SgmlReader is a versatile C# .NET library written by Chris Lovett for parsing SGML files using the XmlReader API including HTML and OFX data. A command line utility is also provided which outputs the well formed XML result.
MindTouch uses the SgmlReader library extensively. They have also made many improvements to it.
The SgmlReader is an implementation of the XmlReader API. So the only thing you really need to know is how to construct it. SgmlReader has a default constructor, then you need to set some of the properties. To load a DTD you must specify DocType="HTML" or you must provide a SystemLiteral. To specify the SGML document you must provide either the InputStream or Href. Everything else is optional. Then you can read from this reader like any other XmlReader class.
This code is packaged and published at https://www.nuget.org/packages/Microsoft.Xml.SgmlReader/ for your convenience. Using Visual Studio, Manage Nuget Packages simply search for the online package named "Microsoft.Xml.SgmlReader" add it to your project and you are good to go.
// setup SgmlReader
Sgml.SgmlReader sgmlReader = new Sgml.SgmlReader()
{
DocType = "HTML",
WhitespaceHandling = WhitespaceHandling.All,
CaseFolding = Sgml.CaseFolding.ToLower,
InputStream = reader
};
// create document
XmlDocument doc = new XmlDocument()
{
PreserveWhitespace = true,
XmlResolver = null
};
doc.Load(sgmlReader);
return doc;| Property Name | Description |
|---|---|
| SgmlDtd Dtd | Specify the SgmlDtd object directly. This allows you to cache the Dtd and share it across multiple SgmlReaders. To load a DTD from a URL use the SystemLiteral property. |
| string DocType | The name of root element specified in the DOCTYPE tag. If you specify "HTML" then the SgmlReader will use the built-in HTML DTD. In this case you do not need to specify the SystemLiteral property. |
| string PublicIdentifier | The PUBLIC identifier in the DOCTYPE tag. This is optional. |
| string SystemLiteral | The SYSTEM literal in the DOCTYPE tag identifying the location of the DTD. |
| string InternalSubset | The DTD internal subset in the DOCTYPE tag. This is optional. |
| TextReader InputStream | The input stream containing SGML data to parse. You must specify this property or the Href property before calling Read(). |
| string Href | Specify the location of the input SGML document as a URL. |
| string WebProxy | Sometimes you need to specify a proxy server in order to load data via HTTP from outside the firewall. For example: "itgproxy:80". |
| string BaseUri | The base Uri is used to resolve relative Uri's like the SystemLiteral and Href properties. |
| TextWriter ErrorLog | DTD validation errors are written to this stream. |
| string ErrorLogFile | DTD validation errors are written to this log file. |
The command line executable version has the following options:
sgmlreader <options> [InputUri] [OutputFile]
| Argument Name | Description |
|---|---|
| -e "file" | Specifies a file to write error output to. The default is to generate no errors. The special name "$stderr" redirects errors to stderr output stream. |
| -proxy | "server" Specifies the proxy server to use to fetch DTD's through the fire wall. |
| -html | Specifies that the input is HTML. |
| -dtd "uri" | Specifies some other SGML DTD. |
| -base | Add an HTML base tag to the output. |
| -pretty | Pretty print the output. |
| -encoding name | Specify an encoding for the output file (default UTF-8) |
| -noxml | Stops generation of XML declaration in output. |
| -doctype | Copy <!DOCTYPE tag to the output. |
| InputUri | The input file name or URL. Default is stdin. If this is a local file name then it also supports wildcards. |
| OutputFile | The optional output file name. Default is stdout. If the InputUri contains wildcards then this just specifies the output file extension, the default being ".xml". |
Converts all .htm files to corresponding .xml files using the built in HTML DTD.
sgmlreader -html *.htm *.xmlConverts all the MSN home page to XML storing the result in the local file msn.xml.
sgmlreader -html http://www.msn.com -proxy myproxy:80 msn.xmlConverts the given OFX file to XML using the SGML DTD "ofx160.dtd" specified in the test.ofx file.
sgmlreader -dtd ofx160.dtd test.ofx ofx.xmlSgmlReaderUniversal provides a custom EntityResolver named UniversalEntityResolver which allows
resolving resources using the UWP Windows.Storage API's. You will need to provide this resolver
by setting the Resolver property on the SgmlReader.
// setup SgmlReader
Sgml.SgmlReader sgmlReader = new Sgml.SgmlReader() {
Resolver = new UniversalEntityResolver()
}If you have questions, please post them on StackOverflow and tag them with sgmlreader. Feel free to post issues in the github issues list.
If you fix an issue, please submit PR and follow these guidelines:
The build versioning system depends on pwsh and you can install that using:
dotnet tool install --global powershell
Now run the setup script:
pwsh -f Common/setup.ps1
And now you can build successfully:
dotnet build SgmlReaderCore.sln
- Open your Visual Studio Developer Command Prompt - or any cmd session where
msbuildis in thePATH. cd <directory containing SgmlReader.sln>BUILD.cmd
The BUILD.cmd script will automatically install pwsh if necessary, run setup.ps1, restore NuGet packages, and build the solution using the default configuration.
To change the version number edit ~\Common\version.txt only and not any other file. This version
will be replicated automatically to the following files:
Common/version.props
SgmlReader.nuspec
SgmlReaderUniversal/Propertyes/AssemblyInfo.cs
The setup.ps1 script has made these file changes transparent to git using the following filters:
git config --add filter.version.smudge "pwsh -f Common/smudge_version.ps1 %f"
git config --add filter.version.clean "pwsh -f Common/clean_version.ps1 %f"
When you "stage" a change to any one of these files the real version number will be replaced by the string "$version". When you checkout or pull an update to one of these files the string "$version" will be replaced
with the real version from versions.txt.
- Make sure the code formatting is identical to the existing code formatting. You know that you're doing it right if your code is indistinguishable from existing code.
- Fix any build warnings including code analysis warnings.
- Run the unit test to make sure no regressions are being introduced.
- Add a unit test to confirm your fix or feature, if needed.
- Submit a pull request on GitHub.
Please make sure all tests pass and new tests are added for areas you work on. See nunit. If you have Visual Studio, just open the Test Explorer and click Run All.
Note: all 1.8.x releases up to 1.8.7 are compatible with 1.8.0. Use assembly redirection to account for newer versions when recompilation is not an option.
- Lots of code cleanups by Jason Nelson bringing the code up to stuff with lovely new C# 8 language features.
- Add direct support for .NET 5.0 version of SgmlReader.
- Fix automated security alert in unit test project by updating log4net version.
Thanks to Jason Nelson for the following code cleanups:
- Makes HWStack generic to reduce casts and improve codegen
- Inlines variable declarations
- Uses pattern matching
- Adds readonly annotations
- Change build system to support .NET Core 3.1 and net45, net46, net47, net48, netstandard2.0 and netstandard2.1.
- Drop "Portable" library since this is replaced by netstandard.
- Redirect http://www.w3.org/TR/html4/loose.dtd to built-in html.dtd. (Chris Lovett)
- Fixed NameTable so SgmlReader can be used with XPathDocument. (Chris Lovett)
- Switched to xbuild and nunit on mono. (Chris Lovett)
- Add portable support by adding IEntityResolver interface. (Chris Lovett)
- Add desktop and universal implementation of IEntityResolver. (Chris Lovett)
- Move to NUnit 3 (which no longer requires VSIX extension to run the tests). (Chris Lovett)
- Move to Nuget 3 (multi-platform support including portable and universal apps). (Chris Lovett)
- Pulled latest psake and nuget tools. (Andy Sherwood)
- Made sure Html.dtd was embedded as a resource in the build script for the nuget package. (Andy Sherwood)
- Fixed AttributeCount and m_state problems for use under Mono. (Max Zhao)
- Decode unicode surrogate pairs. (CaptainCodeman)
- Incomplete entity codes are kept intact and escaped to produce valid HTML. (Marek Stój)
- Converted license from GPL 3 to Apache 2.0.
- Consolidated read-me files.
- Cleaned up solution and project files.
- Converted project files to Visual Studio 2010. (Andy Sherwood)
- Fully nugetized. Get log4net and nunit from nuget. Build nuget package with build script using psake and powershell. (Andy Sherwood)
- Provide setting to ignore DTD in parsed document (again).
- Provide setting to ignore DTD in parsed document.
- An attribute with a missing value should be assumed to have the name of the attribute as value.
- SgmlReader ExpandEntity with entities not ending in ';' and skips a character.
- SgmlReader adds 65535 character at the end of the string.
- Add test showing behavior of > char in string literals in XML.
- Unable to parse UTF-32 entities.
- Use StringComparison.OrdinalIgnoreCase instead of StringComparison.InvariantCultureIgnoreCase.
- Corrupt attributes may lead to invalid attribute names, which make the produced XML unparseable.
- Error when content contains prefixed XML processing instructions (e.g. <?xml:namespace prefix = o ns = "urn:schemas-microsoft-com:office:office" />).
- Added -ignore flag so tests known to fail can be ignored from the suite.
- Added test to re-parse output to make sure it's valid XML (.Net sometimes was able to generate invalid XML).
- Fixed CData section parsing skips over characters.
- Fixed regression introduced by fixing bug 5150.
- An extra open quote/double-quote prevents the entire element from being read properly.
- Replaced == string equality with culture invariant string.Compare().
- Return 'null' as NameTable since none is used.
- Added '-noformat' switch for regression tests to suppress automatic reformatting (useful for formatting tests).
- Unclosed HTML comment causes infinite loop.
- Don't use XmlNameTable with object comparisons; it becomes unreliable after a while.
BREAKING CHANGE: requires .NET 2.0
- Major code clean-up. (thx jamesgmbutler for the contribution!)
- Add XML-only entity ' to HTML DTD.
- Missing quote in attribute value causes catastrophic failure.
- Unknown prefixes cannot be mapped to the same namespace.
- ² entity is not recognized correctly.
- Added test for entities with digits.
- Never close the BODY tag early (it causes loss of content).
- Remove "<![CDATA[" inside CDATA sections.
- Remove "]]>" inside CDATA sections.
- Convert elements with invalid tag names into text (e.g. <foo@bar.com>).
- Fixed bug where parsing CDATA section skipped first character.
- Don't double parse commented out CDATA sections.
- Added support for namespaces on elements and attributes.
- Unknown prefixes on attributes and elements resolve to '#unknown' namespace.
- Fix bug when parsing down-level comments, like <![if IE]>.
- Don't allow attribute with invalid names (e.g. <p foo:="invalid" ;="bad">, etc.).
- Added 'GetLiteralEntitiesLookup()' method.
- Fixed bugs with namespace prefixes on attributes and elements; prefixes are now stripped automatically.
- Added SgmlReader constructor with XmlNameTable argument to avoid failed comparisons when reusing the DTD.
- Ensured that SgmlReader is initialized identically when reusing a DTD.
- Fix bug reported by chriswang - MoveToAttribute didn't save state properly.
- Fix bug reported by starascendent - build on Visual Studio 2003 was broken.
- Fix bug reported by sanchen - ExpandCharEntity was messed up on hex entities.
- Fix bug reported by kojiishi - off by one bug in SniffName().
- Fix bug reported by kojiishi - bug in loading XmlDocument from SgmlReader - this was caused by the HTML document containing an embedded <?xml version='1.0'?> declaration, so the SgmlReader now strips these.
- Added special stripping of punctuation characters between attributes like ",".
- Improve wrapping of HTML content with auto-generated <html></html> container tags.
- Fix detection of ContentType=text/html and switch to HTML mode.
- Fix problems parsing DOCTYPE tag when case folding is on.
- Fix reading of XHTML DTD.
- Fix parsing of content of type CDATA that resulted in the error message 'Cannot have ']]>' inside an XML CDATA block'.
- Fix parsing of http://www.virtuelvis.com/download/162/evilml.html.
- Fix parsing of attributes missing the equals sign: height"4" (thanks to Ulrich Schwanitz for his fix).
- Fix 'SniffWhitespace' thanks to "Windy Winter".
- Added TestSuite project.
- Added UserAgent string "Mozilla/4.0 (compatible;);" so that SgmlReader gets the right content from webservers.
- Fixed handling of HTML that does not start with root <html> element tag.
- Fixed handling of built in HTML entities.
- Changed ToUpper to CaseFolding enum and added support for "auto-folding" based on input
- Added support for <![CDATA[...]]> blocks
- Added proper encoding support, including support for HTML <META http-equiv="content-type". This means output now has the correct XML declaration (unless you specify the new -noxml option) and any existing xml declarations in the input are stripped out so you don't end up with two.
- Added support for ASP <%...%> blocks (thanks to Dan Whalin).
- Now strips out DOCTYPE by default since HTML DocTypes can cause problems for XmlDocument when it tries to load the HTML DTD. but added "-doctype" switch for those who really need it to come through.
- Fix handling of Office 2000 <?xml:namespace .../> declarations.
- Remove bogus attributes that have no name, in cases like <class= "test">.
- Converted back to Visual Studio 7.0 since this is the lowest common denominator.
- Added ToUpper switch for upper case folding, instead of the default lower case.
- Fix handling of UNC paths.
- Added OFX test suite.
- Fixed bug in parsing CDATA type elements (like <script><!-- --></script>).
- Upgraded project to Visual Studio 7.1.
- Fixed bug in accessing https authenticated sites.
- Fixed bug in handling of content that contains nulls.
- Improved handling of <!DOCTYPE with PUBLIC and no SYSTEM literal.
- Fixed bug in losing attributes when auto-closing tags.
- Fixed pretty printing output by adding WhitespaceHandling flag to SgmlReader.
- Added -encoding option so you can change the encoding of the output file.
- Implemented ReadOuterXml and ReadInnerXml and fix some bugs in dealing with xmlns attributes and dealing with non-HTML tags.
- Fixed some CLS compliance problems with using SgmlReader from VB and a null reference exception bug when loading SgmlReader from XmlDocument.
- Fixed bug in handling of encodings. Now uses the correct encoding returned from the HTTP server.
- Fixed bug in handling of input that contains blank lines at the top.
- Added fix for the way IE & Netscape deal with characters in the range 0x80 through 0x9F in HTML.
- Fixed bug in handling of empty elements, like <INPUT>.
- Add wildcard support for command line utility.
- Initial release.
Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License.