From 7bfa03a78ca1143bf11bafa89c1b9cf0c08c942b Mon Sep 17 00:00:00 2001
From: Anton Khirnov <anton@khirnov.net>
Date: Wed, 10 Feb 2021 15:20:53 +0100
Subject: Add common message text sanitization.

---
 alot/settings/manager.py | 36 +++++++++++++++++++++++++++++++++++-
 alot/widgets/search.py   |  4 +---
 alot/widgets/thread.py   |  6 ++++++
 3 files changed, 42 insertions(+), 4 deletions(-)

diff --git a/alot/settings/manager.py b/alot/settings/manager.py
index 1be5a0cc..a1e2ab6d 100644
--- a/alot/settings/manager.py
+++ b/alot/settings/manager.py
@@ -2,8 +2,9 @@
 # This file is released under the GNU GPL, version 3 or a later revision.
 # For further details see the COPYING file
 
-from   datetime           import datetime, timedelta
+from   datetime             import datetime, timedelta
 import email
+from   functools            import cached_property
 import importlib.util
 import itertools
 import logging
@@ -596,3 +597,36 @@ class SettingsManager:
             else:
                 rep = _pretty_datetime(d)
         return rep
+
+    def _make_trans_table(self, tab_width):
+        table = {}
+
+        # remove C1 control codes
+        for i in range(0x80, 0xa0):
+            table[i] = None
+
+        # replace C0 control codes with characters for their graphical
+        # representations i.e. "control pictures", which start at U+2400
+        for i in range(ord(' ')):
+            table[i] = 0x2400 + i
+
+        # "delete" character
+        table[0x7f] = 0x2421
+
+        # delete LF
+        table[ord('\r')] = None
+
+        # handle CR normally
+        del table[ord('\n')]
+
+        # expand tabs
+        table[ord('\t')] = ' ' * tab_width
+
+        return table
+
+    @cached_property
+    def sanitize_text_table(self):
+        return self._make_trans_table(self.get('tabwidth'))
+    @cached_property
+    def sanitize_header_table(self):
+        return self._make_trans_table(1)
diff --git a/alot/widgets/search.py b/alot/widgets/search.py
index 12cf84e8..43a71434 100644
--- a/alot/widgets/search.py
+++ b/alot/widgets/search.py
@@ -212,9 +212,7 @@ def prepare_string(partname, thread, query, maxw):
         # get string
         s = content(thread, query)
 
-    # sanitize
-    s = s.replace('\n', ' ')
-    s = s.replace('\r', '')
+    s = s.translate(settings.sanitize_header_table)
 
     # shorten if max width is requested
     if maxw:
diff --git a/alot/widgets/thread.py b/alot/widgets/thread.py
index 7d3bafca..90eeabdf 100644
--- a/alot/widgets/thread.py
+++ b/alot/widgets/thread.py
@@ -40,6 +40,8 @@ class MessageSummaryWidget(urwid.WidgetWrap):
         if date is not None:
             text += " (%s)" % date
 
+        text = text.translate(settings.sanitize_header_table)
+
         cols = [urwid.Text(text)]
 
         if settings.get('msg_summary_hides_threadwide_tags'):
@@ -288,6 +290,8 @@ class _TextPart(_MIMEPartWidget):
 
         self._fold_context = settings.get('thread_fold_context')
 
+        text = text.translate(settings.sanitize_text_table)
+
         # try highlighting with pygments first
         hilit = self._highlight_pygments(text, part, attr_text)
         if hilit is None:
@@ -562,6 +566,8 @@ class HeadersWidget(urwid.WidgetWrap):
 
         widgets = []
         for key, value in headers:
+            value = value.translate(settings.sanitize_header_table)
+
             # TODO even/odd
             keyw   = ('fixed', max_key_len + 1,
                       urwid.Text((self._key_attr, key)))
-- 
cgit v1.2.3