views:

214

answers:

2

Hi Guys,

Does anybody have an example of spliting a html string (coming from a tiny mce editor) and splitting it into N parts using C#?

I need to split the string evenly without splitting words.

I was thinking of just splitting the html and using the HtmlAgilityPack to try and fix the broken tags. Though I'm not sure how to find the split point, as Ideally it should be based purley on the text rather than the html aswell.

Anybody got any ideas on how to go about this?

UPDATE

As requested, here is an example of input and desired output.

INPUT:

<p><strong>Lorem ipsum dolor sit amet, <em>consectetur adipiscing</em></strong> elit.</p>

OUTPUT (When split into 3 cols):

Part1: <p><strong>Lorem ipsum dolor</strong></p>
Part2: <p><strong>sit amet, <em>consectetur</em></strong></p>
Part3: <p><strong><em>adipiscing</em></strong> elit.</p>

UPDATE 2:

I've just had a play with Tidy HTML and that seems to work well at fixing broken tags, so this may be good option if I can find a way to locate the split pints?

UPDATE 3

Using a method similar to this http://stackoverflow.com/questions/1613896/truncate-string-on-whole-words-in-net-c, I've now managed to get a list of plain text words that will make up each part. So, say using Tidy HTML I have a valid XML structure for the html, and given this list of words, anybody got an idea on what would now be the best way to split it?

UPDATE 4

Can anybody see an issue with using a regex to find the indices with the HTML in the followin way:

Given the plain text string "sit amet, consectetur", replace all spaces with the regex "(\s|<(.|\n)+?>)*", in theory finding that string with any combination of spaces and/or tags

I could then just use Tidy HTML to fix the broken html tags?

Many thanks

Matt

+4  A: 

A Proposed Solution

Man, this is a curse of mine! I apparently cannot walk away from a problem without spending up-to-and-including an unreasonable amount of time on it.

I thought about this. I thought about HTML Tidy, and maybe it would work, but I had trouble wrapping my head around it.

So, I wrote my own solution.

I tested this on your input and on some other input that I threw together myself. It seems to work pretty well. Surely there are holes in it, but it might provide you with a starting point.

Anyway, my approach was this:

  1. Encapsulate the notion of a single word in an HTML document using a class that includes information about that word's position in the HTML document hierarchy, up to a given "top". This I have implemented in the HtmlWord class below.
  2. Create a class that is capable of writing a single line composed of these HTML words above, such that start-element and end-element tags are added in the appropriate places. This I have implemented in the HtmlLine class below.
  3. Write a few extension methods to make these classes immediately and intuitively accessible straight from an HtmlAgilityPack.HtmlNode object. These I have implemented in the HtmlHelper class below.

Am I crazy for doing all this? Probably, yes. But, you know, if you can't figure out any other way, you can give this a try.

Here's how it works with your sample input:

var document = new HtmlDocument();
document.LoadHtml("<p><strong>Lorem ipsum dolor sit amet, <em>consectetur adipiscing</em></strong> elit.</p>");

var nodeToSplit = document.DocumentNode.SelectSingleNode("p");
var lines = nodeToSplit.SplitIntoLines(3);

foreach (var line in lines)
    Console.WriteLine(line.ToString());

Output:

<p><strong>Lorem ipsum dolor </strong></p>
<p><strong>sit amet, <em>consectetur </em></strong></p>
<p><strong><em>adipiscing </em></strong>elit. </p>

And now for the code:

HtmlWord class

using System;
using System.Collections.Generic;
using System.Linq;

using HtmlAgilityPack;

public class HtmlWord {
    public string Text { get; private set; }
    public HtmlNode[] NodeStack { get; private set; }

    // convenience property to display list of ancestors cleanly
    // (for ease of debugging)
    public string NodeList {
        get { return string.Join(", ", NodeStack.Select(n => n.Name).ToArray()); }
    }

    internal HtmlWord(string text, HtmlNode node, HtmlNode top) {
        Text = text;
        NodeStack = GetNodeStack(node, top);
    }

    private static HtmlNode[] GetNodeStack(HtmlNode node, HtmlNode top) {
        var nodes = new Stack<HtmlNode>();

        while (node != null && !node.Equals(top)) {
            nodes.Push(node);
            node = node.ParentNode;
        };

        return nodes.ToArray();
    }
}

HtmlLine class

using System;
using System.Collections.Generic;
using System.IO;
using System.Linq;
using System.Text;
using System.Xml;

using HtmlAgilityPack;

[Flags()]
public enum NodeChange {
    None = 0,
    Dropped = 1,
    Added = 2
}

public class HtmlLine {
    private List<HtmlWord> _words;
    public IList<HtmlWord> Words {
        get { return _words.AsReadOnly(); }
    }

    public int WordCount {
        get { return _words.Count; }
    }

    public HtmlLine(IEnumerable<HtmlWord> words) {
        _words = new List<HtmlWord>(words);
    }

    private static NodeChange CompareNodeStacks(HtmlWord x, HtmlWord y, out HtmlNode[] droppedNodes, out HtmlNode[] addedNodes) {
        var droppedList = new List<HtmlNode>();
        var addedList = new List<HtmlNode>();

        // traverse x's NodeStack backwards to see which nodes
        // do not include y (and are therefore "finished")
        foreach (var node in x.NodeStack.Reverse()) {
            if (!Array.Exists(y.NodeStack, n => n.Equals(node)))
                droppedList.Add(node);
        }

        // traverse y's NodeStack forwards to see which nodes
        // do not include x (and are therefore "new")
        foreach (var node in y.NodeStack) {
            if (!Array.Exists(x.NodeStack, n => n.Equals(node)))
                addedList.Add(node);
        }

        droppedNodes = droppedList.ToArray();
        addedNodes = addedList.ToArray();

        NodeChange change = NodeChange.None;
        if (droppedNodes.Length > 0)
            change &= NodeChange.Dropped;
        if (addedNodes.Length > 0)
            change &= NodeChange.Added;

        // could maybe use this in some later revision?
        // not worth the effort right now...
        return change;
    }

    public override string ToString() {
        if (WordCount < 1)
            return string.Empty;

        var lineBuilder = new StringBuilder();

        using (var lineWriter = new StringWriter(lineBuilder))
        using (var xmlWriter = new XmlTextWriter(lineWriter)) {
            var firstWord = _words[0];
            foreach (var node in firstWord.NodeStack) {
                xmlWriter.WriteStartElement(node.Name);
                foreach (var attr in node.Attributes)
                    xmlWriter.WriteAttributeString(attr.Name, attr.Value);
            }
            xmlWriter.WriteString(firstWord.Text + " ");

            for (int i = 1; i < WordCount; ++i) {
                var previousWord = _words[i - 1];
                var word = _words[i];

                HtmlNode[] droppedNodes;
                HtmlNode[] addedNodes;

                CompareNodeStacks(
                    previousWord,
                    word,
                    out droppedNodes,
                    out addedNodes
                );

                foreach (var dropped in droppedNodes)
                    xmlWriter.WriteEndElement();
                foreach (var added in addedNodes) {
                    xmlWriter.WriteStartElement(added.Name);
                    foreach (var attr in added.Attributes)
                        xmlWriter.WriteAttributeString(attr.Name, attr.Value);
                }

                xmlWriter.WriteString(word.Text + " ");

                if (i == _words.Count - 1) {
                    foreach (var node in word.NodeStack)
                        xmlWriter.WriteEndElement();
                }
            }
        }

        return lineBuilder.ToString();
    }
}

HtmlHelper static class

using System;
using System.Collections.Generic;
using System.Linq;

using HtmlAgilityPack;

public static class HtmlHelper {
    public static IList<HtmlLine> SplitIntoLines(this HtmlNode node, int wordsPerLine) {
        var lines = new List<HtmlLine>();

        var words = node.GetWords(node.ParentNode);

        for (int i = 0; i < words.Count; i += wordsPerLine) {
            lines.Add(new HtmlLine(words.Skip(i).Take(wordsPerLine)));
        }

        return lines.AsReadOnly();
    }

    public static IList<HtmlWord> GetWords(this HtmlNode node, HtmlNode top) {
        var words = new List<HtmlWord>();

        if (node.HasChildNodes) {
            foreach (var child in node.ChildNodes)
                words.AddRange(child.GetWords(top));
        } else {
            var textNode = node as HtmlTextNode;
            if (textNode != null && !string.IsNullOrEmpty(textNode.Text)) {
                string[] singleWords = textNode.Text.Split(
                    new string[] {" "},
                    StringSplitOptions.RemoveEmptyEntries
                );
                words.AddRange(
                    singleWords
                        .Select(w => new HtmlWord(w, node.ParentNode, top)
                    )
                );
            }
        }

        return words.AsReadOnly();
    }
}

Conclusion

Just to reiterate: this is a thrown-together solution; I'm sure it has problems. I present it only as a starting point for you to consider -- again, if you're unable to get the behavior you want through other means.

Dan Tao
He's not splitting on paragraphs, though. He's trying to split a single paragraph into multiple paragraphs, while preserving the font-based formatting of the original.
MusiGenesis
@MusiGenesis: Yeah, I only noticed that after posting my original answer (shame on me). That's a much trickier problem! Now I'm putting a bit more thought into it.
Dan Tao
Yes, this is an interesting one. Good luck.
MusiGenesis
Voted +1 for OCD
kervin
Wowsers! Thats awesome. I'm going to have to give that a go =)Really appreciate all your help.
Matt Brailsford
+1 for volume alone. I feel your pain dude - I spent a couple of days not long ago trying to figure out how to render animation to the Windows desktop (*underneath* the icons). In a rare moment of sanity, I gave up.
MusiGenesis
@MusiGenesis: Those rare moments of sanity are quite rare indeed, for me.
Dan Tao
A: 

This suggestion is only a hack - hopefully there is a better way.

Basically, you want to take a chunk of HTML-formatted text and split it into smaller pieces that still preserve the font etc. of the original. I think you could load the original HTML into an RTF control or a Word object, split it into pieces there that preserve the formatting, and then output the pieces as separate HTML.

There may also be a way of using HtmlAgilityPack like this, if it provides a simple way of extracting text with formatting info from the original HTML.

MusiGenesis