Merge pull request #163 from charlesLoder/refactor-chars

Refactor chars

src/char.ts

@@ -1,9 +1,6 @@
 import { Cluster } from "./cluster";
-import { taamim } from "./utils/regularExpressions";
-const consonants = /[\u{05D0}-\u{05F2}]/u;
-const ligature = /[\u{05C1}-\u{05C2}]/u;
-const dagesh = /[\u{05BC}\u{05BF}]/u; // includes rafe
-const niqqud = /[\u{05B0}-\u{05BB}\u{05C7}]/u;
+import { consonants, dagesh, ligatures, vowels, rafe, sheva, taamim } from "./utils/regularExpressions";
+import { CharToNameMap, charToNameMap, NameToCharMap, nameToCharMap } from "./utils/charMap";
 
 /**
  * A Hebrew character and its positioning number for being sequenced correctly.
@@ -12,45 +9,222 @@ const niqqud = /[\u{05B0}-\u{05BB}\u{05C7}]/u;
 export class Char {
   #text: string;
   #cluster: Cluster | null = null;
+  #sequencePosition: number;
 
   constructor(char: string) {
     this.#text = char;
-  }
-
-  /**
-   * @returns the text of the Char
-   *
-   * ```typescript
-   * const text: Text = new Text("אֱלֹהִ֑ים");
-   * text.chars[0].text;
-   * // "א"
-   * ```
-   */
-  get text(): string {
-    return this.#text;
+    this.#sequencePosition = this.findPos();
   }
 
   private findPos(): number {
     const char = this.text;
-    if (consonants.test(char)) {
+    if (Char.consonants.test(char)) {
       return 0;
     }
-    if (ligature.test(char)) {
+    if (Char.ligatures.test(char)) {
       return 1;
     }
-    if (dagesh.test(char)) {
+    if (Char.dagesh.test(char)) {
       return 2;
     }
-    if (niqqud.test(char)) {
+    if (Char.rafe.test(char)) {
+      return 2;
+    }
+    if (Char.vowels.test(char)) {
+      return 3;
+    }
+    if (Char.sheva.test(char)) {
       return 3;
     }
-    if (taamim.test(char)) {
+    if (Char.taamim.test(char)) {
       return 4;
     }
     // i.e. any non-hebrew char
     return 10;
   }
 
+  private isCharKeyOfCharToNameMap(char: string): char is keyof CharToNameMap {
+    return char in charToNameMap;
+  }
+
+  private static get consonants() {
+    return consonants;
+  }
+
+  private static get dagesh() {
+    return dagesh;
+  }
+
+  private static get ligatures() {
+    return ligatures;
+  }
+
+  private static get rafe() {
+    return rafe;
+  }
+
+  private static get sheva() {
+    return sheva;
+  }
+
+  private static get taamim() {
+    return taamim;
+  }
+
+  private static get vowels() {
+    return vowels;
+  }
+
+  /**
+   * The parent `Cluster` of the `Char`, if any.
+   *
+   * ```typescript
+   * const text: Text = new Text("דָּבָר");
+   * const firstChar = text.chars[0];
+   * firstChar.text;
+   * // "ד"
+   * firstChar.cluster?.text;
+   * // "דָּ"
+   * ```
+   */
+  get cluster(): Cluster | null {
+    return this.#cluster;
+  }
+
+  set cluster(cluster: Cluster | null) {
+    this.#cluster = cluster;
+  }
+
+  isCharacterName(name: keyof NameToCharMap): boolean {
+    if (!nameToCharMap[name]) {
+      throw new Error(`${name} is not a valid value`);
+    }
+
+    const match = this.#text.match(nameToCharMap[name]);
+
+    return !!match;
+  }
+
+  /**
+   * Returns `true` if the `Char` is a consonant
+   *
+   * ```typescript
+   * const text: Text = new Text("אֱלֹהִ֑ים");
+   * text.chars[0].isConsonant;
+   * // true
+   * ```
+   */
+  get isConsonant(): boolean {
+    return Char.consonants.test(this.#text);
+  }
+
+  /**
+   * Returns `true` if the `Char` is a ligature
+   *
+   * ```typescript
+   * const text: Text = new Text("שָׁלֽוֹם");
+   * text.chars[1].isLigature;
+   * // true
+   * ```
+   */
+  get isLigature(): boolean {
+    return Char.ligatures.test(this.#text);
+  }
+
+  /**
+   * Returns `true` if the `Char` is a dagesh
+   *
+   * ```typescript
+   * const text: Text = new Text("בּ");
+   * text.chars[1].isDagesh;
+   * // true
+   */
+  get isDagesh(): boolean {
+    return Char.dagesh.test(this.#text);
+  }
+
+  /**
+   * Returns `true` if the `Char` is a rafe
+   *
+   * ```typescript
+   * const text: Text = new Text("בֿ");
+   * text.chars[1].isRafe;
+   * // true
+   * ```
+   */
+  get isRafe(): boolean {
+    return Char.rafe.test(this.#text);
+  }
+
+  /**
+   * Returns `true` if the `Char` is a sheva
+   * ```typescript
+   * const text: Text = new Text("בְ");
+   * text.chars[1].isSheva;
+   * // true
+   * ```
+   */
+  get isSheva(): boolean {
+    return Char.sheva.test(this.#text);
+  }
+
+  /**
+   * Returns `true` if the `Char` is a sheva
+   *
+   *
+   * ```typescript
+   * const text: Text = new Text("בֺ");
+   * text.chars[1].isVowel;
+   * // true
+   * ```
+   */
+  get isVowel(): boolean {
+    return Char.vowels.test(this.#text);
+  }
+
+  /**
+   * Returns `true` if the `Char` is a taamim
+   *
+   * ```typescript
+   * const text: Text = new Text("בֺ֨");
+   * text.chars[2].isTaamim;
+   * // true
+   * ```
+   */
+  get isTaamim(): boolean {
+    return Char.taamim.test(this.#text);
+  }
+
+  /**
+   * Returns `true` if the `Char` is not a Hebrew character
+   *
+   * ```typescript
+   * const text: Text = new Text("a");
+   * text.chars[0].isNotHebrew;
+   * // true
+   * ```
+   */
+  get isNotHebrew(): boolean {
+    return this.sequencePosition === 10;
+  }
+
+  /**
+   * Returns the name of the character
+   *
+   * ```typescript
+   * const text: Text = new Text("אֱלֹהִ֑ים");
+   * text.chars[0].characterName;
+   * // "ALEF"
+   * ```
+   */
+  get characterName(): CharToNameMap[keyof CharToNameMap] | null {
+    const text = this.#text;
+    if (this.isCharKeyOfCharToNameMap(text)) {
+      return charToNameMap[text];
+    }
+    return null;
+  }
+
   /**
    * @returns a number used for sequencing
    *
@@ -69,26 +243,19 @@ export class Char {
    * ```
    */
   get sequencePosition(): number {
-    return this.findPos();
+    return this.#sequencePosition;
   }
 
   /**
-   * The parent `Cluster` of the `Char`, if any.
+   * @returns the text of the Char
    *
    * ```typescript
-   * const text: Text = new Text("דָּבָר");
-   * const firstChar = text.chars[0];
-   * firstChar.text;
-   * // "ד"
-   * firstChar.cluster?.text;
-   * // "דָּ"
+   * const text: Text = new Text("אֱלֹהִ֑ים");
+   * text.chars[0].text;
+   * // "א"
    * ```
    */
-  get cluster(): Cluster | null {
-    return this.#cluster;
-  }
-
-  set cluster(cluster: Cluster | null) {
-    this.#cluster = cluster;
+  get text(): string {
+    return this.#text;
   }
 }

src/cluster.ts

@@ -2,7 +2,7 @@ import { Char } from "./char";
 import { Node } from "./node";
 import { Syllable } from "./syllable";
 import { taamim, hebChars, vowelsCaptureGroup, punctuation } from "./utils/regularExpressions";
-import { charToNameMap, CharToNameMap, NameToCharMap, nameToCharMap } from "./utils/vowelMap";
+import { charToNameMap, CharToNameMap, NameToCharMap, nameToCharMap } from "./utils/charMap";
 
 /**
  * A cluster is group of Hebrew character constituted by:

src/syllable.ts

@@ -1,6 +1,6 @@
 import { Cluster } from "./cluster";
 import { Char } from "./char";
-import { CharToNameMap, charToNameMap, NameToCharMap, nameToCharMap } from "./utils/vowelMap";
+import { CharToNameMap, charToNameMap, NameToCharMap, nameToCharMap } from "./utils/charMap";
 import { vowelsCaptureGroupWithSheva } from "./utils/regularExpressions";
 import { removeTaamim } from "./utils/removeTaamim";
 import { Node } from "./node";