vandoucli: awk

vandroucli

@@ -1,19 +1,46 @@
 #!/bin/bash
 
 #Создан:  Чт 05 апр 2018 10:26:40
-#Изменён: Чт 05 апр 2018 10:41:39
+#Изменён: Вс 22 сен 2019 08:29:53
 
 # Получение последних записей с сайта vandrouki.ru
 
-URL="https://vandrouki.ru"
+URLS="https://vandrouki.asia https://vandrouki.by https://vandrouki.ru"
 
-curl -s -q "$URL" \
+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]" - \
-	| tr '\n' '\r' \
-	| sed -e 's/\r\r/####/g; s/\r/@/g; s/####/\n/g' \
-	| sed 's/href="/@/; s/" rel=[^>]*>/@/; s/published.>/@/' \
-	| awk -F'@' '{print $5 "@" $3 "@" $2}' \
-	| sed -e :a -e 's/<[^>]*>//g;/</N;//ba' \
-	| column -t -s @
+	| xmllint --html --xpath "//html/body/div[@id='content']/div[@id='primary']/div/*[self::div/span[@class='published'] or self::h2/a]" - 2>&- \
+	| awk 'BEGIN { RS="" }
+	{
+		gsub(/\n+/, "@");
+		gsub(/<h2 class="entry-title">/, "\n");
+		gsub(/(href=")|(" rel=[^>]*>)|(published.>)/, "@");
+		gsub(/(&#[0-9]+;)|(<[^@>]*>*)/, "");
+		gsub(/@\s*@*/,"@");
+
+		print
+	}' \
+	| awk -F'@' -v blue="$blue" -v yellow="$yellow" -v green="$green" -v rstc="$rstc" '
+		/^\s*$/ {
+			next
+		} {
+			if ($4 ~ /^\s*$/) {
+				date = $5
+			} else {
+				date = $4
+			}
+
+			print blue " # " date "\t" yellow toupper($3) "\n\t\t" green $2 rstc
+		}'
+
+	echo -e "\n--------------------------------\n"
+}
 

vandroucli.sed

@@ -0,0 +1,32 @@
+#!/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
+}
+