gff-nostream 1.3.1

package/esm/parse.js

@@ -0,0 +1,317 @@
+import * as GFF3 from './util';
+const containerAttributes = {
+    Parent: 'child_features',
+    Derives_from: 'derived_features',
+};
+export class FASTAParser {
+    constructor(seqCallback) {
+        this.seqCallback = seqCallback;
+        this.currentSequence = undefined;
+    }
+    addLine(line) {
+        const defMatch = /^>\s*(\S+)\s*(.*)/.exec(line);
+        if (defMatch) {
+            this._flush();
+            this.currentSequence = { id: defMatch[1], sequence: '' };
+            if (defMatch[2]) {
+                this.currentSequence.description = defMatch[2].trim();
+            }
+        }
+        else if (this.currentSequence && /\S/.test(line)) {
+            this.currentSequence.sequence += line.replaceAll(/\s/g, '');
+        }
+    }
+    _flush() {
+        if (this.currentSequence) {
+            this.seqCallback(this.currentSequence);
+        }
+    }
+    finish() {
+        this._flush();
+    }
+}
+export default class Parser {
+    constructor(args) {
+        this.fastaParser = undefined;
+        // if this is true, the parser ignores the
+        // rest of the lines in the file.  currently
+        // set when the file switches over to FASTA
+        this.eof = false;
+        this.lineNumber = 0;
+        // features that we have to keep on hand for now because they
+        // might be referenced by something else
+        this._underConstructionTopLevel = [];
+        // index of the above by ID
+        this._underConstructionById = {};
+        this._completedReferences = {};
+        // features that reference something we have not seen yet
+        // structured as:
+        // {  'some_id' : {
+        //     'Parent' : [ orphans that have a Parent attr referencing it ],
+        //     'Derives_from' : [ orphans that have a Derives_from attr referencing it ],
+        //    }
+        // }
+        this._underConstructionOrphans = {};
+        // eslint-disable-next-line @typescript-eslint/no-empty-function
+        const nullFunc = () => { };
+        this.featureCallback = args.featureCallback || nullFunc;
+        this.endCallback = args.endCallback || nullFunc;
+        this.commentCallback = args.commentCallback || nullFunc;
+        this.errorCallback = args.errorCallback || nullFunc;
+        this.directiveCallback = args.directiveCallback || nullFunc;
+        this.sequenceCallback = args.sequenceCallback || nullFunc;
+        this.disableDerivesFromReferences =
+            args.disableDerivesFromReferences || false;
+        // number of lines to buffer
+        this.bufferSize = args.bufferSize === undefined ? 1000 : args.bufferSize;
+    }
+    addLine(line) {
+        // if we have transitioned to a fasta section, just delegate to that parser
+        if (this.fastaParser) {
+            this.fastaParser.addLine(line);
+            return;
+        }
+        if (this.eof) {
+            // otherwise, if we are done, ignore this line
+            return;
+        }
+        this.lineNumber += 1;
+        if (/^\s*[^#\s>]/.test(line)) {
+            // feature line, most common case
+            this._bufferLine(line);
+            return;
+        }
+        const match = /^\s*(#+)(.*)/.exec(line);
+        if (match) {
+            // directive or comment
+            const [, hashsigns] = match;
+            let [, , contents] = match;
+            if (hashsigns.length === 3) {
+                // sync directive, all forward-references are resolved.
+                this._emitAllUnderConstructionFeatures();
+            }
+            else if (hashsigns.length === 2) {
+                const directive = GFF3.parseDirective(line);
+                if (directive) {
+                    if (directive.directive === 'FASTA') {
+                        this._emitAllUnderConstructionFeatures();
+                        this.eof = true;
+                        this.fastaParser = new FASTAParser(this.sequenceCallback);
+                    }
+                    else {
+                        this._emitItem(directive);
+                    }
+                }
+            }
+            else {
+                contents = contents.replace(/\s*/, '');
+                this._emitItem({ comment: contents });
+            }
+        }
+        else if (/^\s*$/.test(line)) {
+            // blank line, do nothing
+        }
+        else if (/^\s*>/.test(line)) {
+            // implicit beginning of a FASTA section
+            this._emitAllUnderConstructionFeatures();
+            this.eof = true;
+            this.fastaParser = new FASTAParser(this.sequenceCallback);
+            this.fastaParser.addLine(line);
+        }
+        else {
+            // it's a parse error
+            const errLine = line.replaceAll(/\r?\n?$/g, '');
+            throw new Error(`GFF3 parse error.  Cannot parse '${errLine}'.`);
+        }
+    }
+    finish() {
+        this._emitAllUnderConstructionFeatures();
+        if (this.fastaParser) {
+            this.fastaParser.finish();
+        }
+        this.endCallback();
+    }
+    _emitItem(i) {
+        if (Array.isArray(i)) {
+            this.featureCallback(i);
+        }
+        else if ('directive' in i) {
+            this.directiveCallback(i);
+        }
+        else if ('comment' in i) {
+            this.commentCallback(i);
+        }
+    }
+    _enforceBufferSizeLimit(additionalItemCount = 0) {
+        const _unbufferItem = (item) => {
+            var _a, _b;
+            if (item && Array.isArray(item) && ((_b = (_a = item[0].attributes) === null || _a === void 0 ? void 0 : _a.ID) === null || _b === void 0 ? void 0 : _b[0])) {
+                const ids = item[0].attributes.ID;
+                ids.forEach(id => {
+                    delete this._underConstructionById[id];
+                    delete this._completedReferences[id];
+                });
+                item.forEach(i => {
+                    if (i.child_features) {
+                        i.child_features.forEach(c => _unbufferItem(c));
+                    }
+                    if (i.derived_features) {
+                        i.derived_features.forEach(d => _unbufferItem(d));
+                    }
+                });
+            }
+        };
+        while (this._underConstructionTopLevel.length + additionalItemCount >
+            this.bufferSize) {
+            const item = this._underConstructionTopLevel.shift();
+            if (item) {
+                this._emitItem(item);
+                _unbufferItem(item);
+            }
+        }
+    }
+    /**
+     * return all under-construction features, called when we know
+     * there will be no additional data to attach to them
+     */
+    _emitAllUnderConstructionFeatures() {
+        this._underConstructionTopLevel.forEach(this._emitItem.bind(this));
+        this._underConstructionTopLevel = [];
+        this._underConstructionById = {};
+        this._completedReferences = {};
+        // if we have any orphans hanging around still, this is a
+        // problem. die with a parse error
+        if (Array.from(Object.values(this._underConstructionOrphans)).length) {
+            throw new Error(`some features reference other features that do not exist in the file (or in the same '###' scope). ${Object.keys(this._underConstructionOrphans).join(',')}`);
+        }
+    }
+    // do the right thing with a newly-parsed feature line
+    _bufferLine(line) {
+        var _a, _b, _c;
+        const rawFeatureLine = GFF3.parseFeature(line);
+        const featureLine = {
+            ...rawFeatureLine,
+            child_features: [],
+            derived_features: [],
+        };
+        // featureLine._lineNumber = this.lineNumber //< debugging aid
+        // NOTE: a feature is an arrayref of one or more feature lines.
+        const ids = ((_a = featureLine.attributes) === null || _a === void 0 ? void 0 : _a.ID) || [];
+        const parents = ((_b = featureLine.attributes) === null || _b === void 0 ? void 0 : _b.Parent) || [];
+        const derives = this.disableDerivesFromReferences
+            ? []
+            : ((_c = featureLine.attributes) === null || _c === void 0 ? void 0 : _c.Derives_from) || [];
+        if (!ids.length && !parents.length && !derives.length) {
+            // if it has no IDs and does not refer to anything, we can just
+            // output it
+            this._emitItem([featureLine]);
+            return;
+        }
+        let feature = undefined;
+        ids.forEach(id => {
+            const existing = this._underConstructionById[id];
+            if (existing) {
+                // another location of the same feature
+                if (existing[existing.length - 1].type !== featureLine.type) {
+                    this._parseError(`multi-line feature "${id}" has inconsistent types: "${featureLine.type}", "${existing[existing.length - 1].type}"`);
+                }
+                existing.push(featureLine);
+                feature = existing;
+            }
+            else {
+                // haven't seen it yet, so buffer it so we can attach
+                // child features to it
+                feature = [featureLine];
+                this._enforceBufferSizeLimit(1);
+                if (!parents.length && !derives.length) {
+                    this._underConstructionTopLevel.push(feature);
+                }
+                this._underConstructionById[id] = feature;
+                // see if we have anything buffered that refers to it
+                this._resolveReferencesTo(feature, id);
+            }
+        });
+        // try to resolve all its references
+        this._resolveReferencesFrom(feature || [featureLine], { Parent: parents, Derives_from: derives }, ids);
+    }
+    _resolveReferencesTo(feature, id) {
+        const references = this._underConstructionOrphans[id];
+        //   references is of the form
+        //   {
+        //     'Parent' : [ orphans that have a Parent attr referencing this feature ],
+        //     'Derives_from' : [ orphans that have a Derives_from attr referencing this feature ],
+        //    }
+        if (!references) {
+            return;
+        }
+        feature.forEach(loc => {
+            loc.child_features.push(...references.Parent);
+        });
+        feature.forEach(loc => {
+            loc.derived_features.push(...references.Derives_from);
+        });
+        delete this._underConstructionOrphans[id];
+    }
+    _parseError(message) {
+        this.eof = true;
+        this.errorCallback(`${this.lineNumber}: ${message}`);
+    }
+    _resolveReferencesFrom(feature, references, ids) {
+        // this is all a bit more awkward in javascript than it was in perl
+        function postSet(obj, slot1, slot2) {
+            let subObj = obj[slot1];
+            if (!subObj) {
+                subObj = {};
+                obj[slot1] = subObj;
+            }
+            const returnVal = subObj[slot2] || false;
+            subObj[slot2] = true;
+            return returnVal;
+        }
+        references.Parent.forEach(toId => {
+            const otherFeature = this._underConstructionById[toId];
+            if (otherFeature) {
+                const pname = containerAttributes.Parent;
+                if (!ids.filter(id => postSet(this._completedReferences, id, `Parent,${toId}`)).length) {
+                    otherFeature.forEach(location => {
+                        location[pname].push(feature);
+                    });
+                }
+            }
+            else {
+                let ref = this._underConstructionOrphans[toId];
+                if (!ref) {
+                    ref = {
+                        Parent: [],
+                        Derives_from: [],
+                    };
+                    this._underConstructionOrphans[toId] = ref;
+                }
+                ref.Parent.push(feature);
+            }
+        });
+        references.Derives_from.forEach(toId => {
+            const otherFeature = this._underConstructionById[toId];
+            if (otherFeature) {
+                const pname = containerAttributes.Derives_from;
+                if (!ids.filter(id => postSet(this._completedReferences, id, `Derives_from,${toId}`)).length) {
+                    otherFeature.forEach(location => {
+                        location[pname].push(feature);
+                    });
+                }
+            }
+            else {
+                let ref = this._underConstructionOrphans[toId];
+                if (!ref) {
+                    ref = {
+                        Parent: [],
+                        Derives_from: [],
+                    };
+                    this._underConstructionOrphans[toId] = ref;
+                }
+                ref.Derives_from.push(feature);
+            }
+        });
+    }
+}
+//# sourceMappingURL=parse.js.map

package/esm/parse.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"parse.js","sourceRoot":"","sources":["../src/parse.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,IAAI,MAAM,QAAQ,CAAA;AAE9B,MAAM,mBAAmB,GAAG;IAC1B,MAAM,EAAE,gBAAyB;IACjC,YAAY,EAAE,kBAA2B;CAC1C,CAAA;AAED,MAAM,OAAO,WAAW;IAMtB,YAAY,WAAkD;QAC5D,IAAI,CAAC,WAAW,GAAG,WAAW,CAAA;QAC9B,IAAI,CAAC,eAAe,GAAG,SAAS,CAAA;IAClC,CAAC;IAED,OAAO,CAAC,IAAY;QAClB,MAAM,QAAQ,GAAG,mBAAmB,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;QAC/C,IAAI,QAAQ,EAAE,CAAC;YACb,IAAI,CAAC,MAAM,EAAE,CAAA;YACb,IAAI,CAAC,eAAe,GAAG,EAAE,EAAE,EAAE,QAAQ,CAAC,CAAC,CAAC,EAAE,QAAQ,EAAE,EAAE,EAAE,CAAA;YACxD,IAAI,QAAQ,CAAC,CAAC,CAAC,EAAE,CAAC;gBAChB,IAAI,CAAC,eAAe,CAAC,WAAW,GAAG,QAAQ,CAAC,CAAC,CAAC,CAAC,IAAI,EAAE,CAAA;YACvD,CAAC;QACH,CAAC;aAAM,IAAI,IAAI,CAAC,eAAe,IAAI,IAAI,CAAC,IAAI,CAAC,IAAI,CAAC,EAAE,CAAC;YACnD,IAAI,CAAC,eAAe,CAAC,QAAQ,IAAI,IAAI,CAAC,UAAU,CAAC,KAAK,EAAE,EAAE,CAAC,CAAA;QAC7D,CAAC;IACH,CAAC;IAEO,MAAM;QACZ,IAAI,IAAI,CAAC,eAAe,EAAE,CAAC;YACzB,IAAI,CAAC,WAAW,CAAC,IAAI,CAAC,eAAe,CAAC,CAAA;QACxC,CAAC;IACH,CAAC;IAED,MAAM;QACJ,IAAI,CAAC,MAAM,EAAE,CAAA;IACf,CAAC;CACF;AAkBD,MAAM,CAAC,OAAO,OAAO,MAAM;IAkCzB,YAAY,IAAgB;QAzB5B,gBAAW,GAA4B,SAAS,CAAA;QAChD,0CAA0C;QAC1C,4CAA4C;QAC5C,2CAA2C;QAC3C,QAAG,GAAG,KAAK,CAAA;QACX,eAAU,GAAG,CAAC,CAAA;QACd,6DAA6D;QAC7D,wCAAwC;QAChC,+BAA0B,GAAuB,EAAE,CAAA;QAC3D,2BAA2B;QACnB,2BAAsB,GAC5B,EAAE,CAAA;QACI,yBAAoB,GAGxB,EAAE,CAAA;QACN,yDAAyD;QACzD,iBAAiB;QACjB,mBAAmB;QACnB,qEAAqE;QACrE,iFAAiF;QACjF,OAAO;QACP,IAAI;QACI,8BAAyB,GAA2C,EAAE,CAAA;QAG5E,gEAAgE;QAChE,MAAM,QAAQ,GAAG,GAAG,EAAE,GAAE,CAAC,CAAA;QAEzB,IAAI,CAAC,eAAe,GAAG,IAAI,CAAC,eAAe,IAAI,QAAQ,CAAA;QACvD,IAAI,CAAC,WAAW,GAAG,IAAI,CAAC,WAAW,IAAI,QAAQ,CAAA;QAC/C,IAAI,CAAC,eAAe,GAAG,IAAI,CAAC,eAAe,IAAI,QAAQ,CAAA;QACvD,IAAI,CAAC,aAAa,GAAG,IAAI,CAAC,aAAa,IAAI,QAAQ,CAAA;QACnD,IAAI,CAAC,iBAAiB,GAAG,IAAI,CAAC,iBAAiB,IAAI,QAAQ,CAAA;QAC3D,IAAI,CAAC,gBAAgB,GAAG,IAAI,CAAC,gBAAgB,IAAI,QAAQ,CAAA;QACzD,IAAI,CAAC,4BAA4B;YAC/B,IAAI,CAAC,4BAA4B,IAAI,KAAK,CAAA;QAE5C,4BAA4B;QAC5B,IAAI,CAAC,UAAU,GAAG,IAAI,CAAC,UAAU,KAAK,SAAS,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,IAAI,CAAC,UAAU,CAAA;IAC1E,CAAC;IAED,OAAO,CAAC,IAAY;QAClB,2EAA2E;QAC3E,IAAI,IAAI,CAAC,WAAW,EAAE,CAAC;YACrB,IAAI,CAAC,WAAW,CAAC,OAAO,CAAC,IAAI,CAAC,CAAA;YAC9B,OAAM;QACR,CAAC;QACD,IAAI,IAAI,CAAC,GAAG,EAAE,CAAC;YACb,8CAA8C;YAC9C,OAAM;QACR,CAAC;QAED,IAAI,CAAC,UAAU,IAAI,CAAC,CAAA;QAEpB,IAAI,aAAa,CAAC,IAAI,CAAC,IAAI,CAAC,EAAE,CAAC;YAC7B,iCAAiC;YACjC,IAAI,CAAC,WAAW,CAAC,IAAI,CAAC,CAAA;YACtB,OAAM;QACR,CAAC;QAED,MAAM,KAAK,GAAG,cAAc,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;QACvC,IAAI,KAAK,EAAE,CAAC;YACV,uBAAuB;YACvB,MAAM,CAAC,EAAE,SAAS,CAAC,GAAG,KAAK,CAAA;YAC3B,IAAI,CAAC,EAAE,AAAD,EAAG,QAAQ,CAAC,GAAG,KAAK,CAAA;YAE1B,IAAI,SAAS,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;gBAC3B,uDAAuD;gBACvD,IAAI,CAAC,iCAAiC,EAAE,CAAA;YAC1C,CAAC;iBAAM,IAAI,SAAS,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;gBAClC,MAAM,SAAS,GAAG,IAAI,CAAC,cAAc,CAAC,IAAI,CAAC,CAAA;gBAC3C,IAAI,SAAS,EAAE,CAAC;oBACd,IAAI,SAAS,CAAC,SAAS,KAAK,OAAO,EAAE,CAAC;wBACpC,IAAI,CAAC,iCAAiC,EAAE,CAAA;wBACxC,IAAI,CAAC,GAAG,GAAG,IAAI,CAAA;wBACf,IAAI,CAAC,WAAW,GAAG,IAAI,WAAW,CAAC,IAAI,CAAC,gBAAgB,CAAC,CAAA;oBAC3D,CAAC;yBAAM,CAAC;wBACN,IAAI,CAAC,SAAS,CAAC,SAAS,CAAC,CAAA;oBAC3B,CAAC;gBACH,CAAC;YACH,CAAC;iBAAM,CAAC;gBACN,QAAQ,GAAG,QAAQ,CAAC,OAAO,CAAC,KAAK,EAAE,EAAE,CAAC,CAAA;gBACtC,IAAI,CAAC,SAAS,CAAC,EAAE,OAAO,EAAE,QAAQ,EAAE,CAAC,CAAA;YACvC,CAAC;QACH,CAAC;aAAM,IAAI,OAAO,CAAC,IAAI,CAAC,IAAI,CAAC,EAAE,CAAC;YAC9B,yBAAyB;QAC3B,CAAC;aAAM,IAAI,OAAO,CAAC,IAAI,CAAC,IAAI,CAAC,EAAE,CAAC;YAC9B,wCAAwC;YACxC,IAAI,CAAC,iCAAiC,EAAE,CAAA;YACxC,IAAI,CAAC,GAAG,GAAG,IAAI,CAAA;YACf,IAAI,CAAC,WAAW,GAAG,IAAI,WAAW,CAAC,IAAI,CAAC,gBAAgB,CAAC,CAAA;YACzD,IAAI,CAAC,WAAW,CAAC,OAAO,CAAC,IAAI,CAAC,CAAA;QAChC,CAAC;aAAM,CAAC;YACN,qBAAqB;YACrB,MAAM,OAAO,GAAG,IAAI,CAAC,UAAU,CAAC,UAAU,EAAE,EAAE,CAAC,CAAA;YAC/C,MAAM,IAAI,KAAK,CAAC,oCAAoC,OAAO,IAAI,CAAC,CAAA;QAClE,CAAC;IACH,CAAC;IAED,MAAM;QACJ,IAAI,CAAC,iCAAiC,EAAE,CAAA;QACxC,IAAI,IAAI,CAAC,WAAW,EAAE,CAAC;YACrB,IAAI,CAAC,WAAW,CAAC,MAAM,EAAE,CAAA;QAC3B,CAAC;QACD,IAAI,CAAC,WAAW,EAAE,CAAA;IACpB,CAAC;IAEO,SAAS,CACf,CAA2D;QAE3D,IAAI,KAAK,CAAC,OAAO,CAAC,CAAC,CAAC,EAAE,CAAC;YACrB,IAAI,CAAC,eAAe,CAAC,CAAC,CAAC,CAAA;QACzB,CAAC;aAAM,IAAI,WAAW,IAAI,CAAC,EAAE,CAAC;YAC5B,IAAI,CAAC,iBAAiB,CAAC,CAAC,CAAC,CAAA;QAC3B,CAAC;aAAM,IAAI,SAAS,IAAI,CAAC,EAAE,CAAC;YAC1B,IAAI,CAAC,eAAe,CAAC,CAAC,CAAC,CAAA;QACzB,CAAC;IACH,CAAC;IAEO,uBAAuB,CAAC,mBAAmB,GAAG,CAAC;QACrD,MAAM,aAAa,GAAG,CAAC,IAAuB,EAAE,EAAE;;YAChD,IAAI,IAAI,IAAI,KAAK,CAAC,OAAO,CAAC,IAAI,CAAC,KAAI,MAAA,MAAA,IAAI,CAAC,CAAC,CAAC,CAAC,UAAU,0CAAE,EAAE,0CAAG,CAAC,CAAC,CAAA,EAAE,CAAC;gBAC/D,MAAM,GAAG,GAAG,IAAI,CAAC,CAAC,CAAC,CAAC,UAAU,CAAC,EAAE,CAAA;gBACjC,GAAG,CAAC,OAAO,CAAC,EAAE,CAAC,EAAE;oBACf,OAAO,IAAI,CAAC,sBAAsB,CAAC,EAAE,CAAC,CAAA;oBACtC,OAAO,IAAI,CAAC,oBAAoB,CAAC,EAAE,CAAC,CAAA;gBACtC,CAAC,CAAC,CAAA;gBACF,IAAI,CAAC,OAAO,CAAC,CAAC,CAAC,EAAE;oBACf,IAAI,CAAC,CAAC,cAAc,EAAE,CAAC;wBACrB,CAAC,CAAC,cAAc,CAAC,OAAO,CAAC,CAAC,CAAC,EAAE,CAAC,aAAa,CAAC,CAAC,CAAC,CAAC,CAAA;oBACjD,CAAC;oBACD,IAAI,CAAC,CAAC,gBAAgB,EAAE,CAAC;wBACvB,CAAC,CAAC,gBAAgB,CAAC,OAAO,CAAC,CAAC,CAAC,EAAE,CAAC,aAAa,CAAC,CAAC,CAAC,CAAC,CAAA;oBACnD,CAAC;gBACH,CAAC,CAAC,CAAA;YACJ,CAAC;QACH,CAAC,CAAA;QAED,OACE,IAAI,CAAC,0BAA0B,CAAC,MAAM,GAAG,mBAAmB;YAC5D,IAAI,CAAC,UAAU,EACf,CAAC;YACD,MAAM,IAAI,GAAG,IAAI,CAAC,0BAA0B,CAAC,KAAK,EAAE,CAAA;YACpD,IAAI,IAAI,EAAE,CAAC;gBACT,IAAI,CAAC,SAAS,CAAC,IAAI,CAAC,CAAA;gBACpB,aAAa,CAAC,IAAI,CAAC,CAAA;YACrB,CAAC;QACH,CAAC;IACH,CAAC;IAED;;;OAGG;IACK,iCAAiC;QACvC,IAAI,CAAC,0BAA0B,CAAC,OAAO,CAAC,IAAI,CAAC,SAAS,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC,CAAA;QAElE,IAAI,CAAC,0BAA0B,GAAG,EAAE,CAAA;QACpC,IAAI,CAAC,sBAAsB,GAAG,EAAE,CAAA;QAChC,IAAI,CAAC,oBAAoB,GAAG,EAAE,CAAA;QAE9B,yDAAyD;QACzD,kCAAkC;QAClC,IAAI,KAAK,CAAC,IAAI,CAAC,MAAM,CAAC,MAAM,CAAC,IAAI,CAAC,yBAAyB,CAAC,CAAC,CAAC,MAAM,EAAE,CAAC;YACrE,MAAM,IAAI,KAAK,CACb,sGAAsG,MAAM,CAAC,IAAI,CAC/G,IAAI,CAAC,yBAAyB,CAC/B,CAAC,IAAI,CAAC,GAAG,CAAC,EAAE,CACd,CAAA;QACH,CAAC;IACH,CAAC;IAED,sDAAsD;IAC9C,WAAW,CAAC,IAAY;;QAC9B,MAAM,cAAc,GAAG,IAAI,CAAC,YAAY,CAAC,IAAI,CAAC,CAAA;QAC9C,MAAM,WAAW,GAAiC;YAChD,GAAG,cAAc;YACjB,cAAc,EAAE,EAAE;YAClB,gBAAgB,EAAE,EAAE;SACrB,CAAA;QACD,8DAA8D;QAE9D,+DAA+D;QAC/D,MAAM,GAAG,GAAG,CAAA,MAAA,WAAW,CAAC,UAAU,0CAAE,EAAE,KAAI,EAAE,CAAA;QAC5C,MAAM,OAAO,GAAG,CAAA,MAAA,WAAW,CAAC,UAAU,0CAAE,MAAM,KAAI,EAAE,CAAA;QACpD,MAAM,OAAO,GAAG,IAAI,CAAC,4BAA4B;YAC/C,CAAC,CAAC,EAAE;YACJ,CAAC,CAAC,CAAA,MAAA,WAAW,CAAC,UAAU,0CAAE,YAAY,KAAI,EAAE,CAAA;QAE9C,IAAI,CAAC,GAAG,CAAC,MAAM,IAAI,CAAC,OAAO,CAAC,MAAM,IAAI,CAAC,OAAO,CAAC,MAAM,EAAE,CAAC;YACtD,+DAA+D;YAC/D,YAAY;YACZ,IAAI,CAAC,SAAS,CAAC,CAAC,WAAW,CAAC,CAAC,CAAA;YAC7B,OAAM;QACR,CAAC;QAED,IAAI,OAAO,GAAiC,SAAS,CAAA;QACrD,GAAG,CAAC,OAAO,CAAC,EAAE,CAAC,EAAE;YACf,MAAM,QAAQ,GAAG,IAAI,CAAC,sBAAsB,CAAC,EAAE,CAAC,CAAA;YAChD,IAAI,QAAQ,EAAE,CAAC;gBACb,uCAAuC;gBACvC,IAAI,QAAQ,CAAC,QAAQ,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,IAAI,KAAK,WAAW,CAAC,IAAI,EAAE,CAAC;oBAC5D,IAAI,CAAC,WAAW,CACd,uBAAuB,EAAE,8BACvB,WAAW,CAAC,IACd,OAAO,QAAQ,CAAC,QAAQ,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,IAAI,GAAG,CAC7C,CAAA;gBACH,CAAC;gBACD,QAAQ,CAAC,IAAI,CAAC,WAAW,CAAC,CAAA;gBAC1B,OAAO,GAAG,QAAQ,CAAA;YACpB,CAAC;iBAAM,CAAC;gBACN,qDAAqD;gBACrD,uBAAuB;gBACvB,OAAO,GAAG,CAAC,WAAW,CAAC,CAAA;gBAEvB,IAAI,CAAC,uBAAuB,CAAC,CAAC,CAAC,CAAA;gBAC/B,IAAI,CAAC,OAAO,CAAC,MAAM,IAAI,CAAC,OAAO,CAAC,MAAM,EAAE,CAAC;oBACvC,IAAI,CAAC,0BAA0B,CAAC,IAAI,CAAC,OAAO,CAAC,CAAA;gBAC/C,CAAC;gBACD,IAAI,CAAC,sBAAsB,CAAC,EAAE,CAAC,GAAG,OAAO,CAAA;gBAEzC,qDAAqD;gBACrD,IAAI,CAAC,oBAAoB,CAAC,OAAO,EAAE,EAAE,CAAC,CAAA;YACxC,CAAC;QACH,CAAC,CAAC,CAAA;QAEF,oCAAoC;QACpC,IAAI,CAAC,sBAAsB,CACzB,OAAO,IAAI,CAAC,WAAW,CAAC,EACxB,EAAE,MAAM,EAAE,OAAO,EAAE,YAAY,EAAE,OAAO,EAAE,EAC1C,GAAG,CACJ,CAAA;IACH,CAAC;IAEO,oBAAoB,CAAC,OAAyB,EAAE,EAAU;QAChE,MAAM,UAAU,GAAG,IAAI,CAAC,yBAAyB,CAAC,EAAE,CAAC,CAAA;QACrD,8BAA8B;QAC9B,MAAM;QACN,+EAA+E;QAC/E,2FAA2F;QAC3F,OAAO;QACP,IAAI,CAAC,UAAU,EAAE,CAAC;YAChB,OAAM;QACR,CAAC;QACD,OAAO,CAAC,OAAO,CAAC,GAAG,CAAC,EAAE;YACpB,GAAG,CAAC,cAAc,CAAC,IAAI,CAAC,GAAG,UAAU,CAAC,MAAM,CAAC,CAAA;QAC/C,CAAC,CAAC,CAAA;QACF,OAAO,CAAC,OAAO,CAAC,GAAG,CAAC,EAAE;YACpB,GAAG,CAAC,gBAAgB,CAAC,IAAI,CAAC,GAAG,UAAU,CAAC,YAAY,CAAC,CAAA;QACvD,CAAC,CAAC,CAAA;QACF,OAAO,IAAI,CAAC,yBAAyB,CAAC,EAAE,CAAC,CAAA;IAC3C,CAAC;IAEO,WAAW,CAAC,OAAe;QACjC,IAAI,CAAC,GAAG,GAAG,IAAI,CAAA;QACf,IAAI,CAAC,aAAa,CAAC,GAAG,IAAI,CAAC,UAAU,KAAK,OAAO,EAAE,CAAC,CAAA;IACtD,CAAC;IAEO,sBAAsB,CAC5B,OAAyB,EACzB,UAAwD,EACxD,GAAa;QAEb,mEAAmE;QACnE,SAAS,OAAO,CACd,GAAoE,EACpE,KAAa,EACb,KAAa;YAEb,IAAI,MAAM,GAAG,GAAG,CAAC,KAAK,CAAC,CAAA;YACvB,IAAI,CAAC,MAAM,EAAE,CAAC;gBACZ,MAAM,GAAG,EAAE,CAAA;gBACX,GAAG,CAAC,KAAK,CAAC,GAAG,MAAM,CAAA;YACrB,CAAC;YACD,MAAM,SAAS,GAAG,MAAM,CAAC,KAAK,CAAC,IAAI,KAAK,CAAA;YACxC,MAAM,CAAC,KAAK,CAAC,GAAG,IAAI,CAAA;YACpB,OAAO,SAAS,CAAA;QAClB,CAAC;QAED,UAAU,CAAC,MAAM,CAAC,OAAO,CAAC,IAAI,CAAC,EAAE;YAC/B,MAAM,YAAY,GAAG,IAAI,CAAC,sBAAsB,CAAC,IAAI,CAAC,CAAA;YACtD,IAAI,YAAY,EAAE,CAAC;gBACjB,MAAM,KAAK,GAAG,mBAAmB,CAAC,MAAM,CAAA;gBACxC,IACE,CAAC,GAAG,CAAC,MAAM,CAAC,EAAE,CAAC,EAAE,CACf,OAAO,CAAC,IAAI,CAAC,oBAAoB,EAAE,EAAE,EAAE,UAAU,IAAI,EAAE,CAAC,CACzD,CAAC,MAAM,EACR,CAAC;oBACD,YAAY,CAAC,OAAO,CAAC,QAAQ,CAAC,EAAE;wBAC9B,QAAQ,CAAC,KAAK,CAAC,CAAC,IAAI,CAAC,OAAO,CAAC,CAAA;oBAC/B,CAAC,CAAC,CAAA;gBACJ,CAAC;YACH,CAAC;iBAAM,CAAC;gBACN,IAAI,GAAG,GAAG,IAAI,CAAC,yBAAyB,CAAC,IAAI,CAAC,CAAA;gBAC9C,IAAI,CAAC,GAAG,EAAE,CAAC;oBACT,GAAG,GAAG;wBACJ,MAAM,EAAE,EAAE;wBACV,YAAY,EAAE,EAAE;qBACjB,CAAA;oBACD,IAAI,CAAC,yBAAyB,CAAC,IAAI,CAAC,GAAG,GAAG,CAAA;gBAC5C,CAAC;gBACD,GAAG,CAAC,MAAM,CAAC,IAAI,CAAC,OAAO,CAAC,CAAA;YAC1B,CAAC;QACH,CAAC,CAAC,CAAA;QAEF,UAAU,CAAC,YAAY,CAAC,OAAO,CAAC,IAAI,CAAC,EAAE;YACrC,MAAM,YAAY,GAAG,IAAI,CAAC,sBAAsB,CAAC,IAAI,CAAC,CAAA;YACtD,IAAI,YAAY,EAAE,CAAC;gBACjB,MAAM,KAAK,GAAG,mBAAmB,CAAC,YAAY,CAAA;gBAC9C,IACE,CAAC,GAAG,CAAC,MAAM,CAAC,EAAE,CAAC,EAAE,CACf,OAAO,CAAC,IAAI,CAAC,oBAAoB,EAAE,EAAE,EAAE,gBAAgB,IAAI,EAAE,CAAC,CAC/D,CAAC,MAAM,EACR,CAAC;oBACD,YAAY,CAAC,OAAO,CAAC,QAAQ,CAAC,EAAE;wBAC9B,QAAQ,CAAC,KAAK,CAAC,CAAC,IAAI,CAAC,OAAO,CAAC,CAAA;oBAC/B,CAAC,CAAC,CAAA;gBACJ,CAAC;YACH,CAAC;iBAAM,CAAC;gBACN,IAAI,GAAG,GAAG,IAAI,CAAC,yBAAyB,CAAC,IAAI,CAAC,CAAA;gBAC9C,IAAI,CAAC,GAAG,EAAE,CAAC;oBACT,GAAG,GAAG;wBACJ,MAAM,EAAE,EAAE;wBACV,YAAY,EAAE,EAAE;qBACjB,CAAA;oBACD,IAAI,CAAC,yBAAyB,CAAC,IAAI,CAAC,GAAG,GAAG,CAAA;gBAC5C,CAAC;gBACD,GAAG,CAAC,YAAY,CAAC,IAAI,CAAC,OAAO,CAAC,CAAA;YAChC,CAAC;QACH,CAAC,CAAC,CAAA;IACJ,CAAC;CACF"}

package/esm/util.d.ts

@@ -0,0 +1,166 @@
+/**
+ * Unescape a string value used in a GFF3 attribute.
+ *
+ * @param stringVal - Escaped GFF3 string value
+ * @returns An unescaped string value
+ */
+export declare function unescape(stringVal: string): string;
+/**
+ * Escape a value for use in a GFF3 attribute value.
+ *
+ * @param rawVal - Raw GFF3 attribute value
+ * @returns An escaped string value
+ */
+export declare function escape(rawVal: string | number): string;
+/**
+ * Escape a value for use in a GFF3 column value.
+ *
+ * @param rawVal - Raw GFF3 column value
+ * @returns An escaped column value
+ */
+export declare function escapeColumn(rawVal: string | number): string;
+/**
+ * Parse the 9th column (attributes) of a GFF3 feature line.
+ *
+ * @param attrString - String of GFF3 9th column
+ * @returns Parsed attributes
+ */
+export declare function parseAttributes(attrString: string): GFF3Attributes;
+/**
+ * Parse a GFF3 feature line
+ *
+ * @param line - GFF3 feature line
+ * @returns The parsed feature
+ */
+export declare function parseFeature(line: string): GFF3FeatureLine;
+/**
+ * Parse a GFF3 directive line.
+ *
+ * @param line - GFF3 directive line
+ * @returns The parsed directive
+ */
+export declare function parseDirective(line: string): GFF3Directive | GFF3SequenceRegionDirective | GFF3GenomeBuildDirective | null;
+/**
+ * Format an attributes object into a string suitable for the 9th column of GFF3.
+ *
+ * @param attrs - Attributes
+ * @returns GFF3 9th column string
+ */
+export declare function formatAttributes(attrs: GFF3Attributes): string;
+/**
+ * Format a feature object or array of feature objects into one or more lines of
+ * GFF3.
+ *
+ * @param featureOrFeatures - A feature object or array of feature objects
+ * @returns A string of one or more GFF3 lines
+ */
+export declare function formatFeature(featureOrFeatures: GFF3FeatureLine | GFF3FeatureLineWithRefs | (GFF3FeatureLine | GFF3FeatureLineWithRefs)[]): string;
+/**
+ * Format a directive into a line of GFF3.
+ *
+ * @param directive - A directive object
+ * @returns A directive line string
+ */
+export declare function formatDirective(directive: GFF3Directive): string;
+/**
+ * Format a comment into a GFF3 comment.
+ * Yes I know this is just adding a # and a newline.
+ *
+ * @param comment - A comment object
+ * @returns A comment line string
+ */
+export declare function formatComment(comment: GFF3Comment): string;
+/**
+ * Format a sequence object as FASTA
+ *
+ * @param seq - A sequence object
+ * @returns Formatted single FASTA sequence string
+ */
+export declare function formatSequence(seq: GFF3Sequence): string;
+/**
+ * Format a directive, comment, sequence, or feature, or array of such items,
+ * into one or more lines of GFF3.
+ *
+ * @param itemOrItems - A comment, sequence, or feature, or array of such items
+ * @returns A formatted string or array of strings
+ */
+export declare function formatItem(itemOrItems: GFF3FeatureLineWithRefs | GFF3Directive | GFF3Comment | GFF3Sequence | (GFF3FeatureLineWithRefs | GFF3Directive | GFF3Comment | GFF3Sequence)[]): string | string[];
+/** A record of GFF3 attribute identifiers and the values of those identifiers */
+export type GFF3Attributes = Record<string, string[] | undefined>;
+/** A representation of a single line of a GFF3 file */
+export interface GFF3FeatureLine {
+    /** The ID of the landmark used to establish the coordinate system for the current feature */
+    seq_id: string | null;
+    /** A free text qualifier intended to describe the algorithm or operating procedure that generated this feature */
+    source: string | null;
+    /** The type of the feature */
+    type: string | null;
+    /** The start coordinates of the feature */
+    start: number | null;
+    /** The end coordinates of the feature */
+    end: number | null;
+    /** The score of the feature */
+    score: number | null;
+    /** The strand of the feature */
+    strand: string | null;
+    /** For features of type "CDS", the phase indicates where the next codon begins relative to the 5' end of the current CDS feature */
+    phase: string | null;
+    /** Feature attributes */
+    attributes: GFF3Attributes | null;
+}
+/**
+ * A GFF3 Feature line that includes references to other features defined in
+ * their "Parent" or "Derives_from" attributes
+ */
+export interface GFF3FeatureLineWithRefs extends GFF3FeatureLine {
+    /** An array of child features */
+    child_features: GFF3Feature[];
+    /** An array of features derived from this feature */
+    derived_features: GFF3Feature[];
+}
+/**
+ * A GFF3 feature, which may include multiple individual feature lines
+ */
+export type GFF3Feature = GFF3FeatureLineWithRefs[];
+/** A GFF3 directive */
+export interface GFF3Directive {
+    /** The name of the directive */
+    directive: string;
+    /** The string value of the directive */
+    value?: string;
+}
+/** A GFF3 sequence-region directive */
+export interface GFF3SequenceRegionDirective extends GFF3Directive {
+    /** The string value of the directive */
+    value: string;
+    /** The sequence ID parsed from the directive */
+    seq_id: string;
+    /** The sequence start parsed from the directive */
+    start: string;
+    /** The sequence end parsed from the directive */
+    end: string;
+}
+/** A GFF3 genome-build directive */
+export interface GFF3GenomeBuildDirective extends GFF3Directive {
+    /** The string value of the directive */
+    value: string;
+    /** The genome build source parsed from the directive */
+    source: string;
+    /** The genome build name parsed from the directive */
+    buildName: string;
+}
+/** A GFF3 comment */
+export interface GFF3Comment {
+    /** The text of the comment */
+    comment: string;
+}
+/** A GFF3 FASTA single sequence */
+export interface GFF3Sequence {
+    /** The ID of the sequence */
+    id: string;
+    /** The description of the sequence */
+    description?: string;
+    /** The sequence */
+    sequence: string;
+}
+export type GFF3Item = GFF3Feature | GFF3Directive | GFF3Comment | GFF3Sequence;