M minpkg/lib/min_xml.nim → minpkg/lib/min_xml.nim

@@ -3,6 +3,7 @@ import
     std/[xmlparser,
     xmltree,
     parsexml,
+    htmlparser,
     strtabs,
     critbits]
 import
@@ -87,6 +88,16 @@ let s = vals[0].getString()
         try:
             let xml = parseXml(s, {reportComments, allowUnquotedAttribs,
                     allowEmptyAttribs})
+            i.push(i.newXDict(xml))
+        except CatchableError:
+            let msg = getCurrentExceptionMsg()
+            raiseInvalid(msg)
+
+    def.symbol("from-html") do (i: In):
+        let vals = i.expect("str")
+        let s = vals[0].getString()
+        try:
+            let xml = parseHtml(s)
             i.push(i.newXDict(xml))
         except CatchableError:
             let msg = getCurrentExceptionMsg()

M next-release.md → next-release.md

@@ -1,4 +1,5 @@ 
 ### New Features
 
 - Added a new `color` symbol to the `io` module to enable/disable terminal color output.
+- Added a new `from-html` symbol to the `xml` module to parse HTML documents and fragments.
 

M site/contents/reference-xml.md → site/contents/reference-xml.md

@@ -4,8 +4,11 @@ title: "xml Module"
 -----
 {@ _defs_.md || 0 @}
 
+{#op||from-html||{{sl}}||{{xnode}}||
+Returns an {{xnode}} representing an HTML string (wrapped in a `<document>` tag unless a valid HTML document is provided as input).#}
+
 {#op||from-xml||{{sl}}||{{xnode}}||
-> Returns an {{xnode}} representing an XML string representing an element or fragment.
+> Returns an {{xnode}} representing an XML string (element or fragment).
 > 
 > > %sidebar%
 > > Example

M tests/xml.min → tests/xml.min

@@ -35,5 +35,10 @@ xnode "li.test" xqueryall
     size 2 ==
    ) *test/assert
 
+   (
+    "<b>this is a test</b><img src='test.jpg'><i>&copy</i>" from-html :xnode
+    xnode to-xml "<document>\n  <b>this is a test</b>\n  <img src=\"test.jpg\" />\n  <i>&amp;copy</i>\n</document>" ==
+   ) *test/assert
+
   *test/report
   clear-stack