Adding a directive parser, and refactoring code a bit

.gitignore

@@ -1,5 +1,6 @@
 # IDE
 .vscode
+.idea
 
 # Logs and other files
 npm-debug.log*

package.json

@@ -45,6 +45,7 @@
   "dependencies": {
     "@types/glob": "^5.0.30",
     "@types/lodash": "^4.14.41",
+    "cheerio": "~0.22.0",
     "cli": "^1.0.1",
     "fs": "0.0.1-security",
     "glob": "^7.1.1",

src/example.ts

@@ -1,6 +1,5 @@
 import { Extractor } from './extractor';
 import { JsonSerializer } from './serializers/json.serializer';
-import { PotSerializer } from './serializers/pot.serializer';
 
 const dir = '/path/to/extract/strings/from';
 const dest = '/path/to/save/template/to/template.pot';

src/extractor.ts

@@ -1,23 +1,20 @@
 import { ParserInterface } from './parsers/parser.interface';
-import { HtmlParser } from './parsers/html.parser';
-import { TypescriptParser } from './parsers/typescript.parser';
+import { PipeParser } from './parsers/pipe.parser';
+import { DirectiveParser } from "./parsers/directive.parser";
+import { ServiceParser } from './parsers/service.parser';
 import { SerializerInterface } from './serializers/serializer.interface';
-import { PotSerializer } from './serializers/pot.serializer';
 
 import * as lodash from 'lodash';
 import * as glob from 'glob';
 import * as fs from 'fs';
 
-export interface TypeParserMap {
-	[ext: string]: ParserInterface
-}
-
 export class Extractor {
 
-	public parsers: TypeParserMap = {
-		html: new HtmlParser(),
-		ts: new TypescriptParser()
-	};
+	public parsers: ParserInterface[] = [
+		new PipeParser(),
+		new ServiceParser(),
+		new DirectiveParser()
+	];
 
 	public globPatterns: string[] = [
 		'/**/*.ts',
@@ -67,15 +64,14 @@ export class Extractor {
 	 * Extract messages from file using specialized parser
 	 */
 	protected _extractMessages(filePath: string): string[] {
-		const ext: string = filePath.split('.').pop();
-		if (!this.parsers.hasOwnProperty(ext)) {
-			return [];
-		}
+		let results = [];
 
 		const contents: string = fs.readFileSync(filePath, 'utf-8');
-		const parser: ParserInterface = this.parsers[ext];
+		this.parsers.forEach((parser: ParserInterface) => {
+			results = results.concat(parser.process(contents));
+		});
 
-		return parser.process(contents);
+		return results;
 	}
 
 }

src/parsers/directive.parser.ts

@@ -0,0 +1,66 @@
+import {ParserInterface} from './parser.interface';
+import * as $ from 'cheerio';
+
+export class DirectiveParser implements ParserInterface {
+
+    public patterns = {
+        template: `template:\\s?(("|'|\`)(.|[\\r\\n])+?[^\\\\]\\2)`
+    };
+
+    protected _parseTemplate(content) {
+        let results: string[] = [],
+            template = content.trim()
+            // hack for cheerio that doesn't support wrapped attributes
+                .replace('[translate]=', '__translate__=');
+
+        $(template).find('[translate],[__translate__]').contents().filter(function() {
+            return this.nodeType === 3; // node type 3 = text node
+        }).each(function() {
+            let key,
+                $this = $(this),
+                element = $(this).parent(),
+                wrappedAttr = element.attr('__translate__'), // previously [translate]=
+                attr = element.attr('translate'); // translate=
+
+            // only support string values for now
+            if(wrappedAttr && wrappedAttr.match(/^['"].*['"]$/)) {
+                key = wrappedAttr.substr(1, wrappedAttr.length - 2);
+            } else if(attr) {
+                key = attr;
+            }
+
+            if(!key) {
+                key = $this.text().replace(/\\n/gi, '').trim();
+            }
+
+            if(key) {
+                results.push(key);
+            }
+        });
+
+        return results;
+    }
+
+    public process(contents: string): string[] {
+        const regExp = new RegExp(this.patterns.template, 'gi');
+
+        let results: string[] = [],
+            hasTemplate = false,
+            matches;
+
+        while(matches = regExp.exec(contents)) {
+            let content = matches[1]
+                .substr(1, matches[1].length - 2);
+
+            hasTemplate = true;
+            results = results.concat(this._parseTemplate(content));
+        }
+
+        if(!hasTemplate) {
+            this._parseTemplate(contents);
+        }
+
+        return results;
+    }
+
+}

src/parsers/pipe.parser.ts

@@ -1,11 +1,10 @@
 import { ParserInterface } from './parser.interface';
 
-export class HtmlParser implements ParserInterface {
+export class PipeParser implements ParserInterface {
 
 	public patterns = {
-		PipeSingleQuote: '\'((?:\\\\.|[^\'\\\\])*)\'\\s*\\|\\s*translate(:.*?)?',
-		PipeDoubleQuote: '"((?:\\\\.|[^"\\\\])*)"\\s*\\|\\s*translate(:.*?)?'
-	}
+		pipe: `(['"\`])([^\\1\\r\\n]*)\\1\\s+\\|\\s*translate(:.*?)?`
+	};
 
 	public process(contents: string): string[] {
 		let results: string[] = [];
@@ -15,7 +14,7 @@ export class HtmlParser implements ParserInterface {
 
 			let matches;
 			while (matches = regExp.exec(contents)) {
-				results.push(matches[1]);
+				results.push(matches[2]);
 			}
 		}
 

src/parsers/service.parser.ts

@@ -1,11 +1,10 @@
 import { ParserInterface } from './parser.interface';
 
-export class TypescriptParser implements ParserInterface {
+export class ServiceParser implements ParserInterface {
 
 	public patterns = {
-		TranslateServiceMethodsSingleQuote: '{{TRANSLATE_SERVICE}}\.(?:get|instant)\\s*\\\(\\s*\'((?:\\\\.|[^\'\\\\])*)\\s*\'',
-		TranslateServiceMethodsDoubleQuote: '{{TRANSLATE_SERVICE}}\.(?:get|instant)\\s*\\\(\\s*"((?:\\\\.|[^"\\\\])*)\\s*"',
-	}
+		translateServiceMethods: `{{TRANSLATE_SERVICE}}\.(?:get|instant)\\s*\\\(\\s*(['"\`])([^\\1\\r\\n]+)\\1`,
+	};
 
 	public process(contents: string): string[] {
 		let results: string[] = [];
@@ -22,7 +21,7 @@ export class TypescriptParser implements ParserInterface {
 
 			let matches;
 			while (matches = regExp.exec(contents)) {
-				results.push(matches[1]);
+				results.push(matches[2]);
 			}
 		}
 
@@ -49,7 +48,7 @@ export class TypescriptParser implements ParserInterface {
 	 * Extract name of TranslateService variable for use in patterns
 	 */
 	protected _extractTranslateServiceVar(contents: string): string {
-		const matches = contents.match(/([a-z0-9_]+)\s*:\s*TranslateService/i)
+		const matches = contents.match(/([a-z0-9_]+)\s*:\s*TranslateService/i);
 		if (matches === null) {
 			return '';
 		}