nlptoolkit-ngram - npm Package Compare versions

Comparing version 1.0.3 to 1.0.4

dist/LaplaceSmoothing.d.ts

@@ -5,2 +5,6 @@ import { SimpleSmoothing } from "./SimpleSmoothing";
     private readonly delta;
+    /**
+     * Constructor for Laplace smoothing. Sets the delta.
+     * @param delta Delta value in Laplace smoothing.
+     */
     constructor(delta?: number);
@@ -7,0 +11,0 @@ /**

dist/LaplaceSmoothing.js

@@ -15,2 +15,6 @@ (function (factory) {
     class LaplaceSmoothing extends SimpleSmoothing_1.SimpleSmoothing {
+        /**
+         * Constructor for Laplace smoothing. Sets the delta.
+         * @param delta Delta value in Laplace smoothing.
+         */
         constructor(delta = 1.0) {
@@ -17,0 +21,0 @@ super();

dist/MultipleFile.d.ts

@@ -5,6 +5,22 @@ export declare class MultipleFile {
     private contents;
+    /**
+     * Constructor for {@link MultipleFile} class. Initializes the buffer reader with the first input file
+     * from the fileNameList. MultipleFile supports simple multipart file system, where a text file is divided
+     * into multiple files.
+     * @param args A list of files given as dynamic parameters.
+     */
     constructor(...args: Array<any>);
+    /**
+     * Reads a single line from the current file. If the end of file is reached for the current file,
+     * next file is opened and a single line from that file is read. If all files are read, the method
+     * returns null.
+     * @return Read line from the current file.
+     */
     readLine(): string;
+    /**
+     * Checks if the current file has more lines to be read.
+     * @returns True if the current file has more lines to be read, false otherwise.
+     */
     hasNextLine(): boolean;
     readCorpus(): Array<Array<string>>;
 }

dist/MultipleFile.js

@@ -15,2 +15,8 @@ (function (factory) {
     class MultipleFile {
+        /**
+         * Constructor for {@link MultipleFile} class. Initializes the buffer reader with the first input file
+         * from the fileNameList. MultipleFile supports simple multipart file system, where a text file is divided
+         * into multiple files.
+         * @param args A list of files given as dynamic parameters.
+         */
         constructor(...args) {
@@ -21,2 +27,8 @@ this.fileIndex = 0;
         }
+        /**
+         * Reads a single line from the current file. If the end of file is reached for the current file,
+         * next file is opened and a single line from that file is read. If all files are read, the method
+         * returns null.
+         * @return Read line from the current file.
+         */
         readLine() {
@@ -29,2 +41,6 @@ if (!this.contents.hasNextLine()) {
         }
+        /**
+         * Checks if the current file has more lines to be read.
+         * @returns True if the current file has more lines to be read, false otherwise.
+         */
         hasNextLine() {
@@ -31,0 +47,0 @@ return this.fileIndex != this.fileNameList.length - 1 || this.contents.hasNextLine();

dist/NGram.d.ts

@@ -20,3 +20,12 @@ import { NGramNode } from "./NGramNode";
     constructor(...args: Array<any>);
+    /**
+     * Reads the header from the input file.
+     * @param multipleFile Input file
+     */
     readHeader(multipleFile: MultipleFile): void;
+    /**
+     * Merges current NGram with the given NGram. If N of the two NGram's are not same, it does not
+     * merge. Merges first the vocabulary, then the NGram trees.
+     * @param toBeMerged NGram to be merged with.
+     */
     merge(toBeMerged: NGram<Symbol>): void;
@@ -192,3 +201,8 @@ /**
     setAdjustedProbability(countsOfCounts: Array<number>, height: number, pZero: number): void;
+    /**
+     * Prunes NGram according to the given threshold. All nodes having a probability less than the threshold will be
+     * pruned.
+     * @param threshold Probability threshold used for pruning.
+     */
     prune(threshold: number): void;
 }

dist/NGram.js

@@ -41,2 +41,6 @@ (function (factory) {
         }
+        /**
+         * Reads the header from the input file.
+         * @param multipleFile Input file
+         */
         readHeader(multipleFile) {
@@ -60,2 +64,7 @@ let line = multipleFile.readLine();
         }
+        /**
+         * Merges current NGram with the given NGram. If N of the two NGram's are not same, it does not
+         * merge. Merges first the vocabulary, then the NGram trees.
+         * @param toBeMerged NGram to be merged with.
+         */
         merge(toBeMerged) {
@@ -417,2 +426,7 @@ if (this.N != toBeMerged.getN()) {
         }
+        /**
+         * Prunes NGram according to the given threshold. All nodes having a probability less than the threshold will be
+         * pruned.
+         * @param threshold Probability threshold used for pruning.
+         */
         prune(threshold) {
@@ -419,0 +433,0 @@ if (threshold > 0.0 && threshold <= 1.0) {

dist/NGramNode.d.ts

@@ -10,2 +10,4 @@ import { MultipleFile } from "./MultipleFile";
     private unknown;
+    constructor1(symbol: any): void;
+    constructor2(symbol: any, multipleFile: MultipleFile): void;
     /**
@@ -18,2 +20,6 @@ * Constructor of {@link NGramNode}
     constructor(symbol: any, multipleFile?: MultipleFile);
+    /**
+     * Merges this NGramNode with the corresponding NGramNode in another NGram.
+     * @param toBeMerged Parallel NGramNode of the parallel NGram tree.
+     */
     merge(toBeMerged: NGramNode<Symbol>): void;
@@ -139,3 +145,9 @@ /**
     generateNextString(s: Array<Symbol>, index: number): Symbol;
+    /**
+     * Prunes the NGramNode according to the given threshold. Removes the child(ren) whose probability is less than the
+     * threshold.
+     * @param threshold Threshold for pruning the NGram tree.
+     * @param N N in N-Gram.
+     */
     prune(threshold: number, N: number): void;
 }

dist/NGramNode.js

@@ -25,27 +25,37 @@ (function (factory) {
             if (multipleFile == undefined) {
-                this.symbol = symbol;
-                this.count = 0;
+                this.constructor1(symbol);
             }
             else {
-                if (!symbol) {
-                    this.symbol = multipleFile.readLine().trim();
+                this.constructor2(symbol, multipleFile);
+            }
+        }
+        constructor1(symbol) {
+            this.symbol = symbol;
+            this.count = 0;
+        }
+        constructor2(symbol, multipleFile) {
+            if (!symbol) {
+                this.symbol = multipleFile.readLine().trim();
+            }
+            let line = multipleFile.readLine().trim();
+            let items = line.split(" ");
+            if (items.length != 4) {
+                Console.log("Error in line -> " + line);
+            }
+            this.count = Number(items[0]);
+            this.probability = Number(items[1]);
+            this.probabilityOfUnseen = Number(items[2]);
+            let numberOfChildren = Number(items[3]);
+            if (numberOfChildren > 0) {
+                this.children = new Map();
+                for (let i = 0; i < numberOfChildren; i++) {
+                    let childNode = new NGramNode(false, multipleFile);
+                    this.children.set(childNode.symbol, childNode);
                 }
-                let line = multipleFile.readLine().trim();
-                let items = line.split(" ");
-                if (items.length != 4) {
-                    Console.log("Error in line -> " + line);
-                }
-                this.count = Number(items[0]);
-                this.probability = Number(items[1]);
-                this.probabilityOfUnseen = Number(items[2]);
-                let numberOfChildren = Number(items[3]);
-                if (numberOfChildren > 0) {
-                    this.children = new Map();
-                    for (let i = 0; i < numberOfChildren; i++) {
-                        let childNode = new NGramNode(false, multipleFile);
-                        this.children.set(childNode.symbol, childNode);
-                    }
-                }
             }
         }
+        /**
+         * Merges this NGramNode with the corresponding NGramNode in another NGram.
+         * @param toBeMerged Parallel NGramNode of the parallel NGram tree.
+         */
         merge(toBeMerged) {
@@ -389,2 +399,8 @@ if (this.children != undefined) {
         }
+        /**
+         * Prunes the NGramNode according to the given threshold. Removes the child(ren) whose probability is less than the
+         * threshold.
+         * @param threshold Threshold for pruning the NGram tree.
+         * @param N N in N-Gram.
+         */
         prune(threshold, N) {
@@ -391,0 +407,0 @@ if (N == 0) {

dist/NoSmoothing.d.ts

 import { SimpleSmoothing } from "./SimpleSmoothing";
 import { NGram } from "./NGram";
 export declare class NoSmoothing<Symbol> extends SimpleSmoothing<Symbol> {
+    /**
+     * Calculates the N-Gram probabilities with no smoothing
+     * @param nGram N-Gram for which no smoothing is done.
+     * @param level Height of the NGram node.
+     */
     setProbabilitiesWithLevel(nGram: NGram<Symbol>, level: number): void;
 }

dist/NoSmoothing.js

@@ -15,2 +15,7 @@ (function (factory) {
     class NoSmoothing extends SimpleSmoothing_1.SimpleSmoothing {
+        /**
+         * Calculates the N-Gram probabilities with no smoothing
+         * @param nGram N-Gram for which no smoothing is done.
+         * @param level Height of the NGram node.
+         */
         setProbabilitiesWithLevel(nGram, level) {
@@ -17,0 +22,0 @@ nGram.setProbabilityWithPseudoCount(0.0, level);

dist/NoSmoothingWithDictionary.d.ts

 import { NoSmoothing } from "./NoSmoothing";
 import { NGram } from "./NGram";
 export declare class NoSmoothingWithDictionary<Symbol> extends NoSmoothing<Symbol> {
-    private dictionary;
+    private readonly dictionary;
     /**
@@ -6,0 +6,0 @@ * Constructor of {@link NoSmoothingWithDictionary}

dist/NoSmoothingWithNonRareWords.d.ts

@@ -5,3 +5,3 @@ import { NoSmoothing } from "./NoSmoothing";
     private dictionary;
-    private probability;
+    private readonly probability;
     /**
@@ -8,0 +8,0 @@ * Constructor of {@link NoSmoothingWithNonRareWords}

dist/SimpleSmoothing.d.ts

 import { NGram } from "./NGram";
 export declare abstract class SimpleSmoothing<Symbol> {
     abstract setProbabilitiesWithLevel(nGram: NGram<Symbol>, level: number): void;
+    /**
+     * Calculates the N-Gram probabilities with simple smoothing.
+     * @param nGram N-Gram for which simple smoothing calculation is done.
+     */
     setProbabilities(nGram: NGram<Symbol>): void;
 }

dist/SimpleSmoothing.js

@@ -14,2 +14,6 @@ (function (factory) {
     class SimpleSmoothing {
+        /**
+         * Calculates the N-Gram probabilities with simple smoothing.
+         * @param nGram N-Gram for which simple smoothing calculation is done.
+         */
         setProbabilities(nGram) {
@@ -16,0 +20,0 @@ this.setProbabilitiesWithLevel(nGram, nGram.getN());

package.json

 {
   "name": "nlptoolkit-ngram",
-  "version": "1.0.3",
+  "version": "1.0.4",
   "description": "NGram library",
@@ -5,0 +5,0 @@ "main": "index.js",

README.md

@@ -27,3 +27,3 @@ N-Gram
 You can also see [Python](https://github.com/starlangsoftware/NGram-Py), [Java](https://github.com/starlangsoftware/NGram), 
-[C++](https://github.com/starlangsoftware/NGram-CPP), [Swift](https://github.com/starlangsoftware/NGram-Swift), 
+[C++](https://github.com/starlangsoftware/NGram-CPP), [C](https://github.com/starlangsoftware/NGram-C), [Swift](https://github.com/starlangsoftware/NGram-Swift), 
 [Cython](https://github.com/starlangsoftware/NGram-Cy) or [C#](https://github.com/starlangsoftware/NGram-CS) repository.
@@ -30,0 +30,0 @@

source/LaplaceSmoothing.ts

@@ -8,2 +8,6 @@ import {SimpleSmoothing} from "./SimpleSmoothing";
 
+    /**
+     * Constructor for Laplace smoothing. Sets the delta.
+     * @param delta Delta value in Laplace smoothing.
+     */
     constructor(delta: number = 1.0) {
@@ -10,0 +14,0 @@ super();

source/MultipleFile.ts

@@ -9,2 +9,8 @@ import {FileContents} from "nlptoolkit-util/dist/FileContents";
 
+    /**
+     * Constructor for {@link MultipleFile} class. Initializes the buffer reader with the first input file
+     * from the fileNameList. MultipleFile supports simple multipart file system, where a text file is divided
+     * into multiple files.
+     * @param args A list of files given as dynamic parameters.
+     */
     constructor(... args: Array<any>) {
@@ -16,2 +22,8 @@ this.fileIndex = 0
 
+    /**
+     * Reads a single line from the current file. If the end of file is reached for the current file,
+     * next file is opened and a single line from that file is read. If all files are read, the method
+     * returns null.
+     * @return Read line from the current file.
+     */
     readLine(): string{
@@ -25,2 +37,6 @@ if (!this.contents.hasNextLine()){
 
+    /**
+     * Checks if the current file has more lines to be read.
+     * @returns True if the current file has more lines to be read, false otherwise.
+     */
     hasNextLine(): boolean {
@@ -27,0 +43,0 @@ return this.fileIndex != this.fileNameList.length - 1 || this.contents.hasNextLine()

source/NGram.ts

@@ -40,2 +40,6 @@ import {NGramNode} from "./NGramNode";
 
+    /**
+     * Reads the header from the input file.
+     * @param multipleFile Input file
+     */
     readHeader(multipleFile: MultipleFile){
@@ -60,2 +64,7 @@ let line = multipleFile.readLine();
 
+    /**
+     * Merges current NGram with the given NGram. If N of the two NGram's are not same, it does not
+     * merge. Merges first the vocabulary, then the NGram trees.
+     * @param toBeMerged NGram to be merged with.
+     */
     merge(toBeMerged: NGram<Symbol>){
@@ -435,2 +444,7 @@ if (this.N != toBeMerged.getN()){
 
+    /**
+     * Prunes NGram according to the given threshold. All nodes having a probability less than the threshold will be
+     * pruned.
+     * @param threshold Probability threshold used for pruning.
+     */
     prune(threshold: number){
@@ -437,0 +451,0 @@ if (threshold > 0.0 && threshold <= 1.0){

source/NGramNode.ts

@@ -14,2 +14,29 @@ import {MultipleFile} from "./MultipleFile";
 
+    constructor1(symbol: any){
+        this.symbol = symbol
+        this.count = 0
+    }
+
+    constructor2(symbol: any, multipleFile: MultipleFile) {
+        if (!symbol) {
+            this.symbol = <Symbol><unknown>multipleFile.readLine().trim()
+        }
+        let line = multipleFile.readLine().trim();
+        let items = line.split(" ");
+        if (items.length != 4){
+            Console.log("Error in line -> " + line);
+        }
+        this.count = Number(items[0])
+        this.probability = Number(items[1]);
+        this.probabilityOfUnseen = Number(items[2]);
+        let numberOfChildren = Number(items[3]);
+        if (numberOfChildren > 0){
+            this.children = new Map<Symbol, NGramNode<Symbol>>()
+            for (let i = 0; i < numberOfChildren; i++){
+                let childNode = new NGramNode<Symbol>(false, multipleFile);
+                this.children.set(childNode.symbol, childNode);
+            }
+        }
+    }
+
     /**
@@ -23,27 +50,12 @@ * Constructor of {@link NGramNode}
         if (multipleFile == undefined){
-            this.symbol = symbol
-            this.count = 0
+            this.constructor1(symbol);
         } else {
-            if (!symbol) {
-                this.symbol = <Symbol><unknown>multipleFile.readLine().trim()
-            }
-            let line = multipleFile.readLine().trim();
-            let items = line.split(" ");
-            if (items.length != 4){
-                Console.log("Error in line -> " + line);
-            }
-            this.count = Number(items[0])
-            this.probability = Number(items[1]);
-            this.probabilityOfUnseen = Number(items[2]);
-            let numberOfChildren = Number(items[3]);
-            if (numberOfChildren > 0){
-                this.children = new Map<Symbol, NGramNode<Symbol>>()
-                for (let i = 0; i < numberOfChildren; i++){
-                    let childNode = new NGramNode<Symbol>(false, multipleFile);
-                    this.children.set(childNode.symbol, childNode);
-                }
-            }
+            this.constructor2(symbol, multipleFile);
         }
     }
 
+    /**
+     * Merges this NGramNode with the corresponding NGramNode in another NGram.
+     * @param toBeMerged Parallel NGramNode of the parallel NGram tree.
+     */
     merge(toBeMerged: NGramNode<Symbol>){
@@ -387,2 +399,8 @@ if (this.children != undefined){
 
+    /**
+     * Prunes the NGramNode according to the given threshold. Removes the child(ren) whose probability is less than the
+     * threshold.
+     * @param threshold Threshold for pruning the NGram tree.
+     * @param N N in N-Gram.
+     */
     prune(threshold: number, N: number){
@@ -389,0 +407,0 @@ if (N == 0){

source/NoSmoothing.ts

@@ -6,2 +6,7 @@ import {SimpleSmoothing} from "./SimpleSmoothing";
 
+    /**
+     * Calculates the N-Gram probabilities with no smoothing
+     * @param nGram N-Gram for which no smoothing is done.
+     * @param level Height of the NGram node.
+     */
     public setProbabilitiesWithLevel(nGram: NGram<Symbol>, level: number): void {
@@ -8,0 +13,0 @@ nGram.setProbabilityWithPseudoCount(0.0, level);

source/NoSmoothingWithDictionary.ts

@@ -6,3 +6,3 @@ import {NoSmoothing} from "./NoSmoothing";
 
-    private dictionary: Set<Symbol>
+    private readonly dictionary: Set<Symbol>
 
@@ -9,0 +9,0 @@ /**

source/NoSmoothingWithNonRareWords.ts

@@ -7,3 +7,3 @@ import {NoSmoothing} from "./NoSmoothing";
     private dictionary: Set<Symbol>
-    private probability: number
+    private readonly probability: number
 
@@ -10,0 +10,0 @@ /**

source/SimpleSmoothing.ts

@@ -6,2 +6,6 @@ import {NGram} from "./NGram";
 
+    /**
+     * Calculates the N-Gram probabilities with simple smoothing.
+     * @param nGram N-Gram for which simple smoothing calculation is done.
+     */
     public setProbabilities(nGram: NGram<Symbol>){
@@ -8,0 +12,0 @@ this.setProbabilitiesWithLevel(nGram, nGram.getN());

No alert changes

Improved metrics

Total package byte prevSize

6848767

increased by0.11%

Lines of code

3757

increased by4.83%

No dependency changes