From 5187d0ac208deb06eff3bafb7ffd2fc32286b46a Mon Sep 17 00:00:00 2001 From: makefu Date: Mon, 2 May 2022 22:25:19 +0200 Subject: ma rss: deploy ratt job --- makefu/2configs/deployment/rss.euer.krebsco.de.nix | 24 --------- makefu/2configs/deployment/rss/ebk.yml | 59 ++++++++++++++++++++++ makefu/2configs/deployment/rss/ratt-hourly.sh | 28 ++++++++++ makefu/2configs/deployment/rss/ratt.nix | 26 ++++++++++ .../deployment/rss/rss.euer.krebsco.de.nix | 30 +++++++++++ makefu/2configs/deployment/rss/urls | 5 ++ 6 files changed, 148 insertions(+), 24 deletions(-) delete mode 100644 makefu/2configs/deployment/rss.euer.krebsco.de.nix create mode 100644 makefu/2configs/deployment/rss/ebk.yml create mode 100755 makefu/2configs/deployment/rss/ratt-hourly.sh create mode 100644 makefu/2configs/deployment/rss/ratt.nix create mode 100644 makefu/2configs/deployment/rss/rss.euer.krebsco.de.nix create mode 100644 makefu/2configs/deployment/rss/urls (limited to 'makefu/2configs') diff --git a/makefu/2configs/deployment/rss.euer.krebsco.de.nix b/makefu/2configs/deployment/rss.euer.krebsco.de.nix deleted file mode 100644 index 19f20f50f..000000000 --- a/makefu/2configs/deployment/rss.euer.krebsco.de.nix +++ /dev/null @@ -1,24 +0,0 @@ -{ pkgs, lib, config, ... }: -let - fqdn = "rss.euer.krebsco.de"; -in { - services.tt-rss = { - enable = true; - virtualHost = fqdn; - selfUrlPath = "https://${fqdn}"; - }; - - services.postgresql.package = pkgs.postgresql_9_6; - state = [ config.services.postgresqlBackup.location ]; - - services.postgresqlBackup = { - enable = true; - databases = [ config.services.tt-rss.database.name ]; - }; - - services.nginx.virtualHosts."${fqdn}" = { - enableACME = true; - forceSSL = true; - }; -} - diff --git a/makefu/2configs/deployment/rss/ebk.yml b/makefu/2configs/deployment/rss/ebk.yml new file mode 100644 index 000000000..3248f5c4e --- /dev/null +++ b/makefu/2configs/deployment/rss/ebk.yml @@ -0,0 +1,59 @@ +regex: https://www.ebay\-kleinanzeigen.de/s\-.* +selectors: + httpsettings: + cookie: {} + header: {} + useragent: Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36 (KHTML, like Gecko) + Chrome/90.0.4430.72 Safari/537.36 + insecure: false + feed: + title: title + authorname: "" + authoremail: "" + item: + container: ul[id='srchrslt-adtable'] li[class='ad-listitem lazyload-item '] + title: | + title = sel:find("h2.text-module-begin"):first():text():gsub("^%s*(.-)%s*$", "%1") + print(title) + link: | + link = sel:find("a"):first():attr("href") + print("https://www.ebay-kleinanzeigen.de" .. link) + created: |- + created = "" + sel:find("div.aditem-main--top--right"):each(function(i, s) + created = s:text():gsub("^%s*(.-)%s*$", "%1") + end) + if created:match("Heute") then + time = created:gsub("^.*,", "") + print(os.date("%d.%m.%Y") .. time .. " CET") + return + end + if created:match("Gestern") then + time = created:gsub("^.*,", "") + print(os.date("%d.%m.%Y", os.time()-24*60*60) .. time .. " CET") + return + end + if created:match("\.") then + print(created .. " 00:00 CET") + return + end + createdformat: 02.01.2006 15:04 MST + description: |- + description = sel:find(".aditem-main--middle"):html() + place = sel:find(".aditem-main--top--left"):html() + print(description .. place) + content: "" + image: | + img = sel:find("div.imagebox"):first():attr("data-imgsrc") + if img ~= "" then + -- prepend host if needed + if not(img:match("https*:\/\/.*")) then + img = "https://www.ebay-kleinanzeigen.de" .. img + end + print(img) + end + nextpage: | + nextpage = sel:find("link[rel=next]"):attr("href") + print("https://www.ebay-kleinanzeigen.de" .. nextpage) + nextpagecount: 5 + sort: "" diff --git a/makefu/2configs/deployment/rss/ratt-hourly.sh b/makefu/2configs/deployment/rss/ratt-hourly.sh new file mode 100755 index 000000000..67f2529bd --- /dev/null +++ b/makefu/2configs/deployment/rss/ratt-hourly.sh @@ -0,0 +1,28 @@ +#!/bin/sh +set -eu +URLS=${1?must provide URLS file} +OUTFILE=${2:-all.xml} + +echo "init, writing to $OUTFILE" + +cat > "$OUTFILE" < + + + makefu Ebay Kleinanzeigen + https://www.ebay-kleinanzeigen.de/ + Feed for all kleinanzeigen + $(date '+%a, %d %b %Y %H:%M:%S %z') +EOF +echo "looping through $URLS" +cat "$URLS" | while read line;do + echo "fetching $line" + ratt auto "$line" | \ + xmlstarlet sel -t -c "//item" >> "$OUTFILE" || : +done + +echo "close" +cat >> "$OUTFILE" < + +EOF diff --git a/makefu/2configs/deployment/rss/ratt.nix b/makefu/2configs/deployment/rss/ratt.nix new file mode 100644 index 000000000..b794d9201 --- /dev/null +++ b/makefu/2configs/deployment/rss/ratt.nix @@ -0,0 +1,26 @@ +{ pkgs, lib, config, ... }: +let + fqdn = "rss.euer.krebsco.de"; + ratt-path = "/var/lib/ratt/"; + out-path = "${ratt-path}/all.xml"; +in { + systemd.tmpfiles.rules = ["d ${ratt-path} 0750 nginx nginx - -" ]; + systemd.services.run-ratt = { + enable = true; + path = with pkgs; [ "/nix/store/vhmzblnaav2lp4lwqdgm13l55qlm79mk-ratt-unstable-2022-01-11" xmlstarlet ]; + script = builtins.readFile ./ratt-hourly.sh; + scriptArgs = "${./urls} ${out-path}"; + + preStart = "install -v -m750 ${./ebk.yml} ${ratt-path}/ebk.yml"; # ratt requires the config file in the cwd + serviceConfig.User = "nginx"; + serviceConfig.WorkingDirectory= ratt-path; + startAt = "00/3:07"; # every 3 hours, fetch latest + }; + + services.nginx.virtualHosts."${fqdn}" = { + locations."=/ratt/all.xml" = { + alias = out-path; + }; + }; +} + diff --git a/makefu/2configs/deployment/rss/rss.euer.krebsco.de.nix b/makefu/2configs/deployment/rss/rss.euer.krebsco.de.nix new file mode 100644 index 000000000..e64a69d9c --- /dev/null +++ b/makefu/2configs/deployment/rss/rss.euer.krebsco.de.nix @@ -0,0 +1,30 @@ +{ pkgs, lib, config, ... }: +let + fqdn = "rss.euer.krebsco.de"; + ratt-path = "/var/lib/ratt/"; +in { + systemd.tmpfiles.rules = ["d ${ratt-path} 0750 nginx nginx - -" ]; + services.tt-rss = { + enable = true; + virtualHost = fqdn; + selfUrlPath = "https://${fqdn}"; + }; + + services.postgresql.package = pkgs.postgresql_9_6; + state = [ config.services.postgresqlBackup.location ]; + + services.postgresqlBackup = { + enable = true; + databases = [ config.services.tt-rss.database.name ]; + }; + + services.nginx.virtualHosts."${fqdn}" = { + enableACME = true; + forceSSL = true; + locations."/ratt/" = { + alias = ratt-path; + extraConfig = "autoindex on;"; + }; + }; +} + diff --git a/makefu/2configs/deployment/rss/urls b/makefu/2configs/deployment/rss/urls new file mode 100644 index 000000000..12d4c092a --- /dev/null +++ b/makefu/2configs/deployment/rss/urls @@ -0,0 +1,5 @@ +https://www.ebay-kleinanzeigen.de/s-muehlhausen/preis:0:45/duplo-eisenbahn/k0l9313r5 +https://www.ebay-kleinanzeigen.de/s-heimwerken/nein/muehlhausen/bohrmaschine/k0c84l9313r5+heimwerken.versand_s:nein +https://www.ebay-kleinanzeigen.de/s-stuttgart/zigbee/k0l9280 +https://www.ebay-kleinanzeigen.de/s-stuttgart/ikea-tradfri-fernbedienung/k0l9280 +https://www.ebay-kleinanzeigen.de/s-70378/d%C3%B6rrautomat/k0l9334r5 -- cgit v1.2.3