From 3c6855e2f382001577d0cc850f8be636dee25f0d Mon Sep 17 00:00:00 2001
From: stuebinm
Date: Mon, 24 Jul 2023 17:14:20 +0200
Subject: htmlq is nice

(thanks to the shork for the hint!)
---
 home/home.nix     | 8 +++++---
 home/packages.nix | 2 +-
 2 files changed, 6 insertions(+), 4 deletions(-)

(limited to 'home')

diff --git a/home/home.nix b/home/home.nix
index 0a5072f..49b5713 100644
--- a/home/home.nix
+++ b/home/home.nix
@@ -101,6 +101,7 @@
       agenda = "CLICOLOR_FORCE=1 almanac month ~/.cache/feedsync/* | less";
       xmlpretty = "xmllint --format - | bat -l xml";
       wormhole = "wormhole-rs";
+      noms = "curl https://www.bahn.de/service/zug/db_lounge -s | htmlq --attribute href a | grep Speisen | xargs curl -s | pdftotext - /dev/stdout | less";
     };
     functions.fetchpdf.body = ''
       set url $argv[1]
@@ -108,12 +109,13 @@
           echo "No url given"
           return
       end
-      set choice (curl -L $url | rg -e '<a[^>]*href="([^"]*\.pdf)"[^>]*>([^<]*)</a>' -o | sed 's/<a[^>]*href="\([^"]*\)"[^>]*>\(.*\)<\/a>/"\1" "\2"/g' | fzf)
+      set anchors (curl -L $url | sd '\n' "" | htmlq a | rg .pdf)
+      set choice (paste (printf '%s\n' $anchors | htmlq -t a | psub) (printf '%s\n' $anchors | htmlq --attribute href a | psub) | fzf)
       if test "$status" != 0
           return
       end
-      set docurl (echo $choice | sed 's/"\(.*\)" ".*"/\1/g')
-      set docname (echo $choice | sed 's/".*" "\(.*\)"/\1/g')
+      set docurl (echo $choice | cut -f 2)
+      set docname (echo $choice | cut -f 1)
       echo Fetching $docname from $docurl
       curl "$docurl" > ~/Downloads/news/"$docname".pdf
       zathura ~/Downloads/news/"$docname".pdf
diff --git a/home/packages.nix b/home/packages.nix
index 998f205..49d7a02 100644
--- a/home/packages.nix
+++ b/home/packages.nix
@@ -43,7 +43,7 @@ in
     (agda.withPackages (p: [ p.standard-library p.cubical ]))
     # html, js & co
     jq html-tidy nodePackages.stylelint nodePackages.js-beautify zola
-    libxml2 fq
+    libxml2 fq htmlq
     # purescript
     spago purescript # nodePackages.purescript-language-server nodejs
     # rust
-- 
cgit v1.2.3