From 5c87fd6cd4663923616b42ba86f49ec8ae669bbc Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Kier=C3=A1n=20Meinhardt?= <kmein@posteo.de>
Date: Mon, 16 Dec 2024 17:17:43 +0100
Subject: [PATCH] celan: use tei edition

---
 configs/bots/celan.nix | 72 ++++++++++++++++++++++++++----------------
 1 file changed, 44 insertions(+), 28 deletions(-)

diff --git a/configs/bots/celan.nix b/configs/bots/celan.nix
index 54596b8..2af5d27 100644
--- a/configs/bots/celan.nix
+++ b/configs/bots/celan.nix
@@ -3,12 +3,7 @@
   lib,
   config,
   ...
-}: let
-  celan = pkgs.fetchzip {
-    url = "http://c.krebsco.de/celan.tar.gz";
-    sha256 = "sha256-nA+EwAH2vkeolsy9AoPLEMt1uGKDZe/aPrS95CZvuus=";
-  };
-in {
+}: {
   niveum.bots.celan = {
     enable = true;
     time = "08:00";
@@ -22,28 +17,49 @@ in {
       tokenFile = config.age.secrets.mastodon-token-celan.path;
       language = "de";
     };
-    command = toString (pkgs.writers.writeDash "random-celan" ''
-      cd ${celan}
-      poem="$(${pkgs.findutils}/bin/find . -type f | ${pkgs.coreutils}/bin/shuf -n1)"
-      source="$(${pkgs.coreutils}/bin/dirname "$poem" \
-        | ${pkgs.gnused}/bin/sed 's#^\./##;s/[-_]/ /g;s!/! › !g;s/0\([0-9]\+\)/\1/g' \
-        | ${pkgs.gnused}/bin/sed 's/Der Sand aus den Urnen/#Der_Sand_aus_den_Urnen (1948)/g' \
-        | ${pkgs.gnused}/bin/sed 's/Mohn und Gedächtnis/#Mohn_und_Gedächtnis (1952)/g' \
-        | ${pkgs.gnused}/bin/sed 's/Von Schwelle zu Schwelle/#Von_Schwelle_zu_Schwelle (1955)/g' \
-        | ${pkgs.gnused}/bin/sed 's/Sprachgitter/#& (1959)/g' \
-        | ${pkgs.gnused}/bin/sed 's/Niemandsrose/#& (1963)/g' \
-        | ${pkgs.gnused}/bin/sed 's/Atemwende/#& (1967)/g' \
-        | ${pkgs.gnused}/bin/sed 's/Fadensonnen/#& (1968)/g' \
-        | ${pkgs.gnused}/bin/sed 's/Lichtzwang/#& (1970)/g' \
-        | ${pkgs.gnused}/bin/sed 's/Schneepart/#& (1971)/g' \
-        | ${pkgs.gnused}/bin/sed 's/Zeitgehöft/#& (1976)/g' \
-        | ${pkgs.gnused}/bin/sed 's/Frühwerk/#& (1989)/g' \
-        | ${pkgs.gnused}/bin/sed 's/Eingedunkelt/#& (1991)/g' \
-        | ${pkgs.gnused}/bin/sed 's/Nachlaß/#& (1997)/g'
-      )"
-      cat "$poem"
-      echo
-      printf "Aus: %s\n\n#PaulCelan #Celan #Lyrik #poetry" "$source"
+    command = toString (pkgs.writers.writePython3 "random-celan.py" { libraries = [pkgs.python3Packages.lxml]; } ''
+      from lxml import etree
+      import random
+
+
+      def xml_text(elements):
+          return "".join("".join(t.itertext()) for t in elements).strip()
+
+
+      tree = etree.parse('${pkgs.fetchurl {
+        url = "http://c.krebsco.de/celan.tei.xml";
+        hash = "sha256-HgNmJYfhuwyfm+FcNtnnYWpJpIIU1ElHLeLiIFjF9mE=";
+      }}')
+      root = tree.getroot()
+
+      tei = {"tei": "http://www.tei-c.org/ns/1.0"}
+
+      poems = root.xpath(".//tei:lg[@type='poem']", namespaces=tei)
+
+      poem = random.choice(poems)
+
+      for stanza in poem.xpath("./tei:lg[@type='stanza']", namespaces=tei):
+          for line in stanza.xpath('./tei:l', namespaces=tei):
+              if line.text:
+                  print(line.text.strip())
+          print()
+
+      current_element = poem
+      while current_element is not None:
+          if current_element.tag == "{http://www.tei-c.org/ns/1.0}text":
+              text_element = current_element
+
+              title = xml_text(text_element.xpath("./tei:front/tei:docTitle",
+                               namespaces=tei))
+              print(f"Aus: #{title.replace(" ", "_")}", end=" ")
+
+              if date := xml_text(text_element.xpath("./tei:front/tei:docDate",
+                                  namespaces=tei)):
+                  print(f"({date})")
+              break
+          current_element = current_element.getparent()
+
+      print("\n\n#PaulCelan #Celan #Lyrik #poetry")
     '');
   };