/**
* Test of extending Tesseract.
*
* @throws java.lang.Exception
*/
@Test
public void testExtendingTesseract() throws Exception {
logger.info("Extends Tesseract");
File imageFile = new File(this.testResourcesDataPath, "eurotext.tif");
String expResult = "The (quick) [brown] {fox} jumps!\nOver the $43,456.78 <lazy> #90 dog";
String[] expResults = expResult.split("\\s");
TessExtension instance1 = new TessExtension();
instance1.setDatapath(new File(datapath).getPath());
int pageIteratorLevel = TessPageIteratorLevel.RIL_WORD;
logger.info(
"PageIteratorLevel: "
+ Utils.getConstantName(pageIteratorLevel, TessPageIteratorLevel.class));
List<Word> result = instance1.getTextElements(imageFile, pageIteratorLevel);
// print the complete result
for (Word word : result) {
logger.info(word.toString());
}
List<String> text = new ArrayList<String>();
for (Word word : result.subList(0, expResults.length)) {
text.add(word.getText());
}
assertArrayEquals(expResults, text.toArray());
}
@Test
public void shouldGetSynonymsFromOtherSide() {
wordRepository.save(earth);
wordRepository.save(globe);
assertThat("globe's synonyms count", globe.getSynonymsCount(), is(0));
// as earth and globe both are under controll of springdata graph, no need to further call
// earth.save()
earth.synonymWith(globe, "the planet we live");
assertThat("globe's synonyms count", globe.getSynonymsCount(), is(equalTo(1)));
assertThat("globe's synonyms contain earth", globe.getSynonyms(), hasItem(earth));
}
@Test
public void shouldGetIdiomFromOtherSide() {
wordRepository.save(earth);
wordRepository.save(onEarth);
assertThat("onEarth's idiom count", onEarth.getIdiomsCount(), is(0));
// as earth and sky both are under controll of springdata graph, no need to further call
// earth.save()
earth.idiomWith(onEarth, "added a prefix 'on'");
assertThat("onEarth's idiom count", onEarth.getIdiomsCount(), is(equalTo(1)));
assertThat("onEarth's idiom contains earth", onEarth.getIdioms(), hasItem(earth));
}
@Test
public void shouldGetExtensionFromOtherSide() {
wordRepository.save(earth);
wordRepository.save(sky);
assertThat("sky's extensions count", sky.getExtensionsCount(), is(0));
// as earth and sky both are under controll of springdata graph, no need to further call
// earth.save()
earth.extendWith(sky, "earth and sky just intuitive");
assertThat("sky's extensions count", sky.getExtensionsCount(), is(equalTo(1)));
assertThat("sky's extensions contains sky", sky.getExtensions(), hasItem(earth));
}
@Test
public void shouldGetSynonymsFromSavingSide() {
wordRepository.save(earth);
assertThat("earth's synonyms count", earth.getSynonymsCount(), is(0));
wordRepository.save(globe);
// as earth and globe both are under controll of springdata graph, no need to further call
// earth.save()
earth.synonymWith(globe, "the planet we live");
assertThat("earth's synonyms count", earth.getSynonymsCount(), is(equalTo(1)));
assertThat("earth's synonym contains globe", earth.getSynonyms(), hasItem(globe));
Relationship relationship = earth.getSynonymRelationshipTo(globe);
assertThat("relationship should work properly", relationship.getWord(), is(earth));
assertThat("relationship should work properly", relationship.getAnotherWord(), is(globe));
assertThat(
"relationship should work properly", relationship.getOnEnglish(), is("the planet we live"));
}
@Test
public void shouldGetIdiomFromSavingSide() {
wordRepository.save(earth);
assertThat("earth's idioms count", earth.getIdiomsCount(), is(0));
wordRepository.save(onEarth);
// as earth and sky both are under controll of springdata graph, no need to further call
// earth.save()
earth.idiomWith(onEarth, "added a prefix 'on'");
assertThat("earth's idioms count", earth.getIdiomsCount(), is(equalTo(1)));
assertThat("earth's idioms contains onEarth", earth.getIdioms(), hasItem(onEarth));
Relationship relationship = earth.getIdiomRelationshipTo(onEarth);
assertThat("relationship should work properly", relationship.getWord(), is(earth));
assertThat("relationship should work properly", relationship.getAnotherWord(), is(onEarth));
assertThat(
"relationship should work properly",
relationship.getOnEnglish(),
is("added a prefix 'on'"));
}
@Test
public void shouldGetExtensionFromSavingSide() {
wordRepository.save(earth);
assertThat("earth's extensions count", earth.getExtensionsCount(), is(0));
wordRepository.save(sky);
// as earth and sky both are under controll of springdata graph, no need to further call
// earth.save()
earth.extendWith(sky, "earth and sky just intuitive");
assertThat("earth's extensions count", earth.getExtensionsCount(), is(equalTo(1)));
assertThat("earth's extensions contains sky", earth.getExtensions(), hasItem(sky));
Relationship relationship = earth.getExtensionRelationshipTo(sky);
assertThat("relationship should work properly", relationship.getWord(), is(earth));
assertThat("relationship should work properly", relationship.getAnotherWord(), is(sky));
assertThat(
"relationship should work properly",
relationship.getOnEnglish(),
is("earth and sky just intuitive"));
}
@Test
public void shouldGetAntonymFromOtherSide() {
wordRepository.save(happy);
wordRepository.save(sad);
wordRepository.save(earth);
assertThat("sad's antonym count", sad.getAntonymsCount(), is(0));
// as earth and sky both are under controll of springdata graph, no need to further call
// earth.save()
happy.antonymWith(sad, "feel good vs feel bad'");
assertThat("sad's antonym count", sad.getAntonymsCount(), is(equalTo(1)));
assertThat("sad's antonym contains happy", sad.getAntonyms(), hasItem(happy));
Relationship relationship = happy.getSynonymRelationshipTo(sad);
assertThat(
"relationship should work properly", relationship, is(CoreMatchers.<Object>nullValue()));
Set<Relationship> relationships = happy.getRelationshipsTo(earth);
assertThat(relationships.size(), is(0));
}
@Test
public void shouldGetAntonymFromSavingSide() {
wordRepository.save(happy);
assertThat("happy's antonym count", happy.getAntonymsCount(), is(0));
wordRepository.save(sad);
// as earth and sky both are under controll of springdata graph, no need to further call
// earth.save()
happy.antonymWith(sad, "feel good vs feel bad");
assertThat("happy's antonym count", happy.getAntonymsCount(), is(equalTo(1)));
assertThat("happy's antonym contains sad", happy.getAntonyms(), hasItem(sad));
Relationship relationship = happy.getAntonymRelationshipTo(sad);
assertThat("relationship should work properly", relationship.getWord(), is(happy));
assertThat("relationship should work properly", relationship.getAnotherWord(), is(sad));
assertThat(
"relationship should work properly",
relationship.getOnEnglish(),
is("feel good vs feel bad"));
Set<Relationship> relationships = happy.getRelationshipsTo(sad);
assertThat(relationships.size(), is(1));
assertThat(relationships, hasItem(relationship));
}
public PalindromeTest() {
ArrayList<String> words = new ArrayList<String>();
ArrayList<String> palis;
try {
for (String line : Files.readAllLines(Paths.get("com/jsanders/web2"))) {
words.add(line);
}
} catch (IOException ex) {
System.out.println("IO error");
System.exit(1);
}
palis = Palindrome.palindromes(words);
assertEquals(palis.size(), 161, 0.01);
int shortest = Word.shortestLength(words);
assertEquals(shortest, 1, 0.01);
int longest = Word.longestLength(words);
assertEquals(longest, 24, 0.01);
ArrayList<String> shortestWords = Word.shortestWords(words);
assertEquals(shortestWords.size(), 52, 0.01);
ArrayList<String> longestWords = Word.longestWords(words);
assertEquals(longestWords.size(), 5, 0.01);
int totalWords = Word.totalWords(words);
double avgLen = Word.averageLength(words);
assertEquals(totalWords, 235886, 0.01);
assertEquals(avgLen, 9.56, 0.01);
ArrayList<Double> letterFreq = Word.letterFrequency(words);
assertEquals(letterFreq.get(0), 0.087, 0.01);
double properFreq = Word.properFrequency(words);
assertEquals(properFreq, 0.106, 0.01);
ArrayList<Integer> startFreq = Word.startFrequency(words);
assertEquals(startFreq.get(0), 17096, 0.01);
ArrayList<String> sameStartEnd = Word.startEndWords(words);
assertEquals(sameStartEnd.size(), 11505, 0.01);
try {
PrintWriter f = new PrintWriter("short.txt");
for (String w : shortestWords) f.println(w);
f.close();
f = new PrintWriter("long.txt");
for (String w : longestWords) f.println(w);
f.close();
f = new PrintWriter("same.txt");
for (String w : sameStartEnd) f.println(w);
f.close();
f = new PrintWriter("statistics.txt");
f.println("avg word len: " + avgLen);
f.println("freq of letters: " + letterFreq);
f.println("freq of proper nouns/names: " + properFreq);
f.println("words that start with each letter:: " + startFreq);
f.close();
} catch (IOException ex) {
System.out.println("IO error");
System.exit(1);
}
}
@Test
public void shouldRetrieveTheSameWordJustSaved() {
Word retrievedWord = wordRepository.save(earth);
assertEquals("retrieved word match persisted one", earth, retrievedWord);
assertEquals("retrieved word name match ", "earth", retrievedWord.getName());
}