From fd1d74ada1596df8f4221316072264f945d63735 Mon Sep 17 00:00:00 2001
From: end-4 <97237370+end-4@users.noreply.github.com>
Date: Tue, 14 Oct 2025 10:19:09 +0200
Subject: [PATCH] refractor gemini-categorize-wallpaper.sh

---
 .../scripts/ai/gemini-categorize-wallpaper.sh | 44 +++++++++++++------
 1 file changed, 30 insertions(+), 14 deletions(-)

diff --git a/.config/quickshell/ii/scripts/ai/gemini-categorize-wallpaper.sh b/.config/quickshell/ii/scripts/ai/gemini-categorize-wallpaper.sh
index 33915fb97..2e6ddf20f 100755
--- a/.config/quickshell/ii/scripts/ai/gemini-categorize-wallpaper.sh
+++ b/.config/quickshell/ii/scripts/ai/gemini-categorize-wallpaper.sh
@@ -1,33 +1,45 @@
 #!/usr/bin/env bash
 
 if [[ -z "$1" ]]; then
-    echo "Usage: $0 <image_path>"
+    echo "Usage: $0 <image_path> [model] [prompt]"
+    echo "Tip: set GEMINI_WALLPAPER_MODEL and/or GEMINI_WALLPAPER_PROMPT to provide defaults."
     exit 1
 fi
 
+# Variables
 SOURCE_IMG_PATH="$1"
+MODEL="${2:-${GEMINI_WALLPAPER_MODEL:-gemini-2.5-flash-lite}}" # We use the flash variant so it's fast
 WALLPAPER_NAME="$(basename "$SOURCE_IMG_PATH")"
+PROMPT="${3:-${GEMINI_WALLPAPER_PROMPT:-Categorize the wallpaper. Its file name is $WALLPAPER_NAME}}"
 RESIZED_IMG_PATH="/tmp/quickshell/ai/wallpaper.jpg"
+
+# Resize image for speed
 magick "$SOURCE_IMG_PATH" -resize 200x -quality 50 "$RESIZED_IMG_PATH"
+
+# Get API key
 API_KEY=$(secret-tool lookup 'application' 'illogical-impulse' | jq -r '.apiKeys.gemini')
 
+# Encode image to base64
 if [[ "$(base64 --version 2>&1)" = *"FreeBSD"* ]]; then
-B64FLAGS="--input"
+    B64FLAGS="--input"
 else
-B64FLAGS="-w0"
+    B64FLAGS="-w0"
 fi
+B64DATA="$(base64 $B64FLAGS $RESIZED_IMG_PATH)"
+# echo $B64DATA
 
+# Prepare request data
 payload='{
     "contents": [{
-    "parts":[
-        {
-            "inline_data": {
-            "mime_type":"image/jpeg",
-            "data": "'"$(base64 $B64FLAGS $RESIZED_IMG_PATH)"'"
-            }
-        },
-        {"text": "Categorize the wallpaper. Its file name is '"$WALLPAPER_NAME"'"}
-    ]
+        "parts":[
+            {
+                "inline_data": {
+                "mime_type":"image/jpeg",
+                "data": "'"$B64DATA"'"
+                }
+            },
+            {"text": "'"$PROMPT"'"}
+        ]
     }],
     "generationConfig": {
         "responseMimeType": "text/x.enum",
@@ -35,14 +47,18 @@ payload='{
             "type": "string",
             "enum": [ "abstract", "anime", "city", "minimalist", "landscape", "plants", "person", "space" ]
         },
-        "temperature": 0,
+        "temperature": 0
     }
 }'
 # echo "$payload" | jq
-response=$(curl "https://generativelanguage.googleapis.com/v1beta/models/gemini-2.0-flash:generateContent" \
+
+# Make the request
+response=$(curl "https://generativelanguage.googleapis.com/v1beta/models/${MODEL}:generateContent" \
 -H "x-goog-api-key: $API_KEY" \
 -H 'Content-Type: application/json' \
 -X POST \
 -d "$payload" 2> /dev/null)
+# echo "$response" | jq
 
+# Write the result
 echo "$response" | jq -r '.candidates[0].content.parts[0].text'