public void AddLocations(XRay xray,
KfxContainer kfx,
bool skipNoLikes,
int minClipLen,
IProgressBar progress,
CancellationToken token)
{
_logger.Log("Scanning book content...");
var paragraphs = _paragraphsService.GetParagraphs(kfx).ToArray();
// Set start and end of content
// TODO Figure out how to identify the first *actual* bit of content after the TOC
var last = paragraphs.Last();
xray.Srl = 1;
xray.Erl = last.Location + last.Length - 1;
progress?.Set(0, paragraphs.Length);
foreach (var paragraph in paragraphs)
{
token.ThrowIfCancellationRequested();
foreach (var character in xray.Terms.Where(term => term.Match))
{
var occurrences = _termsService.FindOccurrences(kfx, character, paragraph);
if (!occurrences.Any())
{
continue;
}
character.Occurrences.UnionWith(occurrences);
ExcerptHelper.EnhanceOrAddExcerpts(xray.Excerpts, character.Id, new IndexLength(paragraph.Location, paragraph.Length));
}
// Attempt to match downloaded notable clips, not worried if no matches occur as some will be added later anyway
if (xray.NotableClips != null)
{
ExcerptHelper.ProcessNotablesForParagraph(paragraph.ContentText, paragraph.Location, xray.NotableClips, xray.Excerpts, skipNoLikes, minClipLen);
}
progress?.Add(1);
}
var missingOccurrences = xray.Terms
.Where(term => term.Match && term.Occurrences.Count == 0)
.Select(term => term.TermName)
.ToArray();
if (!missingOccurrences.Any())
{
return;
}
var termList = string.Join(", ", missingOccurrences);
_logger.Log($"\r\nNo locations were found for the following terms. You should add aliases for them using the book as a reference:\r\n{termList}\r\n");
}
public void DefaultTocTest(string kfxFile, int tocLength)
{
var fs = new FileStream(kfxFile, FileMode.Open, FileAccess.Read);
var kfx = new KfxContainer(fs);
var toc = kfx.GetDefaultToc();
Assert.NotNull(toc);
Assert.AreEqual(tocLength, toc.Count);
}
public void GetPageCountTest(string kfxFile, int pages)
{
var fs = new FileStream(kfxFile, FileMode.Open, FileAccess.Read);
var kfx = new KfxContainer(fs);
var pageCount = kfx.GetPageCount();
Assert.NotNull(pageCount);
Assert.AreEqual(pages, pageCount);
}
public void CoverImageTest(string kfxFile, int height, int width)
{
var fs = new FileStream(kfxFile, FileMode.Open, FileAccess.Read);
var kfx = new KfxContainer(fs);
var coverImage = kfx.CoverImage;
Assert.NotNull(coverImage);
Assert.AreEqual(height, coverImage.Height);
Assert.AreEqual(width, coverImage.Width);
}
public void ContentTest(string kfxFile, string search, int firstOffset, int lastOffset, int chunkCount, long sum)
{
var fs = new FileStream(kfxFile, FileMode.Open, FileAccess.Read);
var kfx = new KfxContainer(fs);
var contentChunks = kfx.GetContentChunks();
var testSearch = FindInChunks(contentChunks, search).ToArray();
Assert.AreEqual(chunkCount, contentChunks.Count);
Assert.AreEqual(firstOffset, testSearch.First());
Assert.AreEqual(lastOffset, testSearch.Last());
Assert.AreEqual(sum, testSearch.Sum());
}
public HashSet <Occurrence> FindOccurrences(IMetadata metadata, Term term, Paragraph paragraph)
{
if (!term.Match)
{
return(new HashSet <Occurrence>());
}
return(metadata switch
{
MobiMetadata _ => FindOccurrencesLegacy(term, paragraph),
KfxContainer _ => FindOccurrences(term, paragraph),
_ => FindOccurrencesLegacy(term, paragraph)
});
public void GetKfxContainerMetadataTest(string kfxFile)
{
var fs = new FileStream(kfxFile, FileMode.Open, FileAccess.Read);
var kfx = new KfxContainer(fs);
Assert.AreEqual("B018LJYLS8", kfx.Asin);
Assert.AreEqual("Shelley, Mary W.", kfx.Author);
Assert.AreEqual("EBOK", kfx.CdeContentType);
Assert.AreEqual(4096, kfx.ContainerInfo.ChunkSize);
Assert.AreEqual(0, kfx.ContainerInfo.CompressionType);
Assert.AreEqual(YjContainer.ContainerFormat.KfxMain, kfx.ContainerInfo.ContainerFormat);
Assert.AreEqual("CR!ZLWPJZFVMQ5HGT49NILVDTAKVNRN", kfx.ContainerInfo.ContainerId);
Assert.AreEqual(0, kfx.ContainerInfo.DrmScheme);
Assert.AreEqual("CONT", kfx.ContainerInfo.Header.Signature);
Assert.AreEqual(2, kfx.ContainerInfo.Header.Version);
Assert.AreEqual("KPR-3.28.1", kfx.ContainerInfo.KfxGenApplicationVersion);
Assert.AreEqual("kfxlib-20181220", kfx.ContainerInfo.KfxGenPackageVersion);
Assert.AreEqual("Frankenstein", kfx.Title);
}
public static IMetadata Load(string file)
{
using var fs = new FileStream(file, FileMode.Open, FileAccess.Read);
IMetadata metadata;
switch (Path.GetExtension(file))
{
case ".azw3":
case ".mobi":
metadata = new Metadata(fs);
break;
case ".kfx":
metadata = new KfxContainer(fs);
break;
default:
throw new NotSupportedException("Unsupported book format");
}
return(metadata);
}
/// <summary>
/// Builds an X-Ray file from the parameters given and returns the path at which the file has been saved (or null if something failed)
/// </summary>
public async Task <string> BuildAsync([NotNull] Request request, CancellationToken cancellationToken)
{
using var metadata = await GetAndValidateMetadataAsync(request.BookPath, cancellationToken);
if (metadata == null)
{
return(null);
}
var dataSource = string.IsNullOrEmpty(request.DataUrl) || request.DataUrl == SecondarySourceRoentgen.FakeUrl
? _secondaryDataSourceFactory.Get(SecondaryDataSourceFactory.Enum.Roentgen)
: _secondaryDataSourceFactory.GetInferredSource(request.DataUrl);
if (dataSource == null)
{
_logger.Log("Data source could not be determined from the given path.");
return(null);
}
Core.XRay.XRay xray;
try
{
xray = await _xrayService.CreateXRayAsync(request.DataUrl, metadata.DbName, metadata.UniqueId, metadata.Asin, request.AmazonTld ?? "com", request.IncludeTopics, dataSource, _progress, cancellationToken);
if (xray.Terms.Count == 0)
{
_logger.Log($"No terms were available on {dataSource.Name}, cancelling the build...");
return(null);
}
var aliasPath = _directoryService.GetAliasPath(xray.Asin);
_xrayService.ExportAndDisplayTerms(xray, dataSource, false, request.SplitAliases);
if (xray.Terms.Any(term => term.Aliases?.Count > 0))
{
_logger.Log("Character aliases read from the XML file.");
}
else if (!File.Exists(aliasPath))
{
_logger.Log("Aliases file not found.");
}
else
{
_aliasesRepository.LoadAliasesForXRay(xray);
_logger.Log($"Character aliases read from {aliasPath}.");
}
_logger.Log("Initial X-Ray built, adding locations and chapters...");
//Expand the X-Ray file from the unpacked mobi
Task buildTask = metadata switch
{
// ReSharper disable AccessToDisposedClosure
MobiMetadata _ => Task.Run(() => _xrayService.ExpandFromRawMl(xray, metadata, metadata.GetRawMlStream(), true, true, 25, true, null, _progress, cancellationToken, true, false), cancellationToken),
KfxContainer kfx => Task.Run(() => _kfxXrayService.AddLocations(xray, kfx, true, 25, _progress, cancellationToken), cancellationToken),
_ => throw new NotSupportedException()
};
await buildTask.ConfigureAwait(false);
}
catch (OperationCanceledException)
{
_logger.Log("Build canceled.");
return(null);
}
catch (Exception ex)
{
_logger.Log($"An error occurred while building the X-Ray:\r\n{ex.Message}\r\n{ex.StackTrace}");
return(null);
}
_logger.Log("Saving X-Ray to file...");
var xrayPath = _directoryService.GetArtifactPath(ArtifactType.XRay, metadata, Path.GetFileNameWithoutExtension(request.BookPath), true);
try
{
var xrayExporter = _xrayExporterFactory.Get(XRayExporterFactory.Enum.Sqlite);
xrayExporter.Export(xray, xrayPath, _progress, cancellationToken);
}
catch (OperationCanceledException)
{
_logger.Log("Building canceled.");
return(null);
}
catch (Exception ex)
{
// TODO: Add option to retry maybe?
_logger.Log($"An error occurred while creating the X-Ray file. Is it opened in another program?\r\n{ex.Message}");
return(null);
}
_logger.Log($"X-Ray file created successfully!\r\nSaved to {xrayPath}");
return(xrayPath);
}
public void AddLocations(XRay xray,
KfxContainer kfx,
bool skipNoLikes,
int minClipLen,
IProgressBar progress,
CancellationToken token)
{
_logger.Log("Scanning book content...");
var contentChunks = kfx.GetContentChunks();
// Set start and end of content
// TODO Figure out how to identify the first *actual* bit of content after the TOC
var last = contentChunks.Last();
xray.Srl = 1;
xray.Erl = last.Pid + last.Length - 1;
var offset = 0;
var excerptId = 0;
progress?.Set(0, contentChunks.Count);
foreach (var contentChunk in contentChunks)
{
token.ThrowIfCancellationRequested();
if (contentChunk.ContentText != null)
{
foreach (var character in xray.Terms.Where(term => term.Match))
{
// If the aliases are not supposed to be in regex format, escape them
var aliases = character.RegexAliases
? character.Aliases
: character.Aliases.Select(Regex.Escape);
var searchList = new[] { character.TermName }.Concat(aliases).ToArray();
//Search content for character name and aliases, respecting the case setting
var regexOptions = character.MatchCase || character.RegexAliases
? RegexOptions.None
: RegexOptions.IgnoreCase;
var currentOffset = offset;
var highlights = searchList
.Select(search => Regex.Matches(contentChunk.ContentText, $@"{Quotes}?\b{search}{_punctuationMarks}", regexOptions))
.SelectMany(matches => matches.Cast <Match>())
.ToLookup(match => currentOffset + match.Index, match => match.Length);
if (highlights.Count == 0)
{
continue;
}
var highlightOccurrences = highlights.SelectMany(highlightGroup => highlightGroup.Select(highlight => new[] { highlightGroup.Key, highlight }));
character.Occurrences.AddRange(highlightOccurrences);
// Check excerpts
var exCheck = xray.Excerpts.Where(t => t.Start.Equals(offset)).ToArray();
if (exCheck.Length > 0)
{
if (!exCheck[0].RelatedEntities.Contains(character.Id))
{
exCheck[0].RelatedEntities.Add(character.Id);
}
}
else
{
var newExcerpt = new Excerpt
{
Id = excerptId++,
Start = offset,
Length = contentChunk.Length
};
newExcerpt.RelatedEntities.Add(character.Id);
xray.Excerpts.Add(newExcerpt);
}
}
// Attempt to match downloaded notable clips, not worried if no matches occur as some will be added later anyway
if (xray.NotableClips != null)
{
foreach (var quote in xray.NotableClips)
{
var index = contentChunk.ContentText.IndexOf(quote.Text, StringComparison.Ordinal);
if (index <= -1)
{
continue;
}
// See if an excerpt already exists at this location
var excerpt = xray.Excerpts.FirstOrDefault(e => e.Start == index);
if (excerpt == null)
{
if (skipNoLikes && quote.Likes == 0 ||
quote.Text.Length < minClipLen)
{
continue;
}
excerpt = new Excerpt
{
Id = excerptId++,
Start = offset,
Length = contentChunk.Length,
Notable = true,
Highlights = quote.Likes
};
excerpt.RelatedEntities.Add(0); // Mark the excerpt as notable
// TODO: also add other related entities
xray.Excerpts.Add(excerpt);
}
else
{
excerpt.RelatedEntities.Add(0);
}
xray.FoundNotables++;
}
}
progress?.Add(1);
}
offset += contentChunk.Length;
}
var missingOccurrences = xray.Terms
.Where(term => term.Match && term.Occurrences.Count == 0)
.Select(term => term.TermName)
.ToArray();
if (!missingOccurrences.Any())
{
return;
}
var termList = string.Join(", ", missingOccurrences);
_logger.Log($"\r\nNo locations were found for the following terms. You should add aliases for them using the book as a reference:\r\n{termList}\r\n");
}
