fix: do not yield molfile that has less than 40 characters long

Author: lpatiny

src/MolfileStream.js

@@ -17,14 +17,16 @@ export class MolfileStream extends TransformStream {
           }
           const eolLength = this.#buffer[endOfDelimiter - 1] === '\r' ? 2 : 1;
           // need to remove the last eol because we will split on eol+'>' in getMolecule
-          controller.enqueue(this.#buffer.slice(begin, index - eolLength));
+          if (index - eolLength - begin > 40) {
+            controller.enqueue(this.#buffer.slice(begin, index - eolLength));
+          }
           index = endOfDelimiter + eolLength;
           begin = index;
         }
         this.#buffer = this.#buffer.slice(begin);
       },
       flush: (controller) => {
-        if (this.#buffer) {
+        if (this.#buffer && this.#buffer.length > 40) {
           controller.enqueue(this.#buffer);
         }
       },

src/__tests__/test2.sdf

@@ -496,3 +496,8 @@ M  END
 0:Chloroform-D1 (CDCl3) 
 
 $$$$
+
+asdfasdf
+iABAAA
+M END
+$$$$

src/iterator.js

@@ -16,7 +16,6 @@ export async function* iterator(readStream, options = {}) {
 
   const moleculeStream = readStream.pipeThrough(new MolfileStream({ eol }));
   for await (const entry of moleculeStream) {
-    if (entry.length < 20) continue;
     const molecule = getMolecule(entry, {
       eol,
       dynamicTyping,

src/parse.js

@@ -13,6 +13,7 @@ import { getMolecule } from './util/getMolecule';
  * @param {object} [options.modifiers] - Object containing callbacks to apply on some specific fields
  * @param {boolean} [options.mixedEOL=false] - Set to true if you know there is a mixture between \r\n and \n
  * @param {string} [options.eol] - Specify the end of line character. Default will be the one found in the file
+ * @returns {object} - Object containing the molecules, the labels and the statistics
  */
 export function parse(sdf, options = {}) {
   options = { ...options };
@@ -53,7 +54,7 @@ export function parse(sdf, options = {}) {
 
   for (let i = 0; i < entriesBoundaries.length; i++) {
     let sdfPart = sdf.slice(...entriesBoundaries[i]);
-
+    if (sdfPart.length < 40) continue;
     let currentLabels = [];
     const molecule = getMolecule(sdfPart, labels, currentLabels, options);
     if (!molecule) continue;