private static int Test_HtmlStringIdIndexOf_One(string html, string id, bool expected)
{
int nerror = 0;
var actual = EpubWizard.HtmlStringIdIndexOf(html, id) >= 0;
if (actual != expected)
{
nerror++;
App.Error($"ERROR: HtmlStringContainsId({html}, {id}) expected {expected} but got {actual}");
}
return(nerror);
}
public static string FindClosestAnchorHelper(string foundHtml, int maxPosition, IList <EpubChapter> chapterList, int currDepth, int maxDepth, ref string closest)
{
foreach (var chapter in chapterList)
{
// FAIL: for many ebooks, the anchors are complex values that unlikely to be found in
// any book. But for Fire at Red Lake, there's a set of transcriber's notes in their own
// chapter. The chapter anchor for these is "tn". No surprise, there's plenty of words
// that includes a 'tn' and therefore the transcribers notes are preferentially found.
var chpos = EpubWizard.HtmlStringIdIndexOf(foundHtml, chapter.HashLocation ?? "***NO ANCHOR***"); //was .Anchor
if (chpos >= 0 && chpos <= maxPosition)
{
// Found a possible match. The same html might be referred to by multiple chapters,
// so keep on going; there might be a better match.
closest = chapter.HashLocation; // .Anchor;
}
if ((currDepth + 1 < maxDepth) && (chapter.SubChapters.Count > 0))
{
FindClosestAnchorHelper(foundHtml, maxPosition, chapter.SubChapters, currDepth + 1, maxDepth, ref closest);
}
}
return(closest);
}
/// <summary>
/// Returns the id of the chapter that contains an anchor. This is used, for example, when
/// selecting an image and wanting to shift the chapter display.
/// </summary>
/// <param name="id"></param>
/// <returns></returns>
public static string GetChapterContainingId(EpubBookExt epubBook, string id, int preferredHtmlIndex)
{
if (string.IsNullOrEmpty(id))
{
// just return the first chapter
return(GetFirstChapter(epubBook.TableOfContents).HashLocation ?? ""); // ?? "";
}
// Step one: find the html with the id
var idList = EpubWizard.GetIdVariants(id);
var(foundHtml, foundIndex, foundHtmlName, foundId) = EpubWizard.FindHtmlContainingId(epubBook, idList, preferredHtmlIndex);
if (foundHtml == null)
{
if (id != "uiLog")
{
// uiLog isn't always findable for ... reasons
App.Error($"IMPOSSIBLE ERROR: completely unable to find id {id} ");
}
return(null);
}
var pos = EpubWizard.HtmlStringIdIndexOf(foundHtml, id);
string closest = null;
FindClosestAnchorHelper(foundHtml, pos, epubBook.TableOfContents, 0, 3, ref closest);
// Fixup #1: try the TOC directly
// FAIL: the order of the fixups is really important. BAEN 2013 short stories doesn't include chapter id values
// and they have one story with nested sections AND they have duplicate ID values (calibre_pb_1 etc.) AND each story is
// in its own HTML page. If you select the first story after the story with sub-stories, then we really want to find
// the story by chapter and don't want the previous story.
if (closest == null)
{
// FAIL: All of me a small town romance: the chapters don't have any anchors at all.
// Instead of looking for the chapter by id, look for it based on a matching
// filename. If it matches, return the Filename as the id.
// No, it's not quite an id, but it is close enough to work :-)
foreach (var chapter in epubBook.TableOfContents)
{
// The chapters here might have names like ../TextFiles/chapter.xhml
// while we're looking for plain TextFiles/chapter.xhml
// We have to return the raw chapter name because we'll use it later on.
// // // BUG: using the wrong name!!! var htmlFileNameVariants = MakeHtmlFileNameVariants(foundHtmlName);
var htmlFileNameVariants = MakeHtmlFileNameVariants(chapter.FileName());
foreach (var fname in htmlFileNameVariants)
{
if (fname == foundHtmlName)
{
closest = chapter.FileName();
}
}
}
}
// Fixup #2: maybe try the previous HTML
if (closest == null)
{
// Didn't find one; that's probably because we're in a gap. We need to find the same thing for the
// previous chapter, but with closest set to the end of the html.
if (foundIndex > 0)
{
foundHtml = FindHtmlByIndex(epubBook, foundIndex - 1);
FindClosestAnchorHelper(foundHtml, int.MaxValue, epubBook.TableOfContents, 0, 3, ref closest);
}
// First html, and still can't find anything? Give up, we're not going to find anything.
}
// All the fixups failed
if (closest == null)
{
App.Error($"ERROR: when asked for matching chapter, can't find it for {id}. Possibly the chapters have no anchors.");
}
return(closest);
}