@teselagen/bio-parsers 0.3.6 → 0.3.8

package/index.mjs

@@ -39,6 +39,7 @@ var __async = (__this, __arguments, generator) => {
     step((generator = generator.apply(__this, __arguments)).next());
   });
 };
+var _a, _b;
 const untitledSequenceName = "Untitled Sequence";
 const gbDivisions = {
   // https://www.ncbi.nlm.nih.gov/Sitemap/samplerecord.html#GenBankDivisionB
@@ -94,6 +95,19 @@ function createInitialSequence(options) {
   };
 }
 __name(createInitialSequence, "createInitialSequence");
+function extractFileExtension(name) {
+  if (typeof name === "string") {
+    let ext = "";
+    const match = name.match(/\.(\w+)$/);
+    if (match && match[1]) {
+      ext = match[1];
+    }
+    return ext;
+  } else {
+    return "";
+  }
+}
+__name(extractFileExtension, "extractFileExtension");
 function splitStringIntoLines(string) {
   let lines = [];
   if (string === "") {
@@ -119,10 +133,7 @@ function getAugmentedNamespace(n) {
   if (typeof f == "function") {
     var a = /* @__PURE__ */ __name(function a2() {
       if (this instanceof a2) {
-        var args = [null];
-        args.push.apply(args, arguments);
-        var Ctor = Function.bind.apply(f, args);
-        return new Ctor();
+        return Reflect.construct(f, arguments, this.constructor);
       }
       return f.apply(this, arguments);
     }, "a");
@@ -19370,7 +19381,7 @@ function validateSequence(sequence, options = {}) {
         (u) => u === "U" ? "T" : "t"
       );
     }
-    if (temp !== sequence.sequence && !sequence.isDNA && !sequence.isProtein) {
+    if (temp !== sequence.sequence && !sequence.isDNA && !sequence.isProtein && sequence.isRNA !== false) {
       sequence.type = "RNA";
       sequence.sequence = temp;
     } else {
@@ -19557,7 +19568,11 @@ function validateSequenceArray(parsingResultArray, options) {
   return parsingResultArray;
 }
 __name(validateSequenceArray, "validateSequenceArray");
-function fastaToJson(fileString, options) {
+function fastaToJson(fileString, options = {}) {
+  const ext = extractFileExtension(options.fileName);
+  if (/^(faa)$/.test(ext)) {
+    options.isProtein = true;
+  }
   let resultArray = [];
   let result = null;
   try {
@@ -20219,7 +20234,7 @@ function getLengthOfWhiteSpaceBeforeStartOfLetters(string) {
   }
 }
 __name(getLengthOfWhiteSpaceBeforeStartOfLetters, "getLengthOfWhiteSpaceBeforeStartOfLetters");
-class Match {
+const _Match = class _Match {
   constructor(props) {
     Object.assign(this, props);
     this.value = this.obj[this.prop];
@@ -20236,8 +20251,9 @@ class Match {
   log() {
     console.info(this.toString());
   }
-}
-__name(Match, "Match");
+};
+__name(_Match, "Match");
+let Match = _Match;
 const GLOBAL = typeof window == "object" ? window : global;
 function searchWholeObjByName(what, where) {
   const searchBy = /* @__PURE__ */ __name((what2, where2, prop) => what2 == prop, "searchBy");
@@ -20283,13 +20299,13 @@ function searchWholeObjByName(what, where) {
 }
 __name(searchWholeObjByName, "searchWholeObjByName");
 function searchWholeObjByNameSimple(what, where) {
-  var _a2, _b2, _c;
-  return (_c = (_b2 = (_a2 = searchWholeObjByName(what, where)) == null ? void 0 : _a2[0]) == null ? void 0 : _b2.value) == null ? void 0 : _c[0];
+  var _a3, _b3, _c;
+  return (_c = (_b3 = (_a3 = searchWholeObjByName(what, where)) == null ? void 0 : _a3[0]) == null ? void 0 : _b3.value) == null ? void 0 : _c[0];
 }
 __name(searchWholeObjByNameSimple, "searchWholeObjByNameSimple");
 function searchWholeObjByNameSimpleArray(what, where) {
-  var _a2, _b2;
-  return (_b2 = (_a2 = searchWholeObjByName(what, where)) == null ? void 0 : _a2[0]) == null ? void 0 : _b2.value;
+  var _a3, _b3;
+  return (_b3 = (_a3 = searchWholeObjByName(what, where)) == null ? void 0 : _a3[0]) == null ? void 0 : _b3.value;
 }
 __name(searchWholeObjByNameSimpleArray, "searchWholeObjByNameSimpleArray");
 var validator$2 = {};
@@ -20717,7 +20733,7 @@ const buildOptions$1 = /* @__PURE__ */ __name(function(options) {
 }, "buildOptions$1");
 OptionsBuilder.buildOptions = buildOptions$1;
 OptionsBuilder.defaultOptions = defaultOptions$1;
-class XmlNode {
+const _XmlNode = class _XmlNode {
   constructor(tagname) {
     this.tagname = tagname;
     this.child = [];
@@ -20737,8 +20753,9 @@ class XmlNode {
       this.child.push({ [node.tagname]: node.child });
     }
   }
-}
-__name(XmlNode, "XmlNode");
+};
+__name(_XmlNode, "XmlNode");
+let XmlNode = _XmlNode;
 var xmlNode$1 = XmlNode;
 const util$2 = util$4;
 function readDocType$1(xmlData, i) {
@@ -20949,7 +20966,7 @@ const xmlNode = xmlNode$1;
 const readDocType = DocTypeReader;
 const toNumber = strnum;
 "<((!\\[CDATA\\[([\\s\\S]*?)(]]>))|((NAME:)?(NAME))([^>]*)>|((\\/)(NAME)\\s*>))([^<]*)".replace(/NAME/g, util$1.nameRegexp);
-let OrderedObjParser$1 = /* @__PURE__ */ __name(class OrderedObjParser {
+let OrderedObjParser$1 = (_a = class {
   constructor(options) {
     this.options = options;
     this.currentNode = null;
@@ -20988,7 +21005,7 @@ let OrderedObjParser$1 = /* @__PURE__ */ __name(class OrderedObjParser {
     this.saveTextToParentTag = saveTextToParentTag;
     this.addChild = addChild;
   }
-}, "OrderedObjParser");
+}, __name(_a, "OrderedObjParser"), _a);
 function addExternalEntities(externalEntities) {
   const entKeys = Object.keys(externalEntities);
   for (let i = 0; i < entKeys.length; i++) {
@@ -21220,6 +21237,7 @@ const parseXml = /* @__PURE__ */ __name(function(xmlData) {
           if (tagExp.length > 0 && tagExp.lastIndexOf("/") === tagExp.length - 1) {
             if (tagName[tagName.length - 1] === "/") {
               tagName = tagName.substr(0, tagName.length - 1);
+              jPath = jPath.substr(0, jPath.length - 1);
               tagExp = tagName;
             } else {
               tagExp = tagExp.substr(0, tagExp.length - 1);
@@ -21537,10 +21555,10 @@ function isLeafTag(obj, options) {
 __name(isLeafTag, "isLeafTag");
 node2json.prettify = prettify$1;
 const { buildOptions } = OptionsBuilder;
-const OrderedObjParser2 = OrderedObjParser_1;
+const OrderedObjParser = OrderedObjParser_1;
 const { prettify } = node2json;
 const validator$1 = validator$2;
-let XMLParser$1 = /* @__PURE__ */ __name(class XMLParser {
+let XMLParser$1 = (_b = class {
   constructor(options) {
     this.externalEntities = {};
     this.options = buildOptions(options);
@@ -21566,7 +21584,7 @@ let XMLParser$1 = /* @__PURE__ */ __name(class XMLParser {
         throw Error(`${result.err.msg}:${result.err.line}:${result.err.col}`);
       }
     }
-    const orderedObjParser = new OrderedObjParser2(this.options);
+    const orderedObjParser = new OrderedObjParser(this.options);
     orderedObjParser.addExternalEntities(this.externalEntities);
     const orderedResult = orderedObjParser.parseXml(xmlData);
     if (this.options.preserveOrder || orderedResult === void 0)
@@ -21590,7 +21608,7 @@ let XMLParser$1 = /* @__PURE__ */ __name(class XMLParser {
       this.externalEntities[key] = value;
     }
   }
-}, "XMLParser");
+}, __name(_b, "XMLParser"), _b);
 var XMLParser_1 = XMLParser$1;
 const EOL = "\n";
 function toXml(jArray, options) {
@@ -21795,13 +21813,18 @@ Builder.prototype.j2x = function(jObj, level) {
   let attrStr = "";
   let val2 = "";
   for (let key in jObj) {
-    if (typeof jObj[key] === "undefined")
-      ;
-    else if (jObj[key] === null) {
-      if (key[0] === "?")
+    if (typeof jObj[key] === "undefined") {
+      if (this.isAttribute(key)) {
+        val2 += "";
+      }
+    } else if (jObj[key] === null) {
+      if (this.isAttribute(key)) {
+        val2 += "";
+      } else if (key[0] === "?") {
         val2 += this.indentate(level) + "<" + key + "?" + this.tagEndChar;
-      else
+      } else {
         val2 += this.indentate(level) + "<" + key + "/" + this.tagEndChar;
+      }
     } else if (jObj[key] instanceof Date) {
       val2 += this.buildTextValNode(jObj[key], key, "", level);
     } else if (typeof jObj[key] !== "object") {
@@ -21887,7 +21910,7 @@ Builder.prototype.buildObjectNode = function(val2, key, attrStr, level) {
       piClosingChar = "?";
       tagEndExp = "";
     }
-    if (attrStr && val2.indexOf("<") === -1) {
+    if ((attrStr || attrStr === "") && val2.indexOf("<") === -1) {
       return this.indentate(level) + "<" + key + attrStr + piClosingChar + ">" + val2 + tagEndExp;
     } else if (this.options.commentPropName !== false && key === this.options.commentPropName && piClosingChar.length === 0) {
       return this.indentate(level) + `<!--${val2}-->` + this.newLine;
@@ -21939,7 +21962,7 @@ function indentate(level) {
 }
 __name(indentate, "indentate");
 function isAttribute(name) {
-  if (name.startsWith(this.options.attributeNamePrefix)) {
+  if (name.startsWith(this.options.attributeNamePrefix) && name !== this.options.textNodeName) {
     return name.substr(this.attrPrefixLen);
   } else {
     return false;
@@ -21948,10 +21971,10 @@ function isAttribute(name) {
 __name(isAttribute, "isAttribute");
 var json2xml = Builder;
 const validator = validator$2;
-const XMLParser2 = XMLParser_1;
+const XMLParser = XMLParser_1;
 const XMLBuilder = json2xml;
 var fxp = {
-  XMLParser: XMLParser2,
+  XMLParser,
   XMLValidator: validator,
   XMLBuilder
 };
@@ -22067,19 +22090,6 @@ function parseSbolJson(sbolJson, options) {
   };
 }
 __name(parseSbolJson, "parseSbolJson");
-function extractFileExtension(name) {
-  if (typeof name === "string") {
-    let ext = "";
-    const match = name.match(/\.(\w+)$/);
-    if (match && match[1]) {
-      ext = match[1];
-    }
-    return ext;
-  } else {
-    return "";
-  }
-}
-__name(extractFileExtension, "extractFileExtension");
 /*!
  *  Copyright 2008 Fair Oaks Labs, Inc.
  *  All rights reserved.
@@ -25882,7 +25892,8 @@ var buffer$1 = {};
   }, "fill");
   const errors = {};
   function E(sym, getMessage, Base) {
-    errors[sym] = /* @__PURE__ */ __name(class NodeError extends Base {
+    var _a3;
+    errors[sym] = (_a3 = class extends Base {
       constructor() {
         super();
         Object.defineProperty(this, "message", {
@@ -25908,7 +25919,7 @@ var buffer$1 = {};
       toString() {
         return `${this.name} [${sym}]: ${this.message}`;
       }
-    }, "NodeError");
+    }, __name(_a3, "NodeError"), _a3);
   }
   __name(E, "E");
   E(
@@ -29391,7 +29402,7 @@ const containerAttributes = {
   Parent: "child_features",
   Derives_from: "derived_features"
 };
-class FASTAParser {
+const _FASTAParser = class _FASTAParser {
   constructor(seqCallback) {
     this.seqCallback = seqCallback;
     this.currentSequence = void 0;
@@ -29414,9 +29425,10 @@ class FASTAParser {
   finish() {
     this._flush();
   }
-}
-__name(FASTAParser, "FASTAParser");
-class Parser {
+};
+__name(_FASTAParser, "FASTAParser");
+let FASTAParser = _FASTAParser;
+const _Parser = class _Parser {
   constructor(args) {
     this.fastaParser = void 0;
     this.eof = false;
@@ -29535,14 +29547,14 @@ class Parser {
   }
   // do the right thing with a newly-parsed feature line
   _bufferLine(line) {
-    var _a2, _b2, _c;
+    var _a3, _b3, _c;
     const rawFeatureLine = parseFeature(line);
     const featureLine = __spreadProps(__spreadValues({}, rawFeatureLine), {
       child_features: [],
       derived_features: []
     });
-    const ids = ((_a2 = featureLine.attributes) === null || _a2 === void 0 ? void 0 : _a2.ID) || [];
-    const parents = ((_b2 = featureLine.attributes) === null || _b2 === void 0 ? void 0 : _b2.Parent) || [];
+    const ids = ((_a3 = featureLine.attributes) === null || _a3 === void 0 ? void 0 : _a3.ID) || [];
+    const parents = ((_b3 = featureLine.attributes) === null || _b3 === void 0 ? void 0 : _b3.Parent) || [];
     const derives = this.disableDerivesFromReferences ? [] : ((_c = featureLine.attributes) === null || _c === void 0 ? void 0 : _c.Derives_from) || [];
     if (!ids.length && !parents.length && !derives.length) {
       this._emitItem([featureLine]);
@@ -29640,8 +29652,9 @@ class Parser {
       }
     });
   }
-}
-__name(Parser, "Parser");
+};
+__name(_Parser, "Parser");
+let Parser = _Parser;
 function _callback(callback) {
   if (process && process.nextTick)
     process.nextTick(callback);
@@ -29668,7 +29681,7 @@ function _processParseOptions(options) {
   return out;
 }
 __name(_processParseOptions, "_processParseOptions");
-class GFFTransform extends streamBrowserify.Transform {
+const _GFFTransform = class _GFFTransform extends streamBrowserify.Transform {
   constructor(inputOptions = {}) {
     super({ objectMode: true });
     this.textBuffer = "";
@@ -29708,8 +29721,9 @@ class GFFTransform extends streamBrowserify.Transform {
     this.parser.finish();
     _callback(callback);
   }
-}
-__name(GFFTransform, "GFFTransform");
+};
+__name(_GFFTransform, "GFFTransform");
+let GFFTransform = _GFFTransform;
 function parseStream(options = {}) {
   return new GFFTransform(options);
 }
@@ -29753,7 +29767,7 @@ function formatSync(items) {
   return str;
 }
 __name(formatSync, "formatSync");
-class FormattingTransform extends streamBrowserify.Transform {
+const _FormattingTransform = class _FormattingTransform extends streamBrowserify.Transform {
   constructor(options = {}) {
     super(Object.assign(options, { objectMode: true }));
     this.linesSinceLastSyncMark = 0;
@@ -29795,8 +29809,9 @@ class FormattingTransform extends streamBrowserify.Transform {
     this.haveWeEmittedData = true;
     _callback(callback);
   }
-}
-__name(FormattingTransform, "FormattingTransform");
+};
+__name(_FormattingTransform, "FormattingTransform");
+let FormattingTransform = _FormattingTransform;
 function formatStream(options = {}) {
   return new FormattingTransform(options);
 }
@@ -29900,8 +29915,8 @@ function geneiousXmlToJson(string, options) {
       const toRet = lodashExports.filter(
         resultArray,
         (r) => {
-          var _a2, _b2;
-          return (_b2 = (_a2 = r == null ? void 0 : r.parsedSequence) == null ? void 0 : _a2.sequence) == null ? void 0 : _b2.length;
+          var _a3, _b3;
+          return (_b3 = (_a3 = r == null ? void 0 : r.parsedSequence) == null ? void 0 : _a3.sequence) == null ? void 0 : _b3.length;
         }
       );
       if (toRet.length)
@@ -30011,8 +30026,8 @@ function jbeiXmlToJson(string, options) {
       const toRet = lodashExports.filter(
         resultArray,
         (r) => {
-          var _a2, _b2;
-          return (_b2 = (_a2 = r == null ? void 0 : r.parsedSequence) == null ? void 0 : _a2.sequence) == null ? void 0 : _b2.length;
+          var _a3, _b3;
+          return (_b3 = (_a3 = r == null ? void 0 : r.parsedSequence) == null ? void 0 : _a3.sequence) == null ? void 0 : _b3.length;
         }
       );
       if (toRet.length)
@@ -30159,9 +30174,9 @@ var freb = /* @__PURE__ */ __name(function(eb, start) {
   }
   return { b, r };
 }, "freb");
-var _a = freb(fleb, 2), fl = _a.b, revfl = _a.r;
+var _a2 = freb(fleb, 2), fl = _a2.b, revfl = _a2.r;
 fl[28] = 258, revfl[258] = 28;
-var _b = freb(fdeb, 0), fd = _b.b;
+var _b2 = freb(fdeb, 0), fd = _b2.b;
 var rev = new u16(32768);
 for (var i = 0; i < 32768; ++i) {
   var x = (i & 43690) >> 1 | (i & 21845) << 1;
@@ -30464,7 +30479,7 @@ function strFromU8(dat, latin1) {
   } else if (td) {
     return td.decode(dat);
   } else {
-    var _a2 = dutf8(dat), s = _a2.s, r = _a2.r;
+    var _a3 = dutf8(dat), s = _a3.s, r = _a3.r;
     if (r.length)
       err(8);
     return s;
@@ -30476,7 +30491,7 @@ var slzh = /* @__PURE__ */ __name(function(d, b) {
 }, "slzh");
 var zh = /* @__PURE__ */ __name(function(d, b, z) {
   var fnl = b2(d, b + 28), fn = strFromU8(d.subarray(b + 46, b + 46 + fnl), !(b2(d, b + 8) & 2048)), es = b + 46 + fnl, bs = b4(d, b + 20);
-  var _a2 = z && bs == 4294967295 ? z64e(d, es) : [bs, b4(d, b + 24), b4(d, b + 42)], sc = _a2[0], su = _a2[1], off = _a2[2];
+  var _a3 = z && bs == 4294967295 ? z64e(d, es) : [bs, b4(d, b + 24), b4(d, b + 42)], sc = _a3[0], su = _a3[1], off = _a3[2];
   return [b2(d, b + 10), sc, su, fn, es + b2(d, b + 30) + b2(d, b + 32), off];
 }, "zh");
 var z64e = /* @__PURE__ */ __name(function(d, b) {
@@ -30506,7 +30521,7 @@ function unzipSync(data, opts) {
   }
   var fltr = opts && opts.filter;
   for (var i = 0; i < c; ++i) {
-    var _a2 = zh(data, o, z), c_2 = _a2[0], sc = _a2[1], su = _a2[2], fn = _a2[3], no = _a2[4], off = _a2[5], b = slzh(data, off);
+    var _a3 = zh(data, o, z), c_2 = _a3[0], sc = _a3[1], su = _a3[2], fn = _a3[3], no = _a3[4], off = _a3[5], b = slzh(data, off);
     o = no;
     if (!fltr || fltr({
       name: fn,
@@ -30565,7 +30580,7 @@ function anyToJson(fileContentStringOrFileObj, options) {
         );
       }
     }
-    if (/^(fasta|fas|fa|fna|ffn)$/.test(ext)) {
+    if (/^(fasta|fas|fa|fna|ffn|faa)$/.test(ext)) {
       return fastaToJson(fileContentString, options);
     } else if (/^(gb|gbk)$/.test(ext)) {
       return genbankToJson(fileContentString, options);
@@ -32394,7 +32409,7 @@ function jsonToGenbank(_serSeq, options) {
 }
 __name(jsonToGenbank, "jsonToGenbank");
 function createGenbankLocus(serSeq, options) {
-  var _a2, _b2;
+  var _a3, _b3;
   if (serSeq.sequence.symbols) {
     serSeq.sequence = serSeq.sequence.symbols.split("");
   }
@@ -32403,9 +32418,9 @@ function createGenbankLocus(serSeq, options) {
   if (serSeq.isProtein) {
     dnaType = "";
   } else if (serSeq.type === "RNA") {
-    dnaType = (serSeq == null ? void 0 : serSeq.doubleStranded) ? "RNA" : (_a2 = serSeq == null ? void 0 : serSeq.sequenceTypeFromLocus) != null ? _a2 : "ss-RNA";
+    dnaType = (serSeq == null ? void 0 : serSeq.doubleStranded) ? "RNA" : (_a3 = serSeq == null ? void 0 : serSeq.sequenceTypeFromLocus) != null ? _a3 : "ss-RNA";
   } else {
-    dnaType = (serSeq == null ? void 0 : serSeq.doubleStranded) ? "DNA" : (_b2 = serSeq == null ? void 0 : serSeq.sequenceTypeFromLocus) != null ? _b2 : "DNA";
+    dnaType = (serSeq == null ? void 0 : serSeq.doubleStranded) ? "DNA" : (_b3 = serSeq == null ? void 0 : serSeq.sequenceTypeFromLocus) != null ? _b3 : "DNA";
   }
   const date = getCurrentDateString();
   let line = StringUtil.rpad("LOCUS", " ", 12);