From: Nicolas Boisselier <nicolas.boisselier@gmail.com>
Date: Sat, 5 Dec 2020 09:39:45 +0000 (+0000)
Subject: html_attribute.sh
X-Git-Url: https://git.nbdom.net/?a=commitdiff_plain;h=4ad4e08f014708e5d93f753d6ff38b94e1cf57f3;p=nb.git

html_attribute.sh
---

diff --git a/lib/postgres/html_attribute.sh b/lib/postgres/html_attribute.sh
index 564442b8..559fccf5 100755
--- a/lib/postgres/html_attribute.sh
+++ b/lib/postgres/html_attribute.sh
@@ -2,8 +2,5 @@
 http_get https://html.com/attributes/ \
 | sed -E "s,<(thrive_headline|header)[^.>]+>([^<]+)</\1>,<div>\2</div>," \
 | xmllint --html --xpath //table - 2>/dev/null \
-| xml2csv - \
 | sed -E \
-  -e "s/”/\"/g" -e "s,^[^<]+(<[^>]+>)[^\t]+,\L\1," \
-  -e "s,^(<([^ ]+) ([^=>]+)),https://html.com/attributes/\2-\3/\t\1," \
-| awk 'BEGIN{FS="\t"}{print $2"\t"$3"\t"$1}' \
+  -e "s,<a[^>]+href=.https://html.com/attributes/([^/]+)[^<]+</a></td>[^<]*<td>,\1,g" \