Commit 30d02b72f6a5b8dd8b72cea593bdc9c883afea1e | snowbound

corpus/native-external-assets/README.md created+16

...	@@ -0,0 +1,16 @@
1	# External payload fixture
2
3	OneNote 2010 authored an empty attachment and a 1 KiB attachment with
4	`tools/native/external-assets.ps1`. The retained native section is in `native/`.
5	The fixture generator converts all file-data declarations into the documented
6	external representation while retaining their identities and exact payloads.
7	The `read/` capture is an independent fresh-cache OneNote reopen; `payloads.json`
8	contains the native attachment hashes. `provenance.json` records source hashes.
9
10	```sh
11	cargo run -p onestore-notebook --example externalize_fixture -- corpus/native-external-assets/native/synthetic.one /tmp/external-notebook
12	python3 tools/native_runner.py /tmp/external-notebook /tmp/external-native --expected-pages 3 --collect-notebook
13	```
14
15	The generator creates a new destination and requires reserved fragment space.
16	It is a fixture converter, not a concurrent publication API.

corpus/native-external-assets/native/synthetic.one created

Binary files /dev/null and b/corpus/native-external-assets/native/synthetic.one differ

corpus/native-external-assets/notebook/Open Notebook.onetoc2 created

Binary files /dev/null and b/corpus/native-external-assets/notebook/Open Notebook.onetoc2 differ

corpus/native-external-assets/notebook/synthetic.one created

Binary files /dev/null and b/corpus/native-external-assets/notebook/synthetic.one differ

corpus/native-external-assets/notebook/synthetic_onefiles/04DC5FD1-10FD-4404-A37C-D035399F9D9E.onebin created

Binary files /dev/null and b/corpus/native-external-assets/notebook/synthetic_onefiles/04DC5FD1-10FD-4404-A37C-D035399F9D9E.onebin differ

corpus/native-external-assets/notebook/synthetic_onefiles/15575137-1653-47BC-9C05-78420499168B.onebin created

Binary files /dev/null and b/corpus/native-external-assets/notebook/synthetic_onefiles/15575137-1653-47BC-9C05-78420499168B.onebin differ

corpus/native-external-assets/notebook/synthetic_onefiles/C5EABC1E-5091-4685-BAC1-2C785D6046DB.onebin created

corpus/native-external-assets/provenance.json created+23

...	@@ -0,0 +1,23 @@
1	{
2	"native_author": "tools/native/external-assets.ps1",
3	"generator": "crates/onestore-notebook/examples/externalize_fixture.rs",
4	"native_capture": "evidence/m10/external-assets-small-native",
5	"cold_capture": "evidence/m10/external-assets-small-cold",
6	"source_sha256": {
7	"native/synthetic.one": "3a4dcb0b6f9e46044a3cdc663ae505878911c2a7255a7787c55ae2c71a7f4fd6",
8	"read/page-002.xml": "4104b7f1b2a8bcc98c5ce73c335d365ae67d565a72f9da2ba78304dd7226944f",
9	"read/page-001.xml": "8c1a503344f3d6893a3df1a3ff5d51d591c0a61e04101c360aa1546d386f5a15",
10	"read/page-000.xml": "6796df2c0594a3263416dfbd1210ec412e0694d30115869f3ceab768c95a3e32",
11	"read/hierarchy.xml": "4a1207f82678cd2020ee175979074601bd9f85c102bd68e5df34c54e37616574",
12	"read/payloads.json": "0b31ae5e576ba7ee5eab7c9a62d8f673659a76865ad07f9e75f45458c06a7e65",
13	"read/environment.json": "1cd7ab391d85fe504344cd63bd786f969770ce09aef0008623eaf0abb880e655",
14	"notebook/Open Notebook.onetoc2": "8fec3cc6f72a9e0fff45ce604f62c184b0bb63d077018f90086b31396c8500b0",
15	"notebook/synthetic.one": "7343061e2a44c67b5f306c747e707fc1b99d126c51db510cc010885c2b016ca2",
16	"notebook/synthetic_onefiles/15575137-1653-47BC-9C05-78420499168B.onebin": "38b0469c7f0cddba0fa7739b44061117b2621bfe477348150cfc98396f41daf1",
17	"notebook/synthetic_onefiles/C5EABC1E-5091-4685-BAC1-2C785D6046DB.onebin": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
18	"notebook/synthetic_onefiles/04DC5FD1-10FD-4404-A37C-D035399F9D9E.onebin": "2bce1ba628720664be4b9fdd77aae0678e5f0f3f02fc6ff641ec879094f6a404"
19	},
20	"cold_teardown": {
21	"absent": true
22	}
23	}
	\ No newline at end of file

corpus/native-external-assets/read/environment.json created+7

...	@@ -0,0 +1,7 @@
1	{
2	"powershell": "5.1.14409.1005",
3	"schema": "xs2010",
4	"hostname": "ONE-M6-3BB259AD",
5	"cold": true,
6	"onenote": "14.0.4763.1000"
7	}

corpus/native-external-assets/read/hierarchy.xml created+2

...	@@ -0,0 +1,2 @@
1	<?xml version="1.0"?>
2	<one:Notebook xmlns:one="http://schemas.microsoft.com/office/onenote/2010/onenote" name="notebook" nickname="notebook" ID="{0EC3B20E-242C-4796-9C75-FA9263E4B53C}{1}{B0}" path="C:\one-tests\runs\capture\notebook" lastModifiedTime="2026-09-08T01:58:45.000Z" color="#B49EDE"><one:Section name="synthetic" ID="{6936E4B1-EE51-08FC-24EA-C42E76D8CE85}{1}{B0}" path="C:\one-tests\runs\capture\notebook\synthetic.one" lastModifiedTime="2026-09-08T01:58:45.000Z" color="#8AA8E4"><one:Page ID="{0895AAA7-E512-03D9-0023-94EDB3ECC902}{1}{B0}" name="Native before 🦀" dateTime="2026-09-07T23:43:30.000Z" lastModifiedTime="2026-09-08T01:58:45.000Z" pageLevel="1"/><one:Page ID="{2001177A-A50B-470C-A3B4-C397629432DC}{1}{B0}" name="Attachment 0" dateTime="2026-09-08T01:55:22.000Z" lastModifiedTime="2026-09-08T01:58:45.000Z" pageLevel="1"/><one:Page ID="{5B886F89-E2CC-4184-AF00-42A8CB7EEDBF}{1}{B0}" name="Attachment 1024" dateTime="2026-09-08T01:55:23.000Z" lastModifiedTime="2026-09-08T01:58:45.000Z" pageLevel="1"/></one:Section></one:Notebook>

corpus/native-external-assets/read/page-000.xml created+2

...	@@ -0,0 +1,2 @@
1	<?xml version="1.0"?>
2	<one:Page xmlns:one="http://schemas.microsoft.com/office/onenote/2010/onenote" ID="{0895AAA7-E512-03D9-0023-94EDB3ECC902}{1}{B0}" name="Native before 🦀" dateTime="2026-09-07T23:43:30.000Z" lastModifiedTime="2026-09-08T01:58:45.000Z" pageLevel="1" style="font-family:Calibri;font-size:11.0pt" lang="en-US"><one:PageSettings RTL="false" color="automatic"><one:PageSize><one:Automatic/></one:PageSize><one:RuleLines visible="false"/></one:PageSettings><one:Outline author="Fixture" lastModifiedBy="Fixture" lastModifiedTime="2026-09-07T23:43:30.000Z" objectID="{6936E4B1-EE51-08FC-24EA-C42E76D8CE85}{23}{B0}"><one:Position x="36.0" y="36.0" z="0"/><one:Size width="136.0371551513672" height="13.4277172088623"/><one:Indents><one:Indent level="0" indent="0.0"/></one:Indents><one:OEChildren><one:OE creationTime="2026-09-07T23:43:30.000Z" lastModifiedTime="2026-09-07T23:43:30.000Z" objectID="{6936E4B1-EE51-08FC-24EA-C42E76D8CE85}{24}{B0}" alignment="left"><one:T><![CDATA[Native before 🦀]]></one:T></one:OE></one:OEChildren></one:Outline></one:Page>

corpus/native-external-assets/read/page-001.xml created+2

...	@@ -0,0 +1,2 @@
1	<?xml version="1.0"?>
2	<one:Page xmlns:one="http://schemas.microsoft.com/office/onenote/2010/onenote" ID="{2001177A-A50B-470C-A3B4-C397629432DC}{1}{B0}" name="Attachment 0" dateTime="2026-09-08T01:55:22.000Z" lastModifiedTime="2026-09-08T01:58:45.000Z" pageLevel="1"><one:QuickStyleDef index="0" name="p" fontColor="automatic" highlightColor="automatic" font="Calibri" fontSize="11.0" spaceBefore="0.0" spaceAfter="0.0"/><one:PageSettings RTL="false" color="automatic"><one:PageSize><one:Automatic/></one:PageSize><one:RuleLines visible="false"/></one:PageSettings><one:Title lang="en-US"><one:OE author="virtual" authorInitials="V" lastModifiedBy="virtual" lastModifiedByInitials="V" creationTime="2026-09-08T01:55:23.000Z" lastModifiedTime="2026-09-08T01:55:23.000Z" objectID="{068F783A-8A63-4688-A2AC-32EB81A6EA31}{33}{B0}" alignment="left" quickStyleIndex="0" style="font-family:Calibri;font-size:17.0pt"><one:T><![CDATA[Attachment 0]]></one:T></one:OE></one:Title><one:Outline author="virtual" authorInitials="V" lastModifiedBy="virtual" lastModifiedByInitials="V" lastModifiedTime="2026-09-08T01:55:23.000Z" objectID="{068F783A-8A63-4688-A2AC-32EB81A6EA31}{39}{B0}"><one:Position x="36.0" y="86.4000015258789" z="0"/><one:Size width="72.0" height="63.0"/><one:OEChildren><one:OE creationTime="2026-09-08T01:55:23.000Z" lastModifiedTime="2026-09-08T01:55:23.000Z" objectID="{068F783A-8A63-4688-A2AC-32EB81A6EA31}{40}{B0}" alignment="left"><one:InsertedFile pathCache="C:\Users\clover\AppData\Local\Microsoft\OneNote\14.0\OneNoteOfflineCache_Files\c5eabc1e-5091-4685-bac1-2c785d6046db.bin" pathSource="C:\one-tests\runs\capture\inputs\0.bin" preferredName="0.bin"/></one:OE></one:OEChildren></one:Outline></one:Page>

corpus/native-external-assets/read/page-002.xml created+2

...	@@ -0,0 +1,2 @@
1	<?xml version="1.0"?>
2	<one:Page xmlns:one="http://schemas.microsoft.com/office/onenote/2010/onenote" ID="{5B886F89-E2CC-4184-AF00-42A8CB7EEDBF}{1}{B0}" name="Attachment 1024" dateTime="2026-09-08T01:55:23.000Z" lastModifiedTime="2026-09-08T01:58:45.000Z" pageLevel="1"><one:QuickStyleDef index="0" name="p" fontColor="automatic" highlightColor="automatic" font="Calibri" fontSize="11.0" spaceBefore="0.0" spaceAfter="0.0"/><one:PageSettings RTL="false" color="automatic"><one:PageSize><one:Automatic/></one:PageSize><one:RuleLines visible="false"/></one:PageSettings><one:Title lang="en-US"><one:OE author="virtual" authorInitials="V" lastModifiedBy="virtual" lastModifiedByInitials="V" creationTime="2026-09-08T01:55:23.000Z" lastModifiedTime="2026-09-08T01:55:23.000Z" objectID="{F876120D-4126-473B-9D5F-F961FAF66D65}{33}{B0}" alignment="left" quickStyleIndex="0" style="font-family:Calibri;font-size:17.0pt"><one:T><![CDATA[Attachment 1024]]></one:T></one:OE></one:Title><one:Outline author="virtual" authorInitials="V" lastModifiedBy="virtual" lastModifiedByInitials="V" lastModifiedTime="2026-09-08T01:55:23.000Z" objectID="{F876120D-4126-473B-9D5F-F961FAF66D65}{39}{B0}"><one:Position x="36.0" y="86.4000015258789" z="0"/><one:Size width="72.0" height="63.0"/><one:OEChildren><one:OE creationTime="2026-09-08T01:55:23.000Z" lastModifiedTime="2026-09-08T01:55:23.000Z" objectID="{F876120D-4126-473B-9D5F-F961FAF66D65}{40}{B0}" alignment="left"><one:InsertedFile pathCache="C:\Users\clover\AppData\Local\Microsoft\OneNote\14.0\OneNoteOfflineCache_Files\04dc5fd1-10fd-4404-a37c-d035399f9d9e.bin" pathSource="C:\one-tests\runs\capture\inputs\1024.bin" preferredName="1024.bin"/></one:OE></one:OEChildren></one:Outline></one:Page>

corpus/native-external-assets/read/payloads.json created+18

...	@@ -0,0 +1,18 @@
1	[
2	{
3	"sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
4	"name": "0.bin",
5	"object": "{068F783A-8A63-4688-A2AC-32EB81A6EA31}{40}{B0}",
6	"kind": "InsertedFile",
7	"page": "{2001177A-A50B-470C-A3B4-C397629432DC}{1}{B0}",
8	"bytes": 0
9	},
10	{
11	"sha256": "2bce1ba628720664be4b9fdd77aae0678e5f0f3f02fc6ff641ec879094f6a404",
12	"name": "1024.bin",
13	"object": "{F876120D-4126-473B-9D5F-F961FAF66D65}{40}{B0}",
14	"kind": "InsertedFile",
15	"page": "{5B886F89-E2CC-4184-AF00-42A8CB7EEDBF}{1}{B0}",
16	"bytes": 1024
17	}
18	]
	\ No newline at end of file

crates/onestore-notebook/README.md+7

crates/onestore-notebook/examples/document.rs created+115

...	@@ -0,0 +1,115 @@
1	use onestore::{
2	FileDataReference, RevisionIndex, Store,
3	document::{Document, Kind},
4	};
5	use std::{
6	borrow::Cow,
7	collections::BTreeSet,
8	env, fs,
9	io::{self, BufWriter, Write},
10	path::{Path, PathBuf},
11	};
12
13	fn write_json(
14	path: impl AsRef<Path>,
15	value: &impl serde::Serialize,
16	) -> Result<(), Box<dyn std::error::Error>> {
17	let mut output = BufWriter::new(fs::File::create(path)?);
18	serde_json::to_writer(&mut output, value)?;
19	output.flush()?;
20	Ok(())
21	}
22
23	fn main() -> Result<(), Box<dyn std::error::Error>> {
24	let mut args = env::args_os().skip(1);
25	let path = args.next().ok_or("Provide a .one or .onetoc2 file.")?;
26	let destination = args.next().map(PathBuf::from);
27	if args.next().is_some() {
28	return Err("Provide a source file and an optional new export directory.".into());
29	}
30	let source_path = PathBuf::from(path);
31	let bytes = onestore::read_file(&source_path)?;
32	let store = Store::parse(&bytes)?;
33	let index = RevisionIndex::parse(&store)?;
34	let document = Document::parse(&index)?;
35	if let Some(destination) = destination {
36	fs::create_dir(&destination)?;
37	fs::create_dir(destination.join("assets"))?;
38	let parent = source_path
39	.parent()
40	.filter(\|parent\| !parent.as_os_str().is_empty())
41	.unwrap_or_else(\|\| Path::new("."));
42	let mut source = onestore_notebook::Local::open(parent)?;
43	let section = source_path
44	.file_name()
45	.and_then(\|name\| name.to_str())
46	.ok_or("Use a UTF-8 section filename")?;
47	let mut assets = Vec::new();
48	let mut seen = BTreeSet::new();
49	for node in document
50	.spaces
51	.values()
52	.flat_map(\|s\| s.revisions.values())
53	.flat_map(\|r\| r.nodes.values())
54	{
55	if let Kind::File {
56	reference, payload, ..
57	} = &node.kind
58	&& seen.insert(serde_json::to_string(reference)?)
59	{
60	let data = match reference {
61	FileDataReference::Internal(_) => {
62	Cow::Borrowed(payload.ok_or("Missing embedded file data")?)
63	}
64	FileDataReference::External(filename) => {
65	match onestore_notebook::read_external_asset(
66	&mut source,
67	section,
68	filename,
69	256 * 1024 * 1024,
70	) {
71	Ok(bytes) => Cow::Owned(bytes),
72	Err(error) => {
73	let kind = match &error {
74	onestore_notebook::Error::Io { error, .. } => {
75	format!("{:?}", error.kind())
76	}
77	_ => "InvalidData".into(),
78	};
79	assets.push(serde_json::json!({"reference":reference,"path":null,"error":{"kind":kind,"message":error.to_string()}}));
80	continue;
81	}
82	}
83	}
84	FileDataReference::Invalid => {
85	assets.push(serde_json::json!({"reference":reference,"path":null,"error":{"kind":"InvalidData","message":"The document marks this payload as unavailable"}}));
86	continue;
87	}
88	};
89	let path = format!("assets/{}.bin", assets.len());
90	fs::write(destination.join(&path), data)?;
91	assets.push(serde_json::json!({"reference":reference,"path":path}));
92	}
93	}
94	write_json(destination.join("assets.json"), &assets)?;
95	let mut text = std::collections::BTreeMap::new();
96	for (sid, space) in &document.spaces {
97	let mut revisions = std::collections::BTreeMap::new();
98	for (rid, revision) in &space.revisions {
99	let mut objects = std::collections::BTreeMap::new();
100	for (oid, node) in &revision.nodes {
101	if matches!(node.kind, Kind::RichText { .. }) {
102	objects.insert(oid, revision.text_runs(oid)?);
103	}
104	}
105	revisions.insert(*rid, objects);
106	}
107	text.insert(*sid, revisions);
108	}
109	write_json(destination.join("text.json"), &text)?;
110	write_json(destination.join("document.json"), &document)?;
111	} else {
112	serde_json::to_writer(io::stdout().lock(), &document)?;
113	}
114	Ok(())
115	}

crates/onestore-notebook/examples/externalize_fixture.rs created+101

...	@@ -0,0 +1,101 @@
1	//! Builds a new external-payload test fixture from a native capture with reserved fragment space.
2	use onestore::{FileDataReference, RevisionIndex, Store};
3	use std::{fs, path::PathBuf};
4	fn main() -> Result<(), Box<dyn std::error::Error>> {
5	let args: Vec<_> = std::env::args_os().skip(1).collect();
6	if args.len() != 2 {
7	return Err("Provide a native source section and a new fixture directory".into());
8	}
9	let source = fs::read(&args[0])?;
10	let destination = PathBuf::from(&args[1]);
11	fs::create_dir(&destination)?;
12	fs::create_dir(destination.join("synthetic_onefiles"))?;
13	let store = Store::parse(&source)?;
14	assert!(store.checksum_mismatches.is_empty());
15	RevisionIndex::parse(&store)?.validate_current()?;
16	let mut output = source.clone();
17	let mut changed = 0;
18	for list in store.lists.values() {
19	for chunk in &list.fragments {
20	let start = chunk.offset as usize;
21	let end = start + chunk.length as usize - 20;
22	let mut body = Vec::new();
23	let mut touched = false;
24	for node in list
25	.nodes
26	.iter()
27	.filter(\|node\| node.offset >= start + 16 && node.offset < end)
28	{
29	let raw = u32::from_le_bytes(source[node.offset..node.offset + 4].try_into()?);
30	let size = ((raw >> 10) & 0x1fff) as usize;
31	let mut encoded = source[node.offset..node.offset + size].to_vec();
32	if matches!(node.id, 0x72 \| 0x73) {
33	let offset = 4 + 4 + 4 + if node.id == 0x72 { 1 } else { 4 };
34	let count =
35	u32::from_le_bytes(encoded[offset..offset + 4].try_into()?) as usize;
36	let text = String::from_utf16(
37	&encoded[offset + 4..offset + 4 + count * 2]
38	.chunks_exact(2)
39	.map(\|v\| u16::from_le_bytes(v.try_into().unwrap()))
40	.collect::<Vec<_>>(),
41	)?;
42	if let FileDataReference::Internal(guid) = text.parse()? {
43	let filename = format!(
44	"{}.onebin",
45	text.strip_prefix("<ifndf>{")
46	.unwrap()
47	.strip_suffix('}')
48	.unwrap()
49	);
50	let payload = store.file_data(guid)?;
51	let path = destination.join("synthetic_onefiles").join(&filename);
52	if path.exists() {
53	assert_eq!(fs::read(&path)?, payload);
54	} else {
55	fs::write(path, payload)?;
56	}
57	let reference = format!("<file>{filename}");
58	let data: Vec<_> = reference
59	.encode_utf16()
60	.flat_map(u16::to_le_bytes)
61	.collect();
62	encoded.splice(offset + 4..offset + 4 + count * 2, data.iter().copied());
63	encoded[offset..offset + 4]
64	.copy_from_slice(&(data.len() as u32 / 2).to_le_bytes());
65	assert!(encoded.len() <= 0x1fff);
66	let header = (raw & !(0x1fff << 10)) \| ((encoded.len() as u32) << 10);
67	encoded[..4].copy_from_slice(&header.to_le_bytes());
68	changed += 1;
69	touched = true;
70	}
71	}
72	body.extend(encoded);
73	}
74	if touched {
75	assert!(
76	body.len() + 4 <= end - start - 16,
77	"Native fragment has insufficient reserved space"
78	);
79	body.extend_from_slice(&(0xff_u32 \| (4 << 10)).to_le_bytes());
80	output[start + 16..end].fill(0);
81	output[start + 16..start + 16 + body.len()].copy_from_slice(&body);
82	}
83	}
84	}
85	assert!(changed > 0);
86	let converted = Store::parse(&output)?;
87	assert!(converted.checksum_mismatches.is_empty());
88	RevisionIndex::parse(&converted)?.validate_current()?;
89	fs::write(destination.join("synthetic.one"), &output)?;
90	fs::write(
91	destination.join("Open Notebook.onetoc2"),
92	onestore::create_table_of_contents(
93	"Open Notebook.onetoc2",
94	&[("synthetic.one", converted.header.file_id)],
95	)?,
96	)?;
97	println!(
98	"Converted {changed} file-data declarations without changing payload identities or bytes"
99	);
100	Ok(())
101	}

crates/onestore-notebook/src/lib.rs+45-1

...	@@ -77,6 +77,46 @@ pub trait Source {	...	@@ -77,6 +77,46 @@ pub trait Source {
77	fn entries(&mut self, path: &str, limit: usize) -> io::Result<Vec<Entry>>;	77	fn entries(&mut self, path: &str, limit: usize) -> io::Result<Vec<Entry>>;
78	/// Return a consistent file snapshot, rejecting images larger than the byte limit.	78	/// Return a consistent file snapshot, rejecting images larger than the byte limit.
79	fn read(&mut self, path: &str, limit: usize) -> io::Result<Vec<u8>>;	79	fn read(&mut self, path: &str, limit: usize) -> io::Result<Vec<u8>>;
		80	/// Reads an external payload with the same completeness and size guarantees.
		81	fn read_asset(&mut self, path: &str, limit: usize) -> io::Result<Vec<u8>> {
		82	self.read(path, limit)
		83	}
		84	}
		85
		86	/// Reads an external file-data reference from the section's sibling `_onefiles` folder.
		87	/// `NotFound` in `Error::Io` is distinct from a successfully read zero-byte payload.
		88	pub fn read_external_asset(
		89	source: &mut impl Source,
		90	section: &str,
		91	filename: &str,
		92	limit: usize,
		93	) -> Result<Vec<u8>, Error> {
		94	let (stem, extension) = section.rsplit_once('.').ok_or_else(\|\| Error::Entry {
		95	path: section.into(),
		96	})?;
		97	if !extension.eq_ignore_ascii_case("one")
		98	\|\| !section.split('/').all(component)
		99	\|\| stem.ends_with('/')
		100	\|\| stem.is_empty()
		101	{
		102	return Err(Error::Entry {
		103	path: section.into(),
		104	});
		105	}
		106	format!("<file>{filename}")
		107	.parse::<onestore::FileDataReference>()
		108	.map_err(\|_\| Error::Entry {
		109	path: filename.into(),
		110	})?;
		111	let path = format!("{stem}_onefiles/{filename}");
		112	let bytes = source.read_asset(&path, limit).map_err(\|error\| Error::Io {
		113	path: path.clone(),
		114	error,
		115	})?;
		116	if bytes.len() > limit {
		117	return Err(Error::Limit { path });
		118	}
		119	Ok(bytes)
80	}	120	}
81		121
82	pub struct Limits {	122	pub struct Limits {
...	@@ -113,7 +153,8 @@ pub enum Error {	...	@@ -113,7 +153,8 @@ pub enum Error {
113	},	153	},
114	}	154	}
115		155
116	/// Discovers the complete rooted directory within caller-specified work and size limits.	156	/// Discovers rooted notebook topology within caller-specified work and size limits.
		157	/// Reserved `_onefiles` directories contain payloads, not section groups.
117	pub fn discover(source: &mut impl Source, limits: Limits) -> Result<Folder, Error> {	158	pub fn discover(source: &mut impl Source, limits: Limits) -> Result<Folder, Error> {
118	let mut remaining = limits.entries;	159	let mut remaining = limits.entries;
119	let mut identities = BTreeMap::new();	160	let mut identities = BTreeMap::new();
...	@@ -158,6 +199,9 @@ fn scan(	...	@@ -158,6 +199,9 @@ fn scan(
158	for entry in &listing {	199	for entry in &listing {
159	let child = join(path, &entry.name);	200	let child = join(path, &entry.name);
160	if entry.kind == EntryKind::Directory {	201	if entry.kind == EntryKind::Directory {
		202	if entry.name.to_ascii_lowercase().ends_with("_onefiles") {
		203	continue;
		204	}
161	result.groups.push(scan(	205	result.groups.push(scan(
162	source,	206	source,
163	&child,	207	&child,

crates/onestore-notebook/src/source.rs+4

...	@@ -114,4 +114,8 @@ impl Source for Smb<'_> {	...	@@ -114,4 +114,8 @@ impl Source for Smb<'_> {
114	fn read(&mut self, path: &str, limit: usize) -> io::Result<Vec<u8>> {	114	fn read(&mut self, path: &str, limit: usize) -> io::Result<Vec<u8>> {
115	self.client.read_storage(&self.path(path)?, limit)	115	self.client.read_storage(&self.path(path)?, limit)
116	}	116	}
		117
		118	fn read_asset(&mut self, path: &str, limit: usize) -> io::Result<Vec<u8>> {
		119	self.client.read_asset(&self.path(path)?, limit)
		120	}
117	}	121	}

crates/onestore-notebook/tests/assets.rs created+162

...	@@ -0,0 +1,162 @@
1	use onestore_notebook::{Error, Local, read_external_asset};
2	use std::{fs, io};
3
4	#[test]
5	fn nested_external_payloads_are_exact_and_empty_is_distinct_from_missing() {
6	let root = tempfile::tempdir().unwrap();
7	let directory = root.path().join("Group 🦀/Section é_onefiles");
8	fs::create_dir_all(&directory).unwrap();
9	let filename = "2a83ae62-6754-4383-8e2b-4033ff3cfba1.onebin";
10	let mut source = Local::open(root.path()).unwrap();
11	let section = "Group 🦀/Section é.ONE";
12	assert!(
13	matches!(read_external_asset(&mut source, section, filename, 0),
14	Err(Error::Io { error, .. }) if error.kind() == io::ErrorKind::NotFound)
15	);
16	fs::write(directory.join(filename), []).unwrap();
17	assert!(
18	read_external_asset(&mut source, section, filename, 0)
19	.unwrap()
20	.is_empty()
21	);
22	let bytes: Vec<_> = (0..65537).map(\|index\| (index % 251) as u8).collect();
23	fs::write(directory.join(filename), &bytes).unwrap();
24	assert!(
25	matches!(read_external_asset(&mut source, section, filename, bytes.len()-1),
26	Err(Error::Io { error, .. }) if error.kind() == io::ErrorKind::FileTooLarge)
27	);
28	assert_eq!(
29	read_external_asset(&mut source, section, filename, bytes.len()).unwrap(),
30	bytes
31	);
32	assert_eq!(fs::read(directory.join(filename)).unwrap(), bytes);
33	}
34
35	#[test]
36	fn invalid_asset_locations_fail_before_accessing_the_source() {
37	struct Unused;
38	impl onestore_notebook::Source for Unused {
39	fn entries(&mut self, _: &str, _: usize) -> io::Result<Vec<onestore_notebook::Entry>> {
40	panic!("Unexpected enumeration")
41	}
42	fn read(&mut self, _: &str, _: usize) -> io::Result<Vec<u8>> {
43	panic!("Unexpected read")
44	}
45	}
46	let filename = "2a83ae62-6754-4383-8e2b-4033ff3cfba1.onebin";
47	for section in [
48	"",
49	".one",
50	"Group/.one",
51	"/Section.one",
52	"../Section.one",
53	"Group/../Section.one",
54	"Group\\Section.one",
55	"Section.onetoc2",
56	] {
57	assert!(matches!(
58	read_external_asset(&mut Unused, section, filename, 100),
59	Err(Error::Entry { .. })
60	));
61	}
62	for name in [
63	"",
64	"attachment.png",
65	"../2a83ae62-6754-4383-8e2b-4033ff3cfba1.onebin",
66	"2a83ae62-6754-4383-8e2b-4033ff3cfba1.onebin:other",
67	"<invfdo>",
68	] {
69	assert!(matches!(
70	read_external_asset(&mut Unused, "Section.one", name, 100),
71	Err(Error::Entry { .. })
72	));
73	}
74	}
75
76	#[test]
77	fn reserved_payload_directories_are_not_notebook_groups() {
78	let root = tempfile::tempdir().unwrap();
79	fs::write(
80	root.path().join("Section.ONE"),
81	onestore::create_section("Section.one", "Fixture", "Author").unwrap(),
82	)
83	.unwrap();
84	for name in ["section_onefiles", "orphan_onefiles", "ordinary"] {
85	fs::create_dir(root.path().join(name)).unwrap();
86	}
87	fs::write(
88	root.path().join("section_onefiles/unfinished.onebin"),
89	b"payload",
90	)
91	.unwrap();
92	let catalog = onestore_notebook::discover(
93	&mut Local::open(root.path()).unwrap(),
94	onestore_notebook::Limits {
95	entries: 4,
96	bytes_per_file: 1 << 20,
97	depth: 1,
98	},
99	)
100	.unwrap();
101	assert_eq!(catalog.sections.len(), 1);
102	assert_eq!(
103	catalog
104	.groups
105	.iter()
106	.map(\|group\| group.path.as_str())
107	.collect::<Vec<_>>(),
108	["ordinary"]
109	);
110	}
111
112	#[test]
113	#[cfg(feature = "smb")]
114	#[ignore = "requires a disposable Samba mirror of corpus/native-external-assets/notebook at ONESTORE_SMB_NOTEBOOK"]
115	fn live_external_payloads() {
116	use onestore::{
117	FileDataReference, RevisionIndex, Store,
118	document::{Document, Kind},
119	};
120	let root = std::path::Path::new("../../corpus/native-external-assets/notebook");
121	let bytes = fs::read(root.join("synthetic.one")).unwrap();
122	let store = Store::parse(&bytes).unwrap();
123	let index = RevisionIndex::parse(&store).unwrap();
124	let document = Document::parse(&index).unwrap();
125	let client = onestore_smb::Client::connect(
126	&std::env::var("ONESTORE_SMB_LAB").unwrap(),
127	"agent",
128	onestore_smb::Credentials::default(),
129	std::time::Duration::from_secs(5),
130	)
131	.unwrap();
132	let mut remote =
133	onestore_notebook::Smb::new(&client, &std::env::var("ONESTORE_SMB_NOTEBOOK").unwrap())
134	.unwrap();
135	let mut local = Local::open(root).unwrap();
136	let mut seen = std::collections::BTreeSet::new();
137	for node in document
138	.spaces
139	.values()
140	.flat_map(\|space\| space.revisions.values())
141	.flat_map(\|revision\| revision.nodes.values())
142	{
143	if let Kind::File {
144	reference: FileDataReference::External(filename),
145	..
146	} = &node.kind
147	&& seen.insert(filename)
148	{
149	let expected = fs::read(root.join("synthetic_onefiles").join(filename)).unwrap();
150	assert_eq!(
151	read_external_asset(&mut local, "synthetic.one", filename, expected.len()).unwrap(),
152	expected
153	);
154	assert_eq!(
155	read_external_asset(&mut remote, "synthetic.one", filename, expected.len())
156	.unwrap(),
157	expected
158	);
159	}
160	}
161	assert_eq!(seen.len(), 3);
162	}

crates/onestore-smb/README.md+5

 @@ -47,6 +47,11 @@ are rejected with `ResourceBusy`. Notebook identities come from the files, not
 directory names or sizes. Missing paths, denied access and non-directory paths
 have distinct I/O error kinds.
+`Client::read_asset(path, byte_limit)` reads an external payload under a read-only
+share handle that excludes writes and deletion. Empty files succeed; limits,
+interrupted reads and failed close never return partial bytes. This payload read
+does not parse a OneStore header or acquire its reader-coordination bytes.
 `python3 tools/test_smb_directory.py VM OUTPUT` checks a caller-owned disposable
 Linux lab VM against its filesystem listing and interrupts directory requests,
 responses and close. It creates synthetic files in that VM; the caller retains

crates/onestore-smb/src/lib.rs+29-1

...	@@ -175,6 +175,10 @@ impl Client {	...	@@ -175,6 +175,10 @@ impl Client {
175	}	175	}
176		176
177	fn open(&self, path: &str, write: bool) -> io::Result<File<'_>> {	177	fn open(&self, path: &str, write: bool) -> io::Result<File<'_>> {
		178	self.open_shared(path, write, if write { 5 } else { 7 })
		179	}
		180
		181	fn open_shared(&self, path: &str, write: bool, sharing: u32) -> io::Result<File<'_>> {
178	if path.is_empty() \|\| path.contains('\0') \|\| path.encode_utf16().count() > 32767 {	182	if path.is_empty() \|\| path.contains('\0') \|\| path.encode_utf16().count() > 32767 {
179	return Err(io::ErrorKind::InvalidInput.into());	183	return Err(io::ErrorKind::InvalidInput.into());
180	}	184	}
...	@@ -185,7 +189,7 @@ impl Client {	...	@@ -185,7 +189,7 @@ impl Client {
185	impersonation_level: ImpersonationLevel::Impersonation,	189	impersonation_level: ImpersonationLevel::Impersonation,
186	desired_access: FileAccessMask::new(if write { 0xc0000000 } else { 0x80000000 }),	190	desired_access: FileAccessMask::new(if write { 0xc0000000 } else { 0x80000000 }),
187	file_attributes: 0,	191	file_attributes: 0,
188	share_access: ShareAccess(if write { 5 } else { 7 }),	192	share_access: ShareAccess(sharing),
189	create_disposition: CreateDisposition::FileOpen,	193	create_disposition: CreateDisposition::FileOpen,
190	create_options: 0x42,	194	create_options: 0x42,
191	name: smb2::encode_path(&path.replace('\\', "/")),	195	name: smb2::encode_path(&path.replace('\\', "/")),
...	@@ -198,6 +202,30 @@ impl Client {	...	@@ -198,6 +202,30 @@ impl Client {
198	})	202	})
199	}	203	}
200		204
		205	/// Reads a bounded external payload while denying concurrent writes and deletion.
		206	/// Empty files succeed; limits, sharing contention and failed close return no payload.
		207	pub fn read_asset(&self, path: &str, limit: usize) -> io::Result<Vec<u8>> {
		208	let mut file = self.open_shared(path, false, 1)?;
		209	let mut bytes = Vec::new();
		210	let mut block = [0; 65536];
		211	loop {
		212	let count = (limit - bytes.len()).min(block.len() - 1) + 1;
		213	let read = file.read_at(
		214	u64::try_from(bytes.len()).map_err(\|_\| io::ErrorKind::InvalidInput)?,
		215	&mut block[..count],
		216	)?;
		217	if read == 0 {
		218	break;
		219	}
		220	bytes.extend_from_slice(&block[..read]);
		221	if bytes.len() > limit {
		222	return Err(io::ErrorKind::FileTooLarge.into());
		223	}
		224	}
		225	file.close()?;
		226	Ok(bytes)
		227	}
		228
201	/// Reads one bounded, consistent snapshot; contention returns WouldBlock.	229	/// Reads one bounded, consistent snapshot; contention returns WouldBlock.
202	pub fn read(&self, path: &str, limit: usize) -> io::Result<Vec<u8>> {	230	pub fn read(&self, path: &str, limit: usize) -> io::Result<Vec<u8>> {
203	self.read_with(path, \|file\| {	231	self.read_with(path, \|file\| {

crates/onestore-smb/src/tests.rs+49

...	@@ -381,6 +381,55 @@ fn live_storage_inspection() {	...	@@ -381,6 +381,55 @@ fn live_storage_inspection() {
381	}	381	}
382	}	382	}
383		383
		384	#[test]
		385	#[ignore = "requires ONESTORE_SMB_LAB pointing to disposable Samba"]
		386	fn live_assets() {
		387	let reader = client();
		388	let writer = client();
		389	for size in [0, 1, 65535, 65536, 65537, 1048577] {
		390	let bytes: Vec<_> = (0..size).map(\|index\| (index % 251) as u8).collect();
		391	let path = format!(
		392	"asset-{size}-{}.onebin",
		393	SystemTime::now()
		394	.duration_since(UNIX_EPOCH)
		395	.unwrap()
		396	.as_nanos()
		397	);
		398	create(&writer, &path, &bytes);
		399	assert_eq!(reader.read_asset(&path, size).unwrap(), bytes);
		400	if size != 0 {
		401	assert_eq!(
		402	reader.read_asset(&path, size - 1).unwrap_err().kind(),
		403	io::ErrorKind::FileTooLarge
		404	);
		405	assert_eq!(reader.read_asset(&path, size).unwrap(), bytes);
		406	}
		407	let writing = writer.open(&path, true).unwrap();
		408	assert_eq!(
		409	reader.read_asset(&path, size).unwrap_err().kind(),
		410	io::ErrorKind::WouldBlock
		411	);
		412	writing.close().unwrap();
		413	let asset = reader.open_shared(&path, false, 1).unwrap();
		414	assert_eq!(
		415	writer.open(&path, true).err().unwrap().kind(),
		416	io::ErrorKind::WouldBlock
		417	);
		418	let other = writer.open_shared(&path, false, 1).unwrap();
		419	other.close().unwrap();
		420	asset.close().unwrap();
		421	writer.open(&path, true).unwrap().close().unwrap();
		422	assert_eq!(reader.read_asset(&path, size).unwrap(), bytes);
		423	}
		424	assert_eq!(
		425	reader
		426	.read_asset("absent-payload.onebin", 0)
		427	.unwrap_err()
		428	.kind(),
		429	io::ErrorKind::NotFound
		430	);
		431	}
		432
384	#[test]	433	#[test]
385	#[ignore = "requires an owned Samba fixture and maintenance controller"]	434	#[ignore = "requires an owned Samba fixture and maintenance controller"]
386	fn live_reader_hold() {	435	fn live_reader_hold() {

crates/onestore-smb/src/tests/faults.rs+77-10

...	@@ -171,14 +171,7 @@ fn configure(output: &Path, state: Value) -> usize {	...	@@ -171,14 +171,7 @@ fn configure(output: &Path, state: Value) -> usize {
171	}	171	}
172	}	172	}
173		173
174	#[test]	174	fn proxy(output: &Path) -> (Proxy, String) {
175	#[ignore = "requires an owned Samba share and a new ONESTORE_SMB_EVIDENCE directory"]
176	fn live_message_loss() {
177	let output = std::path::PathBuf::from(std::env::var("ONESTORE_SMB_EVIDENCE").unwrap());
178	fs::create_dir(&output).unwrap();
179	fs::create_dir(output.join("interrupted")).unwrap();
180	fs::create_dir(output.join("recovered")).unwrap();
181	fs::create_dir(output.join("source")).unwrap();
182	let address = std::env::var("ONESTORE_SMB_LAB").unwrap();	175	let address = std::env::var("ONESTORE_SMB_LAB").unwrap();
183	let (host, port) = address.rsplit_once(':').unwrap();	176	let (host, port) = address.rsplit_once(':').unwrap();
184	let mut proxy = Proxy(	177	let mut proxy = Proxy(
...	@@ -193,7 +186,7 @@ fn live_message_loss() {	...	@@ -193,7 +186,7 @@ fn live_message_loss() {
193	);	186	);
194	let deadline = Instant::now() + Duration::from_secs(5);	187	let deadline = Instant::now() + Duration::from_secs(5);
195	let port = loop {	188	let port = loop {
196	if let Some(port) = records(&output)	189	if let Some(port) = records(output)
197	.iter()	190	.iter()
198	.find_map(\|event\| event["listening"].as_u64())	191	.find_map(\|event\| event["listening"].as_u64())
199	{	192	{
...	@@ -203,7 +196,81 @@ fn live_message_loss() {	...	@@ -203,7 +196,81 @@ fn live_message_loss() {
203	assert!(Instant::now() < deadline, "proxy did not start");	196	assert!(Instant::now() < deadline, "proxy did not start");
204	std::thread::sleep(Duration::from_millis(5));	197	std::thread::sleep(Duration::from_millis(5));
205	};	198	};
206	let proxied = format!("127.0.0.1:{port}");	199	(proxy, format!("127.0.0.1:{port}"))
		200	}
		201
		202	#[test]
		203	#[ignore = "requires an owned Samba share and a new ONESTORE_SMB_EVIDENCE directory"]
		204	fn live_asset_loss() {
		205	let output = std::path::PathBuf::from(std::env::var("ONESTORE_SMB_EVIDENCE").unwrap());
		206	fs::create_dir(&output).unwrap();
		207	let (_proxy, address) = proxy(&output);
		208	let observer = client();
		209	let bytes: Vec<_> = (0..1048577).map(\|index\| (index % 251) as u8).collect();
		210	let path = format!(
		211	"asset-loss-{}.onebin",
		212	SystemTime::now()
		213	.duration_since(UNIX_EPOCH)
		214	.unwrap()
		215	.as_nanos()
		216	);
		217	create(&observer, &path, &bytes);
		218	for (command, occurrence) in [(8, 1), (8, 9), (8, 17), (8, 18), (6, 1)] {
		219	for direction in ["request", "response"] {
		220	let reader = Client::connect(
		221	&address,
		222	"agent",
		223	Credentials::default(),
		224	Duration::from_secs(5),
		225	)
		226	.unwrap();
		227	// Response occurrences count only replies with the selected status.
		228	let begin = configure(
		229	&output,
		230	json!({"cut":command,"occurrence":if command == 8 && occurrence == 18 && direction == "response" { 1 } else { occurrence },
		231	"direction":direction,"status":if command == 8 && occurrence == 18 { "0xc0000011" } else { "0x0" }}),
		232	);
		233	assert!(
		234	reader.read_asset(&path, bytes.len()).is_err(),
		235	"{command}/{occurrence}/{direction}"
		236	);
		237	assert_eq!(
		238	reader.read_asset(&path, bytes.len()).unwrap_err().kind(),
		239	io::ErrorKind::NotConnected
		240	);
		241	assert_eq!(
		242	records(&output)[begin..]
		243	.iter()
		244	.filter(\|row\| row.get("cut").is_some())
		245	.count(),
		246	1
		247	);
		248	configure(
		249	&output,
		250	json!({"phase":format!("{command}-{occurrence}-{direction}-reconnected")}),
		251	);
		252	let reconnected = Client::connect(
		253	&address,
		254	"agent",
		255	Credentials::default(),
		256	Duration::from_secs(5),
		257	)
		258	.unwrap();
		259	assert_eq!(reconnected.read_asset(&path, bytes.len()).unwrap(), bytes);
		260	}
		261	}
		262	assert_eq!(observer.read_asset(&path, bytes.len()).unwrap(), bytes);
		263	}
		264
		265	#[test]
		266	#[ignore = "requires an owned Samba share and a new ONESTORE_SMB_EVIDENCE directory"]
		267	fn live_message_loss() {
		268	let output = std::path::PathBuf::from(std::env::var("ONESTORE_SMB_EVIDENCE").unwrap());
		269	fs::create_dir(&output).unwrap();
		270	fs::create_dir(output.join("interrupted")).unwrap();
		271	fs::create_dir(output.join("recovered")).unwrap();
		272	fs::create_dir(output.join("source")).unwrap();
		273	let (_proxy, proxied) = proxy(&output);
207	let observer = client();	274	let observer = client();
208	let prefix = SystemTime::now()	275	let prefix = SystemTime::now()
209	.duration_since(UNIX_EPOCH)	276	.duration_since(UNIX_EPOCH)

crates/onestore/examples/document.rs deleted-77

...	@@ -1,77 +0,0 @@
1	use onestore::{
2	FileDataReference, RevisionIndex, Store,
3	document::{Document, Kind},
4	};
5	use std::{
6	collections::BTreeSet,
7	env, fs,
8	io::{self, BufWriter, Write},
9	path::{Path, PathBuf},
10	};
11
12	fn write_json(
13	path: impl AsRef<Path>,
14	value: &impl serde::Serialize,
15	) -> Result<(), Box<dyn std::error::Error>> {
16	let mut output = BufWriter::new(fs::File::create(path)?);
17	serde_json::to_writer(&mut output, value)?;
18	output.flush()?;
19	Ok(())
20	}
21
22	fn main() -> Result<(), Box<dyn std::error::Error>> {
23	let mut args = env::args_os().skip(1);
24	let path = args.next().ok_or("Provide a .one or .onetoc2 file.")?;
25	let destination = args.next().map(PathBuf::from);
26	if args.next().is_some() {
27	return Err("Provide a source file and an optional new export directory.".into());
28	}
29	let bytes = onestore::read_file(path)?;
30	let store = Store::parse(&bytes)?;
31	let index = RevisionIndex::parse(&store)?;
32	let document = Document::parse(&index)?;
33	if let Some(destination) = destination {
34	fs::create_dir(&destination)?;
35	fs::create_dir(destination.join("assets"))?;
36	let mut assets = Vec::new();
37	let mut seen = BTreeSet::new();
38	for node in document
39	.spaces
40	.values()
41	.flat_map(\|s\| s.revisions.values())
42	.flat_map(\|r\| r.nodes.values())
43	{
44	if let Kind::File {
45	reference: FileDataReference::Internal(guid),
46	payload: Some(data),
47	..
48	} = &node.kind
49	&& seen.insert(guid)
50	{
51	let path = format!("assets/{}.bin", assets.len());
52	fs::write(destination.join(&path), data)?;
53	assets.push(serde_json::json!({"reference": FileDataReference::Internal(*guid), "path": path}));
54	}
55	}
56	write_json(destination.join("assets.json"), &assets)?;
57	let mut text = std::collections::BTreeMap::new();
58	for (sid, space) in &document.spaces {
59	let mut revisions = std::collections::BTreeMap::new();
60	for (rid, revision) in &space.revisions {
61	let mut objects = std::collections::BTreeMap::new();
62	for (oid, node) in &revision.nodes {
63	if matches!(node.kind, Kind::RichText { .. }) {
64	objects.insert(oid, revision.text_runs(oid)?);
65	}
66	}
67	revisions.insert(*rid, objects);
68	}
69	text.insert(*sid, revisions);
70	}
71	write_json(destination.join("text.json"), &text)?;
72	write_json(destination.join("document.json"), &document)?;
73	} else {
74	serde_json::to_writer(io::stdout().lock(), &document)?;
75	}
76	Ok(())
77	}

crates/onestore/src/files.rs+9-1

 @@ -49,6 +49,14 @@ impl Object<'_> {
             offset: 0,
             message: "Invalid UTF-16 file-data reference",
         })?;
+        text.parse().map(Some)
+    }
+}
+impl std::str::FromStr for FileDataReference {
+    type Err = Error;
+    fn from_str(text: &str) -> Result<Self, Error> {
         let parsed = if let Some(name) = text.strip_prefix("<file>") {
             name.strip_suffix(".onebin")
                 .and_then(guid)
 @@ -63,7 +71,7 @@ impl Object<'_> {
         } else {
             None
         };
-        parsed.map(Some).ok_or(Error {
+        parsed.ok_or(Error {
             offset: 0,
             message: "Invalid file-data reference syntax",
         })

tools/native/external-assets.ps1 created+34

...	@@ -0,0 +1,34 @@
1	param([Parameter(Mandatory=$true)][string]$Root, [string]$CloneHost = '', [int[]]$Lengths = @(0, 1024))
2	Set-StrictMode -Version Latest
3	$ErrorActionPreference = 'Stop'
4	& "$PSScriptRoot\cold-current.ps1" -Root $Root -CloneHost $CloneHost
5	$app = New-Object -ComObject OneNote.Application
6	$notebook = ''
7	try {
8	$inputs = Join-Path $Root 'inputs'
9	New-Item -ItemType Directory -Path $inputs \| Out-Null
10	$app.OpenHierarchy((Join-Path $Root 'notebook'), '', [ref]$notebook, 0)
11	$section = ''
12	$app.OpenHierarchy('synthetic.one', $notebook, [ref]$section, 0)
13	foreach ($length in $Lengths) {
14	$path = Join-Path $inputs ($length.ToString() + '.bin')
15	$stream = [IO.File]::Create($path)
16	try {
17	$block = New-Object byte[] 1024
18	for ($i = 0; $i -lt $block.Length; $i++) { $block[$i] = [byte]($i % 251) }
19	for ($written = 0; $written -lt $length; $written += $block.Length) { $stream.Write($block, 0, [Math]::Min($block.Length, $length - $written)) }
20	} finally { $stream.Dispose() }
21	$page = ''
22	$app.CreateNewPage($section, [ref]$page, 0)
23	$xml = '<one:Page xmlns:one="http://schemas.microsoft.com/office/onenote/2010/onenote" ID="' + $page + '"><one:Title><one:OE><one:T>Attachment ' + $length + '</one:T></one:OE></one:Title><one:Outline><one:OEChildren><one:OE><one:InsertedFile pathSource="' + $path + '" preferredName="' + $length + '.bin"/></one:OE></one:OEChildren></one:Outline></one:Page>'
24	$app.UpdatePageContent($xml, [DateTime]::MinValue, 1, $false)
25	}
26	$app.SyncHierarchy($notebook)
27
28	} finally {
29	if ($notebook) { $app.CloseNotebook($notebook, $false) }
30	[void][Runtime.InteropServices.Marshal]::FinalReleaseComObject($app)
31	$app = $null
32	[GC]::Collect()
33	[GC]::WaitForPendingFinalizers()
34	}

tools/notebook_editor.py+1-1

...	@@ -61,7 +61,7 @@ class Session:	...	@@ -61,7 +61,7 @@ class Session:
61	source = pending / 'snapshot'	61	source = pending / 'snapshot'
62	source.mkdir(parents=True)	62	source.mkdir(parents=True)
63	for path in sorted((self.output / 'notebook').rglob('*')):	63	for path in sorted((self.output / 'notebook').rglob('*')):
64	if path.suffix.lower() not in ('.one', '.onetoc2'): continue	64	if path.suffix.lower() not in ('.one', '.onetoc2', '.onebin'): continue
65	saved = source / path.relative_to(self.output / 'notebook')	65	saved = source / path.relative_to(self.output / 'notebook')
66	saved.parent.mkdir(parents=True, exist_ok=True)	66	saved.parent.mkdir(parents=True, exist_ok=True)
67	result = bridge('snapshot', path, saved)	67	result = bridge('snapshot', path, saved)

tools/notebook_report.py+3

...	@@ -298,6 +298,8 @@ def generate(source, destination, native=None, versions=(), zone=timezone.utc, e	...	@@ -298,6 +298,8 @@ def generate(source, destination, native=None, versions=(), zone=timezone.utc, e
298	exported = destination / 'model' / str(index)	298	exported = destination / 'model' / str(index)
299	reusable = cached.get(relative.as_posix())	299	reusable = cached.get(relative.as_posix())
300	reused = reusable is not None and reusable[0] == manifest[-1]['sha256']	300	reused = reusable is not None and reusable[0] == manifest[-1]['sha256']
		301	if reused and any('External' in asset['reference'] for asset in json.loads((reusable[1] / 'assets.json').read_text())):
		302	reused = False
301	if reused:	303	if reused:
302	exported.mkdir()	304	exported.mkdir()
303	for name in ('document.json', 'text.json', 'assets.json'):	305	for name in ('document.json', 'text.json', 'assets.json'):
...	@@ -317,6 +319,7 @@ def generate(source, destination, native=None, versions=(), zone=timezone.utc, e	...	@@ -317,6 +319,7 @@ def generate(source, destination, native=None, versions=(), zone=timezone.utc, e
317	}	319	}
318	rows = json.loads((exported / 'assets.json').read_text())	320	rows = json.loads((exported / 'assets.json').read_text())
319	for asset in rows:	321	for asset in rows:
		322	if asset['path'] is None: continue
320	reference = json.dumps(asset['reference'], sort_keys=True)	323	reference = json.dumps(asset['reference'], sort_keys=True)
321	if reused:	324	if reused:
322	name = Path(asset['path']).name	325	name = Path(asset['path']).name

tools/test_notebook_discovery.py+1-1

...	@@ -12,7 +12,7 @@ ROOT = Path(__file__).resolve().parent.parent	...	@@ -12,7 +12,7 @@ ROOT = Path(__file__).resolve().parent.parent
12	class NativeDiscovery(unittest.TestCase):	12	class NativeDiscovery(unittest.TestCase):
13	def test_native_hierarchies(self):	13	def test_native_hierarchies(self):
14	fixtures = ['m6/native-features-01', 'native-encrypted/cold-encrypted-02',	14	fixtures = ['m6/native-features-01', 'native-encrypted/cold-encrypted-02',
15	'native-delete/cold-deletion-05']	15	'native-delete/cold-deletion-05', 'native-external-assets']
16	if os.environ.get('ONESTORE_NOTEBOOK_NATIVE'):	16	if os.environ.get('ONESTORE_NOTEBOOK_NATIVE'):
17	fixtures.append(str(Path(os.environ['ONESTORE_NOTEBOOK_NATIVE']).resolve()))	17	fixtures.append(str(Path(os.environ['ONESTORE_NOTEBOOK_NATIVE']).resolve()))
18	for fixture in fixtures:	18	for fixture in fixtures:

tools/test_notebook_report.py+35

...	@@ -1,6 +1,7 @@	...	@@ -1,6 +1,7 @@
1	import base64	1	import base64
2	from io import BytesIO	2	from io import BytesIO
3	import json	3	import json
		4	import hashlib
4	from pathlib import Path	5	from pathlib import Path
5	import re	6	import re
6	from tempfile import TemporaryDirectory	7	from tempfile import TemporaryDirectory
...	@@ -16,6 +17,40 @@ from document_model import EXPORTER	...	@@ -16,6 +17,40 @@ from document_model import EXPORTER
16		17
17		18
18	class NotebookReportTest(unittest.TestCase):	19	class NotebookReportTest(unittest.TestCase):
		20	def test_external_assets_match_native_bytes_and_refresh_without_a_section_edit(self):
		21	from notebook_editor import Session
		22	fixture = Path(__file__).resolve().parent.parent / 'corpus/native-external-assets'
		23	with TemporaryDirectory() as temporary:
		24	root = Path(temporary)
		25	source = root / 'source'
		26	shutil.copytree(fixture / 'notebook', source)
		27	session = Session(source, root / 'session')
		28	first = root / 'session/g/0/report'
		29	section = next(p.parent for p in (first / 'model').glob('*/assets.json')
		30	if json.loads(p.read_text()))
		31	rows = json.loads((section / 'assets.json').read_text())
		32	expected = {row['sha256'] for row in json.loads((fixture / 'read/payloads.json').read_text(encoding='utf-8-sig'))}
		33	actual = {hashlib.sha256((section / row['path']).read_bytes()).hexdigest() for row in rows}
		34	self.assertTrue(expected <= actual)
		35	for payload in source.rglob('*.onebin'):
		36	self.assertEqual(payload.read_bytes(), (root / 'session/g/0/snapshot' / payload.relative_to(source)).read_bytes())
		37	payload, = [p for p in source.rglob('*.onebin') if p.stat().st_size == 1024]
		38	reference = {'External': payload.name}
		39	old, = [row for row in rows if row['reference'] == reference]
		40	previous_bytes = (section / old['path']).read_bytes()
		41	payload.unlink()
		42	generate(source, root / 'missing', previous=first)
		43	missing, = [row for p in (root / 'missing/model').glob('*/assets.json')
		44	for row in json.loads(p.read_text()) if row['reference'] == reference]
		45	self.assertIsNone(missing['path'])
		46	self.assertEqual(missing['error']['kind'], 'NotFound')
		47	payload.write_bytes(b'Changed external payload')
		48	generate(source, root / 'changed', previous=root / 'missing')
		49	changed, = [(p.parent, row) for p in (root / 'changed/model').glob('*/assets.json')
		50	for row in json.loads(p.read_text()) if row['reference'] == reference]
		51	self.assertEqual((changed[0] / changed[1]['path']).read_bytes(), payload.read_bytes())
		52	self.assertEqual((section / old['path']).read_bytes(), previous_bytes)
		53
19	def test_locked_and_unreadable_sections_remain_visible_in_cached_reports(self):	54	def test_locked_and_unreadable_sections_remain_visible_in_cached_reports(self):
20	fixture = Path(__file__).resolve().parent.parent / 'corpus/native-encrypted/cold-encrypted-02/notebook'	55	fixture = Path(__file__).resolve().parent.parent / 'corpus/native-encrypted/cold-encrypted-02/notebook'
21	with TemporaryDirectory() as temporary:	56	with TemporaryDirectory() as temporary:

author	clover caruso <git@paperclover.net> 2026-09-07 18:42:27-07:00
committer	clover caruso <git@paperclover.net> 2026-09-07 18:42:27-07:00
log	30d02b72f6a5b8dd8b72cea593bdc9c883afea1e
tree	c22ca0a9fd436a9f93add677af1fcc6103aaa068
parent	bca7120d8a96e56b4ae7c6c76d1e82ccef0c5d21
signature	Signed by SSH key SHA256:52mNGHRsVFBDED9IAX5pe+LRWUefqTbxEReunq21QvU

...	@@ -7,6 +7,7 @@ Discovery returns ordered sections and nested groups with file identities and	...	@@ -7,6 +7,7 @@ Discovery returns ordered sections and nested groups with file identities and
7	share-relative paths. Section display-name overrides remain distinct from file	7	share-relative paths. Section display-name overrides remain distinct from file
8	names. TOC references whose identities are absent from the directory remain	8	names. TOC references whose identities are absent from the directory remain
9	inspectable; a cached filename never substitutes for an identity match.	9	inspectable; a cached filename never substitutes for an identity match.
		10	Reserved `_onefiles` directories are excluded from section-group traversal.
10	Encrypted sections and valid storage with an unreadable document graph retain	11	Encrypted sections and valid storage with an unreadable document graph retain
11	their identity as `Locked` or `Unreadable`, without being presented as empty pages.	12	their identity as `Locked` or `Unreadable`, without being presented as empty pages.
12	Malformed storage, failed reads and ambiguous identities reject the discovery.	13	Malformed storage, failed reads and ambiguous identities reject the discovery.
...	@@ -16,3 +17,9 @@ observation across multiple files, not an atomic notebook transaction or	...	@@ -16,3 +17,9 @@ observation across multiple files, not an atomic notebook transaction or
16	authorization to publish an edit. Refresh rejects observed topology changes and	17	authorization to publish an edit. Refresh rejects observed topology changes and
17	duplicate physical files with the same logical identity. Retain the last accepted	18	duplicate physical files with the same logical identity. Retain the last accepted
18	catalog if discovery fails; a connection failure does not mean files were deleted.	19	catalog if discovery fails; a connection failure does not mean files were deleted.
		20
		21	`read_external_asset` resolves a validated UUID `.onebin` filename beneath the
		22	selected section's sibling `_onefiles` folder and returns exact bounded bytes.
		23	Missing files, permissions and size failures retain their I/O error kinds; an
		24	empty payload is a successful empty buffer. Embedded payloads remain available
		25	directly from the core document model.

feat: resolve external notebook payloads with bounded SMB reads