Fix reading HTML via XmlReader API when root element is absent

[greatvovan]

Consider folowing snippet:

            string toParse;
            toParse = "<p>one</p><p>two</p><p>three</p>";
            var sr = new StringReader(toParse);
            var NULL = "NULL";

            using (var reader = new SgmlReader()
                {InputStream = sr, DocType = "HTML", IgnoreDtd = true})
            {
                while (!reader.EOF)
                {
                    reader.Read();
                    WriteLine($"[{reader.NodeType,10}] {reader.Name ?? NULL,5}: " +
                        $"{reader.Value ?? NULL}");
                }
            }

On version 1.8.12 SGMLReader fails to read all document since it comes to this section twice:

            if (this.Depth == 1)
            {
                if (this.m_rootCount == 1)
                {
                    // Hmmm, we found another root level tag, soooo, the only
                    // thing we can do to keep this a valid XML document is stop
                    this.m_state = State.Eof;
                    return false;
                }
                this.m_rootCount++;
            }

It only returns the first <p> element and opening tag for second.

It was working in old good version 1.8.6, but I don't have a code of it, as on GitHub history starts from version 1.8.7 wich is already broken.

This commit allows to fix this behavior if user has set DocType to "HTML" explicitly.

[UweKeim]

Awesome, just the same problem as I'm having here, too. Hopefully it get's integrated and published to NuGet, soon.

I'm now wrapping my HTML fragments inside an artificial <div> root tag to fulfil the requirement of one root only. Of course, this only works when reading from the document, not wenn modifying.

[lovettchris]

I'm worried this change is too HTML centric, what about SgmlReader over other types of data, like OFX ?

[greatvovan]

@lovettchris I can not see how it affects OFX and other types. Why did you get this particular doubt?