Added customization options to the urlpreview

urlpreview/README.md

@@ -38,6 +38,13 @@ If the link returns a 404, the bot will return an emoji `no_results_react` (💨
 - `no_results_react` - Adds a reaction emoji to the message to show that no results were returned. Put `''` to disable.
 - `url_blacklist` - Disable urlpreview for an IP range or a Regex entry
 - `user_blacklist` - Disable urlpreview for a user
+- `html_custom_title_style` - Set to a CSS string in order to apply that style as an html attribute to the title tag in a response/reply. Put `''` to disable.
+- `html_custom_description_style` -  Set to a CSS string in order to apply that style as an html attribute to the description tag in a response/reply. Put `''` to disable.
+- `html_custom_image_style` -  '' Set to a CSS string in order to apply that style as an html attribute to the image tag in a response/reply. Put `''` to disable.
+- `respond_instead_of_reply` - Set to True in order to respond rather than reply. A reply quotes the original message that prompted this bot, while a response simply sends the message without quoting.
+- `use_divtag_instead_of_htag_for_title` -  Set to True to use an HTML div tag instead of the default h3 tag in the title.
+- `use_divtag_instead_of_ptag_for_description` -  Set to True to use an HTML div tag instead of the default p tag .
+- `indent_description` -  Set to True to apply a 4-space indent and a right-arrow to the beginning of the description to improve readability when other customizations are used.
 
 <details open>
 {( <summary><b>htmlparser</b></summary> )}

urlpreview/base-config.yaml

@@ -6,6 +6,14 @@ json_max_char: 2000
 max_links: 3
 min_image_width: 475
 max_image_embed: 300
+html_custom_title_style: '' # 'font-size: medium; padding-bottom: 8px; line-height:1.1; margin-bottom: 5px;'
+html_custom_description_style: '' # 'font-size: small; line-height: 1; padding-bottom: 8px; margin-bottom: 8px; padding-top:1px; margin-top:2px;'
+html_custom_image_style: '' # 'max-height: 60px;'
+respond_instead_of_reply: False
+use_divtag_instead_of_htag_for_title: False
+use_divtag_instead_of_ptag_for_description: False
+indent_description: False
+yt_dlp_storage_path: '.'
 no_results_react: 💨
 url_blacklist:
   - ':\/\/matrix\.to\/'

urlpreview/urlpreview/urlpreview.py

@@ -33,6 +33,13 @@ def do_update(self, helper: ConfigUpdateHelper) -> None:
         helper.copy("no_results_react")
         helper.copy("url_blacklist")
         helper.copy("user_blacklist")
+        helper.copy("html_custom_title_style")
+        helper.copy("html_custom_description_style")
+        helper.copy("html_custom_image_style")
+        helper.copy("respond_instead_of_reply")
+        helper.copy("use_divtag_instead_of_htag_for_title")
+        helper.copy("use_divtag_instead_of_ptag_for_description")
+        helper.copy("indent_description")
 
 class UrlPreviewBot(Plugin):
     async def start(self) -> None:
@@ -61,6 +68,14 @@ async def handler(self, evt: MessageEvent, matches: List[str]) -> None:
         MAX_IMAGE_EMBED = self.config["max_image_embed"]
         NO_RESULTS_REACT = self.config["no_results_react"]
         URL_BLACKLIST = self.config["url_blacklist"]
+        self.HTML_CUSTOM_STYLES_DICT = {}
+        self.HTML_CUSTOM_STYLES_DICT['TITLE'] = self.config["html_custom_title_style"]
+        self.HTML_CUSTOM_STYLES_DICT['DESCRIPTION'] = self.config["html_custom_description_style"]
+        self.HTML_CUSTOM_STYLES_DICT['IMAGE'] = self.config["html_custom_image_style"]
+        self.RESPOND_INSTEAD_OF_REPLY = self.config["respond_instead_of_reply"]
+        self.USE_DIVTAG_INSTEAD_OF_HTAG_FOR_TITLE = self.config["use_divtag_instead_of_htag_for_title"]
+        self.USE_DIVTAG_INSTEAD_OF_PTAG_FOR_DESCRIPTION = self.config["use_divtag_instead_of_ptag_for_description"]
+        self.INDENT_DESCRIPTION = self.config['indent_description']
         await evt.mark_read()
 
         embeds = []
@@ -102,7 +117,10 @@ async def handler(self, evt: MessageEvent, matches: List[str]) -> None:
                     pass
             return
         to_send = "".join(embeds)
-        return await evt.reply(to_send, allow_html=True)
+        if self.RESPOND_INSTEAD_OF_REPLY:
+            return await evt.respond(to_send, allow_html=True)
+        else:
+            return await evt.reply(to_send, allow_html=True)
 
 
 # Utility Commands
@@ -142,7 +160,7 @@ async def embed_url_preview(self, url_str, og, html_custom_headers=None, max_ima
         return None
     if all(v is None for v in og):
         return None
-    # Fetch image_mxc
+
     image_mxc = og.get('image_mxc', None)
     if image_mxc is None:
         image_mxc = await process_image(self, image=og.get('image', None), html_custom_headers=html_custom_headers, content_type=og.get('content_type', None))
@@ -153,10 +171,10 @@ async def embed_url_preview(self, url_str, og, html_custom_headers=None, max_ima
             return f"<blockquote>{image_solo}</blockquote>"
         return None # Everything is empty
     # Default message
-    title = format_title(og.get('title', None), url_str)
-    description = format_description(og.get('description', None))
-    image = format_image(image_mxc, url_str, og.get('content_type', None), format_image_width(og.get('image_width', None), max_image_embed))
-    message = "".join(filter(None, [title, description, image]))
+    title = format_title(og.get('title', None), url_str, custom_styles=self.HTML_CUSTOM_STYLES_DICT['TITLE'], use_divtag_instead_of_htag=self.USE_DIVTAG_INSTEAD_OF_HTAG_FOR_TITLE)
+    description = format_description(og.get('description'), custom_styles=self.HTML_CUSTOM_STYLES_DICT['DESCRIPTION'], indent_description=self.INDENT_DESCRIPTION, use_divtag_instead_of_ptag=self.USE_DIVTAG_INSTEAD_OF_PTAG_FOR_DESCRIPTION)
+    image = format_image(image_mxc, url_str, og.get('content_type', None), format_image_width(og.get('image_width', None), max_image_embed), custom_styles=self.HTML_CUSTOM_STYLES_DICT['IMAGE'])
+    message = "".join(filter(None, [title, description,  image]))
     if message:
         return f"<blockquote>{message}</blockquote>"
     return None

urlpreview/urlpreview/urlpreview_utils.py

@@ -2,6 +2,10 @@
 import re
 import socket
 import urllib.parse
+import os
+import subprocess
+from datetime import datetime as dt
+import hashlib
 from urllib.parse import urlparse
 
 IMAGE_TYPES = ["image/gif", "image/jpg", "image/jpeg", "image/png", "image/webp"]
@@ -31,23 +35,26 @@ def check_line_breaks(text: str):
         return None
     return text.replace('\n', '<br />')
 
-def format_title(title, url_str: str=""):
+def format_title(title, url_str: str="", custom_styles: str="",use_divtag_instead_of_htag: bool=False):
     if not title:
         return None
+    tag = 'div' if use_divtag_instead_of_htag else 'h3'
     if url_str:
-        return f'<h3><a href="{url_str}">{str(title)}</a></h3>'
+        return f'<{tag} style="{custom_styles}"><a href="{url_str}">{str(title)}</a></{tag}>'
     else:
-        return f'<h3>{str(title)}</h3>'
+        return f'<{tag} style="{custom_styles}" >{str(title)}</{tag}>'
 
-def format_description(description, preserve_line_breaks: bool=False):
+def format_description(description, preserve_line_breaks: bool=False, custom_styles: str="", indent_description: bool=False, use_divtag_instead_of_ptag: bool=False):
     if not description:
         return None
+    prefix = '&#8250;&nbsp;&nbsp;&nbsp;&nbsp;' if indent_description else ''
+    tag = 'div' if use_divtag_instead_of_ptag else 'p'
     if preserve_line_breaks is False:
-        return f'<p>'+str(description).replace('\r', ' ').replace('\n', ' ')+'</p>'
+        return f'<{tag} style="{custom_styles}">{prefix}'+str(description).replace('\r', ' ').replace('\n', ' ')+f'</{tag}>'
     else:
-        return f'<p>'+str(description)+'</p>'
+        return f'<{tag} style="{custom_styles}">{prefix}'+str(description)+f'</{tag}>'
 
-def format_image(image_mxc, url_str: str='', content_type: str=None, max_image_embed: int=300):
+def format_image(image_mxc, url_str: str='', content_type: str=None, max_image_embed: int=300, custom_styles: str=""):
     if not image_mxc:
         return None
     if not content_type:
@@ -56,9 +63,19 @@ def format_image(image_mxc, url_str: str='', content_type: str=None, max_image_e
     if max_image_embed > 0:
         width = f'width="{str(max_image_embed)}" '
     if url_str:
-        return f'<a href="{url_str}"><img src="{image_mxc}" alt="{content_type}" {width}/></a>'
+        return f'<a href="{url_str}"><img style="{custom_styles}" src="{image_mxc}" alt="{content_type}" {width}/></a>'
     else:
-        return f'<img src="{image_mxc}" alt="{content_type}" {width}/>'
+        return f'<img style="{custom_styles}" src="{image_mxc}" alt="{content_type}" {width}/>'
+def format_video(video_mxc, url_str: str='', content_type: str=None):
+    if not video_mxc:
+        return None
+    if not content_type:
+        content_type = "Video"
+
+    if url_str:
+        return f'<a href="{url_str}"><video width="320" height="240" controls><source src="{video_mxc}" type="video/mp4"></video></a>'
+    else:
+        return f'<video width="320" height="240" controls><source src="{video_mxc}" type="video/mp4"></video>'
 
 def format_image_width(image_width, max_image_embed: int=300):
     if image_width is None:
@@ -104,6 +121,47 @@ async def matrix_get_image(self, image_url: str, html_custom_headers=None, mime_
         self.log.exception(f"[urlpreview] [utils] Error matrix_get_image client.upload_media: {str(err)}")
         return None
     return mxc
+async def process_video(self, video: str, html_custom_headers=None, content_type: str=None):
+    if not video:
+        return None
+    video_url = urlparse(video)
+    # URL is mxc
+    if video_url.scheme == 'mxc':
+        return video
+    # URL is not mxc
+    # yt-dlp puts out either mp4 or webm
+    if not content_type:
+        content_type = 'video/webm'
+    image_mxc = await matrix_get_youtube_video(
+        self,
+        video_url=video,
+        html_custom_headers=html_custom_headers,
+        mime_type=content_type,
+        filename=content_type.replace('/', '.').replace('jpeg', 'jpg')
+    )
+    return image_mxc
+async def matrix_get_youtube_video(self, video_url: str, html_custom_headers=None, mime_type: str="video/webm", filename: str=""):
+    if not filename:
+        h = hashlib.md5(bytes(dt.now()))
+        filename = str(dt.now() + h.hexdigest())
+    if not video_url:
+        return None
+    try:
+        result = subprocess.run(f'yt-dlp -p {self.YT_DLP_STORAGE_PATH} --no-part {video_url} -o "{filename}"')
+    except Exception as err:
+        self.log.exception(f"[urlpreview] [utils] Error matrix_get_youtube_video http.get: {str(err)}")
+        return None
+
+    og_video = None
+    with open(filename, 'rb') as h:
+        og_video = h.read()
+    try:
+        mxc = await self.client.upload_media(og_video, mime_type=mime_type, filename=filename)
+    except Exception as err:
+        self.log.exception(f"[urlpreview] [utils] Error matrix_get_youtube_video client.upload_media: {str(err)}")
+        return None
+    return mxc
+
 
 def url_check_is_in_range(ip, unsafe_url, ranges):
     for r in ranges: