Merge pull request #103 from yukuku/fix-footnote-xref-binary-search

Fix binary search on footnotes and xref: arrays of arif should be tre…

AlkitabYes2/build.gradle

@@ -32,6 +32,7 @@ dependencies {
     implementation project(':BintexReader')
     implementation project(':BintexWriter')
     implementation project(':Snappy')
+    testImplementation 'junit:junit:4.12'
 }
 repositories {
     mavenCentral()

AlkitabYes2/src/main/java/yuku/alkitab/yes2/section/FootnotesSection.java

@@ -1,14 +1,13 @@
 package yuku.alkitab.yes2.section;
 
 import android.util.Log;
+import java.io.IOException;
 import yuku.alkitab.model.FootnoteEntry;
 import yuku.alkitab.yes2.io.RandomInputStream;
 import yuku.alkitab.yes2.section.base.SectionContent;
+import yuku.alkitab.yes2.util.UnsignedBinarySearchKt;
 import yuku.bintex.BintexReader;
 
-import java.io.IOException;
-import java.util.Arrays;
-
 // The writer is in another class (not here to save code amount for Alkitab app)
 // Section format:
 // {
@@ -56,7 +55,7 @@ public class FootnotesSection extends SectionContent {
 	}
 
 	public FootnoteEntry getFootnoteEntry(final int arif) {
-		final int pos = Arrays.binarySearch(index_arifs, arif);
+		final int pos = UnsignedBinarySearchKt.unsignedIntBinarySearch(index_arifs, arif);
 		if (pos < 0) {
 			return null;
 		}

AlkitabYes2/src/main/java/yuku/alkitab/yes2/section/XrefsSection.java

@@ -1,14 +1,13 @@
 package yuku.alkitab.yes2.section;
 
 import android.util.Log;
+import java.io.IOException;
 import yuku.alkitab.model.XrefEntry;
 import yuku.alkitab.yes2.io.RandomInputStream;
 import yuku.alkitab.yes2.section.base.SectionContent;
+import yuku.alkitab.yes2.util.UnsignedBinarySearchKt;
 import yuku.bintex.BintexReader;
 
-import java.io.IOException;
-import java.util.Arrays;
-
 // The writer is in another class (not here to save code amount for Alkitab app)
 // Section format:
 // {
@@ -56,7 +55,7 @@ public class XrefsSection extends SectionContent {
 	}
 
 	public XrefEntry getXrefEntry(final int arif) {
-		final int pos = Arrays.binarySearch(index_arifs, arif);
+		final int pos = UnsignedBinarySearchKt.unsignedIntBinarySearch(index_arifs, arif);
 		if (pos < 0) {
 			return null;
 		}

AlkitabYes2/src/main/java/yuku/alkitab/yes2/util/unsignedBinarySearch.kt

@@ -0,0 +1,30 @@
+package yuku.alkitab.yes2.util
+
+/**
+ * Treat [a] as a sorted array of unsigned ints and do binary search on it.
+ *
+ * @param key element to look for, treated as an unsigned int.
+ */
+fun unsignedIntBinarySearch(a: IntArray, key: Int): Int {
+    var low = 0
+    var high = a.size - 1
+
+    while (low <= high) {
+        val mid = (low + high) ushr 1
+        val midVal = a[mid]
+
+        val cmp = uintCompare(midVal, key)
+        if (cmp < 0) {
+            low = mid + 1
+        } else if (cmp > 0) {
+            high = mid - 1
+        } else {
+            // key found
+            return mid
+        }
+    }
+
+    return -(low + 1) // key not found.
+}
+
+private fun uintCompare(v1: Int, v2: Int): Int = (v1 xor Int.MIN_VALUE).compareTo(v2 xor Int.MIN_VALUE)

AlkitabYes2/src/test/java/yuku/alkitab/yes2/util/UnsignedBinarySearchKtTest.kt

@@ -0,0 +1,65 @@
+@file:OptIn(ExperimentalUnsignedTypes::class)
+
+package yuku.alkitab.yes2.util
+
+import org.junit.Assert.assertArrayEquals
+import org.junit.Assert.assertEquals
+import org.junit.Before
+import org.junit.Test
+
+class UnsignedBinarySearchKtTest {
+
+    private val allPositiveArray = IntArray(100) { it + 10 }
+    private val allNegativeArray = IntArray(100) { (0xc000_0000L + it.toLong()).toInt() }
+    private val mixedSignArray = IntArray(100) { it } + IntArray(100) { (0xc000_0000L + it.toLong()).toInt() }
+    private val skipArray = IntArray(100) { it * 2 } + IntArray(100) { (0xc000_0000L + (it * 2).toLong()).toInt() }
+
+
+    @Before
+    fun checkSorted() {
+        // all arrays under test must be sorted
+        fun assertSorted(a: IntArray) {
+            val u = UIntArray(a.size) { a[it].toUInt() }
+            assertArrayEquals(u.toTypedArray(), u.sorted().toTypedArray())
+        }
+
+        assertSorted(allPositiveArray)
+        assertSorted(allNegativeArray)
+        assertSorted(mixedSignArray)
+    }
+
+    @Test
+    fun unsignedBinarySearch() {
+        for (i in allPositiveArray.indices) {
+            assertEquals(i, unsignedIntBinarySearch(allPositiveArray, i + 10))
+        }
+        assertEquals(0.inv(), unsignedIntBinarySearch(allPositiveArray, 0))
+        assertEquals(100.inv(), unsignedIntBinarySearch(allPositiveArray, -1))
+        assertEquals(100.inv(), unsignedIntBinarySearch(allPositiveArray, -2))
+        assertEquals(100.inv(), unsignedIntBinarySearch(allPositiveArray, 400))
+
+        for (i in allNegativeArray.indices) {
+            assertEquals(i, unsignedIntBinarySearch(allNegativeArray, allNegativeArray[i]))
+        }
+        assertEquals(0.inv(), unsignedIntBinarySearch(allNegativeArray, 0))
+        assertEquals(100.inv(), unsignedIntBinarySearch(allNegativeArray, -1))
+        assertEquals(100.inv(), unsignedIntBinarySearch(allNegativeArray, -2))
+        assertEquals(0.inv(), unsignedIntBinarySearch(allNegativeArray, 400))
+
+        for (i in mixedSignArray.indices) {
+            assertEquals(i, unsignedIntBinarySearch(mixedSignArray, mixedSignArray[i]))
+        }
+        assertEquals(0, unsignedIntBinarySearch(mixedSignArray, 0))
+        assertEquals(200.inv(), unsignedIntBinarySearch(mixedSignArray, -1))
+        assertEquals(200.inv(), unsignedIntBinarySearch(mixedSignArray, -2))
+        // in the middle
+        assertEquals(100.inv(), unsignedIntBinarySearch(mixedSignArray, 400))
+
+        for (i in skipArray.indices) {
+            assertEquals(i, unsignedIntBinarySearch(skipArray, skipArray[i]))
+        }
+        assertEquals(0, unsignedIntBinarySearch(skipArray, 0))
+        assertEquals(2.inv(), unsignedIntBinarySearch(skipArray, 3))
+        assertEquals(103.inv(), unsignedIntBinarySearch(skipArray, (0xc000_0000L + 5L).toInt()))
+    }
+}