Rewrite displayLocationInfo to avoid findNextWord

Simplifies and fixes counting words, especially when selection intersects words. Also moves implementation to engine and adds more tests.

src/main/java/com/maddyhome/idea/vim/group/FileGroup.kt

@@ -32,17 +32,13 @@ import com.maddyhome.idea.vim.api.ExecutionContext
 import com.maddyhome.idea.vim.api.VimEditor
 import com.maddyhome.idea.vim.api.VimFileBase
 import com.maddyhome.idea.vim.api.injector
-import com.maddyhome.idea.vim.common.TextRange
 import com.maddyhome.idea.vim.group.LastTabService.Companion.getInstance
 import com.maddyhome.idea.vim.helper.EditorHelper
 import com.maddyhome.idea.vim.helper.MessageHelper.message
-import com.maddyhome.idea.vim.helper.countWords
-import com.maddyhome.idea.vim.helper.fileSize
 import com.maddyhome.idea.vim.newapi.IjEditorExecutionContext
 import com.maddyhome.idea.vim.newapi.IjVimEditor
 import com.maddyhome.idea.vim.newapi.execute
 import com.maddyhome.idea.vim.newapi.globalIjOptions
-import com.maddyhome.idea.vim.state.mode.Mode.VISUAL
 import java.io.File
 import java.util.*
 
@@ -265,92 +261,6 @@ class FileGroup : VimFileBase() {
     return null
   }
 
-  override fun displayLocationInfo(vimEditor: VimEditor) {
-    val editor = (vimEditor as IjVimEditor).editor
-    val msg = StringBuilder()
-    val doc = editor.document
-
-    if (injector.vimState.mode !is VISUAL) {
-      val lp = editor.caretModel.logicalPosition
-      val col = editor.caretModel.offset - doc.getLineStartOffset(lp.line)
-      var endoff = doc.getLineEndOffset(lp.line)
-      if (endoff < editor.fileSize && doc.charsSequence[endoff] == '\n') {
-        endoff--
-      }
-      val ecol = endoff - doc.getLineStartOffset(lp.line)
-      val elp = editor.offsetToLogicalPosition(endoff)
-
-      msg.append("Col ").append(col + 1)
-      if (col != lp.column) {
-        msg.append("-").append(lp.column + 1)
-      }
-
-      msg.append(" of ").append(ecol + 1)
-      if (ecol != elp.column) {
-        msg.append("-").append(elp.column + 1)
-      }
-
-      val lline = editor.caretModel.logicalPosition.line
-      val total = IjVimEditor(editor).lineCount()
-
-      msg.append("; Line ").append(lline + 1).append(" of ").append(total)
-
-      val cp = countWords(vimEditor)
-
-      msg.append("; Word ").append(cp.position).append(" of ").append(cp.count)
-
-      val offset = editor.caretModel.offset
-      val size = editor.fileSize
-
-      msg.append("; Character ").append(offset + 1).append(" of ").append(size)
-    } else {
-      msg.append("Selected ")
-
-      val vr = TextRange(
-        editor.selectionModel.blockSelectionStarts,
-        editor.selectionModel.blockSelectionEnds
-      )
-      vr.normalize()
-
-      val lines: Int
-      var cp = countWords(vimEditor)
-      val words = cp.count
-      var word = 0
-      if (vr.isMultiple) {
-        lines = vr.size()
-        val cols = vr.maxLength
-
-        msg.append(cols).append(" Cols; ")
-
-        for (i in 0 until vr.size()) {
-          cp = countWords(vimEditor, vr.startOffsets[i], (vr.endOffsets[i] - 1).toLong())
-          word += cp.count
-        }
-      } else {
-        val slp = editor.offsetToLogicalPosition(vr.startOffset)
-        val elp = editor.offsetToLogicalPosition(vr.endOffset)
-
-        lines = elp.line - slp.line + 1
-
-        cp = countWords(vimEditor, vr.startOffset, (vr.endOffset - 1).toLong())
-        word = cp.count
-      }
-
-      val total = IjVimEditor(editor).lineCount()
-
-      msg.append(lines).append(" of ").append(total).append(" Lines")
-
-      msg.append("; ").append(word).append(" of ").append(words).append(" Words")
-
-      val chars = vr.selectionCount
-      val size = editor.fileSize
-
-      msg.append("; ").append(chars).append(" of ").append(size).append(" Characters")
-    }
-
-    VimPlugin.showMessage(msg.toString())
-  }
-
   override fun displayFileInfo(vimEditor: VimEditor, fullPath: Boolean) {
     val editor = (vimEditor as IjVimEditor).editor
     val msg = StringBuilder()

src/main/java/com/maddyhome/idea/vim/helper/SearchHelperKt.kt

@@ -10,11 +10,9 @@ package com.maddyhome.idea.vim.helper
 
 import com.intellij.codeInsight.daemon.impl.DaemonCodeAnalyzerEx
 import com.intellij.codeInsight.daemon.impl.HighlightInfo
-import com.intellij.openapi.diagnostic.logger
 import com.intellij.openapi.editor.Caret
 import com.intellij.openapi.editor.Editor
 import com.intellij.spellchecker.SpellCheckerSeveritiesProvider
-import com.maddyhome.idea.vim.api.VimEditor
 import com.maddyhome.idea.vim.api.getLineEndOffset
 import com.maddyhome.idea.vim.api.globalOptions
 import com.maddyhome.idea.vim.api.injector
@@ -52,48 +50,6 @@ private fun containsUpperCase(pattern: String): Boolean {
   return false
 }
 
-/**
- * This counts all the words in the file.
- */
-fun countWords(
-  vimEditor: VimEditor,
-  start: Int = 0,
-  end: Long = vimEditor.fileSize(),
-): CountPosition {
-  val offset = vimEditor.currentCaret().offset
-
-  var count = 1
-  var position = 0
-  var last = -1
-  var res = start
-  while (true) {
-    res = injector.searchHelper.findNextWord(vimEditor, res, 1, true, false)
-    if (res == start || res == 0 || res > end || res == last) {
-      break
-    }
-
-    count++
-
-    if (res == offset) {
-      position = count
-    } else if (last < offset && res >= offset) {
-      position = if (count == 2) {
-        1
-      } else {
-        count - 1
-      }
-    }
-
-    last = res
-  }
-
-  if (position == 0 && res == offset) {
-    position = count
-  }
-
-  return CountPosition(count, position)
-}
-
 /**
  * Find the word under the cursor or the next word to the right of the cursor on the current line.
  *
@@ -193,9 +149,3 @@ private fun skip(iterator: IntIterator, n: Int) {
   var i = n
   while (i-- != 0 && iterator.hasNext()) iterator.nextInt()
 }
-
-class CountPosition(val count: Int, val position: Int)
-
-private val logger = logger<SearchLogger>()
-
-private class SearchLogger

src/test/java/org/jetbrains/plugins/ideavim/action/FileGetLocationInfoActionTest.kt

@@ -9,80 +9,231 @@
 package org.jetbrains.plugins.ideavim.action
 
 import com.maddyhome.idea.vim.VimPlugin
-import com.maddyhome.idea.vim.api.injector
-import org.jetbrains.plugins.ideavim.VimBehaviorDiffers
 import org.jetbrains.plugins.ideavim.VimTestCase
 import org.junit.jupiter.api.Test
+import kotlin.test.assertEquals
 
+// For all of these tests, note that Vim might show a different total byte count - off by one. This is ok, and not worth
+// adding a VimBehaviorDiffers annotation for.
+// It's because Vim requires each line to end with a linefeed character (otherwise it's not a line!) and adds one to
+// the last line. If the last line ends with a linefeed, that's just the end of the line. In this case, Vim does not
+// draw an empty line after the last line (because there isn't one!). If we hit enter at the end of the last line, Vim
+// adds a second linefeed, and there's now a new (empty) line at the end of the file, and the file ends with two
+// linefeed characters.
+// IntelliJ treats a linefeed at the end of the last line as a line feed, and draws an empty line. When we initialise
+// a test with a trailing empty line, IntelliJ only creates one linefeed char, instead of the two that Vim creates.
+// Maybe we should ensure that each file ends with a linefeed when initialising tests?
+@Suppress("SpellCheckingInspection")
 class FileGetLocationInfoActionTest : VimTestCase() {
-  @VimBehaviorDiffers(originalVimAfter = "Col 1 of 11; Line 1 of 6; Word 1 of 32; Byte 1 of 166")
   @Test
   fun `test get file info`() {
-    val keys = injector.parser.parseKeys("g<C-G>")
     val before = """
-            ${c}Lorem Ipsum
+      |${c}Lorem Ipsum
-
+      |
-            Lorem ipsum dolor sit amet,
+      |Lorem ipsum dolor sit amet,
-            consectetur adipiscing elit
+      |consectetur adipiscing elit
-            Sed in orci mauris.
+      |Sed in orci mauris.
-            Cras id tellus in ex imperdiet egestas.
+      |Cras id tellus in ex imperdiet egestas.
-    """.trimIndent()
+    """.trimMargin()
     configureByText(before)
-    typeText(keys)
+    typeText("g<C-G>")
-    kotlin.test.assertEquals("Col 1 of 11; Line 1 of 6; Word 1 of 23; Character 1 of 128", VimPlugin.getMessage())
+    assertEquals("Col 1 of 11; Line 1 of 6; Word 1 of 21; Byte 1 of 128", VimPlugin.getMessage())
   }
 
-  @VimBehaviorDiffers(originalVimAfter = "Col 1 of 11; Line 1 of 7; Word 1 of 32; Byte 1 of 167")
   @Test
-  fun `test get file info with empty line`() {
+  fun `test get file info with single word`() {
-    val keys = injector.parser.parseKeys("g<C-G>")
+    configureByText("Lorem")
-    val before = """
+    typeText("g<C-G>")
-            ${c}Lorem Ipsum
+    assertEquals("Col 1 of 5; Line 1 of 1; Word 1 of 1; Byte 1 of 5", VimPlugin.getMessage())
-
+  }
-            Lorem ipsum dolor sit amet,
+
-            consectetur adipiscing elit
+  @Test
-            Sed in orci mauris.
+  fun `test get file info of two separate words`() {
-            Cras id tellus in ex imperdiet egestas.
+    configureByText("Lorem ipsum")
-            
+    typeText("g<C-G>")
-    """.trimIndent()
+    assertEquals("Col 1 of 11; Line 1 of 1; Word 1 of 2; Byte 1 of 11", VimPlugin.getMessage())
-    configureByText(before)
+  }
-    typeText(keys)
+
-    kotlin.test.assertEquals("Col 1 of 11; Line 1 of 7; Word 1 of 24; Character 1 of 129", VimPlugin.getMessage())
+  @Test
+  fun `test get file info of one WORD containing non-word characters`() {
+    configureByText("Lorem,,,,,ipsum")
+    typeText("g<C-G>")
+    assertEquals("Col 1 of 15; Line 1 of 1; Word 1 of 1; Byte 1 of 15", VimPlugin.getMessage())
+  }
+
+  @Test
+  fun `test get file info of words on multiple lines`() {
+    val before = """
+      |Lorem ipsum dolor sit amet
+      |cons${c}ectetur adipiscing elit
+    """.trimMargin()
+    configureByText(before)
+    typeText("g<C-G>")
+    assertEquals("Col 5 of 27; Line 2 of 2; Word 6 of 8; Byte 32 of 54", VimPlugin.getMessage())
+  }
+
+  @Test
+  fun `test get file info of words with trailing punctuation on multiple lines`() {
+    val before = """
+      |Lorem ipsum dolor sit amet,
+      |cons${c}ectetur adipiscing elit
+    """.trimMargin()
+    configureByText(before)
+    typeText("g<C-G>")
+    assertEquals("Col 5 of 27; Line 2 of 2; Word 6 of 8; Byte 33 of 55", VimPlugin.getMessage())
+  }
+
+  @Test
+  fun `test get file info of words with empty lines`() {
+    val before = """
+      |Lorem ipsum dolor sit amet,
+      |
+      |
+      |${c}
+      |consectetur adipiscing elit
+    """.trimMargin()
+    configureByText(before)
+    typeText("g<C-G>")
+    assertEquals("Col 1 of 0; Line 4 of 5; Word 5 of 8; Byte 31 of 58", VimPlugin.getMessage())
+  }
+
+  @Test
+  fun `test get file info on empty line shows zero columns`() {
+    val before = """
+      |Lorem Ipsum
+      |${c}
+      |Lorem ipsum dolor sit amet,
+      |consectetur adipiscing elit
+      |Sed in orci mauris.
+      |Cras id tellus in ex imperdiet egestas.
+      |
+    """.trimMargin()
+    configureByText(before)
+    typeText("g<C-G>")
+    assertEquals("Col 1 of 0; Line 2 of 7; Word 2 of 21; Byte 13 of 129", VimPlugin.getMessage())
   }
 
-  @VimBehaviorDiffers(originalVimAfter = "Col 1 of 40; Line 4 of 7; Word 12 of 32; Byte 55 of 167")
   @Test
   fun `test get file info in the middle`() {
-    val keys = injector.parser.parseKeys("g<C-G>")
     val before = """
-            Lorem Ipsum
+      |Lorem Ipsum
-
+      |
-            Lorem ipsum dolor sit amet,
+      |Lorem ipsum dolor sit amet,
-            all rocks ${c}and lavender and tufted grass,
+      |consectetur ${c}adipiscing elit
-            Sed in orci mauris.
+      |Sed in orci mauris.
-            Cras id tellus in ex imperdiet egestas.
+      |Cras id tellus in ex imperdiet egestas.
-            
+      |
-    """.trimIndent()
+    """.trimMargin()
     configureByText(before)
-    typeText(keys)
+    typeText("g<C-G>")
-    kotlin.test.assertEquals("Col 11 of 40; Line 4 of 7; Word 11 of 28; Character 52 of 142", VimPlugin.getMessage())
+    assertEquals("Col 13 of 27; Line 4 of 7; Word 9 of 21; Byte 54 of 129", VimPlugin.getMessage())
   }
 
-  @VimBehaviorDiffers(originalVimAfter = "Col 1 of 0; Line 7 of 7; Word 32 of 32; Byte 167 of 167")
   @Test
   fun `test get file info on the last line`() {
-    val keys = injector.parser.parseKeys("g<C-G>")
     val before = """
-            Lorem Ipsum
+      |Lorem Ipsum
-
+      |
-            Lorem ipsum dolor sit amet,
+      |Lorem ipsum dolor sit amet,
-            consectetur adipiscing elit
+      |consectetur adipiscing elit
-            Sed in orci mauris.
+      |Sed in orci mauris.
-            Cras id tellus in ex imperdiet egestas.
+      |Cras id tellus in ex imperdiet egestas.
-            $c
+      |$c
-    """.trimIndent()
+    """.trimMargin()
     configureByText(before)
-    typeText(keys)
+    typeText("g<C-G>")
-    kotlin.test.assertEquals("Col 1 of 1; Line 7 of 7; Word 24 of 24; Character 130 of 129", VimPlugin.getMessage())
+    assertEquals("Col 1 of 0; Line 7 of 7; Word 21 of 21; Byte 130 of 129", VimPlugin.getMessage())
+  }
+
+  @Test
+  fun `test get file info with single word selected`() {
+    val before = """
+      |Lorem ${c}ipsum dolor sit amet
+      |consectetur adipiscing elit
+    """.trimMargin()
+    configureByText(before)
+    typeText("ve", "g<C-G>")
+    assertEquals("Selected 1 of 2 Lines; 1 of 8 Words; 5 of 54 Bytes", VimPlugin.getMessage())
+  }
+
+  @Test
+  fun `test get file info with multiple words selected`() {
+    val before = """
+      |Lorem ${c}ipsum dolor sit amet
+      |consectetur adipiscing elit
+    """.trimMargin()
+    configureByText(before)
+    typeText("v2e", "g<C-G>")
+    assertEquals("Selected 1 of 2 Lines; 2 of 8 Words; 11 of 54 Bytes", VimPlugin.getMessage())
+  }
+
+  @Test
+  fun `test get file info with single WORD selected`() {
+    val before = """
+      |Lorem ${c}ipsum,,,,dolor sit amet
+      |consectetur adipiscing elit
+    """.trimMargin()
+    configureByText(before)
+    typeText("vE", "g<C-G>")
+    assertEquals("Selected 1 of 2 Lines; 1 of 7 Words; 14 of 57 Bytes", VimPlugin.getMessage())
+  }
+
+  @Test
+  fun `test get file info across multiple selected lines`() {
+    val before = """
+      |Lorem Ipsum
+      |
+      |Lorem ${c}ipsum dolor sit amet,
+      |consectetur adipiscing elit
+      |Sed in orci mauris.
+      |Cras id tellus in ex imperdiet egestas.
+    """.trimMargin()
+    configureByText(before)
+    typeText("v", "jj", "g<C-G>")
+    assertEquals("Selected 3 of 6 Lines; 9 of 21 Words; 57 of 128 Bytes", VimPlugin.getMessage())
+  }
+
+  @Test
+  fun `test get file info with empty lines selected`() {
+    val before = """
+      |Lorem ${c}ipsum dolor sit amet,
+      |
+      |
+      |
+      |consectetur adipiscing elit
+    """.trimMargin()
+    configureByText(before)
+    typeText("vG", "g<C-G>")
+    assertEquals("Selected 5 of 5 Lines; 5 of 8 Words; 26 of 58 Bytes", VimPlugin.getMessage())
+  }
+
+  @Test
+  fun `test get file info with linewise selection`() {
+    val before = """
+      |Lorem Ipsum
+      |
+      |Lorem ${c}ipsum dolor sit amet,
+      |consectetur adipiscing elit
+      |Sed in orci mauris.
+      |Cras id tellus in ex imperdiet egestas.
+    """.trimMargin()
+    configureByText(before)
+    typeText("V", "jj", "g<C-G>")
+    assertEquals("Selected 3 of 6 Lines; 12 of 21 Words; 76 of 128 Bytes", VimPlugin.getMessage())
+  }
+
+  @Test
+  fun `test get file info with blockwise selection`() {
+    val before = """
+      |Lorem Ipsum
+      |
+      |Lorem ${c}ipsum dolor sit amet,
+      |consectetur adipiscing elit
+      |Sed in orci mauris.
+      |Cras id tellus in ex imperdiet egestas.
+    """.trimMargin()
+    configureByText(before)
+    typeText("<C-V>", "jjj", "llll", "g<C-G>")
+    assertEquals("Selected 5 Cols; 4 of 6 Lines; 5 of 21 Words; 20 of 128 Bytes", VimPlugin.getMessage())
   }
 }

vim-engine/src/main/kotlin/com/maddyhome/idea/vim/api/VimFile.kt

@@ -11,7 +11,7 @@ package com.maddyhome.idea.vim.api
 interface VimFile {
   fun displayFileInfo(vimEditor: VimEditor, fullPath: Boolean)
   fun displayHexInfo(editor: VimEditor)
-  fun displayLocationInfo(vimEditor: VimEditor)
+  fun displayLocationInfo(editor: VimEditor)
   fun selectPreviousTab(context: ExecutionContext)
   fun saveFile(editor: VimEditor, context: ExecutionContext)
   fun saveFiles(editor: VimEditor, context: ExecutionContext)

vim-engine/src/main/kotlin/com/maddyhome/idea/vim/api/VimFileBase.kt

@@ -8,6 +8,13 @@
 
 package com.maddyhome.idea.vim.api
 
+import com.maddyhome.idea.vim.group.visual.VimSelection
+import com.maddyhome.idea.vim.helper.CharacterHelper
+import com.maddyhome.idea.vim.helper.CharacterHelper.charType
+import com.maddyhome.idea.vim.helper.endOffsetInclusive
+import com.maddyhome.idea.vim.state.mode.SelectionType.CHARACTER_WISE
+import com.maddyhome.idea.vim.state.mode.inVisualMode
+import com.maddyhome.idea.vim.state.mode.selectionType
 import java.lang.Long.toHexString
 
 abstract class VimFileBase : VimFile {
@@ -17,4 +24,106 @@ abstract class VimFileBase : VimFile {
 
     injector.messages.showStatusBarMessage(editor, toHexString(ch.code.toLong()))
   }
+
+  override fun displayLocationInfo(editor: VimEditor) {
+    val msg = buildString {
+
+      val caret = editor.currentCaret()
+      val offset = editor.currentCaret().offset
+      val totalByteCount = editor.fileSize()
+      val totalWordCount = countBigWords(editor, offset)
+
+      if (!editor.inVisualMode) {
+        val pos = caret.getBufferPosition()
+        val line = pos.line + 1
+        val col = pos.column + 1
+        val lineEndOffset = editor.getLineEndOffset(pos.line)
+        val lineEndCol = editor.offsetToBufferPosition(lineEndOffset).column
+
+        // Note that Vim can have different screen columns to buffer columns, and displays these in the form "Col 1-3".
+        // Vim uses screen columns to insert inlay text and symbols such as word wrap indicators, but has a single line
+        // even when wrapped, unlike IntelliJ, which has a virtual line per screen line. Because of this, it's not clear
+        // how IdeaVim could represent this, or even if it would be worthwhile to do so.
+        append("Col ").append(col).append(" of ").append(lineEndCol)
+        append("; Line ").append(line).append(" of ").append(editor.lineCount())
+        append("; Word ").append(totalWordCount.currentWord).append(" of ").append(totalWordCount.count)
+        append("; Byte ").append(offset + 1).append(" of ").append(totalByteCount)
+      }
+      else {
+
+        append("Selected ")
+
+        val selection = VimSelection.create(
+          caret.vimSelectionStart,
+          caret.offset,
+          editor.mode.selectionType ?: CHARACTER_WISE,
+          editor
+        ).toVimTextRange()
+
+        val selectedLineCount: Int
+        val selectedWordCount: Int
+        if (selection.isMultiple) {
+          selectedLineCount = selection.size()
+
+          var count = 0
+          for (i in 0 until selection.size()) {
+            val wordCount = countBigWords(editor, offset, selection.startOffsets[i], selection.endOffsets[i])
+            count += wordCount.count
+          }
+          selectedWordCount = count
+
+          append(selection.maxLength).append(" Cols; ")
+        }
+        else {
+          val startPos = editor.offsetToBufferPosition(selection.startOffset)
+          val endPos = editor.offsetToBufferPosition(selection.endOffsetInclusive)
+
+          selectedLineCount = endPos.line - startPos.line + 1
+
+          val wordCount = countBigWords(editor, offset, selection.startOffset, selection.endOffset)
+          selectedWordCount = wordCount.count
+        }
+
+        append(selectedLineCount).append(" of ").append(editor.lineCount()).append(" Lines; ")
+        append(selectedWordCount).append(" of ").append(totalWordCount.count).append(" Words; ")
+        append(selection.selectionCount).append(" of ").append(totalByteCount).append(" Bytes")
+      }
+    }
+
+    injector.messages.showStatusBarMessage(editor, msg)
+  }
 }
+
+/**
+ * Count the number of WORDs that intersect the given range, or exist in the whole file
+ */
+private fun countBigWords(
+  editor: VimEditor,
+  caretOffset: Int,
+  start: Int = 0,
+  endExclusive: Int = editor.text().length,
+): WordCount {
+
+  val chars = editor.text()
+  var wordCount = 0
+  var currentWord = 0
+
+  var lastCharacterType: CharacterHelper.CharacterType? = null
+  for (pos in start until endExclusive) {
+    val characterType = charType(editor, chars[pos], punctuationAsLetters = true)
+    if (characterType != lastCharacterType && characterType != CharacterHelper.CharacterType.WHITESPACE) {
+      wordCount++
+    }
+
+    // The current word is the last counted word
+    if (pos <= caretOffset) {
+      currentWord = wordCount
+    }
+
+    lastCharacterType = characterType
+  }
+
+  return WordCount(wordCount, currentWord)
+}
+
+private data class WordCount(val count: Int, val currentWord: Int)