From b8f6b828740c8701b54c8b7786d1c714bce2f349 Mon Sep 17 00:00:00 2001
From: Xi Xu <i@xi-xu.me>
Date: Tue, 10 Jun 2025 15:46:13 +0800
Subject: [PATCH] Improve favicon download logic in workflow

Revised the favicon download logic to use temporary files and associative arrays for better handling of duplicate filenames and sequential numbering. Enhanced error handling and ensured proper file renaming after successful downloads.
---
 .github/workflows/run.yml | 59 ++++++++++++++++++++++++---------------
 1 file changed, 36 insertions(+), 23 deletions(-)

diff --git a/.github/workflows/run.yml b/.github/workflows/run.yml
index 7dae77392..54f5e03d1 100644
--- a/.github/workflows/run.yml
+++ b/.github/workflows/run.yml
@@ -103,8 +103,9 @@ jobs:
           success_count=0
           error_count=0
 
-          # Array to track used filenames for duplicate handling
-          declare -A filename_count
+          # Use associative arrays to track used filenames and next available index
+          declare -A filename_next_index
+          declare -A filename_used
 
           # Read domains file line by line
           while IFS= read -r domain; do
@@ -121,54 +122,66 @@ jobs:
             fi
             
             # Extract second-level domain (remove top-level domain)
-            # For example: google.com -> google, github.io -> github
             second_level_domain=$(echo "$domain" | sed 's/\.[^.]*$//')
             
-            # Handle duplicate filenames by adding sequential numbers
-            base_filename="$second_level_domain"
-            if [[ -n "${filename_count[$base_filename]}" ]]; then
-              # Increment count for this filename
-              filename_count[$base_filename]=$((filename_count[$base_filename] + 1))
-              final_filename="${base_filename}_${filename_count[$base_filename]}"
-            else
-              # First occurrence of this filename
-              filename_count[$base_filename]=0
-              final_filename="$base_filename"
+            # Initialize next index for this domain if not yet set
+            if [[ -z "${filename_next_index[$second_level_domain]}" ]]; then
+              filename_next_index[$second_level_domain]=0
             fi
             
             # Construct favicon URL
             favicon_url="https://t1.gstatic.com/faviconV2?client=SOCIAL&type=FAVICON&fallback_opts=TYPE,SIZE,URL&url=https://${domain}&size=256"
-            favicon_file="favicons/${final_filename}.png"
+            
+            # Temporarily use a temp file for download
+            temp_favicon_file="favicons/temp_favicon.png"
             
             # Download favicon with enhanced error handling and timeout
-            if timeout 15 curl -L -s -o "$favicon_file" "$favicon_url" --max-time 10 --retry 3 --retry-delay 1; then
+            if timeout 15 curl -L -s -o "$temp_favicon_file" "$favicon_url" --max-time 10 --retry 3 --retry-delay 1; then
               # Check if file was actually downloaded and has content
-              if [ -s "$favicon_file" ]; then
+              if [ -s "$temp_favicon_file" ]; then
                 # Calculate SHA-256 hash of the downloaded file
-                file_hash=$(sha256sum "$favicon_file" | cut -d' ' -f1)
+                file_hash=$(sha256sum "$temp_favicon_file" | cut -d' ' -f1)
                 
                 # Check if the hash matches the failure hash
                 if [ "$file_hash" = "59bfe9bc385ad69f50793ce4a53397316d7a875a7148a63c16df9b674c6cda64" ]; then
                   # This is a failed download (generic/default favicon)
-                  rm -f "$favicon_file"
+                  rm -f "$temp_favicon_file"
                   error_count=$((error_count + 1))
-                  echo "Warning: Generic favicon detected for domain: $domain (mapped to $final_filename) (SHA-256: $file_hash)"
+                  echo "Warning: Generic favicon detected for domain: $domain (SHA-256: $file_hash)"
                 else
+                  # Valid favicon downloaded - now assign the proper filename
+                  current_index=${filename_next_index[$second_level_domain]}
+                  
+                  # Determine the final filename
+                  if [ $current_index -eq 0 ]; then
+                    final_filename="$second_level_domain"
+                  else
+                    final_filename="${second_level_domain}_${current_index}"
+                  fi
+                  
+                  # Move the temp file to the final destination
+                  favicon_file="favicons/${final_filename}.png"
+                  mv "$temp_favicon_file" "$favicon_file"
+                  
+                  # Mark this filename as used and increment the next index
+                  filename_used["$final_filename"]=1
+                  filename_next_index[$second_level_domain]=$((current_index + 1))
+                  
                   # Valid favicon downloaded
                   success_count=$((success_count + 1))
                   echo "Success: Downloaded favicon for $domain -> $final_filename.png"
                 fi
               else
                 # Remove empty file
-                rm -f "$favicon_file"
+                rm -f "$temp_favicon_file"
                 error_count=$((error_count + 1))
-                echo "Warning: Empty file for domain: $domain (mapped to $final_filename)"
+                echo "Warning: Empty file for domain: $domain"
               fi
             else
               # Remove any partial file
-              rm -f "$favicon_file"
+              rm -f "$temp_favicon_file"
               error_count=$((error_count + 1))
-              echo "Error: Failed to download favicon for domain: $domain (mapped to $final_filename)"
+              echo "Error: Failed to download favicon for domain: $domain"
             fi
             
             # Add high failure rate detection (stop if >80% failures after 100 domains)