fix(bluesky): pass embed metadata directly to avoid OG parsing issues (#3)

Adds optional embed_title, embed_description, and embed_image_url inputs
to the bluesky-post workflow. When provided, these values are used directly
instead of parsing OG tags from the URL.

This fixes an issue where titles containing apostrophes (e.g., "It's") were
being truncated due to the OG tag regex stopping at quote characters.

Also fixes the OG tag regex to use backreferences for proper quote matching
as a fallback when metadata is not provided.

Author: CalvinAllen

.github/workflows/bluesky-post.yml

@@ -8,7 +8,19 @@ on:
         required: true
         type: string
       embed_url:
-        description: 'Optional URL to create a website card embed from (fetches OG tags)'
+        description: 'Optional URL to create a website card embed from'
+        required: false
+        type: string
+      embed_title:
+        description: 'Optional title for the embed card (avoids fetching from OG tags)'
+        required: false
+        type: string
+      embed_description:
+        description: 'Optional description for the embed card (avoids fetching from OG tags)'
+        required: false
+        type: string
+      embed_image_url:
+        description: 'Optional image URL for the embed card (avoids fetching from OG tags)'
         required: false
         type: string
     secrets:
@@ -25,6 +37,9 @@ jobs:
         env:
           POST_INPUT: ${{ toJSON(inputs.post_text) }}
           EMBED_URL: ${{ inputs.embed_url }}
+          EMBED_TITLE: ${{ inputs.embed_title }}
+          EMBED_DESCRIPTION: ${{ inputs.embed_description }}
+          EMBED_IMAGE_URL: ${{ inputs.embed_image_url }}
         run: |
           # Authenticate with BlueSky
           echo "Authenticating with BlueSky..."
@@ -119,10 +134,17 @@ jobs:
           # Build embed if URL provided
           EMBED=""
           if [ -n "$EMBED_URL" ]; then
-            echo "Fetching OG tags from: $EMBED_URL"
+            # Use provided metadata or fetch from OG tags
+            OG_TITLE="$EMBED_TITLE"
+            OG_DESC="$EMBED_DESCRIPTION"
+            OG_IMAGE="$EMBED_IMAGE_URL"
+
+            # Only fetch OG tags if we're missing required metadata
+            if [ -z "$OG_TITLE" ] || [ -z "$OG_IMAGE" ]; then
+              echo "Fetching OG tags from: $EMBED_URL"
 
-            # Fetch page and extract OG tags using Python
-            EMBED_DATA=$(python3 << 'PYEOF'
+              # Fetch page and extract OG tags using Python
+              EMBED_DATA=$(python3 << 'PYEOF'
           import urllib.request
           import re
           import json
@@ -139,30 +161,34 @@ jobs:
               with urllib.request.urlopen(req, timeout=10) as response:
                   html = response.read().decode('utf-8', errors='ignore')
 
-              # Extract OG tags
-              og_title = re.search(r'<meta[^>]*property=["\']og:title["\'][^>]*content=["\']([^"\']+)["\']', html, re.I)
-              og_title = og_title.group(1) if og_title else ''
+              # Extract OG tags using backreferences to handle quotes in content
+              og_title = re.search(r'<meta[^>]*property=["\']og:title["\'][^>]*content=(["\'])(.+?)\1', html, re.I)
+              og_title = og_title.group(2) if og_title else ''
 
-              og_desc = re.search(r'<meta[^>]*property=["\']og:description["\'][^>]*content=["\']([^"\']+)["\']', html, re.I)
-              og_desc = og_desc.group(1) if og_desc else ''
+              og_desc = re.search(r'<meta[^>]*property=["\']og:description["\'][^>]*content=(["\'])(.+?)\1', html, re.I)
+              og_desc = og_desc.group(2) if og_desc else ''
 
-              og_image = re.search(r'<meta[^>]*property=["\']og:image["\'][^>]*content=["\']([^"\']+)["\']', html, re.I)
-              og_image = og_image.group(1) if og_image else ''
+              og_image = re.search(r'<meta[^>]*property=["\']og:image["\'][^>]*content=(["\'])(.+?)\1', html, re.I)
+              og_image = og_image.group(2) if og_image else ''
 
               print(json.dumps({'title': og_title, 'description': og_desc, 'image': og_image}))
           except Exception as e:
               print(json.dumps({'error': str(e)}), file=sys.stderr)
               print(json.dumps({}))
           PYEOF
-            )
+              )
 
-            OG_TITLE=$(echo "$EMBED_DATA" | jq -r '.title // empty')
-            OG_DESC=$(echo "$EMBED_DATA" | jq -r '.description // empty')
-            OG_IMAGE=$(echo "$EMBED_DATA" | jq -r '.image // empty')
+              # Use fetched values only for missing metadata
+              [ -z "$OG_TITLE" ] && OG_TITLE=$(echo "$EMBED_DATA" | jq -r '.title // empty')
+              [ -z "$OG_DESC" ] && OG_DESC=$(echo "$EMBED_DATA" | jq -r '.description // empty')
+              [ -z "$OG_IMAGE" ] && OG_IMAGE=$(echo "$EMBED_DATA" | jq -r '.image // empty')
+            else
+              echo "Using provided embed metadata"
+            fi
 
-            echo "OG Title: $OG_TITLE"
-            echo "OG Description: $OG_DESC"
-            echo "OG Image: $OG_IMAGE"
+            echo "Embed Title: $OG_TITLE"
+            echo "Embed Description: $OG_DESC"
+            echo "Embed Image: $OG_IMAGE"
 
             if [ -n "$OG_TITLE" ]; then
               # Upload image as blob if present