Adjust embeds to be placed directly before empty end-of-verse paragraph markers (#182)

Author: isaac091

machine/corpora/place_markers_usfm_update_block_handler.py

@@ -73,6 +73,7 @@ def process_block(self, block: UsfmUpdateBlock) -> UsfmUpdateBlock:
         to_place = []
         adj_src_toks = []
         placed_elements = [elements.pop(0)] if elements[0].type == UsfmUpdateBlockElementType.OTHER else []
+        embed_elements = []
         ignored_elements = []
         for element in elements:
             if element.type == UsfmUpdateBlockElementType.TEXT:
@@ -90,8 +91,10 @@ def process_block(self, block: UsfmUpdateBlock) -> UsfmUpdateBlock:
                 else:
                     trg_sent += element.tokens[0].to_usfm()
 
-            if element.marked_for_removal or element.type == UsfmUpdateBlockElementType.EMBED:
+            if element.marked_for_removal:
                 ignored_elements.append(element)
+            elif element.type == UsfmUpdateBlockElementType.EMBED:
+                embed_elements.append(element)
             elif element.type in [UsfmUpdateBlockElementType.PARAGRAPH, UsfmUpdateBlockElementType.STYLE]:
                 to_place.append(element)
                 adj_src_toks.append(src_tok_idx)
@@ -110,7 +113,7 @@ def process_block(self, block: UsfmUpdateBlock) -> UsfmUpdateBlock:
 
             to_insert.append((trg_str_idx, element))
         to_insert.sort(key=lambda x: x[0])
-        to_insert += [(len(trg_sent), element) for element in end_elements]
+        to_insert += [(len(trg_sent), element) for element in embed_elements + end_elements]
 
         # Construct new text tokens to put between markers
         # and reincorporate headers and empty end-of-verse paragraph markers

tests/corpora/test_place_markers_usfm_update_block_handler.py

@@ -161,10 +161,10 @@ def test_trailing_empty_paragraphs() -> None:
     rows = [(scr_ref("MAT 1:1"), "New verse 1")]
     usfm = r"""\id MAT
 \c 1
-\v 1 Verse 1
+\v 1 \f embed 1 \f*Verse 1
 \p
 \b
-\q1 \f embed \f*
+\q1 \f embed 2 \f*
 """
 
     align_info = [
@@ -183,10 +183,10 @@ def test_trailing_empty_paragraphs() -> None:
     )
     result = r"""\id MAT
 \c 1
-\v 1 New verse 1
+\v 1 New verse 1 \f embed 1 \f*\f embed 2 \f*
 \p
 \b
-\q1 \f embed \f*
+\q1
 """
     assess(target, result)
 
@@ -419,6 +419,35 @@ def test_split_tokens() -> None:
     assess(target, result)
 
 
+def test_no_text() -> None:
+    rows = [(scr_ref("MAT 1:1"), "")]
+    usfm = r"""\id MAT
+\c 1
+\v 1 \w \w*
+"""
+
+    align_info = [
+        PlaceMarkersAlignmentInfo(
+            refs=["MAT 1:1"],
+            source_tokens=[],
+            translation_tokens=[],
+            alignment=to_word_alignment_matrix(""),
+        ),
+    ]
+    target = update_usfm(
+        rows,
+        usfm,
+        paragraph_behavior=UpdateUsfmMarkerBehavior.PRESERVE,
+        style_behavior=UpdateUsfmMarkerBehavior.PRESERVE,
+        update_block_handlers=[PlaceMarkersUsfmUpdateBlockHandler(align_info)],
+    )
+    result = r"""\id MAT
+\c 1
+\v 1  \w \w*
+"""
+    assess(target, result)
+
+
 def scr_ref(*refs: str) -> List[ScriptureRef]:
     return [ScriptureRef.parse(ref) for ref in refs]