//=====================================================================
/// <summary>
/// Constructor
/// </summary>
/// <param name="source">The source path containing the HTML files</param>
/// <param name="dest">The destination path for the MAML topics and
/// supporting files</param>
/// <param name="createCompanion">True to create companion files for
/// all topic files or false to not create them.</param>
/// <param name="moveIntroText">If true, move text before the first section into an introduction element.
/// If false, insert a place holder introduction element.</param>
public HtmlToMaml(string source, string dest, bool createCompanion, bool moveIntroText)
{
XPathDocument rulesFile;
XPathNavigator navRules;
StringBuilder sb;
string name;
sourcePath = source;
destPath = dest;
createCompanionFile = createCompanion;
replaceIntro = moveIntroText;
pathProvider = this;
if (sourcePath.EndsWith("\\", StringComparison.Ordinal))
{
sourcePath = sourcePath.Substring(0, sourcePath.Length - 1);
}
if (destPath.EndsWith("\\", StringComparison.Ordinal))
{
destPath = destPath.Substring(0, destPath.Length - 1);
}
sourcePath = Path.GetFullPath(sourcePath);
destPath = Path.GetFullPath(destPath);
topics = new TopicCollection();
images = new ImageReferenceCollection();
topicDictionary = new Dictionary <FilePath, Topic>();
imageDictionary = new Dictionary <FilePath, ImageReference>();
conversionRules = new Dictionary <string, TagOptions>();
markupSections = new List <string>();
entities = new Dictionary <string, string>();
tokens = new Dictionary <string, string>();
// Load the conversion rules
rulesFile = new XPathDocument(Path.Combine(Path.GetDirectoryName(
Assembly.GetExecutingAssembly().Location), "ConversionRules.xml"));
navRules = rulesFile.CreateNavigator();
XPathNavigator bodyExpr = navRules.SelectSingleNode("//BodyExtract");
if (bodyExpr != null)
{
FileParser.BodyExtractExpression = bodyExpr.GetAttribute(
"expression", string.Empty);
}
// Add the tags we will handle internally
conversionRules.Add("a", null);
conversionRules.Add("code", null);
conversionRules.Add("h1", null);
conversionRules.Add("h2", null);
conversionRules.Add("h3", null);
conversionRules.Add("h4", null);
conversionRules.Add("h5", null);
conversionRules.Add("h6", null);
conversionRules.Add("img", null);
conversionRules.Add("see", null);
// Get the rules to process
sb = new StringBuilder();
foreach (XPathNavigator nav in navRules.Select("//Entities/Entity"))
{
entities.Add(nav.GetAttribute("name", String.Empty),
nav.GetAttribute("value", String.Empty));
}
foreach (XPathNavigator nav in navRules.Select("//MarkupWrapper/Tag"))
{
if (sb.Length != 0)
{
sb.Append("|");
}
name = nav.GetAttribute("name", String.Empty).ToLower(
CultureInfo.InvariantCulture);
conversionRules.Add(name, null);
sb.Append(name);
}
name = sb.ToString();
sb.Insert(0, @"<\s*(");
sb.Append(@")[^>]*?>.*?<\s*/\s*(\1)[^>]*?>|<\s*(");
sb.Append(name);
sb.Append(@")\s*?((\s|/)[^>]*?)?>");
reMarkupWrapper = new Regex(sb.ToString(), RegexOptions.IgnoreCase |
RegexOptions.Singleline);
reReplaceMarker = new Regex("\xFF");
matchMarkupWrapper = new MatchEvaluator(OnMatchMarkupWrapper);
matchMarker = new MatchEvaluator(OnMatchMarker);
sb.Remove(0, sb.Length);
foreach (XPathNavigator nav in navRules.Select("//Remove/Tag"))
{
if (sb.Length != 0)
{
sb.Append("|");
}
name = nav.GetAttribute("name", String.Empty).ToLower(
CultureInfo.InvariantCulture);
conversionRules.Add(name, null);
sb.Append(name);
}
sb.Insert(0, @"<\s*/?\s*(");
sb.Append(@")\s*?((\s|/)[^>]*?)?>");
reRemoveTag = new Regex(sb.ToString(), RegexOptions.IgnoreCase |
RegexOptions.Singleline);
sb.Remove(0, sb.Length);
foreach (XPathNavigator nav in navRules.Select("//Replace/Tag"))
{
if (sb.Length != 0)
{
sb.Append("|");
}
name = nav.GetAttribute("name", String.Empty).ToLower(
CultureInfo.InvariantCulture);
conversionRules.Add(name, new TagOptions(nav));
sb.Append(name);
}
sb.Insert(0, @"<\s*(?<Closing>/?)\s*(?<Tag>");
sb.Append(@")\s*(?<Attributes>(\s|/)[^>]*?)?>");
reReplaceTag = new Regex(sb.ToString(), RegexOptions.IgnoreCase |
RegexOptions.Singleline);
matchReplace = new MatchEvaluator(OnMatchReplace);
matchCode = new MatchEvaluator(OnMatchCode);
matchSee = new MatchEvaluator(OnMatchSee);
matchAnchor = new MatchEvaluator(OnMatchAnchor);
matchImage = new MatchEvaluator(OnMatchImage);
matchHeading = new MatchEvaluator(OnMatchHeading);
matchIntroduction = new MatchEvaluator(OnMatchIntroduction);
matchEntity = new MatchEvaluator(OnMatchEntity);
matchToken = new MatchEvaluator(OnMatchToken);
}
//=====================================================================
/// <summary>
/// Constructor
/// </summary>
/// <param name="source">The source path containing the HTML files</param>
/// <param name="dest">The destination path for the MAML topics and
/// supporting files</param>
/// <param name="createCompanion">True to create companion files for
/// all topic files or false to not create them.</param>
/// <param name="moveIntroText">If true, move text before the first section into an introduction element.
/// If false, insert a place holder introduction element.</param>
public HtmlToMaml(string source, string dest, bool createCompanion, bool moveIntroText)
{
XPathDocument rulesFile;
XPathNavigator navRules;
StringBuilder sb;
string name;
sourcePath = source;
destPath = dest;
createCompanionFile = createCompanion;
replaceIntro = moveIntroText;
pathProvider = this;
if(sourcePath.EndsWith("\\", StringComparison.Ordinal))
sourcePath = sourcePath.Substring(0, sourcePath.Length - 1);
if(destPath.EndsWith("\\", StringComparison.Ordinal))
destPath = destPath.Substring(0, destPath.Length - 1);
sourcePath = Path.GetFullPath(sourcePath);
destPath = Path.GetFullPath(destPath);
topics = new TopicCollection();
images = new ImageReferenceCollection();
topicDictionary = new Dictionary<FilePath, Topic>();
imageDictionary = new Dictionary<FilePath, ImageReference>();
conversionRules = new Dictionary<string, TagOptions>();
markupSections = new List<string>();
entities = new Dictionary<string, string>();
tokens = new Dictionary<string, string>();
// Load the conversion rules
rulesFile = new XPathDocument(Path.Combine(Path.GetDirectoryName(
Assembly.GetExecutingAssembly().Location), "ConversionRules.xml"));
navRules = rulesFile.CreateNavigator();
XPathNavigator bodyExpr = navRules.SelectSingleNode("//BodyExtract");
if(bodyExpr != null)
FileParser.BodyExtractExpression = bodyExpr.GetAttribute(
"expression", string.Empty);
// Add the tags we will handle internally
conversionRules.Add("a", null);
conversionRules.Add("code", null);
conversionRules.Add("h1", null);
conversionRules.Add("h2", null);
conversionRules.Add("h3", null);
conversionRules.Add("h4", null);
conversionRules.Add("h5", null);
conversionRules.Add("h6", null);
conversionRules.Add("img", null);
conversionRules.Add("see", null);
// Get the rules to process
sb = new StringBuilder();
foreach(XPathNavigator nav in navRules.Select("//Entities/Entity"))
entities.Add(nav.GetAttribute("name", String.Empty),
nav.GetAttribute("value", String.Empty));
foreach(XPathNavigator nav in navRules.Select("//MarkupWrapper/Tag"))
{
if(sb.Length != 0)
sb.Append("|");
name = nav.GetAttribute("name", String.Empty).ToLower(
CultureInfo.InvariantCulture);
conversionRules.Add(name, null);
sb.Append(name);
}
name = sb.ToString();
sb.Insert(0, @"<\s*(");
sb.Append(@")[^>]*?>.*?<\s*/\s*(\1)[^>]*?>|<\s*(");
sb.Append(name);
sb.Append(@")\s*?((\s|/)[^>]*?)?>");
reMarkupWrapper = new Regex(sb.ToString(), RegexOptions.IgnoreCase |
RegexOptions.Singleline);
reReplaceMarker= new Regex("\xFF");
matchMarkupWrapper = new MatchEvaluator(OnMatchMarkupWrapper);
matchMarker = new MatchEvaluator(OnMatchMarker);
sb.Remove(0, sb.Length);
foreach(XPathNavigator nav in navRules.Select("//Remove/Tag"))
{
if(sb.Length != 0)
sb.Append("|");
name = nav.GetAttribute("name", String.Empty).ToLower(
CultureInfo.InvariantCulture);
conversionRules.Add(name, null);
sb.Append(name);
}
sb.Insert(0, @"<\s*/?\s*(");
sb.Append(@")\s*?((\s|/)[^>]*?)?>");
reRemoveTag = new Regex(sb.ToString(), RegexOptions.IgnoreCase |
RegexOptions.Singleline);
sb.Remove(0, sb.Length);
foreach(XPathNavigator nav in navRules.Select("//Replace/Tag"))
{
if(sb.Length != 0)
sb.Append("|");
name = nav.GetAttribute("name", String.Empty).ToLower(
CultureInfo.InvariantCulture);
conversionRules.Add(name, new TagOptions(nav));
sb.Append(name);
}
sb.Insert(0, @"<\s*(?<Closing>/?)\s*(?<Tag>");
sb.Append(@")\s*(?<Attributes>(\s|/)[^>]*?)?>");
reReplaceTag = new Regex(sb.ToString(), RegexOptions.IgnoreCase |
RegexOptions.Singleline);
matchReplace = new MatchEvaluator(OnMatchReplace);
matchCode = new MatchEvaluator(OnMatchCode);
matchSee = new MatchEvaluator(OnMatchSee);
matchAnchor = new MatchEvaluator(OnMatchAnchor);
matchImage = new MatchEvaluator(OnMatchImage);
matchHeading = new MatchEvaluator(OnMatchHeading);
matchIntroduction = new MatchEvaluator(OnMatchIntroduction);
matchEntity = new MatchEvaluator(OnMatchEntity);
matchToken = new MatchEvaluator(OnMatchToken);
}
/// <summary>
/// Main program entry point
/// </summary>
/// <param name="args">The command line arguments (the source folder,
/// the destination folder, and the optional "/companion" option to
/// create companion files for each topic file).</param>
/// <returns>Zero on success, non-zero on failure</returns>
public static int Main(string[] args)
{
string sourceFolder = null, destFolder = null;
bool success = false, createCompanionFile = false, moveIntro = false, badArgs = false;
Assembly asm = Assembly.GetExecutingAssembly();
FileVersionInfo fvi = FileVersionInfo.GetVersionInfo(asm.Location);
Console.WriteLine("{0}, version {1}\r\n{2}\r\nE-Mail: [email protected]\r\n", fvi.ProductName,
fvi.ProductVersion, fvi.LegalCopyright);
try
{
if(args.Length < 2 || args.Length > 4)
badArgs = true;
else
{
sourceFolder = Path.GetFullPath(args[0]);
destFolder = Path.GetFullPath(args[1]);
for(int i = 2; i < args.Length; i++)
if(args[i].Length > 1 && args[i].Substring(1).Equals("companion",
StringComparison.OrdinalIgnoreCase))
createCompanionFile = true;
else
if(args[i].Length > 1 && args[i].Substring(1).Equals("moveIntro",
StringComparison.OrdinalIgnoreCase))
moveIntro = true;
else
badArgs = true;
}
if(badArgs)
{
Console.WriteLine("Syntax: ConvertHtmlToMaml sourceFolder destFolder " +
"[/companion] [/moveIntro]");
return 1;
}
if(!Directory.Exists(sourceFolder))
{
Console.WriteLine("Source folder does not exist");
return 2;
}
if(String.Compare(sourceFolder, destFolder, StringComparison.OrdinalIgnoreCase) == 0)
{
Console.WriteLine("Destination folder cannot match source folder");
return 3;
}
if(!Directory.Exists(destFolder))
Directory.CreateDirectory(destFolder);
swLog = new StreamWriter(Path.Combine(destFolder, "_ConversionLog.log"));
HtmlToMaml htmlToMaml = new HtmlToMaml(sourceFolder, destFolder, createCompanionFile,
moveIntro);
htmlToMaml.ConversionProgress += htmlToMaml_ConversionProgress;
htmlToMaml.ConvertTopics();
success = true;
}
catch(Exception ex)
{
System.Diagnostics.Debug.WriteLine(ex.ToString());
Console.WriteLine(ex.ToString());
}
finally
{
if(swLog != null)
swLog.Close();
#if DEBUG
if(System.Diagnostics.Debugger.IsAttached)
{
Console.WriteLine("Hit ENTER...");
Console.ReadLine();
}
#endif
}
return (success) ? 0 : 4;
}
/// <summary>
/// Main program entry point
/// </summary>
/// <param name="args">The command line arguments (the source folder, the destination folder, and the
/// optional "/companion" option to create companion files for each topic file).</param>
/// <returns>Zero on success, non-zero on failure</returns>
public static int Main(string[] args)
{
string sourceFolder = null, destFolder = null;
bool success = false, createCompanionFile = false, moveIntro = false, badArgs = false;
Assembly asm = Assembly.GetExecutingAssembly();
FileVersionInfo fvi = FileVersionInfo.GetVersionInfo(asm.Location);
Console.WriteLine("{0}, version {1}\r\n{2}\r\nE-Mail: [email protected]\r\n", fvi.ProductName,
fvi.ProductVersion, fvi.LegalCopyright);
try
{
if (args.Length < 2 || args.Length > 4)
{
badArgs = true;
}
else
{
sourceFolder = Path.GetFullPath(args[0]);
destFolder = Path.GetFullPath(args[1]);
for (int i = 2; i < args.Length; i++)
{
if (args[i].Length > 1 && args[i].Substring(1).Equals("companion", StringComparison.OrdinalIgnoreCase))
{
createCompanionFile = true;
}
else
if (args[i].Length > 1 && args[i].Substring(1).Equals("moveIntro", StringComparison.OrdinalIgnoreCase))
{
moveIntro = true;
}
else
{
badArgs = true;
}
}
}
if (badArgs)
{
Console.WriteLine("Syntax: ConvertHtmlToMaml sourceFolder destFolder [/companion] [/moveIntro]");
return(1);
}
if (!Directory.Exists(sourceFolder))
{
Console.WriteLine("Source folder does not exist");
return(2);
}
if (sourceFolder.Equals(destFolder, StringComparison.OrdinalIgnoreCase))
{
Console.WriteLine("Destination folder cannot match source folder");
return(3);
}
if (!Directory.Exists(destFolder))
{
Directory.CreateDirectory(destFolder);
}
swLog = new StreamWriter(Path.Combine(destFolder, "_ConversionLog.log"));
HtmlToMaml htmlToMaml = new HtmlToMaml(sourceFolder, destFolder, createCompanionFile, moveIntro);
htmlToMaml.ConversionProgress += htmlToMaml_ConversionProgress;
htmlToMaml.ConvertTopics();
success = true;
}
catch (Exception ex)
{
Debug.WriteLine(ex.ToString());
Console.WriteLine(ex.ToString());
}
finally
{
if (swLog != null)
{
swLog.Close();
}
#if DEBUG
if (Debugger.IsAttached)
{
Console.WriteLine("Hit ENTER...");
Console.ReadLine();
}
#endif
}
return((success) ? 0 : 4);
}
