Python 3.10 (#959)

* Dockerfile & requirements

* Fix tests

* Treat all AssertionErrors as HTMLParseErrors

Dockerfile

@@ -1,7 +1,7 @@
 # --- Stage 0 --- #
 # This first stage is responsible for installing any dependencies the app needs
 # to run, and updating any base dependencies.
-FROM ubuntu:20.04 AS stage_0
+FROM ubuntu:22.04 AS stage_0
 
 RUN groupadd -g 5000 sync-engine \
   && useradd -d /home/sync-engine -m -u 5000 -g 5000 sync-engine
@@ -24,7 +24,7 @@ RUN echo $BUILD_WEEK && apt-get update \
     gpg \
     gpg-agent \
     dirmngr \
-    python3.9 \
+    python3.10 \
     gettext-base \
     libmysqlclient21 \
     mysql-client \
@@ -54,16 +54,16 @@ RUN apt-get update \
     gcc \
     git \
     pkg-config \
-    python3.9-dev \
+    python3.10-dev \
     python3-pip \
     libmysqlclient-dev \
   && rm -rf /var/lib/apt/lists/*
 
 COPY /requirements/ /requirements/
-RUN python3.9 -m pip install pip==24.0 virtualenv==20.25.1 && \
-  python3.9 -m virtualenv /opt/venv && \
-  /opt/venv/bin/python3.9 -m pip install --no-cache --no-deps -r /requirements/prod.txt -r /requirements/test.txt && \
-  /opt/venv/bin/python3.9 -m pip check
+RUN python3.10 -m pip install pip==24.3.1 virtualenv==20.27.0 && \
+  python3.10 -m virtualenv /opt/venv && \
+  /opt/venv/bin/python3.10 -m pip install --no-cache --no-deps -r /requirements/prod.txt -r /requirements/test.txt && \
+  /opt/venv/bin/python3.10 -m pip check
 
 
 # --- Stage 2 --- #

inbox/util/html.py

@@ -9,7 +9,7 @@
     from html import escape as html_escape
 
 
-class HTMLParseError(Exception):
+class HTMLParseError(ValueError):
     pass
 
 
@@ -42,11 +42,6 @@ def handle_data(self, d):
         if not self.strip_tag_contents_mode:
             self.fed.append(d)
 
-    if (3,) <= sys.version_info < (3, 10):
-
-        def error(self, message):
-            raise HTMLParseError(message)
-
     def handle_entityref(self, d):
         try:
             val = chr(name2codepoint[d])
@@ -68,7 +63,19 @@ def strip_tags(html: str) -> str:
     You are responsible for handling it in the calling function.
     """
     s = HTMLTagStripper()
-    s.feed(html)
+
+    # `HTMLParser.feed()` raises `AssertionError` if fed invalid HTML.
+    # See the source at https://github.com/python/cpython/blob/main/Lib/html/parser.py
+    # and https://github.com/python/cpython/blob/main/Lib/_markupbase.py#L30.
+    # We want to catch this exception and raise a more informative exception to
+    # tell it apart from general `AssertionError`s that normally mean a programmer error.
+    # Those are later caught in `Message.calculate_html_snippet` so it can recover from
+    # invalid HTML.
+    try:
+        s.feed(html)
+    except AssertionError as e:
+        raise HTMLParseError(*e.args) from e
+
     return s.get_data()