| public GetBookSetting ( string key ) : MultiTextBase | ||
| key | string | |
| return | MultiTextBase | |
private static bool ShouldSetToDefaultCopyrightAndLicense(HtmlDom dom)
{
var hasCopyright = !dom.GetBookSetting("copyright").Empty;
var hasLicenseUrl = !dom.GetBookSetting("licenseUrl").Empty;
var hasLicenseNotes = !dom.GetBookSetting("licenseNotes").Empty;
//Enhance: this logic is perhaps overly restrictive?
return(!hasCopyright && !hasLicenseUrl && !hasLicenseNotes);
}
private static void CopyItemToFieldsInPages(HtmlDom dom, string key, string valueAttribute = null, string[] languagePreferences = null)
{
if (languagePreferences == null)
{
languagePreferences = new[] { "*", "en" }
}
;
MultiTextBase source = dom.GetBookSetting(key);
if (key == "copyright")
{
// For CC0, we store the "copyright", but don't display it in the text of the book.
var licenseUrl = dom.GetBookSetting("licenseUrl").GetExactAlternative("*");
if (licenseUrl == CreativeCommonsLicense.CC0Url)
{
source = new MultiTextBase();
}
}
foreach (XmlElement target in dom.SafeSelectNodes("//*[@data-derived='" + key + "']"))
{
//just put value into the text of the element
if (string.IsNullOrEmpty(valueAttribute))
{
//clear out what's there now
target.RemoveAttribute("lang");
target.InnerText = "";
var form = source.GetBestAlternative(languagePreferences);
if (form != null && !string.IsNullOrWhiteSpace(form.Form))
{
// HtmlDom.GetBookSetting(key) returns the result of XmlNode.InnerXml which will be Html encoded (& < etc).
// HtmlDom.SetElementFromUserStringPreservingLineBreaks() calls XmlNode.InnerText, which Html encodes if necessary.
// So we need to decode here to prevent double encoding. See http://issues.bloomlibrary.org/youtrack/issue/BL-4585.
// Note that HtmlDom.SetElementFromUserStringPreservingLineBreaks() handles embedded <br/> elements, but makes no
// effort to handle p or div elements.
var decoded = System.Web.HttpUtility.HtmlDecode(form.Form);
HtmlDom.SetElementFromUserStringSafely(target, decoded);
target.SetAttribute("lang", form.WritingSystemId); //this allows us to set the font to suit the language
}
}
else //Put the value into an attribute. The license image goes through this path.
{
target.SetAttribute(valueAttribute, source.GetBestAlternativeString(languagePreferences));
if (source.Empty)
{
//if the license image is empty, make sure we don't have some alternative text
//about the image being missing or slow to load
target.SetAttribute("alt", "");
//over in javascript land, @alt will get set appropriately when the image url is not empty.
}
}
}
}
/// <summary>
/// Create a Clearshare.Metadata object by reading values out of the dom's bloomDataDiv
/// </summary>
/// <param name="brandingNameOrFolderPath"> Normally, the branding is just a name, which we look up in the official branding folder
//but unit tests can instead provide a path to the folder.
/// </param>
public static Metadata GetMetadata(HtmlDom dom, string brandingNameOrFolderPath = "")
{
if (ShouldSetToDefaultCopyrightAndLicense(dom))
{
return GetMetadataWithDefaultCopyrightAndLicense(brandingNameOrFolderPath);
}
var metadata = new Metadata();
var copyright = dom.GetBookSetting("copyright");
if (!copyright.Empty)
{
metadata.CopyrightNotice = WebUtility.HtmlDecode(copyright.GetFirstAlternative());
}
var licenseUrl = dom.GetBookSetting("licenseUrl").GetBestAlternativeString(new[] { "*", "en" });
if (string.IsNullOrWhiteSpace(licenseUrl))
{
//NB: we are mapping "RightsStatement" (which comes from XMP-dc:Rights) to "LicenseNotes" in the html.
//custom licenses live in this field, so if we have notes (and no URL) it is a custom one.
var licenseNotes = dom.GetBookSetting("licenseNotes");
if (!licenseNotes.Empty)
{
metadata.License = new CustomLicense { RightsStatement = WebUtility.HtmlDecode(licenseNotes.GetFirstAlternative()) };
}
else
{
// The only remaining current option is a NullLicense
metadata.License = new NullLicense(); //"contact the copyright owner
}
}
else // there is a licenseUrl, which means it is a CC license
{
try
{
metadata.License = CreativeCommonsLicense.FromLicenseUrl(licenseUrl);
}
catch (Exception e)
{
throw new ApplicationException("Bloom had trouble parsing this license url: '" + licenseUrl + "'. (ref BL-4108)", e);
}
//are there notes that go along with that?
var licenseNotes = dom.GetBookSetting("licenseNotes");
if(!licenseNotes.Empty)
{
var s = WebUtility.HtmlDecode(licenseNotes.GetFirstAlternative());
metadata.License.RightsStatement = HtmlDom.ConvertHtmlBreaksToNewLines(s);
}
}
return metadata;
}
/// <summary>
/// Copy the copyright & license info to the originalCopyrightAndLicense,
/// then remove the copyright so the translator can put in their own if they
/// want. We retain the license, but the translator is allowed to change that.
/// If the source is already a translation (already has original copyright or license)
/// we keep them unchanged.
/// </summary>
public static void SetOriginalCopyrightAndLicense(HtmlDom dom, BookData bookData, CollectionSettings collectionSettings)
{
// At least one of these should exist if the source was a derivative, since we don't allow a
// book to have no license, nor to be uploaded without copyright...unless of course it was derived
// before 3.9, when we started doing this. In that case the best we can do is record the earliest
// information we have for this and later adaptations.
if (bookData.GetMultiTextVariableOrEmpty("originalLicenseUrl").Count > 0 ||
bookData.GetMultiTextVariableOrEmpty("originalLicenseNotes").Count > 0 ||
bookData.GetMultiTextVariableOrEmpty("originalCopyright").Count > 0)
{
return; //leave the original there.
}
// If there's no copyright information in a source-collection book, we're presumably making
// a new original book, and shouldn't try to record any original copyright and license information.
// This is somewhat redundant with the check in BookStarter.SetupNewDocumentContents(), the one
// non-unit-test current caller of this method, that doesn't call this at all if the source is
// a template book. I was trying for a minimal reasonable change for BL-5131, and therefore
// put in this extra check, since previously this method was simply NEVER called in a source
// collection.
var copyrightNotice = GetMetadata(dom).CopyrightNotice;
if (string.IsNullOrEmpty(copyrightNotice) && collectionSettings.IsSourceCollection)
{
return;
}
bookData.Set("originalLicenseUrl", GetLicenseUrl(dom), "*");
bookData.Set("originalCopyright", System.Web.HttpUtility.HtmlEncode(copyrightNotice), "*");
bookData.Set("originalLicenseNotes", dom.GetBookSetting("licenseNotes").GetFirstAlternative(), "*");
bookData.RemoveAllForms("copyright"); // RemoveAllForms does modify the dom
}
/// <summary>
/// Create a Clearshare.Metadata object by reading values out of the dom's bloomDataDiv
/// </summary>
/// <param name="brandingNameOrFolderPath"> Normally, the branding is just a name, which we look up in the official branding folder
//but unit tests can instead provide a path to the folder.
/// </param>
public static Metadata GetMetadata(HtmlDom dom, string brandingNameOrFolderPath = "")
{
if (ShouldSetToDefaultCopyrightAndLicense(dom))
{
return(GetMetadataWithDefaultCopyrightAndLicense(brandingNameOrFolderPath));
}
return(CreateMetadata(dom.GetBookSetting("copyright"), GetLicenseUrl(dom), dom.GetBookSetting("licenseNotes")));
}
/// <summary>
/// Create a Clearshare.Metadata object by reading values out of the dom's bloomDataDiv
/// </summary>
public static Metadata GetMetadata(HtmlDom dom, BookData bookData)
{
if (ShouldSetToDefaultCopyrightAndLicense(dom))
{
return(GetMetadataWithDefaultCopyrightAndLicense());
}
return(CreateMetadata(dom.GetBookSetting("copyright"), GetLicenseUrl(dom), dom.GetBookSetting("licenseNotes"), bookData));
}
private static bool ShouldSetToDefaultCopyrightAndLicense(HtmlDom dom)
{
//Enhance: this logic is perhaps overly restrictive?
foreach (var setting in SettingsToCheckForDefaultCopyright)
{
if (!dom.GetBookSetting(setting).Empty)
{
return(false);
}
}
return(true);
}
internal static string GetOriginalCopyrightAndLicenseNotice(BookData bookData, HtmlDom dom)
{
var originalMetadata = GetOriginalMetadata(dom, bookData);
// As of BL-7898, we are using the existence of an original copyright/license to determine if we are working with a derivative.
if (!IsDerivative(originalMetadata))
{
return(null);
}
// The originalTitle strategy used here is not ideal. We would prefer to have a placeholder specifically for it
// in both EditTab.FrontMatter.OriginalCopyrightSentence and EditTab.FrontMatter.OriginalHadNoCopyrightSentence.
// But we don't want to require a new set of translations if we can avoid it.
var encodedTitle = dom.GetBookSetting("originalTitle")?.GetExactAlternative("*");
var originalTitle = HttpUtility.HtmlDecode(encodedTitle);
var titleCitation = "<cite data-book=\"originalTitle\"" +
(string.IsNullOrEmpty(originalTitle) ? " class=\"missingOriginalTitle\">" : ">") + originalTitle +
"</cite>";
var languagePriorityIdsNotLang1 = bookData.GetLanguagePrioritiesForLocalizedTextOnPage(false);
var originalLicenseSentence = GetOriginalLicenseSentence(languagePriorityIdsNotLang1, originalMetadata.License, out string licenseOnly);
var rawCopyright = originalMetadata.CopyrightNotice;
// If we have all the pieces available, we want to use this one.
// At the very least it's easier to localize into the format the language wants to use.
var fullFormatString = LocalizationManager.GetString(
"EditTab.FrontMatter.FullOriginalCopyrightLicenseSentence",
"Adapted from original, {0}, {1}. Licensed under {2}.",
"On the Credits page of a book being translated, Bloom shows the original copyright. {0} is original title, {1} is original copyright, and {2} is license information.",
languagePriorityIdsNotLang1,
out string langUsed);
// The last condition here (langUsed ==...) is meant to detect if the string has been translated
// into the current language or not.
if (!string.IsNullOrEmpty(originalTitle) && !string.IsNullOrEmpty(rawCopyright) &&
!string.IsNullOrEmpty(licenseOnly) && langUsed == languagePriorityIdsNotLang1.First())
{
return(string.Format(fullFormatString, titleCitation, rawCopyright, licenseOnly));
}
var copyrightNotice = GetOriginalCopyrightSentence(
languagePriorityIdsNotLang1,
rawCopyright,
titleCitation).Trim();
return((copyrightNotice + " " + originalLicenseSentence).Trim());
}
private static void CopyItemToFieldsInPages(HtmlDom dom, string key, string valueAttribute = null, string[] languagePreferences = null)
{
if (languagePreferences == null)
{
languagePreferences = new[] { "*", "en" }
}
;
MultiTextBase source = dom.GetBookSetting(key);
var target = dom.SelectSingleNode("//*[@data-derived='" + key + "']");
if (target == null)
{
return;
}
//just put value into the text of the element
if (string.IsNullOrEmpty(valueAttribute))
{
//clear out what's there now
target.RemoveAttribute("lang");
target.InnerText = "";
var form = source.GetBestAlternative(languagePreferences);
if (form != null && !string.IsNullOrWhiteSpace(form.Form))
{
HtmlDom.SetElementFromUserStringPreservingLineBreaks(target, form.Form);
target.SetAttribute("lang", form.WritingSystemId); //this allows us to set the font to suit the language
}
}
else //Put the value into an attribute. The license image goes through this path.
{
target.SetAttribute(valueAttribute, source.GetBestAlternativeString(languagePreferences));
if (source.Empty)
{
//if the license image is empty, make sure we don't have some alternative text
//about the image being missing or slow to load
target.SetAttribute("alt", "");
//over in javascript land, @alt will get set appropriately when the image url is not empty.
}
}
}
/// <summary>
/// Copy the copyright & license info to the originalCopyrightAndLicense,
/// then remove the copyright so the translator can put in their own if they
/// want. We retain the license, but the translator is allowed to change that.
/// If the source is already a translation (already has original copyright or license)
/// we keep them unchanged.
/// </summary>
public static void SetOriginalCopyrightAndLicense(HtmlDom dom, BookData bookData, CollectionSettings collectionSettings)
{
// If it already has some of this information, just keep it.
if (bookData.BookIsDerivative())
{
return; //leave the original there.
}
// If there's no copyright information in a source-collection book, we're presumably making
// a new original book, and shouldn't try to record any original copyright and license information.
// This is somewhat redundant with the check in BookStarter.SetupNewDocumentContents(), the one
// non-unit-test current caller of this method, that doesn't call this at all if the source is
// a template book. I was trying for a minimal reasonable change for BL-5131, and therefore
// put in this extra check, since previously this method was simply NEVER called in a source
// collection.
var copyrightNotice = BookCopyrightAndLicense.GetMetadata(dom).CopyrightNotice;
if (String.IsNullOrEmpty(copyrightNotice) && collectionSettings.IsSourceCollection)
{
return;
}
bookData.Set("originalLicenseUrl", BookCopyrightAndLicense.GetLicenseUrl(dom), "*");
bookData.Set("originalCopyright", System.Web.HttpUtility.HtmlEncode(copyrightNotice), "*");
bookData.Set("originalLicenseNotes", dom.GetBookSetting("licenseNotes").GetFirstAlternative(), "*");
}
public static void LogMetdata(HtmlDom dom)
{
Logger.WriteEvent("LicenseUrl: " + dom.GetBookSetting("licenseUrl"));
Logger.WriteEvent("LicenseNotes: " + dom.GetBookSetting("licenseNotes"));
Logger.WriteEvent("");
}
public static string GetLicenseUrl(HtmlDom dom)
{
return(dom.GetBookSetting("licenseUrl").GetBestAlternativeString(new[] { "*", "en" }));
}
public void GetBookSetting_TwoVariationsWereThere_ReturnsBoth()
{
var bookDom = new HtmlDom(@"<html ><head></head><body>
<div id='bloomDataDiv'>
<div data-book='leaveMe' lang='en'>something unique</div>
<div data-book='getMe' lang='id'>Buku</div>
<div data-book='getMe' lang='tpi'>Buk</div>
</div>
</body></html>");
var result = bookDom.GetBookSetting("getMe");
Assert.AreEqual(2,result.Count);
Assert.AreEqual("Buk", result["tpi"]);
Assert.AreEqual("Buku", result["id"]);
}
public void GetBookSetting_NotThere_ReturnsEmptyMultistring()
{
var bookDom = new HtmlDom(@"<html ><head></head><body>
<div id='bloomDataDiv'>
</div>
</body></html>");
var result = bookDom.GetBookSetting("getMe");
Assert.AreEqual(0, result.Count);
}
private static bool ShouldSetToDefaultCopyrightAndLicense(HtmlDom dom)
{
var hasCopyright = !dom.GetBookSetting("copyright").Empty;
var hasLicenseUrl = !dom.GetBookSetting("licenseUrl").Empty;
var hasLicenseNotes = !dom.GetBookSetting("licenseNotes").Empty;
//Enhance: this logic is perhaps overly restrictive?
return !hasCopyright && !hasLicenseUrl && !hasLicenseNotes;
}
internal static string GetOriginalCopyrightAndLicenseNotice(CollectionSettings collectionSettings, HtmlDom dom)
{
var originalMetadata = GetOriginalMetadata(dom);
// As of BL-7898, we are using the existence of an original copyright/license to determine if we are working with a derivative.
if (!IsDerivative(originalMetadata))
{
return(null);
}
string idOfLanguageUsed;
var languagePriorityIds = collectionSettings.LicenseDescriptionLanguagePriorities;
var license = originalMetadata.License.GetMinimalFormForCredits(languagePriorityIds, out idOfLanguageUsed);
string originalLicenseSentence;
var preferredLanguageIds = new[] { collectionSettings.Language2Iso639Code, LocalizationManager.UILanguageId, "en" };
// The originalTitle strategy used here is not ideal. We would prefer to have a placeholder specifically for it
// in both EditTab.FrontMatter.OriginalCopyrightSentence and EditTab.FrontMatter.OriginalHadNoCopyrightSentence.
// But we don't want to require a new set of translations if we can avoid it.
string encodedTitle = dom.GetBookSetting("originalTitle")?.GetExactAlternative("*");
string originalTitle = HttpUtility.HtmlDecode(encodedTitle);
// Used when we insert into the no-copyright string, typically "Adapted from an original with no copyright"
var originalTitleBeforePeriod = ", <cite data-book=\"originalTitle\">" + originalTitle + "</cite>";
// Used when we insert into the usual string, typicall "Adapted from original, {0}" ahead of the copyright and license
// as part of the {0} replacement.
var originalTitleAfterComma = "<cite data-book=\"originalTitle\">" + originalTitle + "</cite>, ";
if (string.IsNullOrEmpty(originalTitle))
{
// We need to add the "missingOriginalTitle class.
originalTitleBeforePeriod = ", <cite data-book=\"originalTitle\" class=\"missingOriginalTitle\"></cite>";
originalTitleAfterComma = "<cite data-book=\"originalTitle\" class=\"missingOriginalTitle\"></cite>, ";
}
if (originalMetadata.License is CustomLicense)
{
// I can imagine being more fancy... something like "Licensed under custom license:", and get localizations
// for that... but sheesh, these are even now very rare in Bloom-land and should become more rare.
// So for now, let's just print the custom license contents.
originalLicenseSentence = license;
}
else
{
var licenseSentenceTemplate = LocalizationManager.GetString("EditTab.FrontMatter.OriginalLicenseSentence",
"Licensed under {0}.",
"On the Credits page of a book being translated, Bloom puts texts like 'Licensed under CC-BY', so that we have a record of what the license was for the original book. Put {0} in the translation, where the license should go in the sentence.",
preferredLanguageIds, out idOfLanguageUsed);
originalLicenseSentence = string.IsNullOrWhiteSpace(license) ? "" : string.Format(licenseSentenceTemplate, license);
originalLicenseSentence = originalLicenseSentence.Replace("..", "."); // in case had notes which also had a period.
}
string copyrightNotice;
if (string.IsNullOrWhiteSpace(originalMetadata.CopyrightNotice))
{
var noCopyrightSentence = LocalizationManager.GetString("EditTab.FrontMatter.OriginalHadNoCopyrightSentence",
"Adapted from original without a copyright notice.",
"On the Credits page of a book being translated, Bloom shows this if the original book did not have a copyright notice.",
preferredLanguageIds, out idOfLanguageUsed);
noCopyrightSentence = noCopyrightSentence.Substring(0, noCopyrightSentence.Length - 1) +
originalTitleBeforePeriod + ".";
copyrightNotice = noCopyrightSentence + " " + originalLicenseSentence;
}
else
{
var originalCopyrightSentence = LocalizationManager.GetString("EditTab.FrontMatter.OriginalCopyrightSentence",
"Adapted from original, {0}.",
"On the Credits page of a book being translated, Bloom shows the original copyright. Put {0} in the translation where the copyright notice should go. For example in English, 'Adapted from original, {0}.' comes out like 'Adapted from original, Copyright 2011 SIL'.",
preferredLanguageIds, out idOfLanguageUsed);
copyrightNotice = String.Format(originalCopyrightSentence, originalTitleAfterComma + originalMetadata.CopyrightNotice.Trim()) + " " +
originalLicenseSentence;
}
return(copyrightNotice.Trim());
}
public static Metadata GetOriginalMetadata(HtmlDom dom, string brandingNameOrFolderPath = "")
{
return(CreateMetadata(dom.GetBookSetting("originalCopyright"), dom.GetBookSetting("originalLicenseUrl").GetExactAlternative("*"), dom.GetBookSetting("originalLicenseNotes")));
}
public static void LogMetdata(HtmlDom dom)
{
Logger.WriteEvent("LicenseUrl: " + dom.GetBookSetting("licenseUrl"));
Logger.WriteEvent("LicenseNotes: " + dom.GetBookSetting("licenseNotes"));
Logger.WriteEvent("");
}
/// <summary>
/// Create a Clearshare.Metadata object by reading values out of the dom's bloomDataDiv
/// </summary>
/// <param name="brandingNameOrFolderPath"> Normally, the branding is just a name, which we look up in the official branding folder
//but unit tests can instead provide a path to the folder.
/// </param>
public static Metadata GetMetadata(HtmlDom dom, string brandingNameOrFolderPath = "")
{
if (ShouldSetToDefaultCopyrightAndLicense(dom))
{
return(GetMetadataWithDefaultCopyrightAndLicense(brandingNameOrFolderPath));
}
var metadata = new Metadata();
var copyright = dom.GetBookSetting("copyright");
if (!copyright.Empty)
{
metadata.CopyrightNotice = WebUtility.HtmlDecode(copyright.GetFirstAlternative());
}
var licenseUrl = dom.GetBookSetting("licenseUrl").GetBestAlternativeString(new[] { "*", "en" });
if (string.IsNullOrWhiteSpace(licenseUrl))
{
//NB: we are mapping "RightsStatement" (which comes from XMP-dc:Rights) to "LicenseNotes" in the html.
//custom licenses live in this field, so if we have notes (and no URL) it is a custom one.
var licenseNotes = dom.GetBookSetting("licenseNotes");
if (!licenseNotes.Empty)
{
metadata.License = new CustomLicense {
RightsStatement = WebUtility.HtmlDecode(licenseNotes.GetFirstAlternative())
};
}
else
{
// The only remaining current option is a NullLicense
metadata.License = new NullLicense(); //"contact the copyright owner
}
}
else // there is a licenseUrl, which means it is a CC license
{
try
{
metadata.License = CreativeCommonsLicense.FromLicenseUrl(licenseUrl);
}
catch (IndexOutOfRangeException)
{
// Need to handle urls which do not end with the version number.
// Simply set it to the default version.
if (!licenseUrl.EndsWith("/"))
{
licenseUrl += "/";
}
licenseUrl += CreativeCommonsLicense.kDefaultVersion;
metadata.License = CreativeCommonsLicense.FromLicenseUrl(licenseUrl);
}
catch (Exception e)
{
throw new ApplicationException("Bloom had trouble parsing this license url: '" + licenseUrl + "'. (ref BL-4108)", e);
}
//are there notes that go along with that?
var licenseNotes = dom.GetBookSetting("licenseNotes");
if (!licenseNotes.Empty)
{
var s = WebUtility.HtmlDecode(licenseNotes.GetFirstAlternative());
metadata.License.RightsStatement = HtmlDom.ConvertHtmlBreaksToNewLines(s);
}
}
return(metadata);
}
public static Metadata GetOriginalMetadata(HtmlDom dom, BookData bookData)
{
return(CreateMetadata(dom.GetBookSetting("originalCopyright"), dom.GetBookSetting("originalLicenseUrl").GetExactAlternative("*"), dom.GetBookSetting("originalLicenseNotes"), bookData));
}
private static void CopyItemToFieldsInPages(HtmlDom dom, string key, string valueAttribute = null, string[] languagePreferences= null)
{
if (languagePreferences == null)
languagePreferences = new[] {"*", "en"};
MultiTextBase source = dom.GetBookSetting(key);
var target = dom.SelectSingleNode("//*[@data-derived='" + key + "']");
if (target == null)
{
return;
}
//just put value into the text of the element
if (string.IsNullOrEmpty(valueAttribute))
{
//clear out what's there now
target.RemoveAttribute("lang");
target.InnerText = "";
var form = source.GetBestAlternative(languagePreferences);
if (form != null && !string.IsNullOrWhiteSpace(form.Form))
{
HtmlDom.SetElementFromUserStringPreservingLineBreaks(target, form.Form);
target.SetAttribute("lang", form.WritingSystemId); //this allows us to set the font to suit the language
}
}
else //Put the value into an attribute. The license image goes through this path.
{
target.SetAttribute(valueAttribute, source.GetBestAlternativeString(languagePreferences));
if (source.Empty)
{
//if the license image is empty, make sure we don't have some alternative text
//about the image being missing or slow to load
target.SetAttribute("alt", "");
//over in javascript land, @alt will get set appropriately when the image url is not empty.
}
}
}
