Render embedded Bible paragraphs inline

src/embedded_bible.nim

@@ -30,6 +30,17 @@ proc verseKey(code: string, chapter, verse: int): string =
 proc chapterKey(code: string, chapter: int): string =
   code & "\t" & $chapter
 
+proc legacySegment(text: string): tuple[breakKind: SegmentBreak, text: string] =
+  const pilcrow = "\xC2\xB6"
+  let stripped = text.strip(leading = true, trailing = false)
+
+  if stripped.startsWith(pilcrow):
+    result.breakKind = sbParagraph
+    result.text = stripped[pilcrow.len .. ^1].strip(leading = true, trailing = false)
+  else:
+    result.breakKind = sbSame
+    result.text = text
+
 proc parseSegmentBreak(s: string): SegmentBreak =
   case s
   of "same": sbSame
@@ -64,14 +75,15 @@ proc loadBibleIndex(rows, translationName: string): BibleIndex =
         "invalid embedded " & translationName & " row: " & line)
 
     if parts.len == 4:
+      let legacy = legacySegment(parts[3])
       result.addSegment(BibleSegment(
         code: parts[0],
         chapter: parseInt(parts[1]),
         verse: parseInt(parts[2]),
         segment: 0,
-        breakKind: sbLine,
+        breakKind: legacy.breakKind,
         indent: 0,
-        text: parts[3]))
+        text: legacy.text))
     else:
       result.addSegment(BibleSegment(
         code: parts[0],
@@ -151,9 +163,41 @@ proc addFormattedSegment(
       lines[^1].add(" ")
     lines[^1].add(content)
   else:
-    if segment.breakKind == sbParagraph and hasPassageLine:
+    let isMarkerlessParagraphContinuation =
-      lines.add("")
+      segment.breakKind == sbParagraph and marker.len == 0 and hasPassageLine
-    lines.add(prefix & content)
+
+    if isMarkerlessParagraphContinuation:
+      if lines[^1].len > 0 and not lines[^1][^1].isSpaceAscii:
+        lines[^1].add(" ")
+      lines[^1].add(content)
+    else:
+      if segment.breakKind == sbParagraph and hasPassageLine:
+        lines.add("")
+      lines.add(prefix & content)
+
+proc normalizeFirstSelectedSegment(segment: BibleSegment): BibleSegment =
+  result = segment
+  if result.breakKind == sbSame:
+    result.breakKind = sbLine
+  elif result.breakKind == sbParagraph and result.segment == 0:
+    result.breakKind = sbLine
+
+proc addAllSegments(lines: var seq[string], index: BibleIndex, code: string) =
+  var emittedVerses: Table[string, bool]
+  var firstSelected = true
+
+  discard index.requireLastChapter(code)
+  for segment in index.segments:
+    if segment.code == code:
+      var adjusted = segment
+      if firstSelected:
+        adjusted = normalizeFirstSelectedSegment(adjusted)
+      lines.addFormattedSegment(adjusted, emittedVerses)
+      firstSelected = false
+
+proc addRangeSeparator(lines: var seq[string]) =
+  if lines.len > 1 and lines[^1].len > 0:
+    lines.add("")
 
 proc addVerseLines(
     lines: var seq[string],
@@ -170,8 +214,8 @@ proc addVerseLines(
     if segment.code == code and
         selected.hasKey(verseKey(segment.code, segment.chapter, segment.verse)):
       var adjusted = segment
-      if firstSelected and adjusted.breakKind == sbSame:
+      if firstSelected:
-        adjusted.breakKind = sbLine
+        adjusted = normalizeFirstSelectedSegment(adjusted)
       lines.addFormattedSegment(adjusted, emittedVerses)
       firstSelected = false
 
@@ -180,13 +224,11 @@ proc fetchReference(index: BibleIndex, reference: PassageReference): string =
   let code = reference.book.code
 
   if reference.ranges.len == 0:
-    var emittedVerses: Table[string, bool]
+    lines.addAllSegments(index, code)
-    discard index.requireLastChapter(code)
-    for segment in index.segments:
-      if segment.code == code:
-        lines.addFormattedSegment(segment, emittedVerses)
   else:
-    for range in reference.ranges:
+    for idx, range in reference.ranges:
+      if idx > 0:
+        lines.addRangeSeparator()
       lines.addVerseLines(index, reference, range)
 
   lines.join("\n")

tests/test_offline_kjv.nim

@@ -90,6 +90,9 @@ JHN	8	31	0	paragraph	0	Then Jesus said,
 JHN	8	31	1	same	0	"If you remain in My word,
 JHN	8	32	0	same	0	you shall know the truth."
 JHN	8	33	0	paragraph	0	They answered Him.
+JHN	8	39	0	paragraph	0	They answered Him, "Abraham is our father."
+JHN	8	39	1	paragraph	0	Jesus said to them, "If you were Abraham's children,
+JHN	8	40	0	same	0	you would do the works of Abraham."
 HEB	4	7	0	paragraph	0	again He establishes a certain day:
 HEB	4	7	1	line	0	"Today, if you will hear His voice,
 HEB	4	7	2	line	1	do not harden your hearts."
@@ -126,8 +129,28 @@ HEB	4	7	2	line	1	do not harden your hearts."
       "\n" &
       "  [33] They answered Him."
 
+  test "joins markerless paragraph continuations":
+    let passages = embedded_bible.fetchPassages(layoutRows, "John 8:39-40", "TEST")
+
+    check passages[0] == "John 8:39-40\n" &
+      "  [39] They answered Him, \"Abraham is our father.\" " &
+      "Jesus said to them, \"If you were Abraham's children, " &
+      "[40] you would do the works of Abraham.\""
+
   test "supports legacy verse rows":
     const legacyRows = "JHN\t3\t16\tFor God so loved the world.\n"
     let passages = embedded_bible.fetchPassages(legacyRows, "John 3:16", "TEST")
 
     check passages[0] == "John 3:16\n  [16] For God so loved the world."
+
+  test "groups legacy rows into paragraphs using pilcrows":
+    const legacyRows =
+      "JHN\t8\t31\tThen Jesus said.\n" &
+      "JHN\t8\t32\tYou shall know the truth.\n" &
+      "JHN\t8\t33\t\xC2\xB6 They answered Him.\n"
+    let passages = embedded_bible.fetchPassages(legacyRows, "John 8:31-33", "TEST")
+
+    check passages[0] == "John 8:31-33\n" &
+      "  [31] Then Jesus said. [32] You shall know the truth.\n" &
+      "\n" &
+      "  [33] They answered Him."