private Block ProcessChapterLabelNode(string nodeText, UsxNode usxNode, IList <Block> blocks)
{
var lastChapterAnnouncementBlock = blocks.LastOrDefault(b => b.IsChapterAnnouncement);
// Chapter label before the first chapter means we have a chapter label which applies to all chapters
if (lastChapterAnnouncementBlock == null)
{
m_bookLevelChapterLabel = nodeText;
}
else if (m_bookLevelChapterLabel == null && blocks.Last() == lastChapterAnnouncementBlock)
{
// The node before this was the chapter. We already added it, then found this label.
// Remove that block so it will be replaced with this one.
blocks.RemoveAt(blocks.Count - 1);
var block = new Block(usxNode.StyleTag, m_currentChapter)
{
IsParagraphStart = true, BookCode = m_bookId
};
block.SetStandardCharacter(m_bookId, CharacterVerseData.StandardCharacter.BookOrChapter);
block.BlockElements.Add(new ScriptText(nodeText));
m_currentStartVerse = 0;
m_currentEndVerse = 0;
return(block);
}
// Note: In PG-1140, it was reported that errant/misplaced \cl markers (at the end of the Pauline epistles) were causing
// the preceding block to go AWOL. We aren't sure what the text in the \cl field was supposed to be (it was the same in
// all the books and ended with a comma, which seemed pretty weird), but we decided to just ignore any \cl marker coming
// in an unexpected place. Ideally, the Markers check in Paratext should catch this as an error (and I have requested this),
// but at least as of now, it does not.
return(null);
}
private Block ProcessChapterLabelNode(string nodeText, UsxNode usxNode)
{
Block block = null;
// Chapter label before the first chapter means we have a chapter label which applies to all chapters
if (!m_chapterNodeFound)
{
m_bookLevelChapterLabel = nodeText;
}
if (m_bookLevelChapterLabel == null)
{
block = new Block(usxNode.StyleTag, m_currentChapter)
{
IsParagraphStart = true
};
block.SetStandardCharacter(m_bookId, CharacterVerseData.StandardCharacter.BookOrChapter);
block.BlockElements.Add(new ScriptText(nodeText));
m_currentStartVerse = 0;
m_currentEndVerse = 0;
}
return(block);
}
public IEnumerable <Block> Parse()
{
var titleBuilder = new StringBuilder();
IList <Block> blocks = new List <Block>();
foreach (XmlNode node in m_nodeList)
{
Block block = null;
switch (node.Name)
{
case "chapter":
AddMainTitleIfApplicable(blocks, titleBuilder);
m_chapterNodeFound = true;
block = ProcessChapterNode(node);
if (block == null)
{
continue;
}
break;
case "para":
if (!node.HasChildNodes)
{
continue;
}
var usxPara = new UsxNode(node);
IStyle style = m_stylesheet.GetStyle(usxPara.StyleTag);
if (style.IsChapterLabel)
{
block = ProcessChapterLabelNode(node.InnerText, usxPara);
if (block == null)
{
continue;
}
// The node before this was the chapter. We already added it, then found this label.
// Remove that block so it will be replaced with this one.
blocks.RemoveAt(blocks.Count - 1);
break;
}
if (style.IsParallelPassageReference || !style.IsPublishable)
{
continue;
}
if (style.HoldsBookNameOrAbbreviation)
{
if (style.Id.StartsWith("mt"))
{
titleBuilder.Append(node.InnerText).Append(" ");
if (style.Id == "mt1")
{
MainTitle = node.InnerText;
}
}
else if (style.Id == "h")
{
PageHeader = node.InnerText;
}
continue;
}
AddMainTitleIfApplicable(blocks, titleBuilder);
block = new Block(usxPara.StyleTag, m_currentChapter, m_currentStartVerse, m_currentEndVerse)
{
IsParagraphStart = true
};
if (m_currentChapter == 0)
{
block.SetStandardCharacter(m_bookId, CharacterVerseData.StandardCharacter.Intro);
}
else if (style.IsPublishable && !style.IsVerseText)
{
block.SetStandardCharacter(m_bookId, CharacterVerseData.StandardCharacter.ExtraBiblical);
}
var sb = new StringBuilder();
// <verse number="1" style="v" />
// Acakki me lok me kwena maber i kom Yecu Kricito, Wod pa Lubaŋa,
// <verse number="2" style="v" />
// <note caller="-" style="x"><char style="xo" closed="false">1.2: </char><char style="xt" closed="false">Mal 3.1</char></note>
// kit ma gicoyo kwede i buk pa lanebi Icaya ni,</para>
foreach (XmlNode childNode in usxPara.ChildNodes)
{
switch (childNode.Name)
{
case "verse":
if (sb.Length > 0)
{
sb.TrimStart();
block.BlockElements.Add(new ScriptText(sb.ToString()));
sb.Clear();
}
RemoveLastElementIfVerse(block);
var verseNumStr = childNode.Attributes.GetNamedItem("number").Value;
m_currentStartVerse = ScrReference.VerseToIntStart(verseNumStr);
m_currentEndVerse = ScrReference.VerseToIntEnd(verseNumStr);
if (!block.BlockElements.Any() ||
(block.BlockElements.Count == 1 && block.FirstTextElementIsOnlyPunctuation))
{
block.InitialStartVerseNumber = m_currentStartVerse;
block.InitialEndVerseNumber = m_currentEndVerse;
}
block.BlockElements.Add(new Verse(verseNumStr));
break;
case "char":
IStyle charStyle = m_stylesheet.GetStyle((new UsxNode(childNode)).StyleTag);
if (!charStyle.IsInlineQuotationReference && charStyle.IsPublishable)
{
sb.Append(childNode.InnerText);
}
break;
case "#text":
sb.Append(childNode.InnerText);
break;
case "#whitespace":
if (sb.Length > 0 && sb[sb.Length - 1] != ' ')
{
sb.Append(" ");
}
break;
}
}
sb.TrimStart();
if (sb.Length > 0)
{
block.BlockElements.Add(new ScriptText(sb.ToString()));
sb.Clear();
}
RemoveLastElementIfVerse(block);
break;
}
if (block != null && block.BlockElements.Count > 0)
{
blocks.Add(block);
}
}
return(blocks);
}
public void StyleTag_GetsCorrectStyleTagFromUsxNode()
{
var usxNode = new UsxNode(m_nodes[0]);
Assert.AreEqual("c", usxNode.StyleTag);
}
public IEnumerable <Block> Parse()
{
var titleBuilder = new StringBuilder();
IList <Block> blocks = new List <Block>();
foreach (XmlNode node in m_nodeList)
{
Block block = null;
switch (node.Name)
{
case "chapter":
AddMainTitleIfApplicable(blocks, titleBuilder);
block = ProcessChapterNode(node);
if (block == null)
{
continue;
}
break;
case "para":
if (!node.HasChildNodes)
{
continue;
}
var usxPara = new UsxNode(node);
IStyle style = m_stylesheet.GetStyle(usxPara.StyleTag);
if (style.IsChapterLabel)
{
block = ProcessChapterLabelNode(node.InnerText, usxPara, blocks);
break;
}
if (style.IsParallelPassageReference || !style.IsPublishable)
{
continue;
}
if (style.HoldsBookNameOrAbbreviation)
{
if (style.Id.StartsWith("mt"))
{
titleBuilder.Append(node.InnerText).Append(" ");
if (style.Id == "mt1")
{
MainTitle = node.InnerText;
}
}
else if (style.Id == "h")
{
PageHeader = node.InnerText;
}
continue;
}
AddMainTitleIfApplicable(blocks, titleBuilder);
block = new Block(usxPara.StyleTag, m_currentChapter, m_currentStartVerse, m_currentEndVerse)
{
IsParagraphStart = true
};
if (m_currentChapter == 0)
{
block.SetStandardCharacter(m_bookId, CharacterVerseData.StandardCharacter.Intro);
}
else if (style.IsPublishable && !style.IsVerseText)
{
block.SetStandardCharacter(m_bookId, CharacterVerseData.StandardCharacter.ExtraBiblical);
}
var sb = new StringBuilder();
// <verse number="1" style="v" />
// Acakki me lok me kwena maber i kom Yecu Kricito, Wod pa Lubaŋa,
// <verse number="2" style="v" />
// <note caller="-" style="x"><char style="xo" closed="false">1.2: </char><char style="xt" closed="false">Mal 3.1</char></note>
// kit ma gicoyo kwede i buk pa lanebi Icaya ni,</para>
foreach (XmlNode childNode in usxPara.ChildNodes)
{
switch (childNode.Name)
{
case "verse":
var verseNumAttr = childNode.Attributes.GetNamedItem("number");
if (verseNumAttr == null)
{
continue; // presumably this is an end-verse element
}
if (sb.Length > 0)
{
sb.TrimStart();
block.BlockElements.Add(new ScriptText(sb.ToString()));
sb.Clear();
}
var verseNumStr = verseNumAttr.Value;
if (!HandleVerseBridgeInSeparateVerseFields(block, ref verseNumStr))
{
RemoveEmptyTrailingVerse(block);
m_currentStartVerse = BCVRef.VerseToIntStart(verseNumStr);
m_currentEndVerse = BCVRef.VerseToIntEnd(verseNumStr);
}
if (!block.BlockElements.Any() ||
(block.BlockElements.Count == 1 && block.StartsWithScriptTextElementContainingOnlyPunctuation))
{
block.InitialStartVerseNumber = m_currentStartVerse;
block.InitialEndVerseNumber = m_currentEndVerse;
}
block.BlockElements.Add(new Verse(verseNumStr));
break;
case "char":
var charTag = (new UsxNode(childNode)).StyleTag;
IStyle charStyle = m_stylesheet.GetStyle(charTag);
if (!charStyle.IsInlineQuotationReference && charStyle.IsPublishable)
{
// Starting with USFM 3.0, char styles can have attributes separated by |
var tokens = childNode.InnerText.Split('|');
if (tokens.Any())
{
if (ControlCharacterVerseData.TryGetCharacterForCharStyle(charTag, out var character) && block.StyleTag != charTag)
{
FinalizeCharacterStyleBlock(sb, ref block, blocks, charTag);
block.CharacterId = character;
//ControlCharacterVerseData.Singleton.GetCharacters(m_bookNum, block.ChapterNumber,
//block.InitialStartVerseNumber, block.InitialEndVerseNumber, block.LastVerseNum, m_versification, true)
//.FirstOrDefault(cv => cv.Character == character)?.Character ?? CharacterVerseData.kNeedsReview;
}
sb.Append(tokens[0]);
}
}
break;
case "#text":
var textToAppend = childNode.InnerText;
if (ControlCharacterVerseData.IsCharStyleThatMapsToSpecificCharacter(block.StyleTag) && textToAppend.Any(IsLetter))
{
if (sb.Length > 0 && sb[sb.Length - 1] != ' ' && textToAppend.StartsWith(" "))
{
// Not terribly important (in fact, we don't even need the trailing space either), but if blocks have
// leading spaces, it might look funny in some places in the display.
sb.Append(" ");
textToAppend = textToAppend.TrimStart();
}
FinalizeCharacterStyleBlock(sb, ref block, blocks, usxPara.StyleTag);
}
sb.Append(textToAppend);
break;
case "#whitespace":
if (sb.Length > 0 && sb[sb.Length - 1] != ' ')
{
sb.Append(" ");
}
break;
}
}
FlushStringBuilderToBlockElement(sb, block);
if (RemoveEmptyTrailingVerse(block))
{
var lastVerse = block.LastVerse;
m_currentStartVerse = lastVerse.StartVerse;
m_currentEndVerse = lastVerse.EndVerse;
}
break;
}
if (block != null && block.BlockElements.Count > 0)
{
blocks.Add(block);
}
}
return(blocks);
}
