text-categorizer - npm Package Compare versions

Comparing version 1.0.3 to 1.0.4

dist/constants.d.ts

@@ -14,2 +14,4 @@ export declare const PATTERNS: {
     FILE_PATH: RegExp;
+    MEASUREMENT: RegExp;
+    SQL: RegExp;
 };

dist/constants.js

@@ -8,3 +8,3 @@ export const PATTERNS = {
     DATE: /\b(\d{1,4}[-/.]\d{1,2}[-/.]\d{1,4}|\d{1,2}\s+(Jan|Feb|Mar|Apr|May|Jun|Jul|Aug|Sep|Oct|Nov|Dec)[a-z]*\s+\d{1,4})\b/gi,
-    EQUATION: /[\d\s]*[+\-*/()=]+[\d\s]*/,
+    EQUATION: /(?:\d+\s*|\(\s*)[+\-*/()=]+\s*(?:\d+\s*|\)) /,
     PHONE: /(?:\+\d{1,3}\s?)?\(?\d{3}\)?[-.\s]?\d{3}[-.\s]?\d{4}/g,
@@ -14,3 +14,5 @@ ADDRESS: /\d+\s+([A-Za-z]+(\.?\s|\.)){1,}\s*,?\s*([A-Za-z]+\s*,\s*)?[A-Z]{2}\s*\d{5}(-\d{4})?/g,
     PRODUCT_CODE: /^[A-Z0-9]{3,}-[A-Z0-9]{3,}$|^[A-Z]{2,4}\d{4,}$/,
-    FILE_PATH: /^(?:[a-zA-Z]:\\|\/|\.\/|\.\.\/)(?:[^\\\/:*?"<>|\r\n]+\\)*[^\\\/:*?"<>|\r\n]*$/
+    FILE_PATH: /^(?:[a-zA-Z]:\\|\/|\.\/|\.\.\/)(?:[^\\\/:*?"<>|\r\n]+\\)*[^\\\/:*?"<>|\r\n]*$/,
+    MEASUREMENT: /(\d+(\.\d+)?\s*(cm|mm|in|ft|yd|mi|km|g|kg|lb|oz|ml|l|tsp|tbsp|cup|pt|qt|gal|fl oz|in²|ft²|yd²|mi²|km²|ac|ha|sq mi|sq km|°F|°C|K|°|'|"))/gi,
+    SQL: /\b(SELECT|INSERT|UPDATE|DELETE|FROM|WHERE|JOIN|GROUP BY|ORDER BY|HAVING|CREATE|ALTER|DROP|TABLE|INDEX)\b/i
 };

dist/text-categorizer.d.ts

@@ -15,3 +15,3 @@ import type { ContentCategory } from "./types";
     private static extractSocialElements;
-    private static isSearchQuery;
+    private static isSearchParams;
     private static extractDates;
@@ -26,2 +26,3 @@ private static isPhoneNumber;
     private static isProductCode;
+    private static isMeasurement;
     private static parseCurrency;
@@ -28,0 +29,0 @@ private static parseCsv;

dist/text-categorizer.js

@@ -7,3 +7,3 @@ import { PATTERNS } from "./constants";
     static containsLinks(text) {
-        return text.match(PATTERNS.URL) || [];
+        return text.match(PATTERNS.URL) || undefined;
     }
@@ -91,9 +91,5 @@ static isList(text) {
     }
-    static isSearchQuery(text) {
-        const searchPatterns = [
-            /^(what|how|who|where|when|why)\s.+\??$/i,
-            /^["'].+["']\s*(site:|filetype:|OR|AND)/i,
-            /^[^.!?]+\??$/
-        ];
-        return searchPatterns.some((pattern) => pattern.test(text.trim()));
+    static isSearchParams(text) {
+        const searchPattern = /(\?|\&)?[a-zA-Z0-9_]+=[^&]*/;
+        return searchPattern.test(text.trim());
     }
@@ -129,4 +125,3 @@ static extractDates(text) {
     static isSql(text) {
-        const sqlKeywords = /\b(SELECT|INSERT|UPDATE|DELETE|FROM|WHERE|JOIN|GROUP BY|ORDER BY|HAVING|CREATE|ALTER|DROP|TABLE|INDEX)\b/i;
-        return sqlKeywords.test(text) && text.includes(";");
+        return PATTERNS.SQL.test(text) && text.includes(";");
     }
@@ -142,2 +137,5 @@ static isFilePath(text) {
     }
+    static isMeasurement(text) {
+        return PATTERNS.MEASUREMENT.test(text);
+    }
     static parseCurrency(text) {
@@ -249,9 +247,2 @@ const match = text.match(PATTERNS.CURRENCY);
         }
-        if (this.isSearchQuery(content)) {
-            return {
-                type: "query",
-                content,
-                metadata: { confidence: 0.8 }
-            };
-        }
         const dates = this.extractDates(content);
@@ -289,2 +280,26 @@ if (dates.length > 0) {
         }
+        if (this.isEquation(content)) {
+            return {
+                type: "equation",
+                content,
+                metadata: { format: "math" }
+            };
+        }
+        if (this.isSearchParams(content)) {
+            return {
+                type: "search",
+                content,
+                metadata: { links: this.containsLinks(content) }
+            };
+        }
+        if (this.isMeasurement(content)) {
+            return {
+                type: "measurement",
+                content,
+                metadata: {
+                    amount: parseFloat(content.match(PATTERNS.MEASUREMENT)?.[0] || ""),
+                    unit: content.match(/[a-zA-Z]+/)?.[0] || ""
+                }
+            };
+        }
         return { type: "text", content };
@@ -291,0 +306,0 @@ }

dist/types.d.ts

 export type ContentCategory = {
-    type: "link" | "code" | "list" | "text" | "email" | "json" | "markdown" | "equation" | "date" | "social" | "query" | "phone" | "address" | "csv" | "xml" | "sql" | "filepath" | "currency" | "productCode";
+    type: "link" | "code" | "list" | "text" | "email" | "json" | "markdown" | "equation" | "date" | "social" | "search" | "phone" | "address" | "csv" | "xml" | "sql" | "filepath" | "currency" | "productCode" | "measurement";
     content: string;
@@ -18,3 +18,4 @@ metadata?: {
         structured?: any;
+        unit?: string;
     };
 };

package.json

 {
   "name": "text-categorizer",
-  "version": "1.0.3",
+  "version": "1.0.4",
   "description": "Intelligent text content type detection and classification",
@@ -45,6 +45,9 @@ "main": "dist/index.js",
   "devDependencies": {
+    "@types/jest": "^29.5.14",
     "@types/node": "^18.0.0",
-    "typescript": "^4.9.0",
-    "eslint": "^8.0.0"
+    "eslint": "^8.0.0",
+    "jest": "^29.7.0",
+    "ts-jest": "^29.2.5",
+    "typescript": "^4.9.0"
   }
 }

README.md

@@ -183,3 +183,8 @@ # Text Categorizer
 - Product codes
+- Measurement
 
+## TODO
+
+- Measurements
+
 ## Contributing
@@ -186,0 +191,0 @@

src/constants.ts

@@ -8,3 +8,3 @@ export const PATTERNS = {
   DATE: /\b(\d{1,4}[-/.]\d{1,2}[-/.]\d{1,4}|\d{1,2}\s+(Jan|Feb|Mar|Apr|May|Jun|Jul|Aug|Sep|Oct|Nov|Dec)[a-z]*\s+\d{1,4})\b/gi,
-  EQUATION: /[\d\s]*[+\-*/()=]+[\d\s]*/,
+  EQUATION: /(?:\d+\s*|\(\s*)[+\-*/()=]+\s*(?:\d+\s*|\)) /, ///[\d\s]*[+\-*/()=]+[\d\s]*/,
   PHONE: /(?:\+\d{1,3}\s?)?\(?\d{3}\)?[-.\s]?\d{3}[-.\s]?\d{4}/g,
@@ -16,3 +16,6 @@ ADDRESS:
   FILE_PATH:
-    /^(?:[a-zA-Z]:\\|\/|\.\/|\.\.\/)(?:[^\\\/:*?"<>|\r\n]+\\)*[^\\\/:*?"<>|\r\n]*$/
+    /^(?:[a-zA-Z]:\\|\/|\.\/|\.\.\/)(?:[^\\\/:*?"<>|\r\n]+\\)*[^\\\/:*?"<>|\r\n]*$/,
+  MEASUREMENT:
+    /(\d+(\.\d+)?\s*(cm|mm|in|ft|yd|mi|km|g|kg|lb|oz|ml|l|tsp|tbsp|cup|pt|qt|gal|fl oz|in²|ft²|yd²|mi²|km²|ac|ha|sq mi|sq km|°F|°C|K|°|'|"))/gi,
+  SQL: /\b(SELECT|INSERT|UPDATE|DELETE|FROM|WHERE|JOIN|GROUP BY|ORDER BY|HAVING|CREATE|ALTER|DROP|TABLE|INDEX)\b/i
 };

src/text-categorizer.ts

@@ -11,4 +11,4 @@ import { PATTERNS } from "./constants";
 
-  private static containsLinks(text: string): string[] {
-    return text.match(PATTERNS.URL) || [];
+  private static containsLinks(text: string): string[] | undefined {
+    return text.match(PATTERNS.URL) || undefined;
   }
@@ -116,10 +116,6 @@
 
-  private static isSearchQuery(text: string): boolean {
-    const searchPatterns = [
-      /^(what|how|who|where|when|why)\s.+\??$/i,
-      /^["'].+["']\s*(site:|filetype:|OR|AND)/i,
-      /^[^.!?]+\??$/
-    ];
+  private static isSearchParams(text: string): boolean {
+    const searchPattern = /(\?|\&)?[a-zA-Z0-9_]+=[^&]*/;
 
-    return searchPatterns.some((pattern) => pattern.test(text.trim()));
+    return searchPattern.test(text.trim());
   }
@@ -167,5 +163,3 @@
   private static isSql(text: string): boolean {
-    const sqlKeywords =
-      /\b(SELECT|INSERT|UPDATE|DELETE|FROM|WHERE|JOIN|GROUP BY|ORDER BY|HAVING|CREATE|ALTER|DROP|TABLE|INDEX)\b/i;
-    return sqlKeywords.test(text) && text.includes(";");
+    return PATTERNS.SQL.test(text) && text.includes(";");
   }
@@ -185,2 +179,6 @@
 
+  private static isMeasurement(text: string): boolean {
+    return PATTERNS.MEASUREMENT.test(text);
+  }
+
   private static parseCurrency(text: string): {
@@ -313,10 +311,2 @@ amount: number;
 
-    if (this.isSearchQuery(content)) {
-      return {
-        type: "query",
-        content,
-        metadata: { confidence: 0.8 }
-      };
-    }
-
     const dates = this.extractDates(content);
@@ -358,2 +348,29 @@ if (dates.length > 0) {
 
+    if (this.isEquation(content)) {
+      return {
+        type: "equation",
+        content,
+        metadata: { format: "math" }
+      };
+    }
+
+    if (this.isSearchParams(content)) {
+      return {
+        type: "search",
+        content,
+        metadata: { links: this.containsLinks(content) }
+      };
+    }
+
+    if (this.isMeasurement(content)) {
+      return {
+        type: "measurement",
+        content,
+        metadata: {
+          amount: parseFloat(content.match(PATTERNS.MEASUREMENT)?.[0] || ""),
+          unit: content.match(/[a-zA-Z]+/)?.[0] || ""
+        }
+      };
+    }
+
     return { type: "text", content };
@@ -360,0 +377,0 @@ }

src/types.ts

@@ -13,3 +13,3 @@ export type ContentCategory = {
     | "social"
-    | "query"
+    | "search"
     | "phone"
@@ -22,3 +22,4 @@ | "address"
     | "currency"
-    | "productCode";
+    | "productCode"
+    | "measurement";
   content: string;
@@ -39,3 +40,4 @@ metadata?: {
     structured?: any;
+    unit?: string;
   };
 };

No alert changes

Improved metrics

Total package byte prevSize

41974

increased by31.81%

Number of package files

21

increased by31.25%

Lines of code

985

increased by32.04%

Number of lines in readme file

239

increased by2.14%

Worsened metrics

Dev dependency count

6

increased by100%

No dependency changes