From 9d50287dab79fb10a3a9a9d8dae39cbc3b045e52 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Hugo=20H=C3=B6rnquist?= <hugo@lysator.liu.se>
Date: Thu, 1 Dec 2022 05:10:14 +0100
Subject: Fix scanning of maildirs.

---
 mu4web/main.py | 107 ++++++++++++++++++++++++++++++++++++++++-----------------
 setup.cfg      |   1 +
 2 files changed, 77 insertions(+), 31 deletions(-)

diff --git a/mu4web/main.py b/mu4web/main.py
index 68bf144..4d39c1a 100644
--- a/mu4web/main.py
+++ b/mu4web/main.py
@@ -25,6 +25,7 @@ from user.local import LocalUser
 from user.pam import PamUser
 
 import subprocess
+from dataclasses import dataclass
 
 import flask
 from flask import (
@@ -37,6 +38,11 @@ from flask import (
     get_flashed_messages
 )
 
+try:
+    from natsort import natsorted
+except ModuleNotFoundError:
+    natsorted = sorted
+
 #
 # A few operations depend on the index of attachements. These index
 # are all pre-order traversal indexes of the attachement tree, which
@@ -327,53 +333,92 @@ def search_page(q: str, by: Optional[mu.Sortfield],
                                      body=main_body))
 
 
-def find_maildirs(basedir) -> dict[str, list[str]]:
+@dataclass
+class Leaf:
+    name: str
+
+@dataclass
+class Node:
+    name: str
+    children: list[Union[Leaf, 'Node']]
+
+
+def build_tree(items: list[list[str]]) -> Node:
+    groups: dict[str, list[list[str]]] = {}
+    direct: list[Leaf] = []
+    for key, *rest in items:
+        if rest:
+            groups.setdefault(key, []).append(rest)
+        else:
+            direct.append(Leaf(key))
+
+    node = Node('root', [])
+    for key, values in groups.items():
+        next = build_tree(values)
+        next.name = key
+        node.children.append(next)
+    node.children.extend(direct)
+    return node
+
+
+def find(basedir, **flags) -> list[bytes]:
+    cmdline = ['find', basedir]
+    for key, value in flags.items():
+        cmdline += [f'-{key}', value]
+    cmdline.append('-print0')
+
+    cmd = subprocess.run(cmdline, capture_output=True)
+    return cmd.stdout.split(b'\0')[:-1]
+
+
+def find_maildirs(basedir) -> Node:
     """
     Find all maildirs located under basedir.
 
     A maildir is defined as any directory which contains a `cur`
     directory.
-
-    TODO Currently all returns are grouped by their first component, which
-    fails for directories directly in the root.
     """
-    cmd = subprocess.run(['find', basedir,
-                          '-type', 'd',
-                          '-name', 'cur',
-                          '-print0'],
-                         capture_output=True)
-    groups: dict[str, list[str]] = {}
-    # Group by first component
-    for entry in cmd.stdout.split(b'\0'):
-        dir = os.path.split(entry)[0][len(basedir) + 1:].decode('UTF-8')
-        if not dir:
-            continue
-        parts = dir.split(os.path.sep)
-        groups.setdefault(parts[0], []).append(os.path.sep.join(parts[1:]))
-    return groups
+    basedir = basedir.rstrip('/')
+    files = find(basedir, type='d', name='cur')
+    # + 1 removes leading slash
+    # - 4 removes '/cur'
+    dirs = [entry[len(basedir)+1:-4].decode('UTF-8').split(os.path.sep)
+            for entry in files]
+
+    return build_tree(dirs)
+
+
+def serialize_maildir(maildir: Node, path: list[str] = []) -> HTML:
+    """Build a (recursive) list from a maildir node."""
+    entries: list[HTML] = []
+    for node in natsorted(maildir.children, key=lambda n: n.name):
+        if isinstance(node, Leaf):
+            parts = '/'.join(path + [node.name])
+            url = 'search?' + urlencode({'q': f'maildir:"/{parts}"'})
+            entry = ('li', ('a', {'href': url},
+                            node.name or ('i', 'root')))
+        else:
+            entry = ('li',
+                     ('details',
+                      ('summary', node.name),
+                       serialize_maildir(node, path + [node.name])))
+        entries.append(entry)
 
+    return ('ul', *entries)
 
 def index_page():
     data = mu.info()
-    groups = find_maildirs(data['maildir'])
-
-    entries = []
-
-    for key, values in sorted(groups.items(), key=lambda p: p[0]):
-        entries.append(('li',
-                        ('details',
-                         ('summary', key),
-                         ('ul',
-                          [('li',
-                            ('a', {'href': 'search?' + urlencode({'q': f'maildir:"/{key}/{v}"'})}, v))
-                           for v in sorted(values)]))))
+    maildirs = find_maildirs(data['maildir'] + '/')
+
+    entries = serialize_maildir(maildirs)
+
     rows = []
     for key, value in data.items():
         rows.append(('tr',
                      ('td', key),
                      ('td', value)))
     body = [('div', ('table', ('tbody', rows))),
-            ('div', ('ul', entries)),
+            ('div', entries),
             ]
 
     return render_document(page_base(title='Mail index',
diff --git a/setup.cfg b/setup.cfg
index bf73722..aba19bf 100644
--- a/setup.cfg
+++ b/setup.cfg
@@ -15,6 +15,7 @@ install_requires =
 	flask >= 2.2.2
 	flask-login >= 0.6
 	urllib3 >= 1.26
+# optionally natsort >= 8.2
 setup_requires =
 	setuptools
 py_modules = mu4web
-- 
cgit v1.2.3