#!/bin/bash

URLS="https://vandrouki.ru https://vandrouki.by https://vandrouki.asia"
# URLS="https://vandrouki.by"

rstc="\033[00m"
yellow="\033[1;33m"
blue="\033[1;36m"
green="\033[1;32m"

for url in $URLS; {
	echo -e "$green> $url $rstc\n"

	curl -s -q "$url" \
	| sed 's/<html.*xmlns.*$//g' \
	| xmllint --html --xpath "//html/body/div[@id='content']/div[@id='primary']/div/*[self::div/span[@class='published'] or self::h2/a]" - 2>&- \
	| tr '\n' '\r' \
	| sed -e 's/\r\r/####/g; s/\r/@/g; s/####/\n/g; s/<h2 class="entry-title">/\n/g' \
	| sed 's/href="/@/; s/" rel=[^>]*>/@/; s/published.>/@/' \
	| sed -E 's|^</div>@||; s/&#[0-9]+;//g' \
	| sed '/^\s*$/d' \
	| awk -F'@' -v blue="$blue" -v yellow="$yellow" -v green="$green" -v rstc="$rstc" \
		'{
			print blue " # " $6 "@" yellow toupper($3) "\n@ " green $2 rstc
		}' \
	| sed 's/@@@*/@@/g' \
	| sed -e :a -e 's/<[^>]*>//g;/</N;//ba' \
	| column -t -s @

	echo
}