From 9c93fa6bb39b54f70c8e0f15a62bfe89bb578a96 Mon Sep 17 00:00:00 2001
From: Andrew Valencik <andrew.valencik@gmail.com>
Date: Mon, 18 Nov 2024 13:56:41 -0500
Subject: [PATCH] Add FragmentFormatter

---
 .../highlight/FragmentFormatter.scala         | 51 +++++++++++++++++
 .../highlight/FragmentFormatterSuite.scala    | 55 +++++++++++++++++++
 2 files changed, 106 insertions(+)
 create mode 100644 core/src/main/scala/pink/cozydev/protosearch/highlight/FragmentFormatter.scala
 create mode 100644 core/src/test/scala/pink/cozydev/protosearch/highlight/FragmentFormatterSuite.scala

diff --git a/core/src/main/scala/pink/cozydev/protosearch/highlight/FragmentFormatter.scala b/core/src/main/scala/pink/cozydev/protosearch/highlight/FragmentFormatter.scala
new file mode 100644
index 00000000..408d873e
--- /dev/null
+++ b/core/src/main/scala/pink/cozydev/protosearch/highlight/FragmentFormatter.scala
@@ -0,0 +1,51 @@
+package pink.cozydev.protosearch.highlight
+
+case class FragmentFormatter(
+    maxSize: Int,
+    startTag: String,
+    endTag: String,
+) {
+  private val tagSize = startTag.size + endTag.size
+  private val tags: Array[String] = Array(startTag, endTag)
+
+  def format(fragment: String, offsets: Iterable[Int]): String =
+    if (offsets.size == 0) fragment
+    else {
+      val offsetArr = offsets.toArray
+      assert(offsetArr.size % 2 == 0, "even number of offsets required")
+
+      val sb = new StringBuilder()
+      sb.sizeHint(fragment.size + tagSize * (offsets.size / 2))
+      val chars = fragment.toCharArray()
+
+      try {
+        // Add initial characters
+        sb.appendAll(chars, 0, offsetArr(0))
+        var charsOffset = offsetArr(0)
+
+        // Loop through offsets, two at a time
+        var i = 0
+        while (i < offsetArr.size) {
+          val offsetStart = offsetArr(i)
+          // add chars between offsets
+          val inbetweenChars = offsetStart - charsOffset
+          sb.appendAll(chars, charsOffset, inbetweenChars)
+          charsOffset += inbetweenChars
+
+          // start new offset
+          sb.append(startTag)
+          val offsetLength = offsetArr(i + 1)
+          sb.appendAll(chars, charsOffset, offsetLength)
+          sb.append(endTag)
+          charsOffset += offsetLength
+          i += 2
+        }
+        // Add remaining characters
+        sb.appendAll(chars, charsOffset, chars.size - charsOffset)
+        sb.result()
+      } catch {
+        case _: IndexOutOfBoundsException =>
+          throw new IllegalArgumentException("Offset exceeded string length")
+      }
+    }
+}
diff --git a/core/src/test/scala/pink/cozydev/protosearch/highlight/FragmentFormatterSuite.scala b/core/src/test/scala/pink/cozydev/protosearch/highlight/FragmentFormatterSuite.scala
new file mode 100644
index 00000000..d5ac0f56
--- /dev/null
+++ b/core/src/test/scala/pink/cozydev/protosearch/highlight/FragmentFormatterSuite.scala
@@ -0,0 +1,55 @@
+package pink.cozydev.protosearch.highlight
+
+import munit.FunSuite
+
+class FragmentFormatterSuite extends FunSuite {
+  val formatter = FragmentFormatter(100, "<b>", "</b>")
+
+  test("formats string with empty offsets") {
+    val s = "hello world"
+    val actual = formatter.format(s, List.empty)
+    val expected = s
+    assertEquals(actual, expected)
+  }
+
+  test("formats string with one pair of offsets") {
+    val s = "hello world"
+    val actual = formatter.format(s, List(6, 5))
+    val expected = "hello <b>world</b>"
+    assertEquals(actual, expected)
+  }
+
+  test("formats string with offset in the middle") {
+    val s = "hello world, how are you?"
+    val actual = formatter.format(s, List(6, 5))
+    val expected = "hello <b>world</b>, how are you?"
+    assertEquals(actual, expected)
+  }
+
+  test("formats string with two pair of offsets") {
+    val s = "hello world"
+    val actual = formatter.format(s, List(0, 5, 6, 5))
+    val expected = "<b>hello</b> <b>world</b>"
+    assertEquals(actual, expected)
+  }
+
+  test("formats whole string if one offset") {
+    val s = "hello world"
+    val actual = formatter.format(s, List(0, 11))
+    val expected = "<b>hello world</b>"
+    assertEquals(actual, expected)
+  }
+
+  test("throws if offset length exceeds string boundary") {
+    val s = "hello world"
+    intercept[java.lang.IllegalArgumentException] {
+      formatter.format(s, List(6, 5 + 1))
+    }
+  }
+
+  test("throws if odd number of offset integers") {
+    intercept[java.lang.AssertionError] {
+      formatter.format("", List(1))
+    }
+  }
+}