Add privacy mode that doesn't leak addresses

In privacy mode, any addresses not viewable to an anonymous user will
not be included into outgoing mail.

Signed-off-by: Konstantin Ryabitsev <konstantin@linuxfoundation.org>

3 files changed, 39 insertions, 15 deletions

diff --git a/default.config.toml b/default.config.toml
index a490f19..993b9af 100644
--- a/default.config.toml
+++ b/default.config.toml
@@ -35,6 +35,7 @@ pi_must_bz_groups = ['editbugs']
 pi_url = 'https://lore.kernel.org/all/'
 pi_assign_regex = '^bugbot assign to (\S+)'
 bz_new_bugs_quicksearch = 'OPEN flag:bugbot+'
+bz_privacy_mode = true
 alwayscc = ['bugs@lists.linux.dev']
 
 [templates]
diff --git a/peebz/__init__.py b/peebz/__init__.py
index 05c48e2..78d83b8 100644
--- a/peebz/__init__.py
+++ b/peebz/__init__.py
@@ -381,7 +381,7 @@ def bz_get_changed_bugs(since: str, include_untracked: bool = False) -> List:
     logger.debug('Querying for changed bugs since %s', since)
     params = {
         'chfieldfrom': since,
-        'include_fields': 'id,summary',
+        'include_fields': 'id,summary,product,component',
     }
     rdata = bz_rest('bug', params=params)
     if include_untracked:
@@ -406,7 +406,7 @@ def bz_get_changed_bugs(since: str, include_untracked: bool = False) -> List:
 
 def bz_quicksearch_bugs(query: str) -> Dict:
     params = {
-        'include_fields': 'id,summary',
+        'include_fields': 'id,summary,product,component',
         'quicksearch': query,
     }
     return bz_rest('bug', params=params)
@@ -414,7 +414,7 @@ def bz_quicksearch_bugs(query: str) -> Dict:
 
 def bz_get_query_bugs(params: Dict, exclude: Set[int]) -> List[int]:
     if 'include_fields' not in params:
-        params['include_fields'] = 'id,summary'
+        params['include_fields'] = 'id,summary,product,component'
     rdata = bz_rest('bug', params=params)
     bids = list()
     for bdata in rdata['bugs']:
@@ -755,12 +755,19 @@ def get_bug_recipients(bid: int) -> Set[str]:
         logger.debug('No in-database recipients for bid=%s', bid)
     # Now get all bug cc recipients
     bdata = bz_get_bug(bid, resolve_dupes=True)
-    bugr = set(bdata['cc'])
-    bugr.add(bdata['assigned_to'])
-    bugr.add(bdata['creator'])
-    bugr.update(get_recipients_by_product_component(bdata['product'], bdata['component']))
-    allrecip.update(bugr)
+    product = bdata['product']
+    component = bdata['component']
     config = get_config()
+    cconf = get_component_config(product, component)
+    bugr = get_recipients_by_product_component(product, component)
+    if config['bugzilla'].get('privacy_mode', False) or cconf.get('bz_privacy_mode', False):
+        logger.debug('Privacy mode on, not adding addresses from the bug')
+    else:
+        bugr.update(bdata['cc'])
+        bugr.add(bdata['assigned_to'])
+        bugr.add(bdata['creator'])
+
+    allrecip.update(bugr)
     subsystem_cf = config['bugzilla'].get('subsystem_cf')
     if subsystem_cf:
         subr = get_recipients_by_subsystem(bdata[subsystem_cf])
diff --git a/peebz/bz2pi.py b/peebz/bz2pi.py
index b78a648..1726d30 100644
--- a/peebz/bz2pi.py
+++ b/peebz/bz2pi.py
@@ -16,7 +16,7 @@ from fnmatch import fnmatch
 logger = peebz.logger
 
 
-def process_new_comments(bid: int, dry_run: bool = False):
+def process_new_comments(bid: int, privacy_mode: bool = False, dry_run: bool = False):
     config = peebz.get_config()
     cdatas = peebz.bz_get_newest_comments_for_bid(bid)
     for cdata in cdatas:
@@ -47,11 +47,16 @@ def process_new_comments(bid: int, dry_run: bool = False):
             continue
         clines = cdata['text'].strip().splitlines()
         inre_cid = None
+        comment_author = cdata['creator']
+        if privacy_mode:
+            comment_author = comment_author.split('@')[0]
+            logger.debug('Privacy mode, comment_author=%s', comment_author)
+
         bodyvals = {
             'bzname': config['bugzilla'].get('name'),
             'bug_url': config['bugzilla'].get('bugmask', '').format(bug_id=bid),
             'comment_url': config['bugzilla'].get('bugmask', '').format(bug_id=bid) + f"#c{cdata['count']}",
-            'comment_author': cdata['creator'],
+            'comment_author': comment_author,
         }
         if cdata['attachment_id']:
             logger.info('Processing new attachment for bug_id=%s, comment_id=%s', bid, cid)
@@ -80,14 +85,16 @@ def process_new_comments(bid: int, dry_run: bool = False):
             bodytpt = peebz.get_template_by_bid('new_comment_notify', bid)
 
         msg = email.message.EmailMessage()
-        msg['Reply-To'] = b4.format_addrs([('', cdata['creator'])])
+        if not privacy_mode:
+            msg['Reply-To'] = b4.format_addrs([('', cdata['creator'])])
         body = bodytpt.safe_substitute(bodyvals)
         body = peebz.add_bot_signature(body)
         msg.set_payload(body, charset='utf-8')
         msgid = peebz.notify_bug(bid, cid, msg, inre_cid=inre_cid, dry_run=dry_run)
         if msgid and not dry_run:
             peebz.db_store_msgid_bid_cid(msgid, bid, cid)
-            peebz.db_store_recipients(bid, {cdata['creator']})
+            if not privacy_mode:
+                peebz.db_store_recipients(bid, {cdata['creator']})
             # TODO: This assumes that comments are always in incremental order
             lastcheck = cdata['creation_time'].replace('T', ' ').rstrip('Z')
             peebz.db_store_notify_last_check(bid, lastcheck)
@@ -109,17 +116,22 @@ def main(cmdargs: argparse.Namespace) -> None:
     logger.info('Getting a list of changed bugs since %s', since)
     buglist = peebz.bz_get_changed_bugs(since)
     seen = set()
+    config = peebz.get_config()
     if buglist:
         for bdata in buglist:
             logger.debug('Looking at %s: %s', bdata['id'], bdata['summary'])
             bid = bdata['id']
+            cconf = peebz.get_component_config(bdata['product'], bdata['component'])
+            if config['bugzilla'].get('privacy_mode', False) or cconf.get('bz_privacy_mode', False):
+                privacy_mode = True
+            else:
+                privacy_mode = False
+            process_new_comments(bid, privacy_mode=privacy_mode, dry_run=cmdargs.dry_run)
             seen.add(bid)
-            process_new_comments(bid, dry_run=cmdargs.dry_run)
     else:
         logger.info('No changes to any tracked bugs')
 
     # Now go by product/component and handle new bug queries if defined
-    config = peebz.get_config()
     for bz_product, bz_components in config['components'].items():
         for bz_component in bz_components.keys():
             cconf = peebz.get_component_config(bz_product, bz_component)
@@ -136,11 +148,15 @@ def main(cmdargs: argparse.Namespace) -> None:
                 'quicksearch': qs,
             }
             buglist = peebz.bz_get_query_bugs(params, exclude=seen)
+            if config['bugzilla'].get('privacy_mode', False) or cconf.get('bz_privacy_mode', False):
+                privacy_mode = True
+            else:
+                privacy_mode = False
             if buglist:
                 logger.info('Processing %s matching quicksearch bugs', len(buglist))
                 for bid in buglist:
                     seen.add(bid)
-                    process_new_comments(bid, dry_run=cmdargs.dry_run)
+                    process_new_comments(bid, privacy_mode=privacy_mode, dry_run=cmdargs.dry_run)
             else:
                 logger.info('No changed bugs matching these parameters.')