ma rss: init sofa

2configs/deployment/rss/sofa-urls

@@ -0,0 +1 @@
+https://www.ebay-kleinanzeigen.de/s-ditzingen/preis::50/sofa/k0l8863r10

2configs/deployment/rss/sofa.nix

@@ -0,0 +1,26 @@
+{ pkgs, lib, config, ... }:
+let
+  fqdn = "rss.euer.krebsco.de";
+  ratt-path = "/var/lib/ratt/";
+  out-path = "${ratt-path}/sofa.xml";
+in {
+  systemd.tmpfiles.rules = ["d ${ratt-path} 0750 nginx nginx - -" ];
+  systemd.services.run-ratt-sofa = {
+    enable = true;
+    path = with pkgs; [ ratt xmlstarlet ];
+    script = builtins.readFile ./ratt-hourly.sh;
+    scriptArgs = "${./sofa-urls} ${out-path}";
+
+    preStart = "install -v -m750 ${./sofa.yml} ${ratt-path}/sofa.yml"; # ratt requires the config file in the cwd
+    serviceConfig.User = "nginx";
+    serviceConfig.WorkingDirectory = ratt-path;
+    startAt = "00/3:30"; # every 3 hours, fetch latest
+  };
+
+  services.nginx.virtualHosts."${fqdn}" = {
+    locations."=/ratt/sofa.xml" = {
+      alias = out-path;
+    };
+  };
+}
+

2configs/deployment/rss/sofa.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: ""