pub / newsplanet

commit 8c2b3bddef09473c1d9a6b07d912c9903ca2f7a8
parent 3f10fb35fdae739eb66c278555860cf460592ae8
Author: JayVii <jayvii[AT]posteo[DOT]de>
Date:   Thu, 12 Dec 2024 10:01:40 +0100

feat: serve website icons locally

Diffstat:
M
.gitignore
 | 
1
+
M
.htaccess
 | 
2
++
A
fetch_favicon.sh
 | 
55
+++++++++++++++++++++++++++++++++++++++++++++++++++++++
M
run.sh
 | 
7
+++++--

4 files changed, 63 insertions(+), 2 deletions(-)
diff --git a/.gitignore b/.gitignore
@@ -5,3 +5,4 @@ centre.xml
 atom.xml
 rss.xml
 opml.xml
+assets/site-icons/**
diff --git a/.htaccess b/.htaccess
@@ -9,9 +9,11 @@ IndexOptions +Charset=UTF-8
 
 # Disallow browsing of certain sub-directories (redirect to 404)
 RedirectMatch 404 ^/.git/.*$
+RedirectMatch 404 ^/.gitignore$
 RedirectMatch 404 ^/.ssh/.*$
 RedirectMatch 404 ^/.config/.*$
 RedirectMatch 404 ^/.local/.*$
 RedirectMatch 404 ^/.cache/.*$
 RedirectMatch 404 ^/config.yaml$
 RedirectMatch 404 ^/template.html$
+RedirectMatch 404 ^/*.sh
diff --git a/fetch_favicon.sh b/fetch_favicon.sh
@@ -0,0 +1,55 @@
+#!/usr/bin/env sh
+
+# fetch input config file, fall back to "general"
+if [ -z $1 ]; then
+  CONFIG="centre.yaml"
+else
+  CONFIG="$1"
+fi
+
+# read yaml file
+URLS=`grep -E "^\s*-*\ web:" "$CONFIG" | sed -e 's/^.*web:\ //g'`
+
+# get domains
+DOMAINS=`
+  echo "$URLS" | sed -E -e 's/https:\/\/([^\/]+).*$/\1/g' | sort | uniq
+`
+
+# loop through websites
+for DOMAIN in $DOMAINS; do
+
+  # User output
+  printf "Fetching favicon for ${DOMAIN}...\n"
+
+  # read HTML of webpage
+  HTML=`curl --silent --location "$DOMAIN"`
+
+  # Fetch Favicon reference from HTML
+  ICN_URI=`
+    echo "$HTML" | \
+      grep "rel=\"icon\"" | \
+      tail -n 1 | \
+      sed -E -e 's/^.*href=\"([^\"]*)\".*$/\1/'
+  `
+
+  # Post-Process favicon url in order to prepare the download
+  ICN_URL=`
+    echo "$ICN_URI" | \
+      sed -E \
+        -e "s/^\//https:\/\/$DOMAIN\//" \
+        -e "s/^(\.+)/https:\/\/$DOMAIN\/\1/"
+  `
+
+  # # Get file extension
+  # ICN_EXT=`
+  #   echo "$ICN_URI" | \
+  #     sed -E -e 's/^.*\.([^\.]+)$/\1/'
+  # `
+
+  # Download Favicon to assets folder
+  curl --silent --location \
+    --output "./assets/site-icons/${DOMAIN}.${ICN_EXT}" \
+    "$ICN_URL"
+
+done
+
diff --git a/run.sh b/run.sh
@@ -1,12 +1,15 @@
 #!/usr/bin/env sh
 
-# fetch input config file, fall back to "general"
+# fetch input config file, fall back to "centre"
 if [ -z $1 ]; then
   CONFIG="centre.yaml"
 else
   CONFIG="$1"
 fi
 
+# Re-download favicons
+source ./fetch_favicon.sh "$CONFIG" &
+
 # Gather various information from config file
 printf "Gather information from config file...\n"
 HTML=`grep -E '^\s+file:.*?\.html\s*$' "$CONFIG" | awk '{ print $NF }'`
@@ -40,7 +43,7 @@ sed -E -e "s/<!--MANIFEST-->/\"$MANIFEST\"/" -i "$HTML"
 
 # insert favicon image buttons
 printf "Inserting favicons...\n"
-sed -E -e 's/<!--IMG\ src=\"(https:\/\/[^\/]*)[^\"]*(\"[^>]*>)-->/<img src=\"\1\/favicon.ico\2/g' -i "$HTML"
+sed -E -e 's/<!--IMG\ src=\"https:\/\/([^\/]*)[^\"]*(\"[^>]*>)-->/<img src=\"\/assets\/site-icons\/\1\2/g' -i "$HTML"
 
 # Insert feeds list
 printf "Inserting feeds list...\n"