public bool Convert(DitaElement bodyElement, List <DitaPageSectionJson> sections, string fileName, out string body)
{
StringBuilder bodyStringBuilder = new StringBuilder();
Sections = sections;
FileName = fileName;
if (bodyElement != null)
{
if (bodyElement.IsContainer)
{
foreach (DitaElement childElement in bodyElement?.Children)
{
bodyStringBuilder.Append(Convert(childElement));
}
}
else
{
bodyStringBuilder.Append(bodyElement.InnerText);
}
}
body = bodyStringBuilder.ToString();
return(true);
}
// Parse the title from the book map
private void ParseBookMapTitle()
{
DitaElement titleElement = RootMap.RootElement.FindOnlyChild("booktitle");
AddChildDitaElementTextToDictionary(titleElement, "mainbooktitle", BookTitle);
AddChildDitaElementTextToDictionary(titleElement, "booktitlealt", BookTitle);
}
// Build a dictionary of elements in the document, by id
private void ParseElementIds(DitaElement parentElement)
{
if (parentElement != null)
{
// Does this element have an id?
string id = parentElement.AttributeValueOrDefault("id", string.Empty);
if (!string.IsNullOrEmpty(id))
{
if (!ElementsById.ContainsKey(id))
{
ElementsById.Add(id, parentElement);
}
else
{
Trace.TraceWarning($"Duplicate element id {id} found in {FileName}.");
}
}
// Parse the children
if (parentElement.Children != null)
{
foreach (DitaElement childElement in parentElement.Children)
{
ParseElementIds(childElement);
}
}
}
}
protected List <DitaElement> FindChildren(string[] types, DitaElement parentElement)
{
List <DitaElement> result = null;
if (IsContainer)
{
// Are any of our direct children of this type?
result = parentElement?.Children?.Where(e => types.Contains(e.Type)).ToList();
if (result?.Count == 0)
{
// Try finding children of children
foreach (DitaElement childElement in parentElement.Children)
{
result = FindChildren(types, childElement);
if (result?.Count != 0)
{
break;
}
}
}
}
return(result);
}
private void ResolveKeywords(DitaElement parentElement, DitaCollection collection)
{
if (parentElement != null)
{
if (parentElement.Type == "keyword")
{
string keyref = parentElement.AttributeValueOrDefault("keyref", String.Empty);
if (!string.IsNullOrWhiteSpace(keyref))
{
// Try to find the referred to file
DitaKeyDef keyDef = collection.GetKeyDefByKey(keyref);
if (keyDef != null)
{
if (!string.IsNullOrWhiteSpace(keyDef.Keywords))
{
parentElement.SetInnerText(keyDef.Keywords);
}
}
}
}
else
{
// Check the child elements
if (parentElement.Children != null)
{
foreach (DitaElement childElement in parentElement.Children)
{
ResolveKeywords(childElement, collection);
}
}
}
}
}
// Parse the metadata from the given element?
private void ParseMetaCategories(DitaElement parentElement)
{
try
{
List <DitaElement> categoriesData = parentElement?.FindChildren("category");
if (categoriesData != null)
{
foreach (DitaElement categoryData in categoriesData)
{
foreach (DitaElement data in categoryData.Children)
{
if (data?.Attributes.ContainsKey("name") ?? false)
{
if (BookMeta.ContainsKey(data?.Attributes["name"]))
{
Trace.TraceWarning($"BookMeta already contains a value for {data?.Attributes["name"]}");
}
else
{
BookMeta.Add(data?.Attributes["name"], data?.Attributes["value"]);
}
}
}
}
}
}
catch (Exception ex)
{
Trace.TraceError(ex);
}
}
// Replaces the values in this element with those copied from another element
public void Copy(DitaElement sourceElement)
{
Type = sourceElement.Type;
Attributes = sourceElement.Attributes;
IsContainer = sourceElement.IsContainer;
Children = sourceElement.Children;
InnerText = sourceElement.InnerText;
}
// Parse the book meta data from the book map
private void ParseBookMapBookMeta()
{
DitaElement bookMetaElement = RootMap.RootElement.FindOnlyChild("bookmeta");
ParseBookMapVersion(bookMetaElement);
ParseBookMapReleaseDate(bookMetaElement);
// Everything in category
ParseMetaCategories(bookMetaElement);
}
// Construct from a single topic
public DitaPageJson(DitaFile file, DitaCollection collection)
{
Collection = collection;
// Get the title of the page
Title = DitaFile.FixSpecialCharacters(file.Title);
// Create the file name
FileName = file.NewFileName ?? file.FileName;
FileName = Path.ChangeExtension(FileName, ".json");
OriginalFileName = file.FileName;
// Find the body element
string bodyElementName = null;
if (DitaFile.DitaFileBodyElement.ContainsKey(file.GetType()))
{
bodyElementName = DitaFile.DitaFileBodyElement[file.GetType()]();
}
if (!string.IsNullOrEmpty(bodyElementName))
{
DitaElement bodyElement = file.RootElement.FindOnlyChild(bodyElementName);
if (bodyElement != null)
{
Sections = new List <DitaPageSectionJson>();
// Convert the body to html
DitaElementToHtmlConverter htmlConverter = new DitaElementToHtmlConverter(collection);
htmlConverter.Convert(bodyElement, Sections, file.FileName, out string bodyHtml);
BodyHtml = bodyHtml;
// Convert the body to text
DitaElementToTextConverter textConverter = new DitaElementToTextConverter();
textConverter.Convert(bodyElement, out string bodyText);
BodyText = bodyText;
}
else
{
Trace.TraceWarning($"Body element not found in {file.FileName}.");
}
}
else
{
Trace.TraceWarning($"No body element identified in {file.FileName}.");
}
IsEmpty = string.IsNullOrEmpty(BodyText) || string.IsNullOrEmpty(Title);
}
// Wraps a dita element in another dita element, if needed to output correct html
// For instance, wrap tables, sections, figures, etc in an <a name="..."> for instance
private DitaElement PrependDitaElementIfNeeded(DitaElement inputElement)
{
string id = inputElement.AttributeValueOrDefault("id", null);
// If this object has an id, wrap it in an a, name
if (!string.IsNullOrEmpty(id))
{
DitaElement outputElement = new DitaElement("a", false, null, inputElement.Parent, inputElement.PreviousSibling);
outputElement.Attributes.Add("name", id);
return(outputElement);
}
return(null);
}
// Default constructor
public DitaElement(string type, bool isContainer = false, string innerText = null, DitaElement parent = null, DitaElement previousSibling = null)
{
Type = type;
IsContainer = isContainer;
Parent = parent;
PreviousSibling = previousSibling;
if (IsContainer)
{
Children = new List <DitaElement>();
}
else
{
InnerText = innerText;
}
}
private string Convert(DitaElement element)
{
if (element.IsContainer)
{
StringBuilder elementStringBuilder = new StringBuilder();
foreach (DitaElement childElement in element.Children)
{
elementStringBuilder.AppendLine(Convert(childElement));
}
return(elementStringBuilder.ToString());
}
else
{
return($"{element}");
}
}
// Tries to add the text of the given element to the dictionary
private bool AddChildDitaElementTextToDictionary(DitaElement parentElement, string type,
Dictionary <string, string> dictionary)
{
// Try to find the child elements that match the type
List <DitaElement> childElements = parentElement.FindChildren(type);
if (childElements?.Count > 0)
{
foreach (DitaElement childElement in childElements)
{
dictionary.Add(type, childElement.InnerText);
}
return(true);
}
return(false);
}
// Does a given image element refer to SVG?
private bool IsImageElementSvg(DitaElement imageElement)
{
if (imageElement != null)
{
if (imageElement.Type == "image")
{
string extension = Path.GetExtension(imageElement.AttributeValueOrDefault("href", ""));
if ((extension?.Equals(".svg", StringComparison.OrdinalIgnoreCase) ?? false) ||
(extension?.Equals(".image", StringComparison.OrdinalIgnoreCase) ?? false))
{
return(true);
}
}
}
return(false);
}
// Ingest an XmlNode and recursively parse it to create a group of DitaElements
public DitaElement Convert(XmlNode inputNode, DitaElement parentElement = null, DitaElement previousSiblingElement = null)
{
// What type of element are we creating
string type = inputNode?.Name;
// Does this node/element have children
bool isContainer = !IsNodeOnlyText(inputNode, out string innerText);
if (!isContainer)
{
innerText = CleanInnerText(innerText);
}
// Create the new DITA element
DitaElement outputElement = new DitaElement(type, isContainer, innerText, parentElement, previousSiblingElement);
// Add the attributes
if (inputNode?.Attributes != null)
{
foreach (XmlAttribute attribute in inputNode?.Attributes)
{
outputElement.Attributes.Add(attribute.Name, attribute.InnerText);
}
}
// Add the children of this node/element, if any
// ReSharper disable once InvertIf
if (isContainer && inputNode?.ChildNodes != null)
{
DitaElement previousElement = null;
foreach (XmlNode childNode in inputNode.ChildNodes)
{
DitaElement childElement = Convert(childNode, outputElement, previousElement);
outputElement.Children.Add(childElement);
previousElement = childElement;
}
}
return(outputElement);
}
// Parse the version of the document
private void ParseBookMapVersion(DitaElement bookMetaElement)
{
// Try checking the publisher information section
DitaElement publisherInformationElement = bookMetaElement?.FindOnlyChild("publisherinformation");
DitaElement publishedElement = publisherInformationElement?.FindChildren("published")?.Last();
DitaElement revisionIdElement = publishedElement?.FindOnlyChild("revisionid");
string version = revisionIdElement?.ToString();
// Try checking the prodinfo section
if (string.IsNullOrWhiteSpace(version) || version == "ProductVersionNumber")
{
DitaElement prodinfoElement = bookMetaElement?.FindOnlyChild("prodinfo");
DitaElement vrmlistElement = prodinfoElement?.FindOnlyChild("vrmlist");
DitaElement vrmElement = vrmlistElement?.FindChildren("vrm")?[0];
version = vrmElement?.Attributes?["version"];
}
if (!string.IsNullOrWhiteSpace(version))
{
BookMeta.Add("version", version);
}
}
public bool Convert(DitaElement bodyElement, out string body)
{
StringBuilder bodyStringBuilder = new StringBuilder();
if (bodyElement != null)
{
if (bodyElement.IsContainer)
{
foreach (DitaElement childElement in bodyElement?.Children)
{
bodyStringBuilder.Append(Convert(childElement));
}
}
else
{
bodyStringBuilder.Append(bodyElement);
}
}
body = bodyStringBuilder.ToString();
return(true);
}
// Parse the date of the document
private void ParseBookMapReleaseDate(DitaElement bookMetaElement)
{
DitaElement publisherInformationElement = bookMetaElement?.FindOnlyChild("publisherinformation");
DitaElement publishedElement = publisherInformationElement?.FindChildren("published")?.Last();
DitaElement completedElement = publishedElement?.FindOnlyChild("completed");
string year = completedElement?.FindOnlyChild("year")?.ToString();
string month = completedElement?.FindOnlyChild("month")?.ToString();
string day = completedElement?.FindOnlyChild("day")?.ToString();
if (!string.IsNullOrWhiteSpace(year) && !string.IsNullOrWhiteSpace(month) && !string.IsNullOrWhiteSpace(day))
{
try
{
// Is this a valid date?
DateTime publishDate = new DateTime(int.Parse(year), int.Parse(month), int.Parse(day));
BookMeta.Add("published date", $"{publishDate.Day}/{publishDate.Month}/{publishDate.Year} 00:00:00");
}
catch
{
//
}
}
}
private string Convert(DitaElement element)
{
StringBuilder elementStringBuilder = new StringBuilder();
// Does this element need to be wrapped by another element?
DitaElement prependElement = PrependDitaElementIfNeeded(element);
if (prependElement != null)
{
elementStringBuilder.Append(Convert(prependElement));
}
// Determine the new html tag and attributes
string htmlTag = ConvertDitaTagToHtmlTag(element);
Dictionary <string, string> htmlAttributes = ConvertDitaTagAttributesToHtmlTagAttributes(element);
AddHtmlTagAttributes(htmlAttributes, element);
// If this is a parent, then add the children
if (element.IsContainer)
{
elementStringBuilder.Append(HtmlOpeningTag(htmlTag, htmlAttributes));
foreach (DitaElement childElement in element.Children)
{
elementStringBuilder.Append(Convert(childElement));
}
elementStringBuilder.Append(HtmlClosingTag(htmlTag));
}
else
{
elementStringBuilder.Append($"{HtmlOpeningTag(htmlTag, htmlAttributes)}{element.InnerText}{HtmlClosingTag(htmlTag)}");
}
return(elementStringBuilder.ToString());
}
// Parse the meta date from the map
private void ParseMapMeta()
{
DitaElement bookMetaElement = RootMap.RootElement.FindOnlyChild("topicmeta");
ParseMetaCategories(bookMetaElement);
}
// Takes a DITA "tag" and returns the corresponding HTML tag
private string ConvertDitaTagToHtmlTag(DitaElement element)
{
switch (element.Type)
{
case "b": return("strong");
case "colspec":
TableColumnIndex++;
FixUpTableColumnSpecs();
_tableColumnSpecs[TableColumnIndex] = new DitaTableColumnSpec {
Number = (TableColumnIndex + 1)
};
return("");
case "entry":
TableRowColumnIndex++;
if (element.Parent?.Parent?.Type == "thead" || element.AttributeValueOrDefault("class", "") == "th")
{
return("th");
}
return("td");
case "fig": return("figure");
case "image":
// Is this referring to an SVG or other type of image?
if (IsImageElementSvg(element))
{
return("object");
}
return("img");
case "keyword": return("");
case "row":
TableRowColumnIndex = -1;
return("tr");
case "steps":
return("ol");
case "step":
return("li");
case "table":
TableColumnIndex = -1;
_tableColumnSpecs = null;
break;
case "tgroup":
TableColumnIndex = -1;
_tableColumnSpecs = null;
return("");
case "title":
if (element.Parent?.Type == "section")
{
// Create a reference to this section, if this is the title of the section
if (CurrentSection != null)
{
if (string.IsNullOrEmpty(CurrentSection.Title) && !string.IsNullOrEmpty(CurrentSection.Anchor))
{
CurrentSection.Title = element.ToString();
Sections.Add(CurrentSection);
CurrentSection = null;
}
}
return("h3");
}
return("h4");
case "xref":
return("a");
case "#text": return("");
}
return(element.Type);
}
// Converts DITA tag attributes to html tag attributes
private Dictionary <string, string> ConvertDitaTagAttributesToHtmlTagAttributes(DitaElement element)
{
Dictionary <string, string> htmlAttributes = new Dictionary <string, string>();
foreach (string ditaAttributeKey in element.Attributes.Keys)
{
(string newKey, string newValue) = ConvertDitaTagAttributeToHtmlTagAttribute(ditaAttributeKey, element.Attributes[ditaAttributeKey], element);
if (!string.IsNullOrWhiteSpace(newKey))
{
htmlAttributes.Add(newKey, newValue);
}
}
return(htmlAttributes);
}
// Resolves conrefs in this file
private void ResolveConRefs(DitaElement parentElement, DitaCollection collection)
{
if (parentElement != null)
{
bool updated = false;
// Does this element have a conref?
string conref = parentElement.AttributeValueOrDefault("conref", String.Empty);
if (!string.IsNullOrEmpty(conref))
{
// We expect the conref to be in the form of filename.xml#fileid/elementid
Regex conRefRegex = new Regex("^(.*)#(.*)/(.*)$", RegexOptions.CultureInvariant | RegexOptions.IgnoreCase);
MatchCollection conRefMatchCollection = conRefRegex.Matches(conref);
if (conRefMatchCollection?.Count > 0 && (conRefMatchCollection[0].Groups.Count == 4))
{
// Extract the components of the conref
string refFileName = conRefMatchCollection[0].Groups[1].Value;
string refFileId = conRefMatchCollection[0].Groups[2].Value;
string refElementId = conRefMatchCollection[0].Groups[3].Value;
if (Path.GetFileNameWithoutExtension(refFileName) != refFileId)
{
Trace.TraceWarning($"conref file name '{refFileName}' is not equal to file id '{refFileId}'.");
}
// Try to find the file that this conref refers to
DitaFile refFile = collection.GetFileByName(refFileName);
if (refFile != null)
{
// Does the references element exist in this file
if (refFile.ElementsById.ContainsKey(refElementId))
{
DitaElement refElement = refFile.ElementsById[refElementId];
// Copy the refernce element
parentElement.Copy(refElement);
updated = true;
}
else
{
Trace.TraceWarning($"Element '{refElementId}' not found in file '{refFileName}'.");
}
}
else
{
Trace.TraceWarning($"Can't find file '{refFileName}' referenced in file '{FileName}'.");
}
}
else
{
Trace.TraceWarning($"conref {conref} not in expected format.");
}
}
// Update child references
if (!updated && parentElement.Children != null)
{
foreach (DitaElement childElement in parentElement.Children)
{
ResolveConRefs(childElement, collection);
}
}
}
}
// Converts a single dita tag attribute to an html attribute
private (string newKey, string newValue) ConvertDitaTagAttributeToHtmlTagAttribute(string key, string value, DitaElement element)
{
switch (element.Type)
{
case "a":
return(key, value);
case "colspec":
if (key == "colname")
{
FixUpTableColumnSpecs();
_tableColumnSpecs[TableColumnIndex].Name = value;
}
if (key == "colwidth")
{
FixUpTableColumnSpecs();
_tableColumnSpecs[TableColumnIndex].Width = value;
}
if (key == "colnum")
{
if (int.TryParse(value, out int colnum))
{
FixUpTableColumnSpecs();
_tableColumnSpecs[TableColumnIndex].Number = colnum;
}
}
break;
case "entry":
if (key == "morerows")
{
if (int.TryParse(value, out int rowspan))
{
return("rowspan", $"{rowspan + 1}");
}
}
if (key == "valign")
{
return(key, value);
}
break;
case "image":
if (key == "href")
{
if (IsImageElementSvg(element))
{
return("data", ImageUrlFromHref(value));
}
return("src", ImageUrlFromHref(value));
}
break;
case "section":
if (key == "id")
{
CurrentSection = new DitaPageSectionJson {
Anchor = value
};
return(key, value);
}
break;
case "tgroup":
if (key == "cols")
{
if (int.TryParse(value, out int columns))
{
_tableColumnSpecs = new DitaTableColumnSpec[columns];
}
}
break;
case "xref":
if (key == "href")
{
string url = UrlFromXref(element, out string title);
// If we encounter an empty xref, we want to try to generate link text, based on what it links too
if (string.IsNullOrWhiteSpace(element.ToString()))
{
element.SetInnerText(title);
}
return(key, url);
}
break;
}
return("", "");
}
// Add additional attributes to specific html tags
private void AddHtmlTagAttributes(Dictionary <string, string> htmlAttributes, DitaElement element)
{
switch (element.Type)
{
case "image":
if (IsImageElementSvg(element))
{
// Add the type of embedded svg
if (!htmlAttributes.ContainsKey("type"))
{
htmlAttributes.Add("type", "image/svg+xml");
}
}
break;
case "table":
// Add the generic "table" class to all tables
if (!htmlAttributes.ContainsKey("class"))
{
htmlAttributes.Add("class", "table");
}
break;
case "entry":
// If there is a width defined, add it to the entry
string colname = element.Attributes.ContainsKey("colname") ? element.Attributes["colname"] : null;
if (!htmlAttributes.ContainsKey("width"))
{
string widthAsPercent = ColumnWidthAsPercent(colname, TableRowColumnIndex);
if (!string.IsNullOrEmpty(widthAsPercent))
{
htmlAttributes.Add("width", widthAsPercent);
}
}
// If there is a colspan defined, add it to the entry
if (element.Attributes.ContainsKey("namest") && element.Attributes.ContainsKey("nameend"))
{
// Build the colspan
int startColumn = _tableColumnSpecs?.FirstOrDefault(o => o.Name == element.Attributes["namest"])?.Number ?? -1;
int endColumn = _tableColumnSpecs?.FirstOrDefault(o => o.Name == element.Attributes["nameend"])?.Number ?? -1;
if (startColumn >= 0 && endColumn >= 0)
{
if (!htmlAttributes.ContainsKey("colspan"))
{
htmlAttributes.Add("colspan", $"{endColumn - startColumn + 1}");
}
}
}
break;
}
}
// Returns the relative or absolute url from a Dita XREF for use in an html A tag
private string UrlFromXref(DitaElement xrefElement, out string title)
{
// What is the scope
string scope = xrefElement.AttributeValueOrDefault("scope", null);
string format = xrefElement.AttributeValueOrDefault("format", null);
string href = xrefElement.AttributeValueOrDefault("href", null);
title = null;
if (scope == "external")
{
return(href);
}
if (!string.IsNullOrEmpty(href))
{
string result = null;
if (href[0] == '#')
{
// Link to the same page
if (href.Contains('/'))
{
string[] anchorSplit = href.Split('/');
if (anchorSplit.Length > 1)
{
result = $"#{anchorSplit[1]}";
}
}
else
{
result = href.Substring(1);
}
}
else if (href.ToLowerInvariant().StartsWith("http"))
{
result = href;
}
else
{
// Split by hash, if any
string[] hashSplit = href.Split('#');
// Try to find the topic it is linking to
DitaFile referenceFile = Collection?.GetFileByName(hashSplit[0]);
if (referenceFile != null)
{
result = $"%DOCUMENT_ROOT%/{Path.GetFileNameWithoutExtension(referenceFile.NewFileName)}";
if (hashSplit.Length > 1)
{
result += $"#{hashSplit[1]}";
}
title = referenceFile.Title;
}
else
{
Trace.TraceError($"Xref refers to unknown local file {hashSplit[0]} in {FileName}");
}
}
if (!string.IsNullOrEmpty(result))
{
return(result);
}
else
{
return("#");
}
}
Trace.TraceWarning($"Unknown xref scope={scope}, format={format}, href={href} in {FileName}");
return("#");
}
