CATROID-762 - Regex. Assistant Output

-add Regex creator from given keyword in html text
-add Unit tests

Author: UneclStivun

catroid/src/main/java/org/catrobat/catroid/utils/HtmlRegexExtractor.java

@@ -41,7 +41,9 @@ public HtmlRegexExtractor(Context context) {
 	}
 
 	public void searchKeyword(String keyword, String text) {
-		if (findKeyword(keyword, text) == null) {
+		String keywordFound = findKeyword(keyword, text);
+		String regexFound = htmlToRegexConverter(keywordFound, text);
+		if (regexFound == null) {
 			showError();
 		} else {
 			showSuccess();
@@ -60,7 +62,7 @@ private void showError() {
 	@VisibleForTesting
 	public String findKeyword(String keyword, String text) {
 		if (keyword.equals("")) {
-			throw new IllegalArgumentException("No empty keywords allowed");
+			return null;
 		}
 		if (text.indexOf(keyword) >= 0) {
 			return keyword;
@@ -93,4 +95,51 @@ private String findShortestOccurrenceInText(String regex, String text) {
 		}
 		return shortestOccurrence;
 	}
+
+	public String htmlToRegexConverter(String keyword, String htmlText) {
+		int keywordIndex;
+		String regex;
+
+		if (keyword != null) {
+			keywordIndex = htmlText.indexOf(keyword);
+			if (keyword.equals(htmlText)) {
+				regex = "(.*)";
+			} else {
+				regex = "(.*)";
+				int distance = 0;
+				do {
+					distance++;
+
+					String beforeKeyword = "";
+					int beforeKeywordIndex = keywordIndex - distance;
+					if (beforeKeywordIndex >= 0) {
+						beforeKeyword = String.valueOf(htmlText.charAt(beforeKeywordIndex));
+					}
+
+					String afterKeyword = "";
+					int afterKeywordIndex = keywordIndex + keyword.length() + distance - 1;
+					if (afterKeywordIndex < htmlText.length()) {
+						afterKeyword =
+								String.valueOf(htmlText.charAt(afterKeywordIndex));
+					}
+
+					regex = beforeKeyword + regex + afterKeyword;
+				} while (!matchesUniquely(regex, htmlText, keyword));
+			}
+		} else {
+			regex = null;
+		}
+		return regex;
+	}
+	private boolean matchesUniquely(String pattern, String text, String expectedMatch) {
+		int counter = 0;
+		Matcher matcher = Pattern.compile(pattern).matcher(text);
+
+		String matched = null;
+		while (matcher.find()) {
+			matched = matcher.group(1);
+			counter++;
+		}
+		return counter == 1 && expectedMatch.equals(matched);
+	}
 }

catroid/src/test/java/org/catrobat/catroid/test/ui/regexassistant/HtmlRegexExtractorTest.java

@@ -41,9 +41,9 @@ public void setUp() {
 	}
 	private HtmlRegexExtractor htmlExtractor;
 
-	@Test(expected = IllegalArgumentException.class)
+	@Test
 	public void testFindKeywordWithEmptyKeyword() {
-		htmlExtractor.findKeyword("", "abc");
+		assertNull(htmlExtractor.findKeyword("", "abc"));
 	}
 
 	@Test
@@ -62,12 +62,6 @@ public void testFindKeywordInSentence() {
 				+ "statement in long text"));
 	}
 
-	@Test
-	public void testFindKeywordInSentenceWithSpaceInKeyword() {
-		assertEquals("ab c", htmlExtractor.findKeyword("ab c", "Wer are looking for the "
-				+ "ab c statement in long text"));
-	}
-
 	@Test
 	public void testFindKeywordInSentenceWithSpaceInKeywordAndTextWithNBSP() {
 		assertEquals("ab c", htmlExtractor.findKeyword("ab c", "Wer are looking for "
@@ -108,4 +102,53 @@ public void testFindKeywordSmallestOccurrence() {
 		assertEquals("Hello World", htmlExtractor.findKeyword("Hello World",
 				"Hello Banana Animal Text Hello World Ape"));
 	}
+
+	@Test
+	public void testCreateRegexWithOneCharContext() {
+		assertEquals("b(.*)e", htmlExtractor.htmlToRegexConverter("cd", "abcdefg"));
+	}
+
+	@Test
+	public void testCreateRegexWithKeywordAtStart() {
+		assertEquals("(.*)c", htmlExtractor.htmlToRegexConverter("ab", "abcdefg"));
+	}
+
+	@Test
+	public void testCreateRegexWithKeywordAtEnd() {
+		assertEquals("d(.*)", htmlExtractor.htmlToRegexConverter("efg", "abcdefg"));
+	}
+
+	@Test
+	public void testCreateRegexWithDuplicateKeywordFirstOccurrence1CharContext() {
+		assertEquals("ab(.*)defga", htmlExtractor.htmlToRegexConverter("KEY", "abKEYdefgadKEYdefg"));
+	}
+
+	@Test
+	public void testCreateRegexWith2CharContext() {
+		assertEquals("yb(.*)de", htmlExtractor.htmlToRegexConverter("KEY", "abcdefg ybKEYdefg"));
+	}
+
+	@Test
+	public void testCreateRegexWhereKeywordEqualsHtmlText() {
+		assertEquals("(.*)", htmlExtractor.htmlToRegexConverter("abcdefg", "abcdefg"));
+	}
+
+	@Test
+	public void testCreateRegexPostfixInKeyword() {
+		assertEquals("(.*)b", htmlExtractor.htmlToRegexConverter("abc", "abcbc"));
+	}
+
+	@Test
+	public void testCreateRegexOutOfBoundsAfter2CharContext() {
+		assertEquals("b(.*)ba", htmlExtractor.htmlToRegexConverter("abc", "babcbabcb"));
+	}
+	@Test
+	public void testFirstKeyBordersOnSecondKey() {
+		assertEquals("baaaa(.*)aaaaK", htmlExtractor.htmlToRegexConverter("KEY",
+				"baaaaKEYaaaaKEYaaaa"));
+	}
+	@Test
+	public void testCreateRegexWhereTextOnlyKeywords() {
+		assertEquals("(.*)aa", htmlExtractor.htmlToRegexConverter("a", "aaa"));
+	}
 }