From: Nicolas Boisselier <nicolas.boisselier@gmail.com>
Date: Sat, 5 Dec 2020 10:11:45 +0000 (+0000)
Subject: html_attribute.sh
X-Git-Url: https://git.nbdom.net/?a=commitdiff_plain;h=65637eb21bf848b58870229ba5def83e574b06eb;p=nb.git

html_attribute.sh
---

diff --git a/lib/postgres/html_attribute.sh b/lib/postgres/html_attribute.sh
index 8b19c49e..0944487e 100755
--- a/lib/postgres/html_attribute.sh
+++ b/lib/postgres/html_attribute.sh
@@ -2,26 +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>,\1,g" \
+| sed -E "s,<a[^>]+href=.(https://html.com/attributes/([^/]+)[^\"]*)[^<]+</a>,\2</td><td>\1,g" \
 | xml2csv - \
-| awk 'BEGIN{FS="\t"}{tag=$1; sub(/^[^-]+-/,"",tag); sub(/\/$/,"",tag); print tag"\t"$2"\t"$1}' \
-
-# NB 05.12.20 | sed -E "s,^([\t]+/([^/]+)-([^-/]+)[^\t]*),\1\t\3," \
-
-# NB 05.12.20 | xml2csv - \
-# NB 05.12.20 | sed -E \
-# NB 05.12.20   -e "s/”/\"/g" -e "s,^[^<]+(<[^>]+>)[^\t]+,\L\1," \
-# NB 05.12.20   -e "s,^(<([^ ]+) ([^=>]+)),https://html.com/attributes/\2-\3/\t\1," \
-# NB 05.12.20 | awk 'BEGIN{FS="\t"}{print $2"\t"$3"\t"$1}' \
-
-
-
-#!/bin/bash -l
-# NB 05.12.20 http_get https://html.com/attributes/ \
-# NB 05.12.20 | sed -E "s,<(thrive_headline|header)[^.>]+>([^<]+)</\1>,<div>\2</div>," \
-# NB 05.12.20 | xmllint --html --xpath //table - 2>/dev/null \
-# NB 05.12.20 | xml2csv - \
-# NB 05.12.20 | sed -E \
-# NB 05.12.20   -e "s/”/\"/g" -e "s,^[^<]+(<[^>]+>)[^\t]+,\L\1," \
-# NB 05.12.20   -e "s,^(<([^ ]+) ([^=>]+)),https://html.com/attributes/\2-\3/\t\1," \
-# NB 05.12.20 | awk 'BEGIN{FS="\t"}{print $2"\t"$3"\t"$1}' \
+| awk 'BEGIN{FS="\t"}{print $1"\t"$3"\t"$2}' \