| 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 next = args.next(); |
| 27 | let (destination, option) = if next.as_deref() == Some(std::ffi::OsStr::new("--password-file")) |
| 28 | { |
| 29 | (None, next) |
| 30 | } else { |
| 31 | (next.map(PathBuf::from), args.next()) |
| 32 | }; |
| 33 | let password_file = match (option, args.next(), args.next()) { |
| 34 | (None, None, None) => None, |
| 35 | (Some(flag), Some(path), None) if flag == "--password-file" => Some(PathBuf::from(path)), |
| 36 | _ => return Err("Provide a source file, an optional new export directory, and optionally --password-file PATH.".into()), |
| 37 | }; |
| 38 | let source_path = PathBuf::from(path); |
| 39 | let bytes = onestore::read_file(&source_path)?; |
| 40 | let store = Store::parse(&bytes)?; |
| 41 | let index = RevisionIndex::parse(&store)?; |
| 42 | let unlocked = if let Some(path) = &password_file { |
| 43 | use std::io::Read; |
| 44 | let mut password = zeroize::Zeroizing::new(String::new()); |
| 45 | fs::File::open(path)? |
| 46 | .take(65537) |
| 47 | .read_to_string(&mut password)?; |
| 48 | if password.len() > 65536 { |
| 49 | return Err("The password file exceeds 64 KiB.".into()); |
| 50 | } |
| 51 | Some(onestore::protected::UnlockedSection::open( |
| 52 | &index, |
| 53 | &password, |
| 54 | Default::default(), |
| 55 | )?) |
| 56 | } else { |
| 57 | None |
| 58 | }; |
| 59 | let document = match &unlocked { |
| 60 | Some(section) => section.document()?, |
| 61 | None => Document::parse(&index)?, |
| 62 | }; |
| 63 | if let Some(destination) = destination { |
| 64 | #[cfg(unix)] |
| 65 | { |
| 66 | use std::os::unix::fs::DirBuilderExt; |
| 67 | fs::DirBuilder::new() |
| 68 | .mode(if password_file.is_some() { |
| 69 | 0o700 |
| 70 | } else { |
| 71 | 0o777 |
| 72 | }) |
| 73 | .create(&destination)?; |
| 74 | } |
| 75 | #[cfg(not(unix))] |
| 76 | fs::create_dir(&destination)?; |
| 77 | fs::create_dir(destination.join("assets"))?; |
| 78 | let parent = source_path |
| 79 | .parent() |
| 80 | .filter(|parent| !parent.as_os_str().is_empty()) |
| 81 | .unwrap_or_else(|| Path::new(".")); |
| 82 | let mut source = notebook::discover::Local::open(parent)?; |
| 83 | let section = source_path |
| 84 | .file_name() |
| 85 | .and_then(|name| name.to_str()) |
| 86 | .ok_or("Use a UTF-8 section filename")?; |
| 87 | let mut assets = Vec::new(); |
| 88 | let mut seen = BTreeSet::new(); |
| 89 | for node in document |
| 90 | .spaces |
| 91 | .values() |
| 92 | .flat_map(|s| s.revisions.values()) |
| 93 | .flat_map(|r| r.nodes.values()) |
| 94 | { |
| 95 | if let Kind::File { |
| 96 | reference, payload, .. |
| 97 | } = &node.kind |
| 98 | && seen.insert(serde_json::to_string(reference)?) |
| 99 | { |
| 100 | let data = match reference { |
| 101 | FileDataReference::Internal(_) => { |
| 102 | Cow::Borrowed(payload.ok_or("Missing embedded file data")?) |
| 103 | } |
| 104 | FileDataReference::External(filename) => { |
| 105 | if password_file.is_some() { |
| 106 | assets.push(serde_json::json!({"reference":reference,"path":null,"error":{"kind":"Unsupported","message":"Protected external payload export is not supported"}})); |
| 107 | continue; |
| 108 | } |
| 109 | match notebook::discover::read_external_asset( |
| 110 | &mut source, |
| 111 | section, |
| 112 | filename, |
| 113 | 256 * 1024 * 1024, |
| 114 | ) { |
| 115 | Ok(bytes) => Cow::Owned(bytes), |
| 116 | Err(error) => { |
| 117 | let kind = match &error { |
| 118 | notebook::discover::Error::Io { error, .. } => { |
| 119 | format!("{:?}", error.kind()) |
| 120 | } |
| 121 | _ => "InvalidData".into(), |
| 122 | }; |
| 123 | assets.push(serde_json::json!({"reference":reference,"path":null,"error":{"kind":kind,"message":error.to_string()}})); |
| 124 | continue; |
| 125 | } |
| 126 | } |
| 127 | } |
| 128 | FileDataReference::Invalid => { |
| 129 | assets.push(serde_json::json!({"reference":reference,"path":null,"error":{"kind":"InvalidData","message":"The document marks this payload as unavailable"}})); |
| 130 | continue; |
| 131 | } |
| 132 | }; |
| 133 | let path = format!("assets/{}.bin", assets.len()); |
| 134 | fs::write(destination.join(&path), data)?; |
| 135 | assets.push(serde_json::json!({"reference":reference,"path":path})); |
| 136 | } |
| 137 | } |
| 138 | write_json(destination.join("assets.json"), &assets)?; |
| 139 | let mut text = std::collections::BTreeMap::new(); |
| 140 | for (sid, space) in &document.spaces { |
| 141 | let mut revisions = std::collections::BTreeMap::new(); |
| 142 | for (rid, revision) in &space.revisions { |
| 143 | let mut objects = std::collections::BTreeMap::new(); |
| 144 | for (oid, node) in &revision.nodes { |
| 145 | if matches!(node.kind, Kind::RichText { .. }) { |
| 146 | objects.insert(*oid, revision.text_runs(*oid)?); |
| 147 | } |
| 148 | } |
| 149 | revisions.insert(*rid, objects); |
| 150 | } |
| 151 | text.insert(*sid, revisions); |
| 152 | } |
| 153 | write_json(destination.join("text.json"), &text)?; |
| 154 | write_json(destination.join("document.json"), &document)?; |
| 155 | } else { |
| 156 | serde_json::to_writer(io::stdout().lock(), &document)?; |
| 157 | } |
| 158 | Ok(()) |
| 159 | } |