From: Nicolas Boisselier <nicolas.boisselier@gmail.com>
Date: Sat, 5 Dec 2020 09:16:05 +0000 (+0000)
Subject: lib/postgres/html_attribute.sh
X-Git-Url: https://git.nbdom.net/?a=commitdiff_plain;h=b3b1ea0cf8eca2fea3798a89c66df1f35e48cdf8;p=nb.git

lib/postgres/html_attribute.sh
---

diff --git a/lib/postgres/html_attribute.sh b/lib/postgres/html_attribute.sh
index 7ba5b846..564442b8 100755
--- a/lib/postgres/html_attribute.sh
+++ b/lib/postgres/html_attribute.sh
@@ -3,3 +3,7 @@ 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}' \