Added (some) standardization from #121

Author: fbanados

lib/constants/AECD-rewritings.json

@@ -0,0 +1,35 @@
+{
+"S/he": "s/he",
+"A": "a",
+"The": "the",
+"It": "it",
+"An": "an",
+"Being": "being",
+"Her/his": "her/his",
+"They": "they",
+"There": "there",
+"My": "my",
+"Something": "something",
+"Having": "having",
+"In": "in",
+"One": "one",
+"He": "he",
+"She": "she",
+"Someone": "someone",
+"All": "all",
+"What": "what",
+"On": "on",
+"Your": "your",
+"Two": "two",
+"Three": "three",
+"Last": "last",
+"Four": "four",
+"Along": "along",
+"Too": "too",
+"To": "to",
+"That": "that",
+"Right": "right",
+"Our": "our",
+"Close": "close",
+"As": "as"
+}

lib/convert/AECD.js

@@ -11,6 +11,9 @@ const { readJSON } = fs;
  */
 let posMap;
 
+/** A map of rewritings that is to be used later */
+let rewritings;
+
 /**
  * A class representing a Maskwacîs database entry in DLx JSON format.
  */
@@ -80,7 +83,7 @@ class Entry {
     // this.semanticIndices = Entry.#splitSemanticDomains(RapidWordsIndices);
     this.test            = test;
 
-    this.senses = Entry.#splitDefinition(Definition)
+    this.senses = Entry.#splitDefinition(Entry.#applyRewritings(Definition))
     .map(definition => ({
       definition,
       original: definition,
@@ -139,6 +142,14 @@ class Entry {
     return string.replace(/[?!]/gu, ``);
   }
 
+  static #applyRewritings( string ) {
+    let text = string;
+    rewritings.forEach((replaceWith, match) => {
+      text = text.replaceAll(match,replaceWith);
+    });
+    return text;
+  }
+
   /**
    * Splits a string into senses based on sense numbers (1., 2. etc.).
    * @param  {String} definition The string to split.
@@ -174,6 +185,7 @@ export default async function convertAECD(inputPath, outputPath) {
 
   const posTable = await readJSON(`./lib/constants/AECD-pos.json`);
   posMap         = new Map(Object.entries(posTable));
+  rewritings     = new Map(Object.entries(await readJSON(`./lib/constants/AECD-rewritings.json`)));
 
   const errors  = [];
   let   entries = await readTSV(inputPath, { raw: true, relaxColumnCount: true });