From 6a40e7d7f22db794ed97ac1768ab6e5f29aa1c12 Mon Sep 17 00:00:00 2001
From: Thomas Sileo <t@a4.io>
Date: Fri, 1 Nov 2019 16:05:48 +0100
Subject: [PATCH] Add support for inline images

---
 blueprints/tasks.py       | 12 +++++++++++-
 templates/utils.html      |  4 ++--
 utils/template_filters.py | 28 +++++++++++++++++++++++++++-
 3 files changed, 40 insertions(+), 4 deletions(-)

diff --git a/blueprints/tasks.py b/blueprints/tasks.py
index cc40f2a..e47e218 100644
--- a/blueprints/tasks.py
+++ b/blueprints/tasks.py
@@ -7,6 +7,7 @@ from typing import Dict
 
 import flask
 import requests
+from bs4 import BeautifulSoup
 from flask import current_app as app
 from little_boxes import activitypub as ap
 from little_boxes.activitypub import _to_list
@@ -271,7 +272,9 @@ def select_video_to_cache(links):
     return videos[0]
 
 
-@blueprint.route("/task/cache_attachments", methods=["POST"])
+@blueprint.route(
+    "/task/cache_attachments", methods=["POST"]
+)  # noqa: C910  # too complex
 def task_cache_attachments() -> _Response:
     task = p.parse(flask.request)
     app.logger.info(f"task={task!r}")
@@ -286,6 +289,13 @@ def task_cache_attachments() -> _Response:
         else:
             obj = activity
 
+        if obj.content:
+            content_html = BeautifulSoup(obj.content, "html5lib")
+            for img in content_html.find_all("img"):
+                src = img.attrs.get("src")
+                if src:
+                    Tasks.cache_attachment({"url": src}, iri)
+
         if obj.has_type(ap.ActivityType.VIDEO):
             if isinstance(obj.url, list):
                 # TODO: filter only videogt
diff --git a/templates/utils.html b/templates/utils.html
index e0b774e..928466a 100644
--- a/templates/utils.html
+++ b/templates/utils.html
@@ -137,7 +137,7 @@
     {% if obj | has_type(['Article', 'Page']) %}
         {{ obj.name }} <a href="{{ obj | url_or_id | get_url }}">{{ obj | url_or_id | get_url }}</a>
     {% elif obj | has_type('Question') %}
-        {{ obj.content | clean | replace_custom_emojis(obj) | code_highlight | safe }}
+        {{ obj.content | clean | update_inline_imgs | replace_custom_emojis(obj) | code_highlight | safe }}
 
         
         <ul style="list-style:none;padding:0;">
@@ -206,7 +206,7 @@
 
 
     {% else %}
-	{{ obj.content | clean | replace_custom_emojis(obj) | code_highlight | safe }}
+	{{ obj.content | clean | update_inline_imgs | replace_custom_emojis(obj) | code_highlight | safe }}
     {% endif %}
 
     {% if obj | has_place %}
diff --git a/utils/template_filters.py b/utils/template_filters.py
index 6cf6d75..b2fb0eb 100644
--- a/utils/template_filters.py
+++ b/utils/template_filters.py
@@ -9,6 +9,7 @@ import emoji_unicode
 import flask
 import html2text
 import timeago
+from bs4 import BeautifulSoup
 from cachetools import LRUCache
 from little_boxes import activitypub as ap
 from little_boxes.activitypub import _to_list
@@ -104,8 +105,16 @@ ALLOWED_TAGS = [
     "tfoot",
     "colgroup",
     "caption",
+    "img",
 ]
 
+ALLOWED_ATTRIBUTES = {
+    "a": ["href", "title"],
+    "abbr": ["title"],
+    "acronym": ["title"],
+    "img": ["src", "alt", "title"],
+}
+
 
 @filters.app_template_filter()
 def replace_custom_emojis(content, note):
@@ -126,7 +135,9 @@ def replace_custom_emojis(content, note):
 
 def clean_html(html):
     try:
-        return bleach.clean(html, tags=ALLOWED_TAGS, strip=True)
+        return bleach.clean(
+            html, tags=ALLOWED_TAGS, attributes=ALLOWED_ATTRIBUTES, strip=True
+        )
     except Exception:
         return "failed to clean HTML"
 
@@ -333,6 +344,21 @@ def get_attachment_url(url, size):
     return _get_file_url(url, size, Kind.ATTACHMENT)
 
 
+@filters.app_template_filter()
+def update_inline_imgs(content):
+    soup = BeautifulSoup(content)
+    imgs = soup.find_all("img")
+    if not imgs:
+        return content
+    for img in imgs:
+        if not img.attrs.get("src"):
+            continue
+
+        img.attrs["src"] = _get_file_url(img.attrs["src"], 720, Kind.ATTACHMENT)
+
+    return soup.find("body").decode_contents()
+
+
 @filters.app_template_filter()
 def get_video_url(url):
     if isinstance(url, list):