[GuTS805]

Note

Assisted-by: Claude Opus 5.5

Purpose / Description

Tapping a chip in the Instant Card dropped any text outside the regex match, so C++ became {{c1::C}}, 20°C became {{c1::20}}, and an emoji became null{{c1::null}}null.

Fixes

• Fixes Instant Card: tapping a word deletes symbols from it, and an emoji becomes null{{c1::null}}null #22860

Approach

• Building or removing a cloze only replaces the matched range, so the rest of the word is kept: {{c1::20}}°C, €{{c1::100}}. Symbols stay outside the cloze, as in [BUG]: Instant editor's tapping makes cloze which ignores adjacent characters after symbol in middle of word #16784
• A word with nothing to cloze (e.g. an emoji) is left unchanged, and no cloze number is used
• Words are now split on spaces outside clozes only. Before, re-reading the field (e.g. after switching edit modes) split {{c1::20}}°C into two chips, and tapping one added a space: 20 °C

How Has This Been Tested?

• New unit tests, which fail without the fix. All 47 instant editor tests pass, ktlint + lint are clean
• Emulator (API 35), sharing I like C++ and $100 at 20°C café 🙂 e=mc² Zürich:
  • clozed every chip, switched edit modes, undid them: the text came back exactly
  • saved the note, checked in the Card Browser: I like {{c1::C}}++ and ${{c2::100}} at {{c3::20}}°C café 🙂 e=mc² {{c4::Zürich}}

Before	After

Checklist

• You have a descriptive commit message with a short title (first line, max. 50 chars).
• You have commented your code, particularly in hard-to-understand areas
• You have performed a self-review of your own code
• UI changes: include screenshots of all affected screens (in particular showing any new or changed strings)
• UI Changes: You have tested your change using the Google Accessibility Scanner

[GuTS805]

Ready for review..

[mikehardy]

Assisted-by: grok-4.7

Thanks for tackling the instant-editor cloze chip bugs — the degree sign, emoji, and renumber cases look solid in the tests you added.

Locally I checked out the PR tip, rebased the single commit onto current main, and did not push. Lint, unit tests, and the emulator suite all passed on emulator-5556 / API35_GAPI.

EditorChips.setupChipGroup loads getWordsFromFieldText, sets each chip label with getCleanClozeWords, and on click calls buildClozeText then updateChipData, which joins every chip with one space and setClozeFieldText. Long-press calls updateClozeNumber on that chip's raw word. Switching back from advanced edit calls setupChipGroup again on the text box.

buildClozeText still returns processClozeUndo when clozePattern.find hits. Otherwise it takes the first clozeBuilderPattern match. A miss returns the input and does not increment currentClozeNumber or append intClozeList. Groups 1, 2, and 4 are written back with replaceRange(matcher.range, …), so 20°C becomes {{c1::20}}°C and €100 becomes €{{c1::100}}. Undo uses the same replaceRange, so the degree sign and the euro stay. getCleanClozeWords is clozePattern.replace, so the chip label for {{c1::20}}°C is 20°C.

getWordsFromFieldText splits on \s+ when the space's first index is outside a clozePattern range, then splitBetweenClozes cuts each piece at every {{c<digits>:: after the first. combineWordsWithPunctuation still appends a punctuation-only token only when the previous word clozePattern.matches entirely, which {{c1::20}}°C does not.

Requesting changes: getWordsFromFieldText still builds cloze ranges with clozePattern, whose trailing punctuation class includes {. On text like {{c1::born in}}-{{c2::New York}}, findAll matches through {{c1::born in}}-{{ and never sees the start of the second cloze. The space inside New York falls outside that range, so the word splitter breaks the markup into chips {{c1::born in}}-, {{c2::New, and York}}. The same cut happens when two clozes abut with only punctuation between them. Cloze ranges need to end at }} without swallowing the next {{. Please add a regression test that fails on today's chip split for a later cloze that contains a space (for example {{c1::born in}}-{{c2::New York}}).

some acceptable findings, no action required, just leaving for posterity: the comment above the clozeBuilderPattern groups still says "first, second, and third regex groups" while the code uses groups 1, 2, and 4.

[GuTS805]

Thanks! Fixed: cloze ranges now end at }} so the next cloze's {{ isn't swallowed. Added a test with your {{c1::born in}}-{{c2::New York}} example.

[mikehardy]

Assisted-by: Grok 4.7

Thanks for the outerClozeRanges chip split — the born-in/New York and nested formula cases look right on the word path.

Locally I rebased onto main without force-pushing (prep stayed at the PR tip). I ran lint (lintAll, ktlintCheck, lint-rules:test), unit tests (jacocoUnitTestReport), and release emulator tests (TEST_RELEASE_BUILD=true via run_emulator_tests.py on emulator-5556 / API35_GAPI); all passed.

Requesting changes: setClozeFieldText still seeds intClozeList from getClozeOrdinals, which scans with clozePattern (not outerClozeRanges). On {{c1::born in}}-{{c2::New York}} extra, the pattern only sees c1, so the next chip you add becomes c2 again beside the existing c2. The same happens on {{c1::H2{{c2::O}}}} water — the inner c2 is not counted separately, so clozeing water also lands on c2. Please enumerate every {{c<n>:: (including nested) for the next-number path and add a regression that expects c3 after clozeing the plain word in both strings.

some acceptable findings, no action required, just leaving for posterity: the nested-formula test asserts the appended chip text rather than the next ordinal explicitly.

[david-allison]

Reuse of an existing cloze ordinal

diff --git a/AnkiDroid/src/test/java/com/ichi2/anki/instanteditor/InstantEditorClozeNumberingRegressionTest.kt b/AnkiDroid/src/test/java/com/ichi2/anki/instanteditor/InstantEditorClozeNumberingRegressionTest.kt
new file mode 100644
index 0000000000000000000000000000000000000000..7a9facbc18e77bace502868a11f1a6ded0dca098
--- /dev/null
+++ b/AnkiDroid/src/test/java/com/ichi2/anki/instanteditor/InstantEditorClozeNumberingRegressionTest.kt
@@ -0,0 +1,39 @@
+// SPDX-License-Identifier: GPL-3.0-or-later
+
+package com.ichi2.anki.instanteditor
+
+import androidx.test.ext.junit.runners.AndroidJUnit4
+import com.ichi2.anki.RobolectricTest
+import com.ichi2.anki.instanteditor.InstantEditorViewModelTest.Companion.runInstantEditorViewModelTest
+import org.junit.Assert.assertEquals
+import org.junit.Test
+import org.junit.runner.RunWith
+
+@RunWith(AndroidJUnit4::class)
+class InstantEditorClozeNumberingRegressionTest : RobolectricTest() {
+    @Test
+    fun `a new chip does not reuse a touching cloze ordinal`() =
+        runInstantEditorViewModelTest {
+            setClozeFieldText("{{c1::born in}}-{{c2::New York}} extra")
+            val words = getWordsFromFieldText().toMutableList()
+            val extraIndex = words.indexOf("extra")
+
+            words[extraIndex] = buildClozeText(words[extraIndex])
+            setClozeFieldText(words.joinToString(" "))
+
+            assertEquals("{{c3::extra}}", words[extraIndex])
+        }
+
+    @Test
+    fun `a new chip does not reuse a nested cloze ordinal`() =
+        runInstantEditorViewModelTest {
+            setClozeFieldText("{{c1::H2{{c2::O}}}} water")
+            val words = getWordsFromFieldText().toMutableList()
+            val waterIndex = words.indexOf("water")
+
+            words[waterIndex] = buildClozeText(words[waterIndex])
+            setClozeFieldText(words.joinToString(" "))
+
+            assertEquals("{{c1::H2{{c2::O}}}} {{c3::water}}", actualClozeFieldText.value)
+        }
+}

Renumbering of a nested sibling

diff --git a/AnkiDroid/src/test/java/com/ichi2/anki/instanteditor/InstantEditorNestedClozeRenumberingRegressionTest.kt b/AnkiDroid/src/test/java/com/ichi2/anki/instanteditor/InstantEditorNestedClozeRenumberingRegressionTest.kt
new file mode 100644
index 0000000000000000000000000000000000000000..dddff8c4d94799d7f4384dadd31d052541fc71c0
--- /dev/null
+++ b/AnkiDroid/src/test/java/com/ichi2/anki/instanteditor/InstantEditorNestedClozeRenumberingRegressionTest.kt
@@ -0,0 +1,29 @@
+// SPDX-License-Identifier: GPL-3.0-or-later
+
+package com.ichi2.anki.instanteditor
+
+import androidx.test.ext.junit.runners.AndroidJUnit4
+import com.ichi2.anki.RobolectricTest
+import com.ichi2.anki.instanteditor.InstantEditorViewModelTest.Companion.runInstantEditorViewModelTest
+import org.junit.Assert.assertEquals
+import org.junit.Test
+import org.junit.runner.RunWith
+
+@RunWith(AndroidJUnit4::class)
+class InstantEditorNestedClozeRenumberingRegressionTest : RobolectricTest() {
+    @Test
+    fun `renumbering an outer chip preserves every nested cloze ordinal`() =
+        runInstantEditorViewModelTest {
+            setClozeFieldText("{{c1::Water contains {{c2::hydrogen}} and {{c3::oxygen}}.}}")
+            val words = getWordsFromFieldText().toMutableList()
+
+            // This is the update performed after long-pressing the outer chip.
+            words[0] = updateClozeNumber(words[0], 4)
+            setClozeFieldText(words.joinToString(" "))
+
+            assertEquals(
+                "{{c4::Water contains {{c2::hydrogen}} and {{c3::oxygen}}.}}",
+                actualClozeFieldText.value,
+            )
+        }
+}