ma rss: deploy ratt job

5 files changed, 124 insertions, 0 deletions

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" <<EOF
+<?xml version="1.0" encoding="UTF-8"?>
+<rss version="2.0" xmlns:content="http://purl.org/rss/1.0/modules/content/">
+  <channel>
+    <title>makefu Ebay Kleinanzeigen</title>
+    <link>https://www.ebay-kleinanzeigen.de/</link>
+    <description>Feed for all kleinanzeigen</description>
+    <pubDate>$(date '+%a, %d %b %Y %H:%M:%S %z')</pubDate>
+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
+  </channel>
+</rss>
+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.euer.krebsco.de.nix b/makefu/2configs/deployment/rss/rss.euer.krebsco.de.nix
index 19f20f50f..e64a69d9c 100644
--- a/makefu/2configs/deployment/rss.euer.krebsco.de.nix
+++ b/makefu/2configs/deployment/rss/rss.euer.krebsco.de.nix
@@ -1,7 +1,9 @@
 { pkgs, lib, config, ... }:
 let
   fqdn = "rss.euer.krebsco.de";
+  ratt-path = "/var/lib/ratt/";
 in {
+  systemd.tmpfiles.rules = ["d ${ratt-path} 0750 nginx nginx - -" ];
   services.tt-rss = {
     enable = true;
     virtualHost = fqdn;
@@ -19,6 +21,10 @@ in {
   services.nginx.virtualHosts."${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