Preserve layout in embedded Bible data

src/embedded_bible.nim

@@ -2,11 +2,27 @@ import std/[strutils, tables]
 
 import ./reference_parser
 
-type BibleIndex = object
-  verses: Table[string, string]
-  lastVerseByChapter: Table[string, int]
-  lastChapterByBook: Table[string, int]
-  translationName: string
+type
+  SegmentBreak = enum
+    sbSame = "same"
+    sbLine = "line"
+    sbParagraph = "paragraph"
+
+  BibleSegment = object
+    code: string
+    chapter: int
+    verse: int
+    segment: int
+    breakKind: SegmentBreak
+    indent: int
+    text: string
+
+  BibleIndex = object
+    segments: seq[BibleSegment]
+    verses: Table[string, bool]
+    lastVerseByChapter: Table[string, int]
+    lastChapterByBook: Table[string, int]
+    translationName: string
 
 proc verseKey(code: string, chapter, verse: int): string =
   code & "\t" & $chapter & "\t" & $verse
@@ -14,6 +30,27 @@ proc verseKey(code: string, chapter, verse: int): string =
 proc chapterKey(code: string, chapter: int): string =
   code & "\t" & $chapter
 
+proc parseSegmentBreak(s: string): SegmentBreak =
+  case s
+  of "same": sbSame
+  of "line": sbLine
+  of "paragraph": sbParagraph
+  else:
+    raise newException(ValueError, "invalid embedded Bible segment break: " & s)
+
+proc addSegment(index: var BibleIndex, segment: BibleSegment) =
+  index.segments.add(segment)
+  index.verses[verseKey(segment.code, segment.chapter, segment.verse)] = true
+
+  let cKey = chapterKey(segment.code, segment.chapter)
+  if not index.lastVerseByChapter.hasKey(cKey) or
+      segment.verse > index.lastVerseByChapter[cKey]:
+    index.lastVerseByChapter[cKey] = segment.verse
+
+  if not index.lastChapterByBook.hasKey(segment.code) or
+      segment.chapter > index.lastChapterByBook[segment.code]:
+    index.lastChapterByBook[segment.code] = segment.chapter
+
 proc loadBibleIndex(rows, translationName: string): BibleIndex =
   result.translationName = translationName
 
@@ -21,26 +58,29 @@ proc loadBibleIndex(rows, translationName: string): BibleIndex =
     if line.strip.len == 0:
       continue
 
-    let parts = line.split('\t', maxsplit = 3)
-    if parts.len != 4:
+    let parts = line.split('\t', maxsplit = 6)
+    if parts.len != 4 and parts.len != 7:
       raise newException(ValueError,
         "invalid embedded " & translationName & " row: " & line)
 
-    let code = parts[0]
-    let chapter = parseInt(parts[1])
-    let verse = parseInt(parts[2])
-    let text = parts[3]
-
-    result.verses[verseKey(code, chapter, verse)] = text
-
-    let cKey = chapterKey(code, chapter)
-    if not result.lastVerseByChapter.hasKey(cKey) or
-        verse > result.lastVerseByChapter[cKey]:
-      result.lastVerseByChapter[cKey] = verse
-
-    if not result.lastChapterByBook.hasKey(code) or
-        chapter > result.lastChapterByBook[code]:
-      result.lastChapterByBook[code] = chapter
+    if parts.len == 4:
+      result.addSegment(BibleSegment(
+        code: parts[0],
+        chapter: parseInt(parts[1]),
+        verse: parseInt(parts[2]),
+        segment: 0,
+        breakKind: sbLine,
+        indent: 0,
+        text: parts[3]))
+    else:
+      result.addSegment(BibleSegment(
+        code: parts[0],
+        chapter: parseInt(parts[1]),
+        verse: parseInt(parts[2]),
+        segment: parseInt(parts[3]),
+        breakKind: parseSegmentBreak(parts[4]),
+        indent: parseInt(parts[5]),
+        text: parts[6]))
 
 proc requireLastChapter(index: BibleIndex, code: string): int =
   if not index.lastChapterByBook.hasKey(code):
@@ -62,13 +102,10 @@ proc requireVerse(index: BibleIndex, code: string, chapter, verse: int): string
     raise newException(ValueError,
       "no embedded " & index.translationName & " data for " &
         bookInfo(code).name & " " & $chapter & ":" & $verse)
-  index.verses[vKey]
+  ""
 
-proc addVerseLines(
-    lines: var seq[string],
-    index: BibleIndex,
-    reference: PassageReference,
-    range: RefRange) =
+proc selectedVerses(index: BibleIndex, reference: PassageReference, range: RefRange):
+    Table[string, bool] =
 
   let code = reference.book.code
   discard index.requireLastChapter(code)
@@ -90,16 +127,64 @@ proc addVerseLines(
       raise newException(ValueError, "reference range starts after it ends")
 
     for verse in startVerse .. endVerse:
-      lines.add("  [" & $verse & "] " & index.requireVerse(code, chapter, verse))
+      discard index.requireVerse(code, chapter, verse)
+      result[verseKey(code, chapter, verse)] = true
+
+proc addFormattedSegment(
+    lines: var seq[string],
+    segment: BibleSegment,
+    emittedVerses: var Table[string, bool]) =
+
+  let vKey = verseKey(segment.code, segment.chapter, segment.verse)
+  let marker =
+    if emittedVerses.hasKey(vKey): ""
+    else:
+      emittedVerses[vKey] = true
+      "[" & $segment.verse & "] "
+
+  let prefix = repeat("  ", segment.indent + 1)
+  let content = marker & segment.text
+  let hasPassageLine = lines.len > 1 and lines[^1].len > 0
+
+  if segment.breakKind == sbSame and hasPassageLine:
+    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 addVerseLines(
+    lines: var seq[string],
+    index: BibleIndex,
+    reference: PassageReference,
+    range: RefRange) =
+
+  let code = reference.book.code
+  let selected = index.selectedVerses(reference, range)
+  var emittedVerses: Table[string, bool]
+  var firstSelected = true
+
+  for segment in index.segments:
+    if segment.code == code and
+        selected.hasKey(verseKey(segment.code, segment.chapter, segment.verse)):
+      var adjusted = segment
+      if firstSelected and adjusted.breakKind == sbSame:
+        adjusted.breakKind = sbLine
+      lines.addFormattedSegment(adjusted, emittedVerses)
+      firstSelected = false
 
 proc fetchReference(index: BibleIndex, reference: PassageReference): string =
   var lines = @[$reference]
   let code = reference.book.code
 
   if reference.ranges.len == 0:
-    for chapter in 1 .. index.requireLastChapter(code):
-      for verse in 1 .. index.requireLastVerse(code, chapter):
-        lines.add("  [" & $verse & "] " & index.requireVerse(code, chapter, verse))
+    var emittedVerses: Table[string, bool]
+    discard index.requireLastChapter(code)
+    for segment in index.segments:
+      if segment.code == code:
+        lines.addFormattedSegment(segment, emittedVerses)
   else:
     for range in reference.ranges:
       lines.addVerseLines(index, reference, range)