private void ProcessHtmlChunks(HtmlEnumerator en, String endTag)
{
while (en.MoveUntilMatch(endTag))
{
if (en.IsCurrentHtmlTag)
{
Action <HtmlEnumerator> action;
if (knownTags.TryGetValue(en.CurrentTag, out action))
{
if (Logging.On)
{
Logging.PrintVerbose(en.Current);
}
action(en);
}
// else unknown or not yet implemented - we ignore
}
else
{
Run run = new Run(
new Text(HttpUtility.HtmlDecode(en.Current))
{
Space = SpaceProcessingModeValues.Preserve
}
);
// apply the previously discovered style
htmlStyles.Runs.ApplyTags(run);
elements.Add(run);
}
}
}
public void BeginList(HtmlEnumerator en)
{
int prevAbsNumId = numInstances.Peek().Value;
var absNumId = -1;
// lookup for a predefined list style in the template collection
String type = en.StyleAttributes["list-style-type"];
bool orderedList = en.CurrentTag.Equals("<ol>", StringComparison.OrdinalIgnoreCase);
if (type == null || !knonwAbsNumIds.TryGetValue(type.ToLowerInvariant(), out absNumId))
{
if (orderedList)
{
absNumId = knonwAbsNumIds["decimal"];
}
else
{
absNumId = knonwAbsNumIds["disc"];
}
}
firstItem = true;
levelDepth++;
// save a NumberingInstance if the nested list style is the same as its ancestor.
// this allows us to nest <ol> and restart the indentation to 1.
int currentInstanceId = this.InstanceID;
if (levelDepth > 1 && absNumId == prevAbsNumId && orderedList)
{
EnsureMultilevel(absNumId);
}
else
{
currentInstanceId = ++nextInstanceID;
Numbering numbering = mainPart.NumberingDefinitionsPart.Numbering;
numbering.Append(
new NumberingInstance(
new AbstractNumId()
{
Val = absNumId
},
new LevelOverride(
new StartOverrideNumberingValue()
{
Val = 1
}
)
{
LevelIndex = 0,
}
)
{
NumberID = currentInstanceId
});
}
numInstances.Push(new KeyValuePair <int, int>(currentInstanceId, absNumId));
}
/// <summary>
/// Save the actual list and restart with a new one.
/// Continue to process until we found endTag.
/// </summary>
private void AlternateProcessHtmlChunks(HtmlEnumerator en, string endTag)
{
if (elements.Count > 0)
{
CompleteCurrentParagraph();
}
ProcessHtmlChunks(en, endTag);
}
public void BeginList(HtmlEnumerator en)
{
// lookup for a predefined list style in the template collection
String type = en.StyleAttributes["list-style-type"];
bool orderedList = en.CurrentTag.Equals("<ol>", StringComparison.OrdinalIgnoreCase);
CreateList(type, orderedList);
}
/// <summary>
/// There is a few attributes shared by a large number of tags. This method will check them for a limited
/// number of tags (<p>, <pre>, <div>, <span> and <body>).
/// </summary>
/// <returns>Returns true if the processing of this tag should generate a new paragraph.</returns>
private bool ProcessContainerAttributes(HtmlEnumerator en, IList <OpenXmlElement> styleAttributes)
{
bool newParagraph = false;
// Not applicable to a table : page break
if (!tables.HasContext || en.CurrentTag == "<pre>")
{
String attrValue = en.StyleAttributes["page-break-after"];
if (attrValue == "always")
{
paragraphs.Add(new Paragraph(
new Run(
new Break()
{
Type = BreakValues.Page
})));
}
attrValue = en.StyleAttributes["page-break-before"];
if (attrValue == "always")
{
elements.Add(
new Run(
new Break()
{
Type = BreakValues.Page
})
);
elements.Add(new Run(
new LastRenderedPageBreak())
);
}
}
// support left and right padding
var padding = en.StyleAttributes.GetAsMargin("padding");
if (!padding.IsEmpty && (padding.Left.IsFixed || padding.Right.IsFixed))
{
Indentation indentation = new Indentation();
if (padding.Left.Value > 0)
{
indentation.Left = padding.Left.ValueInDxa.ToString(CultureInfo.InvariantCulture);
}
if (padding.Right.Value > 0)
{
indentation.Right = padding.Right.ValueInDxa.ToString(CultureInfo.InvariantCulture);
}
currentParagraph.InsertInProperties(prop => prop.Indentation = indentation);
}
newParagraph |= htmlStyles.Paragraph.ProcessCommonAttributes(en, styleAttributes);
return(newParagraph);
}
/// <summary>
/// Start the parse processing.
/// </summary>
/// <returns>Returns a list of parsed paragraph.</returns>
public IList <OpenXmlCompositeElement> Parse(String html)
{
if (String.IsNullOrEmpty(html))
{
return(new Paragraph[0]);
}
// ensure a body exists to avoid any errors when trying to access it
if (mainPart.Document == null)
{
new Document(new Body()).Save(mainPart);
}
else if (mainPart.Document.Body == null)
{
mainPart.Document.Body = new Body();
}
// Reset:
elements = new List <OpenXmlElement>();
paragraphs = new List <OpenXmlCompositeElement>();
tables = new TableContext();
htmlStyles.Runs.Reset();
currentParagraph = null;
// Start a new processing
paragraphs.Add(currentParagraph = htmlStyles.Paragraph.NewParagraph());
if (htmlStyles.DefaultStyles.ParagraphStyle != null)
{
currentParagraph.ParagraphProperties = new ParagraphProperties {
ParagraphStyleId = new ParagraphStyleId {
Val = htmlStyles.DefaultStyles.ParagraphStyle
}
};
}
HtmlEnumerator en = new HtmlEnumerator(html);
ProcessHtmlChunks(en, null);
if (elements.Count > 0)
{
this.currentParagraph.Append(elements);
}
// As the Parse method is public, to avoid changing the type of the return value, I use this proxy
// that will allow me to call the recursive method RemoveEmptyParagraphs with no major changes, impacting the client.
RemoveEmptyParagraphs();
return(paragraphs);
}
/// <summary>
/// Converts some common styling attributes to their OpenXml equivalence.
/// </summary>
/// <param name="en">The Html parser.</param>
/// <param name="styleAttributes">The collection of attributes where to store new discovered attributes.</param>
public void ProcessCommonAttributes(HtmlEnumerator en, IList <OpenXmlElement> styleAttributes)
{
if (en.Attributes.Count == 0)
{
return;
}
var colorValue = en.StyleAttributes.GetAsColor("color");
if (colorValue.IsEmpty)
{
colorValue = en.Attributes.GetAsColor("color");
}
if (!colorValue.IsEmpty)
{
styleAttributes.Add(new Color {
Val = colorValue.ToHexString()
});
}
colorValue = en.StyleAttributes.GetAsColor("background-color");
if (!colorValue.IsEmpty)
{
// change the way the background-color renders. It now uses Shading instead of Highlight.
// Changes brought by Wude on http://html2openxml.codeplex.com/discussions/277570
styleAttributes.Add(new Shading {
Val = ShadingPatternValues.Clear, Fill = colorValue.ToHexString()
});
}
var decorations = Converter.ToTextDecoration(en.StyleAttributes["text-decoration"]);
if ((decorations & TextDecoration.Underline) != 0)
{
styleAttributes.Add(new Underline {
Val = UnderlineValues.Single
});
}
if ((decorations & TextDecoration.LineThrough) != 0)
{
styleAttributes.Add(new Strike());
}
String[] classes = en.Attributes.GetAsClass();
if (classes != null)
{
for (int i = 0; i < classes.Length; i++)
{
string className = documentStyle.GetStyle(classes[i], StyleValues.Character, ignoreCase: true);
if (className != null) // only one Style can be applied in OpenXml and dealing with inheritance is out of scope
{
styleAttributes.Add(new RunStyle()
{
Val = className
});
break;
}
}
}
HtmlFont font = en.StyleAttributes.GetAsFont("font");
if (!font.IsEmpty)
{
if (font.Style == FontStyle.Italic)
{
styleAttributes.Add(new Italic());
}
if (font.Weight == FontWeight.Bold || font.Weight == FontWeight.Bolder)
{
styleAttributes.Add(new Bold());
}
if (font.Variant == FontVariant.SmallCaps)
{
styleAttributes.Add(new SmallCaps());
}
if (font.Family != null)
{
styleAttributes.Add(new RunFonts()
{
Ascii = font.Family, HighAnsi = font.Family
});
}
// size are half-point font size
if (font.Size.IsFixed)
{
styleAttributes.Add(new FontSize()
{
Val = (font.Size.ValueInPoint * 2).ToString(CultureInfo.InvariantCulture)
});
}
}
}
public int ProcessItem(HtmlEnumerator en)
{
if (!firstItem)
{
return(this.InstanceID);
}
firstItem = false;
// in case a margin has been specifically specified, we need to create a new list template
// on the fly with a different AbsNumId, in order to let Word doesn't merge the style with its predecessor.
Margin margin = en.StyleAttributes.GetAsMargin("margin");
if (margin.Left.Value > 0 && margin.Left.Type == UnitMetric.Pixel)
{
Numbering numbering = mainPart.NumberingDefinitionsPart.Numbering;
foreach (AbstractNum absNum in numbering.Elements <AbstractNum>())
{
if (absNum.AbstractNumberId == numInstances.Peek().Value)
{
Level lvl = absNum.GetFirstChild <Level>();
Int32 currentNumId = ++nextInstanceID;
numbering.Append(
new AbstractNum(
new MultiLevelType()
{
Val = MultiLevelValues.SingleLevel
},
new Level {
StartNumberingValue = new StartNumberingValue()
{
Val = 1
},
NumberingFormat = new NumberingFormat()
{
Val = lvl.NumberingFormat.Val
},
LevelIndex = 0,
LevelText = new LevelText()
{
Val = lvl.LevelText.Val
}
}
)
{
AbstractNumberId = currentNumId
});
numbering.Save(mainPart.NumberingDefinitionsPart);
numbering.Append(
new NumberingInstance(
new AbstractNumId()
{
Val = currentNumId
}
)
{
NumberID = currentNumId
});
numbering.Save(mainPart.NumberingDefinitionsPart);
mainPart.NumberingDefinitionsPart.Numbering.Reload();
break;
}
}
}
return(this.InstanceID);
}
/// <summary>
/// There is a few attributes shared by a large number of tags. This method will check them for a limited
/// number of tags (<p>, <pre>, <div>, <span> and <body>).
/// </summary>
/// <returns>Returns true if the processing of this tag should generate a new paragraph.</returns>
public bool ProcessCommonAttributes(HtmlEnumerator en, IList <OpenXmlElement> styleAttributes)
{
if (en.Attributes.Count == 0)
{
return(false);
}
bool newParagraph = false;
List <OpenXmlElement> containerStyleAttributes = new List <OpenXmlElement>();
string attrValue = en.Attributes["lang"];
if (attrValue != null && attrValue.Length > 0)
{
try
{
#if !NET_CORE
var ci = System.Globalization.CultureInfo.GetCultureInfo(attrValue);
#else
var ci = new System.Globalization.CultureInfo(attrValue);
#endif
bool rtl = ci.TextInfo.IsRightToLeft;
Languages lang = new Languages()
{
Val = ci.TwoLetterISOLanguageName
};
if (rtl)
{
lang.Bidi = ci.Name;
styleAttributes.Add(new Languages()
{
Bidi = ci.Name
});
// notify table
documentStyle.Tables.BeginTag(en.CurrentTag, new TableJustification()
{
Val = TableRowAlignmentValues.Right
});
}
containerStyleAttributes.Add(new ParagraphMarkRunProperties(lang));
containerStyleAttributes.Add(new BiDi()
{
Val = OnOffValue.FromBoolean(rtl)
});
}
catch (ArgumentException exc)
{
// lang not valid, ignore it
if (Logging.On)
{
Logging.PrintError($"lang attribute {attrValue} not recognized: " + exc.Message, exc);
}
}
}
attrValue = en.StyleAttributes["text-align"];
if (attrValue != null && en.CurrentTag != "<font>")
{
JustificationValues?align = Converter.ToParagraphAlign(attrValue);
if (align.HasValue)
{
containerStyleAttributes.Add(new Justification {
Val = align
});
}
}
// according to w3c, dir should be used in conjonction with lang. But whatever happens, we'll apply the RTL layout
attrValue = en.Attributes["dir"];
if (attrValue != null)
{
if (attrValue.Equals("rtl", StringComparison.OrdinalIgnoreCase))
{
styleAttributes.Add(new RightToLeftText());
containerStyleAttributes.Add(new Justification()
{
Val = JustificationValues.Right
});
}
else if (attrValue.Equals("ltr", StringComparison.OrdinalIgnoreCase))
{
containerStyleAttributes.Add(new Justification()
{
Val = JustificationValues.Left
});
}
}
// <span> and <font> are considered as semi-container attribute. When converted to OpenXml, there are Runs but not Paragraphs
if (en.CurrentTag == "<p>" || en.CurrentTag == "<div>" || en.CurrentTag == "<pre>")
{
var border = en.StyleAttributes.GetAsBorder("border");
if (!border.IsEmpty)
{
ParagraphBorders borders = new ParagraphBorders();
if (border.Top.IsValid)
{
borders.Append(
new TopBorder()
{
Val = border.Top.Style, Color = border.Top.Color.ToHexString(), Size = (uint)border.Top.Width.ValueInPx * 4, Space = 1U
});
}
if (border.Left.IsValid)
{
borders.Append(
new LeftBorder()
{
Val = border.Left.Style, Color = border.Left.Color.ToHexString(), Size = (uint)border.Left.Width.ValueInPx * 4, Space = 1U
});
}
if (border.Bottom.IsValid)
{
borders.Append(
new BottomBorder()
{
Val = border.Bottom.Style, Color = border.Bottom.Color.ToHexString(), Size = (uint)border.Bottom.Width.ValueInPx * 4, Space = 1U
});
}
if (border.Right.IsValid)
{
borders.Append(
new RightBorder()
{
Val = border.Right.Style, Color = border.Right.Color.ToHexString(), Size = (uint)border.Right.Width.ValueInPx * 4, Space = 1U
});
}
containerStyleAttributes.Add(borders);
newParagraph = true;
}
}
else if (en.CurrentTag == "<span>" || en.CurrentTag == "<font>")
{
// OpenXml limits the border to 4-side of the same color and style.
SideBorder border = en.StyleAttributes.GetAsSideBorder("border");
if (border.IsValid)
{
styleAttributes.Add(new DocumentFormat.OpenXml.Wordprocessing.Border()
{
Val = border.Style,
Color = border.Color.ToHexString(),
Size = (uint)border.Width.ValueInPx * 4,
Space = 1U
});
}
}
String[] classes = en.Attributes.GetAsClass();
if (classes != null)
{
for (int i = 0; i < classes.Length; i++)
{
string className = documentStyle.GetStyle(classes[i], StyleValues.Paragraph, ignoreCase: true);
if (className != null)
{
containerStyleAttributes.Add(new ParagraphStyleId()
{
Val = className
});
newParagraph = true;
break;
}
}
}
Margin margin = en.StyleAttributes.GetAsMargin("margin");
Indentation indentation = null;
if (!margin.IsEmpty)
{
if (margin.Top.IsFixed || margin.Bottom.IsFixed)
{
SpacingBetweenLines spacing = new SpacingBetweenLines();
if (margin.Top.IsFixed)
{
spacing.Before = margin.Top.ValueInDxa.ToString(CultureInfo.InvariantCulture);
}
if (margin.Bottom.IsFixed)
{
spacing.After = margin.Bottom.ValueInDxa.ToString(CultureInfo.InvariantCulture);
}
containerStyleAttributes.Add(spacing);
}
if (margin.Left.IsFixed || margin.Right.IsFixed)
{
indentation = new Indentation();
if (margin.Left.IsFixed)
{
indentation.Left = margin.Left.ValueInDxa.ToString(CultureInfo.InvariantCulture);
}
if (margin.Right.IsFixed)
{
indentation.Right = margin.Right.ValueInDxa.ToString(CultureInfo.InvariantCulture);
}
containerStyleAttributes.Add(indentation);
}
}
// implemented by giorand (feature #13787)
Unit textIndent = en.StyleAttributes.GetAsUnit("text-indent");
if (textIndent.IsValid && (en.CurrentTag == "<p>" || en.CurrentTag == "<div>"))
{
if (indentation == null)
{
indentation = new Indentation();
}
indentation.FirstLine = textIndent.ValueInDxa.ToString(CultureInfo.InvariantCulture);
containerStyleAttributes.Add(indentation);
}
this.BeginTag(en.CurrentTag, containerStyleAttributes);
// Process general run styles
documentStyle.Runs.ProcessCommonAttributes(en, styleAttributes);
return(newParagraph);
}
/// <summary>
/// Move inside the current tag related to table (td, thead, tr, ...) and converts some common
/// attributes to their OpenXml equivalence.
/// </summary>
/// <param name="en">The Html enumerator positionned on a <i>table (or related)</i> tag.</param>
/// <param name="runStyleAttributes">The collection of attributes where to store new discovered attributes.</param>
public void ProcessCommonAttributes(HtmlEnumerator en, IList <OpenXmlElement> runStyleAttributes)
{
List <OpenXmlElement> containerStyleAttributes = new List <OpenXmlElement>();
var colorValue = en.StyleAttributes.GetAsColor("background-color");
if (colorValue.IsEmpty) //We do a first try, if it's empty the background-color, I try it with background
{
colorValue = en.StyleAttributes.GetAsColor("background");
}
//
// "background-color" is also handled by RunStyleCollection which duplicate this attribute (bug #13212).
// Also apply on <th> (issue #20).
// As on 05 Jan 2018, the duplication was due to the wrong argument passed during the td/th processing.
// It was the runStyle and not the containerStyle that was provided. The code has been removed as no more useful
if (colorValue.IsEmpty)
{
colorValue = en.Attributes.GetAsColor("bgcolor");
}
if (!colorValue.IsEmpty)
{
containerStyleAttributes.Add(
new Shading()
{
Val = ShadingPatternValues.Clear, Color = "auto", Fill = colorValue.ToHexString()
});
}
var htmlAlign = en.StyleAttributes["vertical-align"];
if (htmlAlign == null)
{
htmlAlign = en.Attributes["valign"];
}
if (htmlAlign != null)
{
TableVerticalAlignmentValues?valign = Converter.ToVAlign(htmlAlign);
if (valign.HasValue)
{
containerStyleAttributes.Add(new TableCellVerticalAlignment()
{
Val = valign
});
}
}
htmlAlign = en.StyleAttributes["text-align"];
if (htmlAlign == null)
{
htmlAlign = en.Attributes["align"];
}
if (htmlAlign != null)
{
JustificationValues?halign = Converter.ToParagraphAlign(htmlAlign);
if (halign.HasValue)
{
this.BeginTagForParagraph(en.CurrentTag, new KeepNext(), new Justification {
Val = halign
});
}
}
// implemented by ddforge
String[] classes = en.Attributes.GetAsClass();
if (classes != null)
{
for (int i = 0; i < classes.Length; i++)
{
string className = documentStyle.GetStyle(classes[i], StyleValues.Table, ignoreCase: true);
if (className != null) // only one Style can be applied in OpenXml and dealing with inheritance is out of scope
{
containerStyleAttributes.Add(new RunStyle()
{
Val = className
});
break;
}
}
}
this.BeginTag(en.CurrentTag, containerStyleAttributes);
// Process general run styles
documentStyle.Runs.ProcessCommonAttributes(en, runStyleAttributes);
}
