Merge pull request #27 from usegalaxy-eu/better-previews

Improve preview generation, matrix mentions and image support

.github/workflows/preview.yml

@@ -13,7 +13,7 @@ jobs:
       pull-requests: write
     steps:
       - name: Checkout
-        uses: actions/checkout@v3
+        uses: actions/checkout@v4
 
       - name: Get changed files in posts folder
         id: get_changed_files
@@ -30,9 +30,9 @@ jobs:
 
       - name: Set up Python
         if: steps.get_changed_files.outputs.any_changed == 'true'
-        uses: actions/setup-python@v3
+        uses: actions/setup-python@v5
         with:
-          python-version: 3.9
+          python-version: '3.12'
 
       - name: Install dependencies
         if: steps.get_changed_files.outputs.any_changed == 'true'

.github/workflows/publish_content.yml

@@ -14,7 +14,7 @@ jobs:
       pull-requests: write
     steps:
       - name: Checkout
-        uses: actions/checkout@v3
+        uses: actions/checkout@v4
 
       - name: Get changed files in posts folder
         id: get_changed_files
@@ -31,9 +31,9 @@ jobs:
 
       - name: Set up Python
         if: steps.get_changed_files.outputs.any_changed == 'true'
-        uses: actions/setup-python@v3
+        uses: actions/setup-python@v5
         with:
-          python-version: 3.9
+          python-version: '3.12'
 
       - name: Install dependencies
         if: steps.get_changed_files.outputs.any_changed == 'true'

github_run.py

@@ -31,14 +31,14 @@ def comment(self, comment_text):
         url = (
             f"https://api.github.com/repos/{self.repo}/issues/{self.pr_number}/comments"
         )
-        data = {"body": str(comment_text)}
-        response = requests.post(url, headers=headers, json=data)
-        if response.status_code == 201:
-            return True
-        else:
-            raise Exception(
-                f"Failed to create github comment!, {response.json().get('message')}"
-            )
+        for comment_body in comment_text.split("\n\n---\n"):
+            data = {"body": str(comment_body)}
+            response = requests.post(url, headers=headers, json=data)
+            if response.status_code != 201:
+                raise Exception(
+                    f"Failed to create github comment!, {response.json().get('message')}"
+                )
+        return True
 
     def get_files(self):
         url = f"https://api.github.com/repos/{self.repo}/pulls/{self.pr_number}/files"

lib/galaxy_social.py

@@ -22,9 +22,6 @@ def __init__(self, preview: bool, json_out: str):
 
         self.plugins = {}
         for plugin in self.plugins_config["plugins"]:
-            if preview and plugin["name"].lower() != "markdown":
-                continue
-
             if plugin["enabled"]:
                 module_name, class_name = plugin["class"].rsplit(".", 1)
                 try:
@@ -112,35 +109,45 @@ def parse_markdown_file(self, file_path):
 
     def process_markdown_file(self, file_path, processed_files):
         content, metadata = self.parse_markdown_file(file_path)
-        if self.preview:
+        formatting_results = {}
+        for media in metadata["media"]:
             try:
-                _, _, message = self.plugins["markdown"].create_post(
+                formatting_results[media] = self.plugins[media].format_content(
                     content=content,
-                    mentions=[],
-                    hashtags=[],
+                    mentions=metadata.get("mentions", {}).get(media, []),
+                    hashtags=metadata.get("hashtags", {}).get(media, []),
                     images=metadata.get("images", []),
-                    media=metadata["media"],
-                    preview=True,
-                    file_path=file_path,
                 )
-                return processed_files, message
             except Exception as e:
-                raise Exception(f"Failed to create preview for {file_path}.\n{e}")
+                raise Exception(f"Failed to format post for {file_path}.\n{e}")
+        if self.preview:
+            message = f'Hi, I\'m your friendly social media assistant. In the following, you will see a preview of this post "{file_path}"'
+            for media in metadata["media"]:
+                formatted_content, preview, warning = formatting_results[media]
+                message += f"\n\n## {media}\n\n"
+                message += preview
+                if warning:
+                    message += f"\nWARNING: {warning}"
+            return processed_files, message.strip()
+
         stats = {}
         url = {}
         if file_path in processed_files:
             stats = processed_files[file_path]
         for media in metadata["media"]:
-            if file_path in processed_files and media in processed_files[file_path]:
+            if stats.get(media):
+                print("Skipping previous post to", media)
                 continue
-            mentions = metadata.get("mentions", {}).get(media, [])
-            hashtags = metadata.get("hashtags", {}).get(media, [])
-            images = metadata.get("images", [])
+            formatted_content, _, _ = formatting_results[media]
             stats[media], url[media] = self.plugins[media].create_post(
-                content, mentions, hashtags, images, file_path=file_path
+                formatted_content, file_path=file_path
             )
         url_text = "\n".join(
-            [f"[{media}]({link})" for media, link in url.items() if link]
+            [
+                f"- [{media}]({link})" if link else f"- {media}"
+                for media, link in url.items()
+                if stats[media]
+            ]
         )
         message = f"Posted to:\n\n{url_text}" if url_text else "No posts created."
 
@@ -150,7 +157,7 @@ def process_markdown_file(self, file_path, processed_files):
 
     def process_files(self, files_to_process):
         processed_files = {}
-        messages = "---\n"
+        messages = ""
         processed_files_path = self.json_out
         if os.path.exists(processed_files_path):
             with open(processed_files_path, "r") as file:

lib/plugins/bluesky.py

@@ -149,11 +149,52 @@ def handle_url_card(
             )
         return embed_external
 
-    def create_post(
-        self, content, mentions, hashtags, images, **kwargs
-    ) -> Tuple[bool, Optional[str]]:
+    def wrap_text_with_index(self, content):
+        if len(content) <= self.max_content_length:
+            return [content]
+        urls = re.findall(r"https?://\S+", content)
+        placeholder_content = re.sub(
+            r"https?://\S+", lambda m: "~" * len(m.group()), content
+        )
+        wrapped_lines = textwrap.wrap(
+            placeholder_content, self.max_content_length - 8, replace_whitespace=False
+        )
+        final_lines = []
+        url_index = 0
+        for i, line in enumerate(wrapped_lines, 1):
+            while "~~~~~~~~~~" in line and url_index < len(urls):
+                placeholder = "~" * len(urls[url_index])
+                line = line.replace(placeholder, urls[url_index], 1)
+                url_index += 1
+            final_lines.append(f"{line} ({i}/{len(wrapped_lines)})")
+        return final_lines
+
+    def format_content(self, content, mentions, hashtags, images, **kwargs):
+        mentions = " ".join([f"@{v}" for v in mentions])
+        hashtags = " ".join([f"#{v}" for v in hashtags])
+        if len(images) > 4:
+            warnings = f"A maximum of four images, not {len(images)}, can be included in a single bluesky post."
+            images = images[:4]
+        else:
+            warnings = ""
+
+        chunks = self.wrap_text_with_index(f"{content}\n\n{mentions}\n{hashtags}")
+
+        formatted_content = {
+            "body": "\n\n".join(chunks),
+            "images": images,
+            "chunks": chunks,
+        }
+        preview = formatted_content["body"]
+        images_preview = "\n".join(
+            [f'![{image.get("alt_text", "")}]({image["url"]})' for image in images]
+        )
+        preview += "\n\n" + images_preview
+        return formatted_content, preview, warnings
+
+    def create_post(self, content, **kwargs) -> Tuple[bool, Optional[str]]:
         embed_images = []
-        for image in images[:4]:
+        for image in content["images"][:4]:
             response = requests.get(image["url"])
             if response.status_code == 200 and response.headers.get(
                 "Content-Type", ""
@@ -172,17 +213,11 @@ def create_post(
             else None
         )
 
-        status = []
         reply_to = None
-        mentions = " ".join([f"@{v}" for v in mentions])
-        hashtags = " ".join([f"#{v}" for v in hashtags])
-        for text in textwrap.wrap(
-            content + "\n" + mentions + "\n" + hashtags,
-            self.max_content_length,
-            replace_whitespace=False,
-        ):
+
+        for text in content["chunks"]:
             facets, last_url = self.parse_facets(text)
-            if not images or reply_to:
+            if not content["images"] or reply_to:
                 embed = self.handle_url_card(cast(str, last_url))
 
             post = self.blueskysocial.send_post(
@@ -192,13 +227,14 @@ def create_post(
             for _ in range(5):
                 data = self.blueskysocial.get_posts([post.uri]).posts
                 if data:
-                    status.append(data[0].record.text == text)
                     break
+            else:
+                return False, None
 
             if reply_to is None:
                 link = f"https://bsky.app/profile/{self.blueskysocial.me.handle}/post/{post.uri.split('/')[-1]}"
                 root = atproto.models.create_strong_ref(post)
             parent = atproto.models.create_strong_ref(post)
             reply_to = atproto.models.AppBskyFeedPost.ReplyRef(parent=parent, root=root)
 
-        return all(status), link
+        return True, link

lib/plugins/markdown.py

@@ -10,42 +10,26 @@ def __init__(self, **kwargs):
             else os.path.join(os.getcwd(), kwargs["save_path"])
         )
 
-    def create_post(self, content, mentions, hashtags, images, **kwargs):
+    def format_content(self, content, mentions, hashtags, images, **kwargs):
+        _images = "\n".join(
+            [f'![{image.get("alt_text", "")}]({image["url"]})' for image in images]
+        )
+        mentions = " ".join([f"@{v}" for v in mentions])
+        hashtags = " ".join([f"#{v}" for v in hashtags])
+        warnings = ""
+        formatted_content = "\n\n".join([content, mentions, hashtags, _images])
+        preview = formatted_content
+        return formatted_content, preview, warnings
+
+    def create_post(self, formatted_content, **kwargs):
         try:
-            _images = (
-                "\n"
-                + "\n".join(
-                    [
-                        f'![{image.get("alt_text", "")}]({image["url"]})'
-                        for image in images
-                    ]
-                )
-                if images
-                else ""
-            )
-            mentions = "\n" + " ".join([f"@{v}" for v in mentions]) if mentions else ""
-            hashtags = "\n" + " ".join([f"#{v}" for v in hashtags]) if hashtags else ""
-            text = f"{content}{mentions}{hashtags}{_images}"
             if self.save_path:
                 os.makedirs(self.save_path, exist_ok=True)
                 prefix = kwargs.get("file_path", "").replace(".md", "")
-                file_name = (
-                    f"{self.save_path}/{prefix.replace('/', '-')}_{time.strftime('%Y%m%d-%H%M%S')}.md"
-                )
+                file_name = f"{self.save_path}/{prefix.replace('/', '-')}_{time.strftime('%Y%m%d-%H%M%S')}.md"
                 with open(file_name, "w") as f:
-                    f.write(text)
-            if kwargs.get("preview"):
-                social_media = ", ".join(kwargs.get("media", []))
-                pre_comment_text = ""
-                if len(images) > 4 and (
-                    "mastodon" in social_media or "bluesky" in social_media
-                ):
-                    pre_comment_text = f"Please note that Mastodon and Bluesky only support up to 4 images in a single post. The first 4 images will be included in the post, and the rest will be ignored.\n"
-                comment_text = f"{pre_comment_text}This is a preview from {prefix.split('/')[-1]} that will be posted to {social_media}:\n\n{text}"
-                return True, None, comment_text
+                    f.write(formatted_content)
             return True, None
         except Exception as e:
-            if kwargs.get("preview", False):
-                print(e)
-                return False, None, e
+            print(e)
             return False, None