LT-22674: Cache text analysis only for text NFC leaves unchanged

Src/Common/RootSite/RootSiteTests/RenderBenchmarkTestsBase.cs

@@ -25,6 +25,8 @@ namespace SIL.FieldWorks.Common.RootSites.RenderBenchmark
 	public abstract class RenderBenchmarkTestsBase : RealDataTestsBase
 	{
 		protected const string DeterministicRenderFontFamily = "Segoe UI";
+		// Second Latin font for scenarios whose runs alternate font family.
+		protected const string SecondaryRenderFontFamily = "Times New Roman";
 		// Pinned Arabic font (loaded privately by RenderTestAssemblySetup). Used for Arabic runs so
 		// they don't depend on the host's Segoe UI Arabic version / font fallback.
 		protected const string ArabicRenderFontFamily = "Scheherazade New";
@@ -320,6 +322,15 @@ protected void SetupScenarioData(string scenarioId)
 				case "multi-ws":
 					CreateMultiWsScenario();
 					break;
+				case "single-para-mixed-ws":
+					CreateSingleParaMixedWsScenario();
+					break;
+				case "nfc-composable-diacritics":
+					CreateNfcComposableDiacriticsScenario();
+					break;
+				case "multi-line-wrap-single-ws":
+					CreateMultiLineWrapSingleWsScenario();
+					break;
 				case "lex-shallow":
 					CreateLexEntryScenario(depth: 2, breadth: 3);
 					break;
@@ -539,6 +550,27 @@ private void CreateMultiWsScenario()
 			AddMultiWsSections(book, 5, versesPerSection: 8, chapterStart: 1);
 		}
 
+		private void CreateSingleParaMixedWsScenario()
+		{
+			var book = CreateBook(19); // PSA
+			m_hvoRoot = book.Hvo;
+			AddSingleMixedWsParagraph(book, sentenceCount: 236);
+		}
+
+		private void CreateNfcComposableDiacriticsScenario()
+		{
+			var book = CreateBook(15); // EZR
+			m_hvoRoot = book.Hvo;
+			AddNfcComposableDiacriticsParagraph(book, wordCount: 80);
+		}
+
+		private void CreateMultiLineWrapSingleWsScenario()
+		{
+			var book = CreateBook(17); // EST
+			m_hvoRoot = book.Hvo;
+			AddSingleWsProseParagraph(book, sentenceCount: 200);
+		}
+
 		#region Rich Data Factories
 
 		protected IScrBook CreateBook(int bookNum)
@@ -928,6 +960,147 @@ protected void AddMultiWsSections(IScrBook book, int sectionCount,
 			}
 		}
 
+		/// <summary>
+		/// Returns the canonical decomposition (NFD) of <paramref name="text"/>, and fails the
+		/// fixture unless the result contains combining marks that NFC would compose. A scenario
+		/// built from it therefore always exercises text that normalization rewrites, whatever
+		/// normalization form the source file itself is stored in.
+		/// </summary>
+		protected static string Decomposed(string text)
+		{
+			string nfd = CustomIcu.GetIcuNormalizer(FwNormalizationMode.knmNFD).Normalize(text);
+			string nfc = CustomIcu.GetIcuNormalizer(FwNormalizationMode.knmNFC).Normalize(text);
+			Assert.That(nfd, Is.Not.EqualTo(nfc),
+				$"Scenario text '{text}' must contain characters that decompose.");
+			return nfd;
+		}
+
+		/// <summary>
+		/// Adds a section whose content is one paragraph of <paramref name="sentenceCount"/>
+		/// sentences. Each sentence's text and run properties come from the two callbacks, which
+		/// receive the sentence index. With <paramref name="withChapterNumber"/> the paragraph
+		/// opens with a chapter-number run.
+		/// </summary>
+		private void AddSingleParagraphSection(IScrBook book, string heading, int sentenceCount,
+			Func<int, string> sentence, Func<int, ITsTextProps> runProps, bool withChapterNumber = false)
+		{
+			var section = Cache.ServiceLocator.GetInstance<IScrSectionFactory>().Create();
+			book.SectionsOS.Add(section);
+
+			var stTextFactory = Cache.ServiceLocator.GetInstance<IStTextFactory>();
+
+			section.HeadingOA = stTextFactory.Create();
+			var headingBldr = new StTxtParaBldr(Cache) { ParaStyleName = ScrStyleNames.SectionHead };
+			headingBldr.AppendRun(heading, StyleUtils.CharStyleTextProps(null, m_wsEng));
+			headingBldr.CreateParagraph(section.HeadingOA);
+
+			section.ContentOA = stTextFactory.Create();
+			var paraBldr = new StTxtParaBldr(Cache) { ParaStyleName = ScrStyleNames.NormalParagraph };
+			if (withChapterNumber)
+				paraBldr.AppendRun("1", StyleUtils.CharStyleTextProps(ScrStyleNames.ChapterNumber, m_wsEng));
+
+			for (int i = 0; i < sentenceCount; i++)
+				paraBldr.AppendRun(sentence(i), runProps(i));
+
+			paraBldr.CreateParagraph(section.ContentOA);
+		}
+
+		/// <summary>
+		/// Adds one paragraph of short decomposed sentences whose runs alternate between two
+		/// writing systems and two font families.
+		/// </summary>
+		protected void AddSingleMixedWsParagraph(IScrBook book, int sentenceCount)
+		{
+			string[] subjects =
+			{
+				"the \u00E9lder",
+				"the h\u00E8rder",
+				"the s\u00EFnger",
+				"the tea\u00E7her",
+				"the trav\u00EAler",
+				"the wom\u00E3n"
+			};
+			string[] predicates =
+			{
+				"spoke of the l\u00F3ng rains",
+				"walked to the f\u00E0r well",
+				"named the sev\u00EBn hills",
+				"counted the cattl\u00E9 at dusk",
+				"kept the \u00F4ld story",
+				"asked for a bless\u0129ng"
+			};
+
+			AddSingleParagraphSection(book, "Single Paragraph, Mixed Writing Systems", sentenceCount,
+				i => Decomposed($"{subjects[i % subjects.Length]} {predicates[i % predicates.Length]} {i + 1}. "),
+				i => AlternatingFontRunProps(i % 2 == 0),
+				withChapterNumber: true);
+		}
+
+		private ITsTextProps AlternatingFontRunProps(bool first)
+		{
+			var bldr = TsStringUtils.MakePropsBldr();
+			bldr.SetIntPropValues((int)FwTextPropType.ktptWs, (int)FwTextPropVar.ktpvDefault,
+				first ? m_wsEng : m_wsFr);
+			bldr.SetStrPropValue((int)FwTextPropType.ktptFontFamily,
+				first ? DeterministicRenderFontFamily : SecondaryRenderFontFamily);
+			return bldr.GetTextProps();
+		}
+
+		/// <summary>
+		/// Adds one wrapped paragraph of Latin sentences, each naming a word spelled with
+		/// decomposed diacritics.
+		/// </summary>
+		protected void AddNfcComposableDiacriticsParagraph(IScrBook book, int wordCount)
+		{
+			string[] words =
+			{
+				"caf\u00E9",
+				"d\u00E9j\u00E0",
+				"no\u00EBl",
+				"fran\u00E7ais",
+				"gar\u00E7on",
+				"h\u00F4tel",
+				"a\u00F1o",
+				"cr\u00E9\u00E9e",
+				"\u00E9l\u00E9gant",
+				"fa\u00E7ade"
+			};
+
+			AddSingleParagraphSection(book, "Decomposed Diacritics Microbenchmark", wordCount,
+				i => Decomposed($"The {words[i % words.Length]} recorded here is entry {i + 1}. "),
+				i => StyleUtils.CharStyleTextProps(null, m_wsEng));
+		}
+
+		/// <summary>
+		/// Adds one long wrapped paragraph of decomposed sentences in a single writing system
+		/// and font.
+		/// </summary>
+		protected void AddSingleWsProseParagraph(IScrBook book, int sentenceCount)
+		{
+			string[] subjects =
+			{
+				"the mer\u00E7hant",
+				"the mas\u00F3n",
+				"the scrib\u00EB",
+				"the sheph\u00E8rd",
+				"the weav\u00EAr",
+				"the pott\u00E9r"
+			};
+			string[] predicates =
+			{
+				"measured the grain by the riv\u00E9r",
+				"repaired the east\u00E8rn wall before dusk",
+				"copied the ledger onto fresh parchm\u00EBnt",
+				"counted the flock past the old gat\u00EA",
+				"dyed the cloth a deep saffr\u00F5n",
+				"shaped the jar on the slow whe\u00EBl"
+			};
+
+			AddSingleParagraphSection(book, "Single Writing-System Line-Wrap Microbenchmark", sentenceCount,
+				i => Decomposed($"{subjects[i % subjects.Length]} {predicates[i % predicates.Length]} on day {i + 1}. "),
+				i => StyleUtils.CharStyleTextProps(null, m_wsEng));
+		}
+
 		#endregion
 
 		#region Lex Entry Scenario Data

Src/Common/RootSite/RootSiteTests/TestData/RenderBenchmarkScenarios.json

@@ -89,6 +89,21 @@
       "description": "Lex entry with senses nested 6 levels deep, 2-wide (depth 6, breadth 2 = 126 senses)",
       "tags": ["lex-entry", "nested-senses", "exponential-cost", "stress"],
       "viewType": "LexEntry"
+    },
+    {
+      "id": "single-para-mixed-ws",
+      "description": "One paragraph of 236 short decomposed (NFD) sentences whose runs alternate two writing systems and two font families",
+      "tags": ["stress", "layout-stress", "multi-ws", "single-paragraph", "nfd"]
+    },
+    {
+      "id": "nfc-composable-diacritics",
+      "description": "One wrapped paragraph of Latin prose spelled with decomposed (NFD) diacritics that NFC composes",
+      "tags": ["stress", "layout-stress", "nfd", "diacritics", "line-breaking"]
+    },
+    {
+      "id": "multi-line-wrap-single-ws",
+      "description": "One long paragraph of 200 decomposed (NFD) sentences in one writing system and font, wrapping over many lines",
+      "tags": ["stress", "layout-stress", "single-paragraph", "line-breaking", "nfd"]
     }
   ]
 }

Src/views/Test/TestUniscribeEngine.h

@@ -16,6 +16,7 @@ Last reviewed:
 
 #include "testViews.h"
 #include "RenderEngineTestBase.h"
+#include "LayoutCache.h"
 
 namespace TestViews
 {
@@ -396,6 +397,94 @@ namespace TestViews
 #endif
 		}
 
+		// Installs a fresh layout-pass cache for the lifetime of the object, restoring the
+		// previous one even when a test assertion throws.
+		class ScopedLayoutPassCache
+		{
+		public:
+			ScopedLayoutPassCache() : m_pPrev(SetCurrentLayoutPassCache(&m_cache)) {}
+			~ScopedLayoutPassCache() { SetCurrentLayoutPassCache(m_pPrev); }
+			LayoutPassCache & Cache() { return m_cache; }
+
+		private:
+			LayoutPassCache m_cache;
+			LayoutPassCache * m_pPrev;
+		};
+
+		// Breaks one text source twice in one layout pass and reports cache hits and misses.
+		// The engine is built here, not via COM, so it sees the cache this module installs.
+		void BreakTwiceInOneLayoutPass(const wchar_t * pszText, int * pcHit, int * pcMiss)
+		{
+			*pcHit = *pcMiss = 0;
+#if defined(WIN32) || defined(_M_X64)
+			HWND hwndDesktop = ::GetDesktopWindow();
+			HDC hdcDesktop = ::GetDC(hwndDesktop);
+			HDC hdc = ::CreateCompatibleDC(hdcDesktop);
+			HBITMAP hbm = ::CreateCompatibleBitmap(hdcDesktop, 100, 100);
+			HGDIOBJ hbmOld = ::SelectObject(hdc, hbm);
+
+			IVwGraphicsWin32Ptr qvg;
+			qvg.CreateInstance(CLSID_VwGraphicsWin32);
+			qvg->Initialize(hdc);
+
+			ILgWritingSystemFactoryPtr qwsf;
+			m_qre->get_WritingSystemFactory(&qwsf);
+			IRenderEnginePtr qre;
+			UniscribeEngine::CreateCom(NULL, IID_IRenderEngine, (void **)&qre);
+			CheckHr(qre->putref_WritingSystemFactory(qwsf));
+			CheckHr(qre->putref_RenderEngineFactory(m_qref));
+			TxtSrc ts(pszText, qwsf);
+			IVwTextSourcePtr qts;
+			ts.QueryInterface(IID_IVwTextSource, (void **)&qts);
+			int cch;
+			CheckHr(qts->get_Length(&cch));
+
+			{
+				ScopedLayoutPassCache scope;
+				for (int ibreak = 0; ibreak < 2; ++ibreak)
+				{
+					ILgSegmentPtr qseg;
+					int dichLimSeg;
+					int dxWidth;
+					LgEndSegmentType est;
+					CheckHr(qre->FindBreakPoint(qvg, qts, NULL, 0, cch, cch, TRUE, TRUE, 4000,
+						klbWordBreak, klbLetterBreak, ktwshAll, FALSE, &qseg, &dichLimSeg,
+						&dxWidth, &est, NULL));
+				}
+				*pcHit = scope.Cache().AnalysisCache().HitCount();
+				*pcMiss = scope.Cache().AnalysisCache().MissCount();
+			}
+
+			qts.Clear();
+			qre.Clear();
+			qvg.Clear();
+			::SelectObject(hdc, hbmOld);
+			::DeleteObject(hbm);
+			::DeleteDC(hdc);
+			::ReleaseDC(hwndDesktop, hdcDesktop);
+#endif
+		}
+
+		void testNfcTextAnalysisIsReusedWithinLayoutPass()
+		{
+#if defined(WIN32) || defined(_M_X64)
+			int cHit, cMiss;
+			BreakTwiceInOneLayoutPass(L"cafe deja vu", &cHit, &cMiss);
+			unitpp::assert_true("text NFC leaves unchanged should be reused on the second break",
+				cHit > 0);
+#endif
+		}
+
+		void testDecomposedTextAnalysisIsReusedWithinLayoutPass()
+		{
+#if defined(WIN32) || defined(_M_X64)
+			int cHit, cMiss;
+			// Decomposed e-acute and a-grave, which NFC composes.
+			BreakTwiceInOneLayoutPass(L"cafe\u0301 de\u0301ja\u0300 vu", &cHit, &cMiss);
+			unitpp::assert_true("text NFC rewrites should be reused on the second break", cHit > 0);
+#endif
+		}
+
 		virtual void Setup()
 		{
 			RenderEngineTestBase::Setup();