wader/fq
1
1
Fork 0
mirror of https://github.com/wader/fq.git synced 2024-11-28 03:02:55 +03:00
Code Issues Projects Releases Wiki Activity
cae288e6be
fq/format/xml/testdata/html.fqtest
Mattias Wadman cae288e6be format,intepr: Refactor json, yaml, etc into formats also move out related functions 
json, yaml, toml, xml, html, csv are now normal formats and most of them also particiate
in probing (not html and csv).

Also fixes a bunch of bugs in to/fromxml, to/fromjq etc.
2022-07-23 21:48:45 +02:00

651 lines
9.8 KiB
Plaintext
Raw Blame History

/test:
test
$ fq -d html . /test
{
"html": {
"body": "test",
"head": ""
}
}
$ fq -d raw -ni . all.xml multi_diff.xml multi_same.xml ns.xml simple.xml escape.xml noscript.html
null> inputs | {name: input_filename, str: (tobytes | tostring)} | slurp("files")
null> spew("files") | .name, (.str | fromhtml | ., (toxml({indent: 2}) | println))
"all.xml"
{
"html": {
"body": {
"elm": {
"first": {
"#comment": "comment"
},
"last": {
"#comment": "comment1 comment2",
"#text": "text1\n \n text2",
"-attr1": "v1",
"-attr2": "v2"
},
"middle": "text"
}
},
"head": ""
}
}
<html>
<body>
<elm>
<first>
<!--comment--></first>
<last attr1="v1" attr2="v2">text1&#xA; &#xA; text2
<!--comment1 comment2--></last>
<middle>text</middle>
</elm>
</body>
<head></head>
</html>
"multi_diff.xml"
{
"html": {
"body": {
"elm1": "",
"elm2": ""
},
"head": ""
}
}
<html>
<body>
<elm1></elm1>
<elm2></elm2>
</body>
<head></head>
</html>
"multi_same.xml"
{
"html": {
"body": {
"elm": [
"",
""
]
},
"head": ""
}
}
<html>
<body>
<elm></elm>
<elm></elm>
</body>
<head></head>
</html>
"ns.xml"
{
"html": {
"body": {
"elm": {
"-xmlns:ns1": "http://test1",
"-xmlns:ns2": "http://test2",
"aaa": "3",
"ns1:aaa": {
"#text": "1",
"-ns1:attr1": "v1"
},
"ns2:aaa": {
"#text": "2",
"-ns2:attr2": "v2",
"ns1:ccc": {
"-ns1:attr3": "v3"
},
"ns2:ccc": {
"-ns2:attr4": "v4"
},
"ns3:ccc": {
"-ns2:attr5": "v5"
}
}
}
},
"head": ""
}
}
<html>
<body>
<elm xmlns:ns1="http://test1" xmlns:ns2="http://test2">
<aaa>3</aaa>
<ns1:aaa ns1:attr1="v1">1</ns1:aaa>
<ns2:aaa ns2:attr2="v2">2
<ns1:ccc ns1:attr3="v3"></ns1:ccc>
<ns2:ccc ns2:attr4="v4"></ns2:ccc>
<ns3:ccc ns2:attr5="v5"></ns3:ccc>
</ns2:aaa>
</elm>
</body>
<head></head>
</html>
"simple.xml"
{
"html": {
"body": {
"elm": ""
},
"head": ""
}
}
<html>
<body>
<elm></elm>
</body>
<head></head>
</html>
"escape.xml"
{
"html": {
"body": {
"a": {
"#text": "&<>",
"-attr": "&<>"
}
},
"head": ""
}
}
<html>
<body>
<a attr="&amp;&lt;&gt;">&amp;&lt;&gt;</a>
</body>
<head></head>
</html>
"noscript.html"
{
"html": {
"body": {
"a": "text"
},
"head": {
"noscript": ""
}
}
}
<html>
<body>
<a>text</a>
</body>
<head>
<noscript></noscript>
</head>
</html>
null> spew("files") | .name, (.str | fromhtml({seq: true}) | ., (toxml({indent: 2}) | println))
"all.xml"
{
"html": {
"body": {
"#seq": 1,
"elm": {
"first": {
"#comment": "comment",
"#seq": 0
},
"last": {
"#comment": "comment1 comment2",
"#seq": 2,
"#text": "text1\n \n text2",
"-attr1": "v1",
"-attr2": "v2"
},
"middle": {
"#seq": 1,
"#text": "text"
}
}
},
"head": {
"#seq": 0
}
}
}
<html>
<head></head>
<body>
<elm>
<first>
<!--comment--></first>
<middle>text</middle>
<last attr1="v1" attr2="v2">text1&#xA; &#xA; text2
<!--comment1 comment2--></last>
</elm>
</body>
</html>
"multi_diff.xml"
{
"html": {
"body": {
"#seq": 1,
"elm1": {
"#seq": 0
},
"elm2": {
"#seq": 1
}
},
"head": {
"#seq": 0
}
}
}
<html>
<head></head>
<body>
<elm1></elm1>
<elm2></elm2>
</body>
</html>
"multi_same.xml"
{
"html": {
"body": {
"#seq": 1,
"elm": [
{
"#seq": 0
},
{
"#seq": 1
}
]
},
"head": {
"#seq": 0
}
}
}
<html>
<head></head>
<body>
<elm></elm>
<elm></elm>
</body>
</html>
"ns.xml"
{
"html": {
"body": {
"#seq": 1,
"elm": {
"-xmlns:ns1": "http://test1",
"-xmlns:ns2": "http://test2",
"aaa": {
"#seq": 2,
"#text": "3"
},
"ns1:aaa": {
"#seq": 0,
"#text": "1",
"-ns1:attr1": "v1"
},
"ns2:aaa": {
"#seq": 1,
"#text": "2",
"-ns2:attr2": "v2",
"ns1:ccc": {
"#seq": 0,
"-ns1:attr3": "v3"
},
"ns2:ccc": {
"#seq": 1,
"-ns2:attr4": "v4"
},
"ns3:ccc": {
"#seq": 2,
"-ns2:attr5": "v5"
}
}
}
},
"head": {
"#seq": 0
}
}
}
<html>
<head></head>
<body>
<elm xmlns:ns1="http://test1" xmlns:ns2="http://test2">
<ns1:aaa ns1:attr1="v1">1</ns1:aaa>
<ns2:aaa ns2:attr2="v2">2
<ns1:ccc ns1:attr3="v3"></ns1:ccc>
<ns2:ccc ns2:attr4="v4"></ns2:ccc>
<ns3:ccc ns2:attr5="v5"></ns3:ccc>
</ns2:aaa>
<aaa>3</aaa>
</elm>
</body>
</html>
"simple.xml"
{
"html": {
"body": {
"#seq": 1,
"elm": ""
},
"head": {
"#seq": 0
}
}
}
<html>
<head></head>
<body>
<elm></elm>
</body>
</html>
"escape.xml"
{
"html": {
"body": {
"#seq": 1,
"a": {
"#text": "&<>",
"-attr": "&<>"
}
},
"head": {
"#seq": 0
}
}
}
<html>
<head></head>
<body>
<a attr="&amp;&lt;&gt;">&amp;&lt;&gt;</a>
</body>
</html>
"noscript.html"
{
"html": {
"body": {
"#seq": 1,
"a": "text"
},
"head": {
"#seq": 0,
"noscript": ""
}
}
}
<html>
<head>
<noscript></noscript>
</head>
<body>
<a>text</a>
</body>
</html>
null> spew("files") | .name, (.str | fromhtml({array: true}) | ., (toxml({indent: 2}) | println))
"all.xml"
[
"html",
[
[
"head"
],
[
"body",
[
[
"elm",
[
[
"first",
{
"#comment": "comment"
}
],
[
"middle",
{
"#text": "text"
}
],
[
"last",
{
"#comment": "comment1 comment2",
"#text": "text1\n \n text2",
"attr1": "v1",
"attr2": "v2"
}
]
]
]
]
]
]
]
<html>
<head></head>
<body>
<elm>
<first>
<!--comment--></first>
<middle>text</middle>
<last attr1="v1" attr2="v2">text1&#xA; &#xA; text2
<!--comment1 comment2--></last>
</elm>
</body>
</html>
"multi_diff.xml"
[
"html",
[
[
"head"
],
[
"body",
[
[
"elm1"
],
[
"elm2"
]
]
]
]
]
<html>
<head></head>
<body>
<elm1></elm1>
<elm2></elm2>
</body>
</html>
"multi_same.xml"
[
"html",
[
[
"head"
],
[
"body",
[
[
"elm"
],
[
"elm"
]
]
]
]
]
<html>
<head></head>
<body>
<elm></elm>
<elm></elm>
</body>
</html>
"ns.xml"
[
"html",
[
[
"head"
],
[
"body",
[
[
"elm",
{
"xmlns:ns1": "http://test1",
"xmlns:ns2": "http://test2"
},
[
[
"ns1:aaa",
{
"#text": "1",
"ns1:attr1": "v1"
}
],
[
"ns2:aaa",
{
"#text": "2",
"ns2:attr2": "v2"
},
[
[
"ns1:ccc",
{
"ns1:attr3": "v3"
}
],
[
"ns2:ccc",
{
"ns2:attr4": "v4"
}
],
[
"ns3:ccc",
{
"ns2:attr5": "v5"
}
]
]
],
[
"aaa",
{
"#text": "3"
}
]
]
]
]
]
]
]
<html>
<head></head>
<body>
<elm xmlns:ns1="http://test1" xmlns:ns2="http://test2">
<ns1:aaa ns1:attr1="v1">1</ns1:aaa>
<ns2:aaa ns2:attr2="v2">2
<ns1:ccc ns1:attr3="v3"></ns1:ccc>
<ns2:ccc ns2:attr4="v4"></ns2:ccc>
<ns3:ccc ns2:attr5="v5"></ns3:ccc>
</ns2:aaa>
<aaa>3</aaa>
</elm>
</body>
</html>
"simple.xml"
[
"html",
[
[
"head"
],
[
"body",
[
[
"elm"
]
]
]
]
]
<html>
<head></head>
<body>
<elm></elm>
</body>
</html>
"escape.xml"
[
"html",
[
[
"head"
],
[
"body",
[
[
"a",
{
"#text": "&<>",
"attr": "&<>"
}
]
]
]
]
]
<html>
<head></head>
<body>
<a attr="&amp;&lt;&gt;">&amp;&lt;&gt;</a>
</body>
</html>
"noscript.html"
[
"html",
[
[
"head",
[
[
"noscript"
]
]
],
[
"body",
[
[
"a",
{
"#text": "text"
}
]
]
]
]
]
<html>
<head>
<noscript></noscript>
</head>
<body>
<a>text</a>
</body>
</html>
null> ^D
Reference in New Issue View Git Blame Copy Permalink