gff-nostream 3.0.5 → 3.0.8

package/src/api.ts

@@ -14,8 +14,10 @@ import type {
 export interface LineRecord {
   line: string
   lineHash?: string | number
-  start: number
-  end: number
+  /** Optional passthrough byte offsets — not used by the parser */
+  start?: number
+  /** Optional passthrough byte offsets — not used by the parser */
+  end?: number
   hasEscapes: boolean
 }
 
@@ -44,19 +46,14 @@ function stringToRecords(str: string) {
   const lines = str.split(/\r?\n/)
   const records: LineRecord[] = []
   for (const line of lines) {
+    if (line.startsWith('##FASTA') || line.startsWith('>')) {
+      break
+    }
     if (line.length === 0 || line.startsWith('#')) {
-      if (line.startsWith('##FASTA')) {
-        break
-      }
       continue
     }
-    if (line.startsWith('>')) {
-      break
-    }
     records.push({
       line,
-      start: 0,
-      end: 0,
       hasEscapes: line.includes('%'),
     })
   }
@@ -95,48 +92,52 @@ export function parseRecords(records: LineRecord[]): GFF3Feature[] {
 
     if (!ids && !parents) {
       items.push([featureLine])
-      continue
-    }
-
-    let feature: GFF3Feature
-    if (ids) {
-      const id = ids[0]!
-      const existing = byId.get(id)
-      if (existing) {
-        existing.push(featureLine)
-        feature = existing
-      } else {
-        feature = [featureLine]
-        if (!parents) {
-          items.push(feature)
-        }
-        byId.set(id, feature)
-        const waiting = orphans.get(id)
-        if (waiting) {
-          for (const w of waiting) {
-            featureLine.child_features.push(w)
+    } else {
+      let feature: GFF3Feature
+      if (ids) {
+        const id = ids[0]!
+        const existing = byId.get(id)
+        if (existing) {
+          // Multi-location continuation: share child_features/derived_features
+          // with the first line so children remain visible across all lines
+          // regardless of arrival order.
+          featureLine.child_features = existing[0]!.child_features
+          featureLine.derived_features = existing[0]!.derived_features
+          existing.push(featureLine)
+          feature = existing
+        } else {
+          feature = [featureLine]
+          if (!parents) {
+            items.push(feature)
+          }
+          byId.set(id, feature)
+          const waiting = orphans.get(id)
+          if (waiting) {
+            for (const w of waiting) {
+              featureLine.child_features.push(w)
+            }
+            orphans.delete(id)
           }
-          orphans.delete(id)
         }
+      } else {
+        feature = [featureLine]
       }
-    } else {
-      feature = [featureLine]
-    }
 
-    if (parents) {
-      for (const parentId of parents) {
-        const parent = byId.get(parentId)
-        if (parent) {
-          for (const p of parent) {
-            p.child_features.push(feature)
+      if (parents) {
+        for (const parentId of parents) {
+          const parent = byId.get(parentId)
+          if (parent) {
+            // child_features is shared across all parent feature lines,
+            // so push once via the first line.
+            parent[0]!.child_features.push(feature)
+          } else {
+            let arr = orphans.get(parentId)
+            if (!arr) {
+              arr = []
+              orphans.set(parentId, arr)
+            }
+            arr.push(feature)
           }
-        } else {
-          let arr = orphans.get(parentId)
-          if (!arr) {
-            arr = []
-            orphans.set(parentId, arr)
-          }
-          arr.push(feature)
         }
       }
     }
@@ -166,17 +167,16 @@ export function parseRecordsJBrowse(records: LineRecord[]): JBrowseFeature[] {
       feature._lineHash = String(record.lineHash)
     }
 
-    const id = feature.id as string | undefined
+    // attribute parsing collapses single-element arrays to scalars, so id can
+    // be string | string[]; defensively take the first if multi-valued.
+    const rawId = feature.id as string | string[] | undefined
+    const id = Array.isArray(rawId) ? rawId[0] : rawId
     const parent = feature.parent as string | string[] | undefined
 
     if (!id && !parent) {
       items.push(feature)
-      continue
-    }
-
-    if (id) {
-      const existing = byId.get(id)
-      if (!existing) {
+    } else if (!id || !byId.has(id)) {
+      if (id) {
         if (!parent) {
           items.push(feature)
         }
@@ -189,21 +189,21 @@ export function parseRecordsJBrowse(records: LineRecord[]): JBrowseFeature[] {
           orphans.delete(id)
         }
       }
-    }
 
-    if (parent) {
-      const parents = Array.isArray(parent) ? parent : [parent]
-      for (const parentId of parents) {
-        const parentFeature = byId.get(parentId)
-        if (parentFeature) {
-          parentFeature.subfeatures.push(feature)
-        } else {
-          let arr = orphans.get(parentId)
-          if (!arr) {
-            arr = []
-            orphans.set(parentId, arr)
+      if (parent) {
+        const parents = Array.isArray(parent) ? parent : [parent]
+        for (const parentId of parents) {
+          const parentFeature = byId.get(parentId)
+          if (parentFeature) {
+            parentFeature.subfeatures.push(feature)
+          } else {
+            let arr = orphans.get(parentId)
+            if (!arr) {
+              arr = []
+              orphans.set(parentId, arr)
+            }
+            arr.push(feature)
           }
-          arr.push(feature)
         }
       }
     }

package/src/util.ts

@@ -2,7 +2,6 @@
 // JavaScript port of Robert Buels's Bio::GFF3::LowLevel Perl module.
 
 const directiveRegex = /^\s*##\s*(\S+)\s*(.*)/
-const lineEndRegex = /\r?\n$/
 const whitespaceRegex = /\s+/
 const nonDigitRegex = /\D/g
 
@@ -123,11 +122,20 @@ export function parseAttributesNoUnescape(attrString: string): GFF3Attributes {
   return parseAttributesImpl(attrString, false)
 }
 
-function normImpl(s: string, shouldUnescape: boolean) {
-  if (s.length === 0 || s === '.') {
-    return null
-  }
-  return shouldUnescape ? unescape(s) : s
+function isEmpty(s: string) {
+  return s.length === 0 || s === '.'
+}
+
+function strField<E extends null | ''>(
+  s: string,
+  shouldUnescape: boolean,
+  empty: E,
+) {
+  return isEmpty(s) ? empty : shouldUnescape ? unescape(s) : s
+}
+
+function numField(s: string) {
+  return isEmpty(s) ? null : +s
 }
 
 function parseFeatureImpl(
@@ -135,26 +143,19 @@ function parseFeatureImpl(
   shouldUnescape: boolean,
 ): GFF3FeatureLine {
   const f = line.split('\t')
-  const startStr = f[3]!
-  const endStr = f[4]!
-  const scoreStr = f[5]!
-  const strand = f[6]!
-  const phase = f[7]!
   const attrString = f[8]!
-
   return {
-    seq_id: normImpl(f[0]!, shouldUnescape),
-    source: normImpl(f[1]!, shouldUnescape),
-    type: normImpl(f[2]!, shouldUnescape),
-    start: startStr.length === 0 || startStr === '.' ? null : +startStr,
-    end: endStr.length === 0 || endStr === '.' ? null : +endStr,
-    score: scoreStr.length === 0 || scoreStr === '.' ? null : +scoreStr,
-    strand: normImpl(strand, false),
-    phase: normImpl(phase, false),
-    attributes:
-      attrString.length === 0 || attrString === '.'
-        ? null
-        : parseAttributesImpl(attrString, shouldUnescape),
+    seq_id: strField(f[0]!, shouldUnescape, null),
+    source: strField(f[1]!, shouldUnescape, null),
+    type: strField(f[2]!, shouldUnescape, null),
+    start: numField(f[3]!),
+    end: numField(f[4]!),
+    score: numField(f[5]!),
+    strand: strField(f[6]!, false, null),
+    phase: strField(f[7]!, false, null),
+    attributes: isEmpty(attrString)
+      ? null
+      : parseAttributesImpl(attrString, shouldUnescape),
   }
 }
 
@@ -179,61 +180,6 @@ export function parseFeatureNoUnescape(line: string): GFF3FeatureLine {
   return parseFeatureImpl(line, false)
 }
 
-function parseFieldsArrayImpl(
-  f: (string | null | undefined)[],
-  shouldUnescape: boolean,
-): GFF3FeatureLine {
-  const seq_id = f[0]
-  const source = f[1]
-  const type = f[2]
-  const startStr = f[3]
-  const endStr = f[4]
-  const scoreStr = f[5]
-  const strand = f[6]
-  const phase = f[7]
-  const attrString = f[8]
-
-  return {
-    seq_id: seq_id ? normImpl(seq_id, shouldUnescape) : null,
-    source: source ? normImpl(source, shouldUnescape) : null,
-    type: type ? normImpl(type, shouldUnescape) : null,
-    start: !startStr || startStr === '.' ? null : +startStr,
-    end: !endStr || endStr === '.' ? null : +endStr,
-    score: !scoreStr || scoreStr === '.' ? null : +scoreStr,
-    strand: strand && strand !== '.' ? strand : null,
-    phase: phase && phase !== '.' ? phase : null,
-    attributes:
-      !attrString || attrString === '.'
-        ? null
-        : parseAttributesImpl(attrString, shouldUnescape),
-  }
-}
-
-/**
- * Parse a GFF3 feature from a pre-split fields array
- *
- * @param f - Array of 9 GFF3 column values (use null or '.' for empty values)
- * @returns The parsed feature
- */
-export function parseFieldsArray(
-  f: (string | null | undefined)[],
-): GFF3FeatureLine {
-  return parseFieldsArrayImpl(f, true)
-}
-
-/**
- * Parse a GFF3 feature from a pre-split fields array without unescaping.
- * Fast path for data known to contain no escaped characters.
- *
- * @param f - Array of 9 GFF3 column values (use null or '.' for empty values)
- * @returns The parsed feature
- */
-export function parseFieldsArrayNoUnescape(
-  f: (string | null | undefined)[],
-): GFF3FeatureLine {
-  return parseFieldsArrayImpl(f, false)
-}
-
 /**
  * Parse a GFF3 directive line.
  *
@@ -252,17 +198,16 @@ export function parseDirective(
     return null
   }
 
-  const [, name] = match
-  let [, , contents] = match
+  const name = match[1]!
+  const contents = match[2]!
 
-  const parsed: GFF3Directive = { directive: name! }
-  if (contents!.length) {
-    contents = contents!.replace(lineEndRegex, '')
-    parsed.value = contents
+  const parsed: GFF3Directive = { directive: name }
+  if (contents.length) {
+    parsed.value = contents.trimEnd()
   }
 
   if (name === 'sequence-region') {
-    const c = contents!.split(whitespaceRegex, 3)
+    const c = contents.split(whitespaceRegex, 3)
     return {
       ...parsed,
       seq_id: c[0]!,
@@ -270,7 +215,7 @@ export function parseDirective(
       end: c[2]!.replaceAll(nonDigitRegex, ''),
     }
   } else if (name === 'genome-build') {
-    const [source, buildName] = contents!.split(whitespaceRegex, 2)
+    const [source, buildName] = contents.split(whitespaceRegex, 2)
     return {
       ...parsed,
       source: source!,
@@ -446,11 +391,6 @@ function parseAttributesJBrowseImpl(
       const eqIdx = attrString.indexOf('=', start)
       if (eqIdx !== -1 && eqIdx < semiIdx && eqIdx + 1 < semiIdx) {
         const tag = attrString.slice(start, eqIdx)
-        if (tag === '_lineHash') {
-          start = semiIdx + 1
-          continue
-        }
-
         let key = COMMON_ATTRS[tag]
         if (key === undefined) {
           key = tag.toLowerCase()
@@ -494,46 +434,32 @@ export function parseAttributesJBrowseNoUnescape(
   parseAttributesJBrowseImpl(attrString, result, false)
 }
 
+const STRAND_MAP: Record<string, number | undefined> = {
+  '+': 1,
+  '-': -1,
+  '.': 0,
+}
+
 function parseFeatureJBrowseImpl(
   line: string,
   shouldUnescape: boolean,
 ): JBrowseFeature {
   const f = line.split('\t')
-  const seq_id = f[0]!
-  const source = f[1]!
-  const type = f[2]!
   const startStr = f[3]!
   const endStr = f[4]!
   const scoreStr = f[5]!
-  const strand = f[6]!
   const phase = f[7]!
   const attrString = f[8]!
 
   const result: JBrowseFeature = {
-    refName:
-      seq_id.length === 0 || seq_id === '.'
-        ? ''
-        : shouldUnescape
-          ? unescape(seq_id)
-          : seq_id,
-    source:
-      source.length === 0 || source === '.'
-        ? null
-        : shouldUnescape
-          ? unescape(source)
-          : source,
-    type:
-      type.length === 0 || type === '.'
-        ? null
-        : shouldUnescape
-          ? unescape(type)
-          : type,
-    start: startStr.length === 0 || startStr === '.' ? 0 : +startStr - 1,
-    end: endStr.length === 0 || endStr === '.' ? 0 : +endStr,
-    score: scoreStr.length === 0 || scoreStr === '.' ? undefined : +scoreStr,
-    strand:
-      strand === '+' ? 1 : strand === '-' ? -1 : strand === '.' ? 0 : undefined,
-    phase: phase.length === 0 || phase === '.' ? undefined : +phase,
+    refName: strField(f[0]!, shouldUnescape, ''),
+    source: strField(f[1]!, shouldUnescape, null),
+    type: strField(f[2]!, shouldUnescape, null),
+    start: isEmpty(startStr) ? 0 : +startStr - 1,
+    end: isEmpty(endStr) ? 0 : +endStr,
+    score: isEmpty(scoreStr) ? undefined : +scoreStr,
+    strand: STRAND_MAP[f[6]!],
+    phase: isEmpty(phase) ? undefined : +phase,
     subfeatures: [],
   }