From: Nicolas Boisselier <nicolas.boisselier@gmail.com>
Date: Sun, 14 May 2023 23:16:10 +0000 (+0200)
Subject: Fix
X-Git-Url: https://git.nbdom.net/?a=commitdiff_plain;h=f9666221239873b316a60a348f0a02b331045d2d;p=nb.git

Fix
---

diff --git a/lib/postgres/html_attribute.sh b/lib/postgres/html_attribute.sh
index 0944487e..fadb700e 100755
--- a/lib/postgres/html_attribute.sh
+++ b/lib/postgres/html_attribute.sh
@@ -2,6 +2,6 @@
 http_get https://html.com/attributes/ \
 | sed -E "s,<(thrive_headline|header)[^.>]+>([^<]+)</\1>,<div>\2</div>," \
 | xmllint --html --xpath //table - 2>/dev/null \
-| sed -E "s,<a[^>]+href=.(https://html.com/attributes/([^/]+)[^\"]*)[^<]+</a>,\2</td><td>\1,g" \
+| sed -E "s,<a[^>]+href=\"((https://html.com)?/attributes/([^\"\/]+)\/?)[^<]+</a>,\3</td><td>https://html.com/\1,g" \
 | xml2csv - \
 | awk 'BEGIN{FS="\t"}{print $1"\t"$3"\t"$2}' \