4 files changed, 72 insertions(+), 8 deletions(-)

A .golden/shapedRuns/intraWordBreak.fontInfo
A .golden/shapedRuns/intraWordBreak.golden
M test/Data/Text/ParagraphLayout/Rich/ParagraphData.hs
M test/Data/Text/ParagraphLayout/RichSpec.hs

@@ 0,0 1,1 @@
+assets/fonts/plex/IBMPlexSansArabic-Regular.ttf 0 32,32 32,32

@@ 0,0 1,19 @@
+[ (0, -35, 
+    [ (GlyphInfo {codepoint = 373, cluster = 15, unsafeToBreak = False, unsafeToConcat = False, safeToInsertTatweel = False}, GlyphPos {x_advance = 24, y_advance = 0, x_offset = 0, y_offset = 0})
+    , (GlyphInfo {codepoint = 971, cluster = 13, unsafeToBreak = False, unsafeToConcat = False, safeToInsertTatweel = False}, GlyphPos {x_advance = 14, y_advance = 0, x_offset = 0, y_offset = 0})
+    , (GlyphInfo {codepoint = 407, cluster = 11, unsafeToBreak = False, unsafeToConcat = False, safeToInsertTatweel = False}, GlyphPos {x_advance = 8, y_advance = 0, x_offset = 0, y_offset = 0})
+    , (GlyphInfo {codepoint = 89, cluster = 10, unsafeToBreak = False, unsafeToConcat = False, safeToInsertTatweel = False}, GlyphPos {x_advance = 18, y_advance = 0, x_offset = 0, y_offset = 0})
+    , (GlyphInfo {codepoint = 1363, cluster = 6, unsafeToBreak = False, unsafeToConcat = False, safeToInsertTatweel = False}, GlyphPos {x_advance = 31, y_advance = 0, x_offset = 0, y_offset = 0})
+    , (GlyphInfo {codepoint = 373, cluster = 4, unsafeToBreak = False, unsafeToConcat = False, safeToInsertTatweel = False}, GlyphPos {x_advance = 24, y_advance = 0, x_offset = 0, y_offset = 0})
+    , (GlyphInfo {codepoint = 971, cluster = 2, unsafeToBreak = False, unsafeToConcat = False, safeToInsertTatweel = False}, GlyphPos {x_advance = 14, y_advance = 0, x_offset = 0, y_offset = 0})
+    , (GlyphInfo {codepoint = 407, cluster = 0, unsafeToBreak = False, unsafeToConcat = False, safeToInsertTatweel = False}, GlyphPos {x_advance = 8, y_advance = 0, x_offset = 0, y_offset = 0})
+    ])
+, (0, -83, 
+    [ (GlyphInfo {codepoint = 1363, cluster = 28, unsafeToBreak = False, unsafeToConcat = False, safeToInsertTatweel = False}, GlyphPos {x_advance = 31, y_advance = 0, x_offset = 0, y_offset = 0})
+    , (GlyphInfo {codepoint = 373, cluster = 26, unsafeToBreak = False, unsafeToConcat = False, safeToInsertTatweel = False}, GlyphPos {x_advance = 24, y_advance = 0, x_offset = 0, y_offset = 0})
+    , (GlyphInfo {codepoint = 971, cluster = 24, unsafeToBreak = False, unsafeToConcat = False, safeToInsertTatweel = False}, GlyphPos {x_advance = 14, y_advance = 0, x_offset = 0, y_offset = 0})
+    , (GlyphInfo {codepoint = 407, cluster = 22, unsafeToBreak = False, unsafeToConcat = False, safeToInsertTatweel = False}, GlyphPos {x_advance = 8, y_advance = 0, x_offset = 0, y_offset = 0})
+    , (GlyphInfo {codepoint = 89, cluster = 21, unsafeToBreak = False, unsafeToConcat = False, safeToInsertTatweel = False}, GlyphPos {x_advance = 18, y_advance = 0, x_offset = 0, y_offset = 0})
+    , (GlyphInfo {codepoint = 1363, cluster = 17, unsafeToBreak = False, unsafeToConcat = False, safeToInsertTatweel = False}, GlyphPos {x_advance = 31, y_advance = 0, x_offset = 0, y_offset = 0})
+    ])
+]<
\ No newline at end of file

@@ 1,6 1,7 @@
 module Data.Text.ParagraphLayout.Rich.ParagraphData
     ( hardBoxBreakLTRParagraph
     , hardBoxBreakRTLParagraph
+    , intraWordBreakParagraph
     , loremIpsumParagraph
     , mixedDirectionComplexParagraph
     , mixedDirectionSimpleParagraph


@@ 79,6 80,19 @@ hardBoxBreakRTLParagraph font = constructParagraph
         b2 = b_ { boxSpacing = BoxSpacingLeftRight 150 200 }
         t = tRTL { textFont = font, textLanguage = "zxx" }
 
+-- | For testing that soft wrap preserves the form of Arabic characters.
+-- Source: <https://www.w3.org/TR/css-text-3/#word-break-shaping>
+intraWordBreakParagraph :: Font -> ParagraphOptions -> Paragraph String
+intraWordBreakParagraph font = constructParagraph
+    empty
+    (rootBox t
+        [ text "text" "نوشتن_نوشتن_نوشتن"
+        ]
+    )
+    empty
+    where
+        t = tRTL { textFont = font, textLanguage = "fa" }
+
 -- | Filler text using the Latin script.
 -- Source: <https://www.lipsum.com/>
 loremIpsumParagraph :: Font -> ParagraphOptions -> Paragraph String

@@ 1,7 1,8 @@
 module Data.Text.ParagraphLayout.RichSpec (spec) where
 
 import Data.Int (Int32)
-import Data.Text.Glyphize (Direction (DirLTR, DirRTL))
+import Data.Word (Word32)
+import Data.Text.Glyphize (Direction (DirLTR, DirRTL), codepoint)
 
 import Test.Hspec
 import System.FilePath ((</>))


@@ 19,6 20,9 @@ fragmentRects :: ParagraphLayout d -> [(d, Rect Int32)]
 fragmentRects p = map toItem $ paragraphFragments p
     where toItem (Fragment { fragmentUserData = d, fragmentRect = r }) = (d, r)
 
+glyphRuns :: ParagraphLayout d -> [[Word32]]
+glyphRuns = map (map (codepoint . fst) . fragmentGlyphs) . paragraphFragments
+
 spec :: Spec
 spec = do
 


@@ 158,6 162,27 @@ spec = do
                 let result = layoutRich input
                 result `shouldBeGolden` "mixedDirectionComplexLTR"
 
+            it "preserves joining forms across soft breaks" $ do
+                let opts = defaultParagraphOptions { paragraphMaxWidth = 4500 }
+                let input = intraWordBreakParagraph font opts
+                let glyphRunsVisual = glyphRuns $ layoutRich input
+                let glyphRunsLogical = map reverse glyphRunsVisual
+                -- Glyph indexes for the corresponding Arabic letters/ligatures:
+                let n_initial = 407
+                let v_final = 971
+                let sh_initial = 373
+                let tn_final = 1363
+                -- Glyph index for the underscore used to break repetitions:
+                let sp = 89
+                -- Unconventional code formatting here shows that there should
+                -- be three repetitions of the same glyph sequence, unaffected
+                -- by the line break inserted after an initial form and before
+                -- a final form.
+                glyphRunsLogical `shouldBe`
+                    [[n_initial, v_final, sh_initial, tn_final, sp,
+                    n_initial, v_final, sh_initial], [tn_final, sp,
+                    n_initial, v_final, sh_initial, tn_final]]
+
     describe "paginate" $ do
         let
             goldenDir = ".golden" </> "paginatedRichParagraphLayout"


@@ 195,13 220,18 @@ spec = do
             let fontPath = arabicFont
             font <- runIO $ loadFont fontPath 0 demoOptions
 
-            let
-                opts = defaultParagraphOptions
-                input = mixedDirectionComplexParagraph DirRTL font opts
-                result = layoutRich input
-                shapedRunsSpec = shapedRunsSpecWithFont fontPath font
+            let shapedRunsSpec = shapedRunsSpecWithFont fontPath font
 
             shapedRunsSpec
                 "handles complex mixed direction in RTL paragraph"
-                "mixedDirectionComplexRTL"
-                result
+                "mixedDirectionComplexRTL" $
+                layoutRich $
+                mixedDirectionComplexParagraph DirRTL font $
+                defaultParagraphOptions
+
+            shapedRunsSpec
+                "preserves joining forms across soft breaks"
+                "intraWordBreak" $
+                layoutRich $
+                intraWordBreakParagraph font $
+                defaultParagraphOptions { paragraphMaxWidth = 144 }