|
0981a08…
|
noreply
|
1 |
"""Apple Notes source connector via osascript (macOS only).""" |
|
0981a08…
|
noreply
|
2 |
|
|
0981a08…
|
noreply
|
3 |
import logging |
|
0981a08…
|
noreply
|
4 |
import re |
|
0981a08…
|
noreply
|
5 |
import subprocess |
|
0981a08…
|
noreply
|
6 |
import sys |
|
0981a08…
|
noreply
|
7 |
from pathlib import Path |
|
0981a08…
|
noreply
|
8 |
from typing import List, Optional |
|
0981a08…
|
noreply
|
9 |
|
|
0981a08…
|
noreply
|
10 |
from video_processor.sources.base import BaseSource, SourceFile |
|
0981a08…
|
noreply
|
11 |
|
|
0981a08…
|
noreply
|
12 |
logger = logging.getLogger(__name__) |
|
0981a08…
|
noreply
|
13 |
|
|
0981a08…
|
noreply
|
14 |
|
|
0981a08…
|
noreply
|
15 |
class AppleNotesSource(BaseSource): |
|
0981a08…
|
noreply
|
16 |
""" |
|
0981a08…
|
noreply
|
17 |
Fetch notes from Apple Notes using osascript (AppleScript). |
|
0981a08…
|
noreply
|
18 |
|
|
0981a08…
|
noreply
|
19 |
Only works on macOS. Requires the Notes app to be available |
|
0981a08…
|
noreply
|
20 |
and permission for osascript to access it. |
|
0981a08…
|
noreply
|
21 |
""" |
|
0981a08…
|
noreply
|
22 |
|
|
0981a08…
|
noreply
|
23 |
def __init__(self, folder: Optional[str] = None): |
|
0981a08…
|
noreply
|
24 |
self.folder = folder |
|
0981a08…
|
noreply
|
25 |
|
|
0981a08…
|
noreply
|
26 |
def authenticate(self) -> bool: |
|
0981a08…
|
noreply
|
27 |
"""Check that we are running on macOS.""" |
|
0981a08…
|
noreply
|
28 |
if sys.platform != "darwin": |
|
0981a08…
|
noreply
|
29 |
logger.error("Apple Notes is only available on macOS (current: %s)", sys.platform) |
|
0981a08…
|
noreply
|
30 |
return False |
|
0981a08…
|
noreply
|
31 |
return True |
|
0981a08…
|
noreply
|
32 |
|
|
0981a08…
|
noreply
|
33 |
def list_videos( |
|
0981a08…
|
noreply
|
34 |
self, |
|
0981a08…
|
noreply
|
35 |
folder_id: Optional[str] = None, |
|
0981a08…
|
noreply
|
36 |
folder_path: Optional[str] = None, |
|
0981a08…
|
noreply
|
37 |
patterns: Optional[List[str]] = None, |
|
0981a08…
|
noreply
|
38 |
) -> List[SourceFile]: |
|
0981a08…
|
noreply
|
39 |
"""List notes from Apple Notes via osascript.""" |
|
0981a08…
|
noreply
|
40 |
if not self.authenticate(): |
|
0981a08…
|
noreply
|
41 |
return [] |
|
0981a08…
|
noreply
|
42 |
|
|
0981a08…
|
noreply
|
43 |
if self.folder: |
|
0981a08…
|
noreply
|
44 |
script = ( |
|
0981a08…
|
noreply
|
45 |
'tell application "Notes"\n' |
|
0981a08…
|
noreply
|
46 |
" set noteList to {}\n" |
|
0981a08…
|
noreply
|
47 |
f" repeat with f in folders of default account\n" |
|
0981a08…
|
noreply
|
48 |
f' if name of f is "{self.folder}" then\n' |
|
0981a08…
|
noreply
|
49 |
" repeat with n in notes of f\n" |
|
0981a08…
|
noreply
|
50 |
' set end of noteList to (id of n) & "|||" & (name of n)\n' |
|
0981a08…
|
noreply
|
51 |
" end repeat\n" |
|
0981a08…
|
noreply
|
52 |
" end if\n" |
|
0981a08…
|
noreply
|
53 |
" end repeat\n" |
|
0981a08…
|
noreply
|
54 |
" return noteList\n" |
|
0981a08…
|
noreply
|
55 |
"end tell" |
|
0981a08…
|
noreply
|
56 |
) |
|
0981a08…
|
noreply
|
57 |
else: |
|
0981a08…
|
noreply
|
58 |
script = ( |
|
0981a08…
|
noreply
|
59 |
'tell application "Notes"\n' |
|
0981a08…
|
noreply
|
60 |
" set noteList to {}\n" |
|
0981a08…
|
noreply
|
61 |
" repeat with n in notes of default account\n" |
|
0981a08…
|
noreply
|
62 |
' set end of noteList to (id of n) & "|||" & (name of n)\n' |
|
0981a08…
|
noreply
|
63 |
" end repeat\n" |
|
0981a08…
|
noreply
|
64 |
" return noteList\n" |
|
0981a08…
|
noreply
|
65 |
"end tell" |
|
0981a08…
|
noreply
|
66 |
) |
|
0981a08…
|
noreply
|
67 |
|
|
0981a08…
|
noreply
|
68 |
try: |
|
0981a08…
|
noreply
|
69 |
result = subprocess.run( |
|
0981a08…
|
noreply
|
70 |
["osascript", "-e", script], |
|
0981a08…
|
noreply
|
71 |
capture_output=True, |
|
0981a08…
|
noreply
|
72 |
text=True, |
|
0981a08…
|
noreply
|
73 |
timeout=30, |
|
0981a08…
|
noreply
|
74 |
) |
|
0981a08…
|
noreply
|
75 |
except FileNotFoundError: |
|
0981a08…
|
noreply
|
76 |
logger.error("osascript not found. Apple Notes requires macOS.") |
|
0981a08…
|
noreply
|
77 |
return [] |
|
0981a08…
|
noreply
|
78 |
except subprocess.TimeoutExpired: |
|
0981a08…
|
noreply
|
79 |
logger.error("osascript timed out while listing notes.") |
|
0981a08…
|
noreply
|
80 |
return [] |
|
0981a08…
|
noreply
|
81 |
|
|
0981a08…
|
noreply
|
82 |
if result.returncode != 0: |
|
0981a08…
|
noreply
|
83 |
logger.error("Failed to list notes: %s", result.stderr.strip()) |
|
0981a08…
|
noreply
|
84 |
return [] |
|
0981a08…
|
noreply
|
85 |
|
|
0981a08…
|
noreply
|
86 |
return self._parse_note_list(result.stdout.strip()) |
|
0981a08…
|
noreply
|
87 |
|
|
0981a08…
|
noreply
|
88 |
def _parse_note_list(self, output: str) -> List[SourceFile]: |
|
0981a08…
|
noreply
|
89 |
"""Parse osascript output into SourceFile objects. |
|
0981a08…
|
noreply
|
90 |
|
|
0981a08…
|
noreply
|
91 |
Expected format: comma-separated items of 'id|||name' pairs. |
|
0981a08…
|
noreply
|
92 |
""" |
|
0981a08…
|
noreply
|
93 |
files: List[SourceFile] = [] |
|
0981a08…
|
noreply
|
94 |
if not output: |
|
0981a08…
|
noreply
|
95 |
return files |
|
0981a08…
|
noreply
|
96 |
|
|
0981a08…
|
noreply
|
97 |
# AppleScript returns a flat comma-separated list |
|
0981a08…
|
noreply
|
98 |
entries = output.split(", ") |
|
0981a08…
|
noreply
|
99 |
for entry in entries: |
|
0981a08…
|
noreply
|
100 |
entry = entry.strip() |
|
0981a08…
|
noreply
|
101 |
if "|||" not in entry: |
|
0981a08…
|
noreply
|
102 |
continue |
|
0981a08…
|
noreply
|
103 |
note_id, _, name = entry.partition("|||") |
|
0981a08…
|
noreply
|
104 |
note_id = note_id.strip() |
|
0981a08…
|
noreply
|
105 |
name = name.strip() |
|
0981a08…
|
noreply
|
106 |
if note_id and name: |
|
0981a08…
|
noreply
|
107 |
files.append( |
|
0981a08…
|
noreply
|
108 |
SourceFile( |
|
0981a08…
|
noreply
|
109 |
name=name, |
|
0981a08…
|
noreply
|
110 |
id=note_id, |
|
0981a08…
|
noreply
|
111 |
mime_type="text/plain", |
|
0981a08…
|
noreply
|
112 |
) |
|
0981a08…
|
noreply
|
113 |
) |
|
0981a08…
|
noreply
|
114 |
|
|
0981a08…
|
noreply
|
115 |
logger.info("Found %d notes", len(files)) |
|
0981a08…
|
noreply
|
116 |
return files |
|
0981a08…
|
noreply
|
117 |
|
|
0981a08…
|
noreply
|
118 |
def download(self, file: SourceFile, destination: Path) -> Path: |
|
0981a08…
|
noreply
|
119 |
"""Download a note's content as plain text.""" |
|
0981a08…
|
noreply
|
120 |
destination = Path(destination) |
|
0981a08…
|
noreply
|
121 |
destination.parent.mkdir(parents=True, exist_ok=True) |
|
0981a08…
|
noreply
|
122 |
|
|
0981a08…
|
noreply
|
123 |
script = ( |
|
0981a08…
|
noreply
|
124 |
'tell application "Notes"\n' |
|
0981a08…
|
noreply
|
125 |
f' set theNote to note id "{file.id}" of default account\n' |
|
0981a08…
|
noreply
|
126 |
" return body of theNote\n" |
|
0981a08…
|
noreply
|
127 |
"end tell" |
|
0981a08…
|
noreply
|
128 |
) |
|
0981a08…
|
noreply
|
129 |
|
|
0981a08…
|
noreply
|
130 |
try: |
|
0981a08…
|
noreply
|
131 |
result = subprocess.run( |
|
0981a08…
|
noreply
|
132 |
["osascript", "-e", script], |
|
0981a08…
|
noreply
|
133 |
capture_output=True, |
|
0981a08…
|
noreply
|
134 |
text=True, |
|
0981a08…
|
noreply
|
135 |
timeout=30, |
|
0981a08…
|
noreply
|
136 |
) |
|
0981a08…
|
noreply
|
137 |
except FileNotFoundError: |
|
0981a08…
|
noreply
|
138 |
raise RuntimeError("osascript not found. Apple Notes requires macOS.") |
|
0981a08…
|
noreply
|
139 |
except subprocess.TimeoutExpired: |
|
0981a08…
|
noreply
|
140 |
raise RuntimeError(f"osascript timed out fetching note {file.id}") |
|
0981a08…
|
noreply
|
141 |
|
|
0981a08…
|
noreply
|
142 |
if result.returncode != 0: |
|
0981a08…
|
noreply
|
143 |
raise RuntimeError(f"Failed to fetch note {file.id}: {result.stderr.strip()}") |
|
0981a08…
|
noreply
|
144 |
|
|
0981a08…
|
noreply
|
145 |
html_body = result.stdout.strip() |
|
0981a08…
|
noreply
|
146 |
text = self._html_to_text(html_body) |
|
0981a08…
|
noreply
|
147 |
|
|
0981a08…
|
noreply
|
148 |
# Prepend title |
|
0981a08…
|
noreply
|
149 |
content = f"# {file.name}\n\n{text}" |
|
0981a08…
|
noreply
|
150 |
destination.write_text(content, encoding="utf-8") |
|
0981a08…
|
noreply
|
151 |
logger.info("Saved Apple Note to %s", destination) |
|
0981a08…
|
noreply
|
152 |
return destination |
|
0981a08…
|
noreply
|
153 |
|
|
0981a08…
|
noreply
|
154 |
@staticmethod |
|
0981a08…
|
noreply
|
155 |
def _html_to_text(html: str) -> str: |
|
0981a08…
|
noreply
|
156 |
"""Strip HTML tags and return plain text. |
|
0981a08…
|
noreply
|
157 |
|
|
0981a08…
|
noreply
|
158 |
Apple Notes returns note bodies as HTML. This uses regex-based |
|
0981a08…
|
noreply
|
159 |
stripping similar to web_source._strip_html_tags. |
|
0981a08…
|
noreply
|
160 |
""" |
|
0981a08…
|
noreply
|
161 |
if not html: |
|
0981a08…
|
noreply
|
162 |
return "" |
|
0981a08…
|
noreply
|
163 |
# Replace <br> variants with newlines |
|
0981a08…
|
noreply
|
164 |
text = re.sub(r"<br\s*/?>", "\n", html, flags=re.IGNORECASE) |
|
0981a08…
|
noreply
|
165 |
# Replace block-level closing tags with newlines |
|
0981a08…
|
noreply
|
166 |
text = re.sub(r"</(?:p|div|li|tr|h[1-6])>", "\n", text, flags=re.IGNORECASE) |
|
0981a08…
|
noreply
|
167 |
# Remove all remaining tags |
|
0981a08…
|
noreply
|
168 |
text = re.sub(r"<[^>]+>", "", text) |
|
0981a08…
|
noreply
|
169 |
# Decode common HTML entities |
|
0981a08…
|
noreply
|
170 |
text = text.replace("&", "&") |
|
0981a08…
|
noreply
|
171 |
text = text.replace("<", "<") |
|
0981a08…
|
noreply
|
172 |
text = text.replace(">", ">") |
|
0981a08…
|
noreply
|
173 |
text = text.replace(""", '"') |
|
0981a08…
|
noreply
|
174 |
text = text.replace("'", "'") |
|
0981a08…
|
noreply
|
175 |
text = text.replace(" ", " ") |
|
0981a08…
|
noreply
|
176 |
# Collapse excessive blank lines |
|
0981a08…
|
noreply
|
177 |
text = re.sub(r"\n{3,}", "\n\n", text) |
|
0981a08…
|
noreply
|
178 |
return text.strip() |