use our own bidi detector to drop 10Mb from gwt

This commit is contained in:
Athou
2024-09-13 15:54:45 +02:00
parent 19bcc2c0da
commit 31cf4d8bb2
5 changed files with 115 additions and 11 deletions

View File

@@ -447,11 +447,6 @@
<artifactId>urlcanon</artifactId>
<version>0.4.0</version>
</dependency>
<dependency>
<groupId>org.gwtproject</groupId>
<artifactId>gwt-servlet</artifactId>
<version>2.11.0</version>
</dependency>
<dependency>
<groupId>org.apache.httpcomponents.client5</groupId>
<artifactId>httpclient5</artifactId>

View File

@@ -16,11 +16,10 @@ import org.netpreserve.urlcanon.Canonicalizer;
import org.netpreserve.urlcanon.ParsedUrl;
import com.commafeed.backend.feed.FeedEntryKeyword.Mode;
import com.commafeed.backend.feed.parser.TextDirectionDetector;
import com.commafeed.backend.model.FeedEntry;
import com.commafeed.backend.model.FeedSubscription;
import com.commafeed.frontend.model.Entry;
import com.google.gwt.i18n.client.HasDirection.Direction;
import com.google.gwt.i18n.shared.BidiUtils;
import lombok.extern.slf4j.Slf4j;
@@ -109,8 +108,8 @@ public class FeedUtils {
return false;
}
Direction direction = BidiUtils.get().estimateDirection(text);
return direction == Direction.RTL;
TextDirectionDetector.Direction direction = TextDirectionDetector.detect(text);
return direction == TextDirectionDetector.Direction.RIGHT_TO_LEFT;
}
public static String removeTrailingSlash(String url) {

View File

@@ -0,0 +1,57 @@
package com.commafeed.backend.feed.parser;
import java.text.Bidi;
import java.util.concurrent.atomic.AtomicLong;
import java.util.regex.Pattern;
import org.apache.commons.lang3.math.NumberUtils;
public class TextDirectionDetector {
private static final Pattern WORDS_PATTERN = Pattern.compile("\\s+");
private static final Pattern URL_PATTERN = Pattern.compile("^https?://.*");
private static final double RTL_THRESHOLD = 0.4D;
public enum Direction {
LEFT_TO_RIGHT, RIGHT_TO_LEFT
}
public static Direction detect(String input) {
if (input == null || input.isBlank()) {
return Direction.LEFT_TO_RIGHT;
}
AtomicLong rtl = new AtomicLong();
AtomicLong total = new AtomicLong();
for (String token : WORDS_PATTERN.split(input)) {
// skip urls
if (URL_PATTERN.matcher(token).matches()) {
continue;
}
// skip numbers
if (NumberUtils.isCreatable(token)) {
continue;
}
boolean requiresBidi = Bidi.requiresBidi(token.toCharArray(), 0, token.length());
if (requiresBidi) {
Bidi bidi = new Bidi(token, Bidi.DIRECTION_DEFAULT_LEFT_TO_RIGHT);
if (bidi.getBaseLevel() == 1) {
rtl.incrementAndGet();
}
}
total.incrementAndGet();
}
if (total.longValue() == 0) {
return Direction.LEFT_TO_RIGHT;
}
double ratio = rtl.doubleValue() / total.doubleValue();
return ratio > RTL_THRESHOLD ? Direction.RIGHT_TO_LEFT : Direction.LEFT_TO_RIGHT;
}
}

View File

@@ -11,12 +11,12 @@ import org.mockito.Mock;
import org.mockito.Mockito;
import org.mockito.junit.jupiter.MockitoExtension;
import com.commafeed.backend.Digests;
import com.commafeed.backend.HttpGetter;
import com.commafeed.backend.HttpGetter.HttpResult;
import com.commafeed.backend.HttpGetter.NotModifiedException;
import com.commafeed.backend.feed.parser.FeedParser;
import com.commafeed.backend.urlprovider.FeedURLProvider;
import com.google.gwt.thirdparty.guava.common.hash.Hashing;
@ExtendWith(MockitoExtension.class)
class FeedFetcherTest {
@@ -43,7 +43,7 @@ class FeedFetcherTest {
String lastModified = "last-modified-1";
String etag = "etag-1";
byte[] content = "content".getBytes();
String lastContentHash = Hashing.sha1().hashBytes(content).toString();
String lastContentHash = Digests.sha1Hex(content);
Mockito.when(getter.getBinary(url, lastModified, etag))
.thenReturn(new HttpResult(content, "content-type", "last-modified-2", "etag-2", null));

View File

@@ -0,0 +1,53 @@
package com.commafeed.backend.feed.parser;
import org.junit.jupiter.api.Assertions;
import org.junit.jupiter.api.Test;
class TextDirectionDetectorTest {
@Test
public void testEstimateDirection() {
Assertions.assertEquals(TextDirectionDetector.Direction.LEFT_TO_RIGHT, TextDirectionDetector.detect(""));
Assertions.assertEquals(TextDirectionDetector.Direction.LEFT_TO_RIGHT, TextDirectionDetector.detect(" "));
Assertions.assertEquals(TextDirectionDetector.Direction.LEFT_TO_RIGHT, TextDirectionDetector.detect("! (...)"));
Assertions.assertEquals(TextDirectionDetector.Direction.LEFT_TO_RIGHT, TextDirectionDetector.detect("Pure Ascii content"));
Assertions.assertEquals(TextDirectionDetector.Direction.LEFT_TO_RIGHT, TextDirectionDetector.detect("-17.0%"));
Assertions.assertEquals(TextDirectionDetector.Direction.LEFT_TO_RIGHT, TextDirectionDetector.detect("http://foo/bar/"));
Assertions.assertEquals(TextDirectionDetector.Direction.LEFT_TO_RIGHT,
TextDirectionDetector.detect("http://foo/bar/?s=\u05d0\u05d0\u05d0\u05d0\u05d0\u05d0\u05d0\u05d0"
+ "\u05d0\u05d0\u05d0\u05d0\u05d0\u05d0\u05d0\u05d0\u05d0\u05d0\u05d0" + "\u05d0\u05d0\u05d0\u05d0\u05d0\u05d0"));
Assertions.assertEquals(TextDirectionDetector.Direction.RIGHT_TO_LEFT, TextDirectionDetector.detect("\u05d0"));
Assertions.assertEquals(TextDirectionDetector.Direction.RIGHT_TO_LEFT, TextDirectionDetector.detect("\u05d0"));
Assertions.assertEquals(TextDirectionDetector.Direction.RIGHT_TO_LEFT,
TextDirectionDetector.detect("http://foo/bar/ \u05d0 http://foo2/bar2/ http://foo3/bar3/"));
Assertions.assertEquals(TextDirectionDetector.Direction.RIGHT_TO_LEFT,
TextDirectionDetector.detect("\u05d0\u05d9\u05df \u05de\u05de\u05e9 " + "\u05de\u05d4 \u05dc\u05e8\u05d0\u05d5\u05ea: "
+ "\u05dc\u05d0 \u05e6\u05d9\u05dc\u05de\u05ea\u05d9 " + "\u05d4\u05e8\u05d1\u05d4 \u05d5\u05d2\u05dd \u05d0"
+ "\u05dd \u05d4\u05d9\u05d9\u05ea\u05d9 \u05de\u05e6\u05dc" + "\u05dd, \u05d4\u05d9\u05d4 \u05e9\u05dd"));
Assertions.assertEquals(TextDirectionDetector.Direction.RIGHT_TO_LEFT,
TextDirectionDetector.detect("\u05db\u05d0\u05df - http://geek.co.il/gallery/v/2007-06"
+ " - \u05d0\u05d9\u05df \u05de\u05de\u05e9 \u05de\u05d4 " + "\u05dc\u05e8\u05d0\u05d5\u05ea: \u05dc\u05d0 \u05e6"
+ "\u05d9\u05dc\u05de\u05ea\u05d9 \u05d4\u05e8\u05d1\u05d4 "
+ "\u05d5\u05d2\u05dd \u05d0\u05dd \u05d4\u05d9\u05d9\u05ea"
+ "\u05d9 \u05de\u05e6\u05dc\u05dd, \u05d4\u05d9\u05d4 "
+ "\u05e9\u05dd \u05d1\u05e2\u05d9\u05e7\u05e8 \u05d4\u05e8" + "\u05d1\u05d4 \u05d0\u05e0\u05e9\u05d9\u05dd. \u05de"
+ "\u05d4 \u05e9\u05db\u05df - \u05d0\u05e4\u05e9\u05e8 " + "\u05dc\u05e0\u05e6\u05dc \u05d0\u05ea \u05d4\u05d4 "
+ "\u05d3\u05d6\u05de\u05e0\u05d5\u05ea \u05dc\u05d4\u05e1" + "\u05ea\u05db\u05dc \u05e2\u05dc \u05db\u05de\u05d4 "
+ "\u05ea\u05de\u05d5\u05e0\u05d5\u05ea \u05de\u05e9\u05e2"
+ "\u05e9\u05e2\u05d5\u05ea \u05d9\u05e9\u05e0\u05d5\u05ea " + "\u05d9\u05d5\u05ea\u05e8 \u05e9\u05d9\u05e9 \u05dc"
+ "\u05d9 \u05d1\u05d0\u05ea\u05e8"));
Assertions.assertEquals(TextDirectionDetector.Direction.RIGHT_TO_LEFT,
TextDirectionDetector.detect("CAPTCHA \u05de\u05e9\u05d5\u05db\u05dc\u05dc " + "\u05de\u05d3\u05d9?"));
Assertions.assertEquals(TextDirectionDetector.Direction.RIGHT_TO_LEFT,
TextDirectionDetector.detect("Yes Prime Minister \u05e2\u05d3\u05db\u05d5\u05df. "
+ "\u05e9\u05d0\u05dc\u05d5 \u05d0\u05d5\u05ea\u05d9 " + "\u05de\u05d4 \u05d0\u05e0\u05d9 \u05e8\u05d5\u05e6"
+ "\u05d4 \u05de\u05ea\u05e0\u05d4 \u05dc\u05d7\u05d2"));
Assertions.assertEquals(TextDirectionDetector.Direction.RIGHT_TO_LEFT, TextDirectionDetector
.detect("17.4.02 \u05e9\u05e2\u05d4:13-20 .15-00 .\u05dc\u05d0 " + "\u05d4\u05d9\u05d9\u05ea\u05d9 \u05db\u05d0\u05df."));
Assertions.assertEquals(TextDirectionDetector.Direction.RIGHT_TO_LEFT,
TextDirectionDetector.detect("5710 5720 5730. \u05d4\u05d3\u05dc\u05ea. " + "\u05d4\u05e0\u05e9\u05d9\u05e7\u05d4"));
Assertions.assertEquals(TextDirectionDetector.Direction.RIGHT_TO_LEFT,
TextDirectionDetector.detect("\u05d4\u05d3\u05dc\u05ea http://www.google.com " + "http://www.gmail.com"));
}
}