tdf#163105 Use HB data while selecting kashida insertion positions

Previously, Writer and Edit Engine would skip inserting kashida in words
if the highest-priority candidate position is marked as invalid by
HarfBuzz. This would happen even if the word contained multiple
lower-ranked valid candidate positions.

This change updates Writer and Edit Engine to pass HarfBuzz kashida
position data to the selection algorithm. The algorithm has been updated
to return the highest-priority valid position, if any. The algorithm has
also been updated to use raw positions marked as valid by HarfBuzz as a
fallback, if no better positions could be found.

Change-Id: I40c6432c4607aee197e8767e5667db504469956a
Reviewed-on: https://gerrit.libreoffice.org/c/core/+/173963
Tested-by: Jenkins
Reviewed-by: Jonathan Clark <jonathan@libreoffice.org>

2 files changed, 78 insertions, 22 deletions

diff --git a/i18nutil/qa/cppunit/test_kashida.cxx b/i18nutil/qa/cppunit/test_kashida.cxx
index 1ab2729cb06a..46b40c2a5b7a 100644
--- a/i18nutil/qa/cppunit/test_kashida.cxx
+++ b/i18nutil/qa/cppunit/test_kashida.cxx
@@ -22,13 +22,17 @@ class KashidaTest : public CppUnit::TestFixture
 {
 public:
     void testCharacteristic();
+    void testManualKashida();
     void testFinalYeh();
     void testNoZwnjExpansion();
+    void testExcludeInvalid();
 
     CPPUNIT_TEST_SUITE(KashidaTest);
     CPPUNIT_TEST(testCharacteristic);
+    CPPUNIT_TEST(testManualKashida);
     CPPUNIT_TEST(testFinalYeh);
     CPPUNIT_TEST(testNoZwnjExpansion);
+    CPPUNIT_TEST(testExcludeInvalid);
     CPPUNIT_TEST_SUITE_END();
 };
 
@@ -54,6 +58,14 @@ void KashidaTest::testCharacteristic()
     CPPUNIT_ASSERT_EQUAL(sal_Int32(3), GetWordKashidaPosition(u"تمثیل"_ustr).value().nIndex);
 }
 
+void KashidaTest::testManualKashida()
+{
+    CPPUNIT_ASSERT_EQUAL(sal_Int32(2), GetWordKashidaPosition(u"برـای"_ustr).value().nIndex);
+
+    // Normally, a kashida would not be inserted after a final Yeh.
+    CPPUNIT_ASSERT_EQUAL(sal_Int32(4), GetWordKashidaPosition(u"نیمِـي"_ustr).value().nIndex);
+}
+
 // tdf#65344: Do not insert kashida before a final Yeh
 void KashidaTest::testFinalYeh()
 {
@@ -73,6 +85,32 @@ void KashidaTest::testNoZwnjExpansion()
     CPPUNIT_ASSERT(!GetWordKashidaPosition(u"مت\u200Cن"_ustr).has_value());
 }
 
+// tdf#163105: Do not insert kashida if the position is invalid
+void KashidaTest::testExcludeInvalid()
+{
+    std::vector<bool> aValid;
+    aValid.resize(5, true);
+
+    CPPUNIT_ASSERT_EQUAL(sal_Int32(3),
+                         GetWordKashidaPosition(u"نویسه"_ustr, aValid).value().nIndex);
+
+    aValid[3] = false;
+    CPPUNIT_ASSERT_EQUAL(sal_Int32(0),
+                         GetWordKashidaPosition(u"نویسه"_ustr, aValid).value().nIndex);
+
+    // Calls after this use the last resort (positions in aValid from end to start)
+    aValid[0] = false;
+    CPPUNIT_ASSERT_EQUAL(sal_Int32(2),
+                         GetWordKashidaPosition(u"نویسه"_ustr, aValid).value().nIndex);
+
+    aValid[2] = false;
+    CPPUNIT_ASSERT_EQUAL(sal_Int32(1),
+                         GetWordKashidaPosition(u"نویسه"_ustr, aValid).value().nIndex);
+
+    aValid[1] = false;
+    CPPUNIT_ASSERT(!GetWordKashidaPosition(u"نویسه"_ustr, aValid).has_value());
+}
+
 CPPUNIT_TEST_SUITE_REGISTRATION(KashidaTest);
 }
 
diff --git a/i18nutil/source/utility/kashida.cxx b/i18nutil/source/utility/kashida.cxx
index d016e96294fb..6a6c7adde690 100644
--- a/i18nutil/source/utility/kashida.cxx
+++ b/i18nutil/source/utility/kashida.cxx
@@ -46,6 +46,7 @@ namespace
 
    - tdf#65344: Kashida must not be inserted before the final form of Yeh, unless
                 preceded by an initial or medial Seen.
+   - tdf#163105: As a last resort, use the last valid insertion position from VCL.
 */
 
 #define IS_JOINING_GROUP(c, g) (u_getIntPropertyValue((c), UCHAR_JOINING_GROUP) == U_JG_##g)
@@ -134,7 +135,8 @@ bool CanConnectToPrev(sal_Unicode cCh, sal_Unicode cPrevCh)
 }
 }
 
-std::optional<i18nutil::KashidaPosition> i18nutil::GetWordKashidaPosition(const OUString& rWord)
+std::optional<i18nutil::KashidaPosition>
+i18nutil::GetWordKashidaPosition(const OUString& rWord, const std::vector<bool>& pValidPositions)
 {
     sal_Int32 nIdx = 0;
     sal_Int32 nPrevIdx = 0;
@@ -142,35 +144,45 @@ std::optional<i18nutil::KashidaPosition> i18nutil::GetWordKashidaPosition(const
     sal_Unicode cCh = 0;
     sal_Unicode cPrevCh = 0;
 
-    int nPriorityLevel = 7; // 0..6 = level found, 7 not found
+    int nPriorityLevel = 8; // 0..7 = level found, 8 not found
 
     sal_Int32 nWordLen = rWord.getLength();
 
+    SAL_WARN_IF(!pValidPositions.empty() && pValidPositions.size() != static_cast<size_t>(nWordLen),
+                "i18n", "Kashida valid position array wrong size");
+
     // ignore trailing vowel chars
     while (nWordLen && isTransparentChar(rWord[nWordLen - 1]))
     {
         --nWordLen;
     }
 
-    auto fnTryInsertBefore = [&rWord, &nIdx, &nPrevIdx, &nKashidaPos, &nPriorityLevel,
-                              &nWordLen](sal_Int32 nNewPriority, bool bIgnoreFinalYeh = false) {
-        // Exclusions:
-
-        // #i98410#: prevent ZWNJ expansion
-        if (rWord[nPrevIdx] == 0x200C || rWord[nPrevIdx + 1] == 0x200C)
-        {
-            return;
-        }
-
-        // tdf#65344: Do not insert kashida before a final Yeh
-        if (!bIgnoreFinalYeh && nIdx == (nWordLen - 1) && isYehChar(rWord[nIdx]))
-        {
-            return;
-        }
-
-        nKashidaPos = nPrevIdx;
-        nPriorityLevel = nNewPriority;
-    };
+    auto fnTryInsertBefore
+        = [&rWord, &nIdx, &nPrevIdx, &nKashidaPos, &nPriorityLevel, &nWordLen,
+           &pValidPositions](sal_Int32 nNewPriority, bool bIgnoreFinalYeh = false) {
+              // Exclusions:
+
+              // tdf#163105: Do not insert kashida if the position is invalid
+              if (!pValidPositions.empty() && !pValidPositions[nPrevIdx])
+              {
+                  return;
+              }
+
+              // #i98410#: prevent ZWNJ expansion
+              if (rWord[nPrevIdx] == 0x200C || rWord[nPrevIdx + 1] == 0x200C)
+              {
+                  return;
+              }
+
+              // tdf#65344: Do not insert kashida before a final Yeh
+              if (!bIgnoreFinalYeh && nIdx == (nWordLen - 1) && isYehChar(rWord[nIdx]))
+              {
+                  return;
+              }
+
+              nKashidaPos = nPrevIdx;
+              nPriorityLevel = nNewPriority;
+          };
 
     while (nIdx < nWordLen)
     {
@@ -270,7 +282,7 @@ std::optional<i18nutil::KashidaPosition> i18nutil::GetWordKashidaPosition(const
             }
         }
 
-        // other connecting possibilities
+        // 7. Other connecting possibilities
         if (nPriorityLevel >= 6 && nIdx > 0)
         {
             // Reh, Zain (right joining) final form may appear in the middle of word
@@ -286,6 +298,12 @@ std::optional<i18nutil::KashidaPosition> i18nutil::GetWordKashidaPosition(const
             }
         }
 
+        // 8. If valid position data exists, use the last legal position
+        if (nPriorityLevel >= 7 && nIdx > 0 && !pValidPositions.empty())
+        {
+            fnTryInsertBefore(7);
+        }
+
         // Do not consider vowel marks when checking if a character
         // can be connected to previous character.
         if (!isTransparentChar(cCh))