@teselagen/bio-parsers 0.3.6 → 0.3.8

package/index.umd.js

@@ -1,7 +1,4 @@
-(function(global2, factory) {
-  typeof exports === "object" && typeof module !== "undefined" ? factory(exports) : typeof define === "function" && define.amd ? define(["exports"], factory) : (global2 = typeof globalThis !== "undefined" ? globalThis : global2 || self, factory(global2.bioParsers = {}));
-})(this, function(exports2) {
-  "use strict";var __defProp = Object.defineProperty;
+var __defProp = Object.defineProperty;
 var __defProps = Object.defineProperties;
 var __getOwnPropDescs = Object.getOwnPropertyDescriptors;
 var __getOwnPropSymbols = Object.getOwnPropertySymbols;
@@ -42,7 +39,11 @@ var __async = (__this, __arguments, generator) => {
     step((generator = generator.apply(__this, __arguments)).next());
   });
 };
-
+(function(global2, factory) {
+  typeof exports === "object" && typeof module !== "undefined" ? factory(exports) : typeof define === "function" && define.amd ? define(["exports"], factory) : (global2 = typeof globalThis !== "undefined" ? globalThis : global2 || self, factory(global2.bioParsers = {}));
+})(this, function(exports2) {
+  var _a2, _b2;
+  "use strict";
   const untitledSequenceName = "Untitled Sequence";
   const gbDivisions = {
     // https://www.ncbi.nlm.nih.gov/Sitemap/samplerecord.html#GenBankDivisionB
@@ -98,6 +99,19 @@ var __async = (__this, __arguments, generator) => {
     };
   }
   __name(createInitialSequence, "createInitialSequence");
+  function extractFileExtension(name2) {
+    if (typeof name2 === "string") {
+      let ext = "";
+      const match = name2.match(/\.(\w+)$/);
+      if (match && match[1]) {
+        ext = match[1];
+      }
+      return ext;
+    } else {
+      return "";
+    }
+  }
+  __name(extractFileExtension, "extractFileExtension");
   function splitStringIntoLines(string) {
     let lines = [];
     if (string === "") {
@@ -123,10 +137,7 @@ var __async = (__this, __arguments, generator) => {
     if (typeof f == "function") {
       var a = /* @__PURE__ */ __name(function a2() {
         if (this instanceof a2) {
-          var args = [null];
-          args.push.apply(args, arguments);
-          var Ctor = Function.bind.apply(f, args);
-          return new Ctor();
+          return Reflect.construct(f, arguments, this.constructor);
         }
         return f.apply(this, arguments);
       }, "a");
@@ -19374,7 +19385,7 @@ var __async = (__this, __arguments, generator) => {
           (u) => u === "U" ? "T" : "t"
         );
       }
-      if (temp !== sequence.sequence && !sequence.isDNA && !sequence.isProtein) {
+      if (temp !== sequence.sequence && !sequence.isDNA && !sequence.isProtein && sequence.isRNA !== false) {
         sequence.type = "RNA";
         sequence.sequence = temp;
       } else {
@@ -19561,7 +19572,11 @@ var __async = (__this, __arguments, generator) => {
     return parsingResultArray;
   }
   __name(validateSequenceArray, "validateSequenceArray");
-  function fastaToJson(fileString, options) {
+  function fastaToJson(fileString, options = {}) {
+    const ext = extractFileExtension(options.fileName);
+    if (/^(faa)$/.test(ext)) {
+      options.isProtein = true;
+    }
     let resultArray = [];
     let result = null;
     try {
@@ -20223,7 +20238,7 @@ var __async = (__this, __arguments, generator) => {
     }
   }
   __name(getLengthOfWhiteSpaceBeforeStartOfLetters, "getLengthOfWhiteSpaceBeforeStartOfLetters");
-  class Match {
+  const _Match = class _Match {
     constructor(props) {
       Object.assign(this, props);
       this.value = this.obj[this.prop];
@@ -20240,8 +20255,9 @@ var __async = (__this, __arguments, generator) => {
     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");
@@ -20287,13 +20303,13 @@ var __async = (__this, __arguments, generator) => {
   }
   __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 = {};
@@ -20721,7 +20737,7 @@ var __async = (__this, __arguments, generator) => {
   }, "buildOptions$1");
   OptionsBuilder.buildOptions = buildOptions$1;
   OptionsBuilder.defaultOptions = defaultOptions$1;
-  class XmlNode {
+  const _XmlNode = class _XmlNode {
     constructor(tagname) {
       this.tagname = tagname;
       this.child = [];
@@ -20741,8 +20757,9 @@ var __async = (__this, __arguments, generator) => {
         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, i2) {
@@ -20953,7 +20970,7 @@ var __async = (__this, __arguments, generator) => {
   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 = (_a2 = class {
     constructor(options) {
       this.options = options;
       this.currentNode = null;
@@ -20992,7 +21009,7 @@ var __async = (__this, __arguments, generator) => {
       this.saveTextToParentTag = saveTextToParentTag;
       this.addChild = addChild;
     }
-  }, "OrderedObjParser");
+  }, __name(_a2, "OrderedObjParser"), _a2);
   function addExternalEntities(externalEntities) {
     const entKeys = Object.keys(externalEntities);
     for (let i2 = 0; i2 < entKeys.length; i2++) {
@@ -21224,6 +21241,7 @@ var __async = (__this, __arguments, generator) => {
             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);
@@ -21544,7 +21562,7 @@ var __async = (__this, __arguments, generator) => {
   const OrderedObjParser = OrderedObjParser_1;
   const { prettify } = node2json;
   const validator$1 = validator$2;
-  let XMLParser$1 = /* @__PURE__ */ __name(class XMLParser {
+  let XMLParser$1 = (_b2 = class {
     constructor(options) {
       this.externalEntities = {};
       this.options = buildOptions(options);
@@ -21594,7 +21612,7 @@ var __async = (__this, __arguments, generator) => {
         this.externalEntities[key] = value;
       }
     }
-  }, "XMLParser");
+  }, __name(_b2, "XMLParser"), _b2);
   var XMLParser_1 = XMLParser$1;
   const EOL = "\n";
   function toXml(jArray, options) {
@@ -21799,13 +21817,18 @@ var __async = (__this, __arguments, generator) => {
     let attrStr = "";
     let val2 = "";
     for (let key in jObj) {
-      if (typeof jObj[key] === "undefined")
-        ;
-      else if (jObj[key] === null) {
-        if (key[0] === "?")
+      if (typeof jObj[key] === "undefined") {
+        if (this.isAttribute(key)) {
+          val2 += "";
+        }
+      } else if (jObj[key] === null) {
+        if (this.isAttribute(key)) {
+          val2 += "";
+        } else if (key[0] === "?") {
           val2 += this.indentate(level) + "<" + key + "?" + this.tagEndChar;
-        else
+        } else {
           val2 += this.indentate(level) + "<" + key + "/" + this.tagEndChar;
+        }
       } else if (jObj[key] instanceof Date) {
         val2 += this.buildTextValNode(jObj[key], key, "", level);
       } else if (typeof jObj[key] !== "object") {
@@ -21891,7 +21914,7 @@ var __async = (__this, __arguments, generator) => {
         piClosingChar = "?";
         tagEndExp = "";
       }
-      if (attrStr && val2.indexOf("<") === -1) {
+      if ((attrStr || attrStr === "") && val2.indexOf("<") === -1) {
         return this.indentate(level) + "<" + key + attrStr + piClosingChar + ">" + val2 + tagEndExp;
       } else if (this.options.commentPropName !== false && key === this.options.commentPropName && piClosingChar.length === 0) {
         return this.indentate(level) + `<!--${val2}-->` + this.newLine;
@@ -21943,7 +21966,7 @@ var __async = (__this, __arguments, generator) => {
   }
   __name(indentate, "indentate");
   function isAttribute(name2) {
-    if (name2.startsWith(this.options.attributeNamePrefix)) {
+    if (name2.startsWith(this.options.attributeNamePrefix) && name2 !== this.options.textNodeName) {
       return name2.substr(this.attrPrefixLen);
     } else {
       return false;
@@ -22071,19 +22094,6 @@ var __async = (__this, __arguments, generator) => {
     };
   }
   __name(parseSbolJson, "parseSbolJson");
-  function extractFileExtension(name2) {
-    if (typeof name2 === "string") {
-      let ext = "";
-      const match = name2.match(/\.(\w+)$/);
-      if (match && match[1]) {
-        ext = match[1];
-      }
-      return ext;
-    } else {
-      return "";
-    }
-  }
-  __name(extractFileExtension, "extractFileExtension");
   /*!
    *  Copyright 2008 Fair Oaks Labs, Inc.
    *  All rights reserved.
@@ -25886,7 +25896,8 @@ var __async = (__this, __arguments, generator) => {
     }, "fill");
     const errors = {};
     function E(sym, getMessage, Base) {
-      errors[sym] = /* @__PURE__ */ __name(class NodeError extends Base {
+      var _a3;
+      errors[sym] = (_a3 = class extends Base {
         constructor() {
           super();
           Object.defineProperty(this, "message", {
@@ -25912,7 +25923,7 @@ var __async = (__this, __arguments, generator) => {
         toString() {
           return `${this.name} [${sym}]: ${this.message}`;
         }
-      }, "NodeError");
+      }, __name(_a3, "NodeError"), _a3);
     }
     __name(E, "E");
     E(
@@ -29395,7 +29406,7 @@ ${seq.sequence}
     Parent: "child_features",
     Derives_from: "derived_features"
   };
-  class FASTAParser {
+  const _FASTAParser = class _FASTAParser {
     constructor(seqCallback) {
       this.seqCallback = seqCallback;
       this.currentSequence = void 0;
@@ -29418,9 +29429,10 @@ ${seq.sequence}
     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;
@@ -29539,14 +29551,14 @@ ${seq.sequence}
     }
     // 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]);
@@ -29644,8 +29656,9 @@ ${seq.sequence}
         }
       });
     }
-  }
-  __name(Parser, "Parser");
+  };
+  __name(_Parser, "Parser");
+  let Parser = _Parser;
   function _callback(callback) {
     if (process && process.nextTick)
       process.nextTick(callback);
@@ -29672,7 +29685,7 @@ ${seq.sequence}
     return out;
   }
   __name(_processParseOptions, "_processParseOptions");
-  class GFFTransform extends streamBrowserify.Transform {
+  const _GFFTransform = class _GFFTransform extends streamBrowserify.Transform {
     constructor(inputOptions = {}) {
       super({ objectMode: true });
       this.textBuffer = "";
@@ -29712,8 +29725,9 @@ ${seq.sequence}
       this.parser.finish();
       _callback(callback);
     }
-  }
-  __name(GFFTransform, "GFFTransform");
+  };
+  __name(_GFFTransform, "GFFTransform");
+  let GFFTransform = _GFFTransform;
   function parseStream(options = {}) {
     return new GFFTransform(options);
   }
@@ -29757,7 +29771,7 @@ ${seq.sequence}
     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;
@@ -29799,8 +29813,9 @@ ${seq.sequence}
       this.haveWeEmittedData = true;
       _callback(callback);
     }
-  }
-  __name(FormattingTransform, "FormattingTransform");
+  };
+  __name(_FormattingTransform, "FormattingTransform");
+  let FormattingTransform = _FormattingTransform;
   function formatStream(options = {}) {
     return new FormattingTransform(options);
   }
@@ -29904,8 +29919,8 @@ ${seq.sequence}
         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)
@@ -30015,8 +30030,8 @@ ${seq.sequence}
         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)
@@ -30468,7 +30483,7 @@ ${seq.sequence}
     } 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;
@@ -30480,7 +30495,7 @@ ${seq.sequence}
   }, "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) {
@@ -30510,7 +30525,7 @@ ${seq.sequence}
     }
     var fltr = opts && opts.filter;
     for (var i2 = 0; i2 < c; ++i2) {
-      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,
@@ -30569,7 +30584,7 @@ ${seq.sequence}
           );
         }
       }
-      if (/^(fasta|fas|fa|fna|ffn)$/.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);
@@ -32398,7 +32413,7 @@ ${seq.sequence}
   }
   __name(jsonToGenbank, "jsonToGenbank");
   function createGenbankLocus(serSeq, options) {
-    var _a2, _b2;
+    var _a3, _b3;
     if (serSeq.sequence.symbols) {
       serSeq.sequence = serSeq.sequence.symbols.split("");
     }
@@ -32407,9 +32422,9 @@ ${seq.sequence}
     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);

package/package.json

@@ -1,10 +1,10 @@
 {
   "name": "@teselagen/bio-parsers",
-  "version": "0.3.6",
+  "version": "0.3.8",
   "type": "commonjs",
   "dependencies": {
-    "@teselagen/sequence-utils": "0.3.6",
-    "@teselagen/range-utils": "0.3.6",
+    "@teselagen/sequence-utils": "0.3.7",
+    "@teselagen/range-utils": "0.3.7",
     "@gmod/gff": "^1.2.1",
     "buffer": "^6.0.3",
     "bufferpack": "^0.0.6",

package/src/anyToJson.js

@@ -71,7 +71,7 @@ async function anyToJson(fileContentStringOrFileObj, options) {
   }
   // console.log(`fileContentString.includes("seq:seq"):`,fileContentString.includes("seq:seq"))
   // console.log(`fileContentString.includes("jbei")):`,fileContentString.includes("jbei"))
-  if (/^(fasta|fas|fa|fna|ffn)$/.test(ext)) {
+  if (/^(fasta|fas|fa|fna|ffn|faa)$/.test(ext)) {
     // FASTA
     return fastaToJson(fileContentString, options);
   } else if (/^(gb|gbk)$/.test(ext)) {

package/src/fastaToJson.js

@@ -1,4 +1,5 @@
 import createInitialSequence from "./utils/createInitialSequence";
+import extractFileExtension from "./utils/extractFileExtension";
 import splitStringIntoLines from "./utils/splitStringIntoLines.js";
 import validateSequenceArray from "./utils/validateSequenceArray";
 
@@ -9,7 +10,12 @@ import validateSequenceArray from "./utils/validateSequenceArray";
  * @author Joshua P Nixon
  */
 
-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 {
@@ -28,8 +34,8 @@ function fastaToJson(fileString, options) {
     resultArray = [
       {
         success: false,
-        messages: ["Import Error: Invalid File"],
-      },
+        messages: ["Import Error: Invalid File"]
+      }
     ];
   }
   return validateSequenceArray(resultArray, options);
@@ -71,10 +77,9 @@ function fastaToJson(fileString, options) {
   }
 
   function parseTitle(line) {
-
-    if (options && 'parseName' in options && !options.parseName){
-      result.parsedSequence.name = line.slice(1)
-      return
+    if (options && "parseName" in options && !options.parseName) {
+      result.parsedSequence.name = line.slice(1);
+      return;
     }
 
     const pipeIndex = line.indexOf("|");

package/src/utils/validateSequence.js

@@ -114,7 +114,12 @@ export default function validateSequence(sequence, options = {}) {
         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 {