Opentk/Source/Bind/DocProcessor.cs

127 lines
4.6 KiB
C#
Raw Normal View History

using System;
using System.Collections;
using System.Collections.Generic;
using System.IO;
using System.Linq;
using System.Text;
using System.Text.RegularExpressions;
using System.Xml;
using System.Xml.Linq;
using System.Xml.XPath;
using Bind.Structures;
namespace Bind
{
class DocProcessor
{
static readonly Regex remove_mathml = new Regex(
@"<(mml:math|inlineequation)[^>]*?>(?:.|\n)*?</\s*\1\s*>",
RegexOptions.Compiled | RegexOptions.Multiline | RegexOptions.IgnorePatternWhitespace);
static readonly Regex remove_doctype = new Regex(
@"<!DOCTYPE[^>\[]*(\[.*\])?>", RegexOptions.Compiled | RegexOptions.Multiline);
Documentation Cached;
string LastFile;
// Strips MathML tags from the source and replaces the equations with the content
// found in the <!-- eqn: :--> comments in the docs.
// Todo: Some simple MathML tags do not include comments, find a solution.
// Todo: Some files include more than 1 function - find a way to map these extra functions.
public Documentation ProcessFile(string file)
{
string text;
if (LastFile == file)
return Cached;
LastFile = file;
text = File.ReadAllText(file);
text = text
.Replace("&epsi;", "epsilon"); // Fix unrecognized &epsi; entities
text = remove_doctype.Replace(text, String.Empty);
Match m = remove_mathml.Match(text);
while (m.Length > 0)
{
string removed = text.Substring(m.Index, m.Length);
text = text.Remove(m.Index, m.Length);
int equation = removed.IndexOf("eqn");
if (equation > 0)
{
// Find the start and end of the equation string
int eqn_start = equation + 4;
int eqn_end = removed.IndexOf(":-->") - equation - 4;
if (eqn_end < 0)
{
// Note: a few docs from man4 delimit eqn end with ": -->"
eqn_end = removed.IndexOf(": -->") - equation - 4;
}
if (eqn_end < 0)
{
Console.WriteLine("[Warning] Failed to find equation for mml.");
goto next;
}
string eqn_substring = removed.Substring(eqn_start, eqn_end);
text = text.Insert(m.Index, "<![CDATA[" + eqn_substring + "]]>");
}
next:
m = remove_mathml.Match(text);
}
//XmlReader doc = null;
XDocument doc = null;
try
{
// The pure XmlReader is ~20x faster than the XmlTextReader.
//doc = XmlReader.Create(new StringReader(text), settings);
//XmlReader reader =
doc = XDocument.Parse(text);
Cached = ToInlineDocs(doc);
return Cached;
}
catch (Exception e)
{
Console.WriteLine(e.ToString());
Console.WriteLine(doc.ToString());
return null;
}
}
Documentation ToInlineDocs(XDocument doc)
{
var inline = new Documentation
{
Summary =
Cleanup(
((IEnumerable)doc.XPathEvaluate("//*[name()='refentry']/*[name()='refnamediv']/*[name()='refpurpose']"))
.Cast<XElement>().First().Value),
Parameters =
((IEnumerable)doc.XPathEvaluate("*[name()='refentry']/*[name()='refsect1'][@id='parameters']/*[name()='variablelist']/*[name()='varlistentry']"))
.Cast<XNode>()
.Select(p =>
new KeyValuePair<string, string>(
p.XPathSelectElement("*[name()='term']/*[name()='parameter']").Value.Trim(),
Cleanup(p.XPathSelectElement("*[name()='listitem']").Value)))
.ToList()
};
inline.Summary = Char.ToUpper(inline.Summary[0]) + inline.Summary.Substring(1);
return inline;
}
static readonly char[] newline = new char[] { '\n' };
static string Cleanup(string text)
{
return
String.Join(" ", text
.Replace("\r", "\n")
.Split(newline, StringSplitOptions.RemoveEmptyEntries)
.Select(s => s.Trim()).ToArray());
}
}
}