Ready to review - almost done testing.

sillsdev · Jan 17, 2025 · 034e8bc · 034e8bc
1 parent ecead42
commit 034e8bc
Show file tree

Hide file tree

Showing 9 changed files with 109 additions and 75 deletions.
diff --git a/src/SIL.Machine/Corpora/IUsfmParserHandler.cs b/src/SIL.Machine/Corpora/IUsfmParserHandler.cs
@@ -76,7 +76,12 @@ IReadOnlyList<UsfmAttribute> attributes
         /// <summary>
         /// End of a sub component
         /// </summary>
-        void EndSubComponent(UsfmParserState state, string marker, bool closed);
+        void EndSubComponent(
+            UsfmParserState state,
+            string marker,
+            IReadOnlyList<UsfmAttribute> attributes,
+            bool closed
+        );
 
         /// <summary>
         /// Start of a sub component text

diff --git a/src/SIL.Machine/Corpora/ScriptureRefUsfmParserHandlerBase.cs b/src/SIL.Machine/Corpora/ScriptureRefUsfmParserHandlerBase.cs
@@ -154,6 +154,9 @@ public override void EndSidebar(UsfmParserState state, string marker, bool close
 
         public override void StartSubComponent(UsfmParserState state, string marker, string caller, string category)
         {
+            if (_curVerseRef.IsDefault)
+                UpdateVerseRef(state.VerseRef, marker, startAsChildElement: true);
+
             if (CurrentTextType != ScriptureTextType.None && !_duplicateVerse)
             {
                 // if we hit a note in a verse paragraph and we aren't in a verse, then start a non-verse segment
@@ -237,12 +240,13 @@ private void EndNonVerseText(UsfmParserState state)
             _curTextType.Pop();
         }
 
-        private void UpdateVerseRef(VerseRef verseRef, string marker)
+        private void UpdateVerseRef(VerseRef verseRef, string marker, bool startAsChildElement = false)
         {
             if (!VerseRef.AreOverlappingVersesRanges(verseRef, _curVerseRef))
             {
                 _curElements.Clear();
-                _curElements.Push(new ScriptureElement(0, marker));
+                int position = startAsChildElement ? 1 : 0;
+                _curElements.Push(new ScriptureElement(position, marker));
             }
             _curVerseRef = verseRef;
         }

diff --git a/src/SIL.Machine/Corpora/UpdateUsfmParserHandler.cs b/src/SIL.Machine/Corpora/UpdateUsfmParserHandler.cs
@@ -207,15 +207,20 @@ public override void StartSubComponent(UsfmParserState state, string marker, str
             base.StartSubComponent(state, marker, caller, category);
         }
 
-        public override void EndSubComponent(UsfmParserState state, string marker, bool closed)
+        public override void EndSubComponent(
+            UsfmParserState state,
+            string marker,
+            IReadOnlyList<UsfmAttribute> attributes,
+            bool closed
+        )
         {
             // strip out notes in verses that are being replaced
             if (ReplaceWithNewTokens(state, closed: closed, endSubComponent: true))
                 SkipTokens(state);
             else
                 CollectTokens(state);
 
-            base.EndSubComponent(state, marker, closed);
+            base.EndSubComponent(state, marker, attributes, closed);
         }
 
         public override void Ref(UsfmParserState state, string marker, string display, string target)

diff --git a/src/SIL.Machine/Corpora/UsfmParser.cs b/src/SIL.Machine/Corpora/UsfmParser.cs
@@ -536,7 +536,7 @@ public bool ProcessToken()
                         State.SpecialTokenCount += 3;
                     }
 
-                    State.Push(new UsfmParserElement(UsfmElementType.SubComponent, token.Marker));
+                    State.Push(new UsfmParserElement(UsfmElementType.SubComponent, token.Marker, token.Attributes));
 
                     Handler?.StartSubComponent(State, token.Marker, token.Data, noteCategory);
                     break;
@@ -657,7 +657,7 @@ private void CloseElement(bool closed = false)
                     Handler?.EndChar(State, element.Marker, element.Attributes, closed);
                     break;
                 case UsfmElementType.SubComponent:
-                    Handler?.EndSubComponent(State, element.Marker, closed);
+                    Handler?.EndSubComponent(State, element.Marker, element.Attributes, closed);
                     break;
                 case UsfmElementType.SubComponentText:
                     Handler?.EndSubComponentText(State);

diff --git a/src/SIL.Machine/Corpora/UsfmParserHandlerBase.cs b/src/SIL.Machine/Corpora/UsfmParserHandlerBase.cs
@@ -55,7 +55,12 @@ bool closed
 
         public virtual void StartSubComponent(UsfmParserState state, string marker, string caller, string category) { }
 
-        public virtual void EndSubComponent(UsfmParserState state, string marker, bool closed) { }
+        public virtual void EndSubComponent(
+            UsfmParserState state,
+            string marker,
+            IReadOnlyList<UsfmAttribute> attributes,
+            bool closed
+        ) { }
 
         public virtual void StartSubComponentText(UsfmParserState state) { }
 

diff --git a/src/SIL.Machine/Corpora/UsfmTextBase.cs b/src/SIL.Machine/Corpora/UsfmTextBase.cs
@@ -206,9 +206,17 @@ public override void StartSubComponent(UsfmParserState state, string marker, str
                 OutputMarker(state);
             }
 
-            public override void EndSubComponent(UsfmParserState state, string marker, bool closed)
+            public override void EndSubComponent(
+                UsfmParserState state,
+                string marker,
+                IReadOnlyList<UsfmAttribute> attributes,
+                bool closed
+            )
             {
-                base.EndSubComponent(state, marker, closed);
+                base.EndSubComponent(state, marker, attributes, closed);
+
+                if (_text._includeMarkers && attributes != null && state.PrevToken?.Type == UsfmTokenType.Attribute)
+                    _rowTexts.Peek().Append(state.PrevToken);
 
                 if (closed)
                     OutputMarker(state);

diff --git a/tests/SIL.Machine.Tests/Corpora/TestData/usfm/Tes/41MATTes.SFM b/tests/SIL.Machine.Tests/Corpora/TestData/usfm/Tes/41MATTes.SFM
@@ -28,7 +28,7 @@
 \s1 Chapter \it Two \it*
 \p
 \p
-\v 1 Chapter \add two\add*, verse \f + \fr 2:1: \ft This is a\bd footnote.\bd*\f*one.
+\v 1 Chapter \add two\add*, verse \f + \fr 2:1: \ft This is a \bd footnote.\bd*\f*one.
 \v 2-3 Chapter two, // verse \fm ∆\fm*two.
 \esb
 \ms This is a sidebar
@@ -41,7 +41,7 @@
 \p
 \v 6 Bad verse. \x - \xo 2:3-4 \xt Cool Book 3:24 \xta The annotation \x* and more content.
 \p
-\v 5 Chapter two, verse five \rq (MAT 3:1)\rq*.
+\v 5 Chapter two, verse five\rq (MAT 3:1)\rq*.
 \v 7a Chapter two, verse seven A,
 \s Section header \ts-s\*
 \p

diff --git a/tests/SIL.Machine.Tests/Corpora/UpdateUsfmParserHandlerTests.cs b/tests/SIL.Machine.Tests/Corpora/UpdateUsfmParserHandlerTests.cs
@@ -325,7 +325,7 @@ public void GetUsfm_NonVerse_CharStyle()
     [Test]
     public void GetUsfm_NonVerse_Paragraph()
     {
-        var rows = new List<(IReadOnlyList<ScriptureRef>, string)> { (ScrRef("MAT 1:0/8:s"), "The first chapter.") };
+        var rows = new List<(IReadOnlyList<ScriptureRef>, string)> { (ScrRef("MAT 1:0/9:s"), "The first chapter.") };
 
         string target = UpdateUsfm(rows);
         Assert.That(target, Contains.Substring("\\s The first chapter.\r\n"));
@@ -422,7 +422,7 @@ public void GetUsfm_NonVerse_KeepNote()
     {
         var rows = new List<(IReadOnlyList<ScriptureRef>, string)>
         {
-            (ScrRef("MAT 1:0/3:ip"), "The introductory paragraph.")
+            (ScrRef("MAT 1:0/4:ip"), "The introductory paragraph.")
         };
 
         string target = UpdateUsfm(rows, subComponentBehavior: UpdateUsfmIntraVerseMarkerBehavior.Preserve);
@@ -437,7 +437,7 @@ public void GetUsfm_NonVerse_SkipNote()
     {
         var rows = new List<(IReadOnlyList<ScriptureRef>, string)>
         {
-            (ScrRef("MAT 1:0/3:ip"), "The introductory paragraph.")
+            (ScrRef("MAT 1:0/4:ip"), "The introductory paragraph.")
         };
 
         string target = UpdateUsfm(rows, subComponentBehavior: UpdateUsfmIntraVerseMarkerBehavior.Strip);
@@ -449,8 +449,8 @@ public void GetUsfm_NonVerse_ReplaceNote()
     {
         var rows = new List<(IReadOnlyList<ScriptureRef>, string)>
         {
-            (ScrRef("MAT 1:0/3:ip"), "The introductory paragraph."),
-            (ScrRef("MAT 1:0/3:ip/1:fe"), "This is a new endnote.")
+            (ScrRef("MAT 1:0/4:ip"), "The introductory paragraph."),
+            (ScrRef("MAT 1:0/4:ip/1:fe"), "This is a new endnote.")
         };
 
         string target = UpdateUsfm(rows);
@@ -509,7 +509,7 @@ public void GetUsfm_Verse_PretranslationsBeforeText()
             (ScrRef("GEN 1:3"), "Pretranslations before the start"),
             (ScrRef("GEN 1:4"), "Pretranslations before the start"),
             (ScrRef("GEN 1:5"), "Pretranslations before the start"),
-            (ScrRef("MAT 1:0/3:ip"), "The introductory paragraph.")
+            (ScrRef("MAT 1:0/4:ip"), "The introductory paragraph.")
         };
 
         string target = UpdateUsfm(rows);