arno/pup
1
0
Fork 0
mirror of https://github.com/ericchiang/pup synced 2025-01-15 10:11:16 +00:00
Code Issues Releases Wiki Activity

Move node attributes under their own key in JSON displayer

Browse Source
This commit is contained in:
Andre Marianiello 2019-02-23 14:49:57 -05:00
parent 14e452d641
commit ae0b54d785
3 changed files with 32 additions and 20 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

26
README.md
View File

@ -281,18 +281,23 @@ $ cat robots.html | pup 'div#p-namespaces a'
$ cat robots.html | pup 'div#p-namespaces a json{}' $ cat robots.html | pup 'div#p-namespaces a json{}'
[ [
{ {
"attrs": {
"accesskey": "c", "accesskey": "c",
"href": "/wiki/Robots_exclusion_standard", "href": "/wiki/Robots_exclusion_standard",
"tag": "a",
"text": "Article",
"title": "View the content page [c]" "title": "View the content page [c]"
}, },
"tag": "a",
"text": "Article"
},
{ {
"attrs": {
"accesskey": "t", "accesskey": "t",
"href": "/wiki/Talk:Robots_exclusion_standard", "href": "/wiki/Talk:Robots_exclusion_standard",
"tag": "a", "rel": "discussion",
"text": "Talk",
"title": "Discussion about the content page [t]" "title": "Discussion about the content page [t]"
},
"tag": "a",
"text": "Talk"
} }
] ]
``` ```
@ -303,18 +308,23 @@ Use the `-i` / `--indent` flag to control the intent level.
$ cat robots.html | pup -i 4 'div#p-namespaces a json{}' $ cat robots.html | pup -i 4 'div#p-namespaces a json{}'
[ [
{ {
"attrs": {
"accesskey": "c", "accesskey": "c",
"href": "/wiki/Robots_exclusion_standard", "href": "/wiki/Robots_exclusion_standard",
"tag": "a",
"text": "Article",
"title": "View the content page [c]" "title": "View the content page [c]"
}, },
"tag": "a",
"text": "Article"
},
{ {
"attrs": {
"accesskey": "t", "accesskey": "t",
"href": "/wiki/Talk:Robots_exclusion_standard", "href": "/wiki/Talk:Robots_exclusion_standard",
"tag": "a", "rel": "discussion",
"text": "Talk",
"title": "Discussion about the content page [t]" "title": "Discussion about the content page [t]"
},
"tag": "a",
"text": "Talk"
} }
] ]
``` ```

6
display.go
View File

@ -264,13 +264,15 @@ type JSONDisplayer struct{}
func jsonify(node *html.Node) map[string]interface{} { func jsonify(node *html.Node) map[string]interface{} {
vals := map[string]interface{}{} vals := map[string]interface{}{}
if len(node.Attr) > 0 { if len(node.Attr) > 0 {
attrs := map[string]interface{}{}
for _, attr := range node.Attr { for _, attr := range node.Attr {
if pupEscapeHTML { if pupEscapeHTML {
vals[attr.Key] = html.EscapeString(attr.Val) attrs[attr.Key] = html.EscapeString(attr.Val)
} else { } else {
vals[attr.Key] = attr.Val attrs[attr.Key] = attr.Val
} }
} }
vals["attrs"] = attrs
} }
vals["tag"] = node.DataAtom.String() vals["tag"] = node.DataAtom.String()
children := []interface{}{} children := []interface{}{}

4
tests/expected_output.txt
View File

@ -10,7 +10,7 @@ a92e50c09cd56970625ac3b74efbddb83b2731bb table li
66950e746590d7f4e9cfe3d1adef42cd0addcf1d table li:last-of-type 66950e746590d7f4e9cfe3d1adef42cd0addcf1d table li:last-of-type
0a37d612cd4c67a42bd147b1edc5a1128456b017 table a[title="The Practice of Programming"] 0a37d612cd4c67a42bd147b1edc5a1128456b017 table a[title="The Practice of Programming"]
0d3918d54f868f13110262ffbb88cbb0b083057d table a[title="The Practice of Programming"] text{} 0d3918d54f868f13110262ffbb88cbb0b083057d table a[title="The Practice of Programming"] text{}
ecb542a30fc75c71a0c6380692cbbc4266ccbce4 json{} 199188dc8f1522426a628e41d96264bffb8beb0f json{}
95ef88ded9dab22ee3206cca47b9c3a376274bda text{} 95ef88ded9dab22ee3206cca47b9c3a376274bda text{}
e4f7358fbb7bb1748a296fa2a7e815fa7de0a08b .after-portlet e4f7358fbb7bb1748a296fa2a7e815fa7de0a08b .after-portlet
da39a3ee5e6b4b0d3255bfef95601890afd80709 .after da39a3ee5e6b4b0d3255bfef95601890afd80709 .after
@ -34,7 +34,7 @@ d314e83b059bb876b0e5ee76aa92d54987961f9a .navbox-list li:nth-last-child(1)
613bf65ac4042b6ee0a7a47f08732fdbe1b5b06b #toc 613bf65ac4042b6ee0a7a47f08732fdbe1b5b06b #toc
da39a3ee5e6b4b0d3255bfef95601890afd80709 #toc li + a da39a3ee5e6b4b0d3255bfef95601890afd80709 #toc li + a
da39a3ee5e6b4b0d3255bfef95601890afd80709 #toc li + a text{} da39a3ee5e6b4b0d3255bfef95601890afd80709 #toc li + a text{}
97d170e1550eee4afc0af065b78cda302a97674c #toc li + a json{} cd0d4cc32346750408f7d4f5e78ec9a6e5b79a0d #toc li + a json{}
da39a3ee5e6b4b0d3255bfef95601890afd80709 #toc li + a + span da39a3ee5e6b4b0d3255bfef95601890afd80709 #toc li + a + span
da39a3ee5e6b4b0d3255bfef95601890afd80709 #toc li + span da39a3ee5e6b4b0d3255bfef95601890afd80709 #toc li + span
da39a3ee5e6b4b0d3255bfef95601890afd80709 #toc li > li da39a3ee5e6b4b0d3255bfef95601890afd80709 #toc li > li