]
+    return [config, { content, offsetLine }]
 }
 
-function getIndentation (line: string): number {
-    return (line.match(/^ */) as string[])[0].length
-}
+function parseResourceContent (content: ResourceDiff, resource: Resource, oldIds: number[], offsetLine: number): Resource {
+    const leafTaxonIndex = resource.metadata.levels.reduce((last, rank, i) => MAIN_RANKS.includes(rank) ? i : last, 0)
+    const data = resource.taxa as Record<TaxonId, WorkingTaxon>
+    const errors = []
 
-function isIndetLine (line: string, indent?: number): boolean {
-    if (indent === undefined) {
-        indent = getIndentation(line)
-    }
+    let id = 0
+    let newId = Math.max(...oldIds)
+    let lineNumber = offsetLine
 
-    line = line.slice(indent)
+    const parents: Array<TaxonId | null> = []
+    const previous = { id: '', indent: 0, group: { isLeaf: false, indent: 0 } }
 
-    return line.startsWith('[indet]')
-}
+    for (const line of content) {
+        const hasOriginalId = line.type !== ResourceDiffType.Added && !/^\s*(\[indet\]|> )/.test(line.original ?? line.text as string)
+        if (hasOriginalId) {
+            id++
+        }
 
-function parseResourceContent (content: ResourceDiff, resource: Resource, oldIds: number[]): Resource {
-    const idBase = `${resource.id}:`
+        if (line.type === ResourceDiffType.Deleted) {
+            continue
+        } else {
+            lineNumber++
+        }
 
-    const data = resource.taxa as Record<TaxonId, WorkingTaxon>
-    let id = 0
-    let parents: Array<TaxonId | null> = []
-    let groupIndent = 0
-    let previousId = ''
-    let newIdOffset = Math.max(...oldIds)
+        const [indentation, name] = (line.text as string).match(/^(\s*)(.*)/)!.slice(1)
+        const lineIndent = indentation.length
 
-    for (const line of content) {
-        if (line.type === ResourceDiffType.Deleted) {
-            // Increase id counter for removed line unless it was an "indet line"
-            if (!isIndetLine(line.original as string)) {
-                id++
-            }
+        // Validate line
+        if (lineIndent % INDENT === 1) {
+            errors.push(makeParseError('Too much or little indentation', lineNumber))
+            continue
+        } else if (lineIndent / INDENT >= resource.metadata.levels.length && !/^[+=>] /.test(name)) {
+            errors.push(makeParseError('Too much indentation', lineNumber))
+            continue
+        } else if (lineIndent <= previous.group.indent && (data[previous.id] && !previous.group.isLeaf)) {
+            errors.push(makeParseError('Missing leaf taxon', lineNumber - 1))
             continue
         }
 
-        const lineIndent = getIndentation(line.text as string)
-        if (lineIndent > groupIndent) {
+        // Update parentage
+        if (lineIndent > previous.indent) {
             // Do not count synonyms as parents (unless this is correcting a typo in the synonym)
-            if (data[previousId] && data[previousId].taxonomicStatus === 'accepted' || /^( {2})+> /.test(line.text as string)) {
-                parents.push(previousId)
+            if (data[previous.id] && data[previous.id].taxonomicStatus === 'accepted' || name.startsWith('> ')) {
+                parents.push(previous.id)
             } else {
                 parents.push(null)
             }
+
             // Handle skips in indentation levels,
             // e.g. if a certain genus has only species
             // whereas other genera in the same key also
             // have subgenera
-            if ((lineIndent - groupIndent) > 2) {
-                const gap = (lineIndent - groupIndent - 2) / 2
-                for (let i = 0; i < gap; i++) { parents.push(null) }
+            for (let i = previous.indent + INDENT; i < lineIndent; i += INDENT) {
+                parents.push(null)
             }
-            groupIndent = lineIndent
-        } else if (lineIndent < groupIndent) {
-            parents = parents.slice(0, lineIndent / 2)
-            groupIndent = lineIndent
+        } else if (lineIndent < previous.indent) {
+            parents.splice(lineIndent / INDENT)
         }
 
+        previous.indent = lineIndent
+
         // Do not process "indet" lines further, as they only serve to indicate
         // that subtaxa are explicitely omitted
-        if (isIndetLine(line.text as string, lineIndent)) {
-            // If the line was previously not and ndet line, increase the id counter
-            if (line.type === ResourceDiffType.Modified && !isIndetLine(line.original as string)) {
-                id++
-            }
+        if (name.startsWith('[indet]')) {
+            previous.group.isLeaf = lineIndent / INDENT >= leafTaxonIndex
             continue
         }
 
-        const parentId = parents.reduce((grandparent, parent) => parent || grandparent, null)
+        const parentId = parents.reduce((grandparent, parent) => parent ?? grandparent, null)
         const parent = parentId === null ? {} as WorkingTaxon : data[parentId]
-        const name = (line.text as string).slice(groupIndent)
-        const rank = resource.metadata.levels[groupIndent / 2]
-        const item = parseName(name, rank, parent)
-        const isSynonym = item.taxonomicStatus !== 'accepted'
+        const rank = resource.metadata.levels[parents.length]
+        let item
+
+        try {
+            item = parseName(name, rank, parent)
+        } catch (error) {
+            errors.push(makeParseError(error.message, lineNumber))
+            continue
+        }
 
         // Add higher classification info
         const itemAsObject = item as { [index: string]: unknown }
@@ -575,15 +234,7 @@ function parseResourceContent (content: ResourceDiff, resource: Resource, oldIds
             item.subgenus = item.infragenericEpithet
         }
 
-        if (isSynonym) {
-            item.higherClassification = parent.higherClassification
-        } else if (parent.higherClassification) {
-            item.higherClassification = parent.higherClassification + ` | ${parent.scientificNameOnly}`
-        } else if (parentId) {
-            item.higherClassification = parent.scientificNameOnly
-        }
-
-        // Amend "parent" with corrections
+        // Amend "parent" with corrections, exit
         if (item.taxonomicStatus === 'incorrect') {
             parent.incorrect = { ...parent }
             for (const key in item) {
@@ -594,40 +245,65 @@ function parseResourceContent (content: ResourceDiff, resource: Resource, oldIds
             continue
         }
 
-        // Set identifiers
-        if (line.type === ResourceDiffType.Added) {
-            newIdOffset++
-            item.scientificNameID = idBase + newIdOffset.toString()
-        } else if (line.type === ResourceDiffType.Modified && isIndetLine(line.original as string)) {
-            newIdOffset++
-            item.scientificNameID = idBase + newIdOffset.toString()
-        } else {
-            id++
-            item.scientificNameID = idBase + (oldIds[id - 1] || id).toString()
+        // Add more classification info
+        const isSynonym = item.taxonomicStatus !== 'accepted'
+        if (isSynonym) {
+            item.higherClassification = parent.higherClassification
+        } else if (parent.higherClassification) {
+            item.higherClassification = parent.higherClassification + ` | ${parent.scientificNameOnly}`
+        } else if (parentId) {
+            item.higherClassification = parent.scientificNameOnly
         }
-        previousId = item.scientificNameID
+
+        // Set identifiers
+        item.scientificNameID = `${resource.id}:${hasOriginalId ? (oldIds[id - 1] ?? id) : ++newId}`
 
         item.parentNameUsageID = isSynonym ? undefined : parent.scientificNameID
         item.parentNameUsage = isSynonym ? undefined : parent.scientificName
         item.acceptedNameUsageID = isSynonym ? parent.scientificNameID : undefined
         item.acceptedNameUsage = isSynonym ? parent.scientificName : undefined
-        item.collectionCode = idBase.slice(0, -1)
+        item.collectionCode = resource.id
 
         data[item.scientificNameID] = item
+
+        // Update loop state
+        previous.id = item.scientificNameID
+        if (item.taxonomicStatus === 'accepted') {
+            previous.group.indent = previous.indent
+            previous.group.isLeaf = lineIndent / INDENT >= leafTaxonIndex
+        }
+    }
+
+    if (errors.length) {
+        throw mergeParserErrors(errors)
     }
 
     return resource
 }
 
-function splitResources (file: string): string[] {
-    return file.split('\n\n===\n\n')
+interface FilePart {
+    content: string
+    offsetLine: number
+}
+
+function splitResources (file: string): FilePart[] {
+    const resources: FilePart[] = []
+    let offsetLine = 0
+    for (const content of file.split(RESOURCE_DELIMITER)) {
+        resources.push({ content, offsetLine })
+        offsetLine += (content + RESOURCE_DELIMITER).split('\n').length - 1
+    }
+    return resources
 }
 
 export function parseFile (file: string, id: WorkId, old?: ResourceHistory): Resource[] {
     const oldResources = old ? splitResources(old.txt) : []
-    return splitResources(file).map((resource, index) => {
-        const [config, content] = parseResource(resource)
-        validateResource(config, content)
+    const newResources = splitResources(file)
+    const resources: Resource[] = []
+    const errors = []
+
+    for (let index = 0; index < newResources.length; index++) {
+        const [config, content] = parseResource(newResources[index])
         const template: Resource = {
             id: `${id}:${index + 1}`,
             file: `${id}-${index + 1}`,
@@ -638,11 +314,11 @@ export function parseFile (file: string, id: WorkId, old?: ResourceHistory): Res
 
         let diff
         if (oldResources[index]) {
-            diff = createDiff(content, parseResource(oldResources[index])[1])
+            diff = createDiff(content.content, parseResource(oldResources[index])[1].content)
             // Ignore empty lines
             diff = diff.filter(line => line.text !== '')
         } else {
-            diff = createDiff(content, content)
+            diff = createDiff(content.content, content.content)
         }
 
         const oldIds = []
@@ -652,8 +328,18 @@ export function parseFile (file: string, id: WorkId, old?: ResourceHistory): Res
             }
         }
 
-        return parseResourceContent(diff, template, oldIds)
-    })
+        try {
+            resources.push(parseResourceContent(diff, template, oldIds, content.offsetLine))
+        } catch (error) {
+            errors.push(error)
+        }
+    }
+
+    if (errors.length) {
+        throw mergeParserErrors(errors)
+    }
+
+    return resources
 }
 
 export function parseFileHeader (file: string): ResourceMetadata[] {