diff options
| -rw-r--r-- | lib/pure/parsexml.nim | 9 | ||||
| -rw-r--r-- | tests/stdlib/thtmlparser.nim | 20 |
2 files changed, 26 insertions, 3 deletions
diff --git a/lib/pure/parsexml.nim b/lib/pure/parsexml.nim index 3b77f9c62..686aad110 100644 --- a/lib/pure/parsexml.nim +++ b/lib/pure/parsexml.nim @@ -472,16 +472,19 @@ proc parseEntity(my: var XmlParser, dest: var string) = inc(pos, 4) else: my.bufpos = pos - parseName(my, dest) + var name = "" + parseName(my, name) pos = my.bufpos - if my.err != errNameExpected: + if my.err != errNameExpected and my.buf[pos] == ';': my.kind = xmlEntity else: add(dest, '&') + add(dest, name) if my.buf[pos] == ';': inc(pos) else: - markError(my, errSemicolonExpected) + my.err = errSemicolonExpected + # do not overwrite 'my.state' here, it's a benign error my.bufpos = pos proc parsePI(my: var XmlParser) = diff --git a/tests/stdlib/thtmlparser.nim b/tests/stdlib/thtmlparser.nim index 0457585d0..ccf2f6202 100644 --- a/tests/stdlib/thtmlparser.nim +++ b/tests/stdlib/thtmlparser.nim @@ -2,6 +2,9 @@ discard """ output: ''' @[] true +https://example.com/test?format=jpg&name=orig## +https://example.com/test?format=jpg&name=orig##text +https://example.com/test?format=jpg##text ''' """ import htmlparser @@ -136,3 +139,20 @@ block t6154: doAssert ps[6].attr("quux") == "" doAssert ps[6].attr("whatever") == "" doassert ps[6].len == 0 + +# bug #11713, #1034 +var content = """ +# with & +<img src="https://example.com/test?format=jpg&name=orig" alt=""> +<img src="https://example.com/test?format=jpg&name=orig" alt="text"> + +# without & +<img src="https://example.com/test?format=jpg" alt="text"> +""" + +var + stream = newStringStream(content) + body = parseHtml(stream) + +for y in body.findAll("img"): + echo y.attr("src"), "##", y.attr("alt") |