diff options
| author | rec <44084068+recloser@users.noreply.github.com> | 2018-10-29 11:10:00 +0100 |
|---|---|---|
| committer | Andreas Rumpf <rumpf_a@web.de> | 2018-10-29 11:10:00 +0100 |
| commit | 9fd0a71e4d4c6b3d451a137372973dcb4f4b47cc (patch) | |
| tree | e60a2b3c247973c47c2446ce3d0b42df64f9d997 /lib | |
| parent | 680f5eeb15583f45dc2dcdbc4f05a6557d09f31a (diff) | |
| download | Nim-9fd0a71e4d4c6b3d451a137372973dcb4f4b47cc.tar.gz | |
Make htmlparser parse unquoted attrib values (#9537)
Fixes #6154
Diffstat (limited to 'lib')
| -rw-r--r-- | lib/pure/htmlparser.nim | 2 | ||||
| -rw-r--r-- | lib/pure/parsexml.nim | 20 |
2 files changed, 20 insertions, 2 deletions
diff --git a/lib/pure/htmlparser.nim b/lib/pure/htmlparser.nim index fbf2b8e73..9e1a5a101 100644 --- a/lib/pure/htmlparser.nim +++ b/lib/pure/htmlparser.nim @@ -2014,7 +2014,7 @@ proc parseHtml*(s: Stream, filename: string, ## Parses the XML from stream `s` and returns a ``XmlNode``. Every ## occurred parsing error is added to the `errors` sequence. var x: XmlParser - open(x, s, filename, {reportComments, reportWhitespace}) + open(x, s, filename, {reportComments, reportWhitespace, allowUnquotedAttribs}) next(x) # skip the DOCTYPE: if x.kind == xmlSpecial: next(x) diff --git a/lib/pure/parsexml.nim b/lib/pure/parsexml.nim index d8d5a7a2d..39b117d40 100644 --- a/lib/pure/parsexml.nim +++ b/lib/pure/parsexml.nim @@ -180,6 +180,7 @@ type errEqExpected, ## ``=`` expected errQuoteExpected, ## ``"`` or ``'`` expected errEndOfCommentExpected ## ``-->`` expected + errAttributeValueExpected ## non-empty attribute value expected ParserState = enum stateStart, stateNormal, stateAttr, stateEmptyElementTag, stateError @@ -187,6 +188,7 @@ type XmlParseOption* = enum ## options for the XML parser reportWhitespace, ## report whitespace reportComments ## report comments + allowUnquotedAttribs ## allow unquoted attribute values (for HTML) XmlParser* = object of BaseLexer ## the parser object. a, b, c: string @@ -207,7 +209,8 @@ const "'>' expected", "'=' expected", "'\"' or \"'\" expected", - "'-->' expected" + "'-->' expected", + "attribute value expected" ] proc open*(my: var XmlParser, input: Stream, filename: string, @@ -669,6 +672,21 @@ proc parseAttribute(my: var XmlParser) = pendingSpace = false add(my.b, buf[pos]) inc(pos) + elif allowUnquotedAttribs in my.options: + const disallowedChars = {'"', '\'', '`', '=', '<', '>', ' ', + '\0', '\t', '\L', '\F', '\f'} + let startPos = pos + while (let c = buf[pos]; c notin disallowedChars): + if c == '&': + my.bufpos = pos + parseEntity(my, my.b) + my.kind = xmlAttribute # parseEntity overwrites my.kind! + pos = my.bufpos + else: + add(my.b, c) + inc(pos) + if pos == startPos: + markError(my, errAttributeValueExpected) else: markError(my, errQuoteExpected) # error corrections: guess what was meant |