#![allow(unused_variables)]
-use endian::{Endian, Parse};
-use std::io::{BufReader, Error as IoError, Read};
+use endian::{Endian, Parse, ToBytes};
+use flate2::read::ZlibDecoder;
+use num::Integer;
+use num_derive::FromPrimitive;
+use std::{
+ collections::VecDeque,
+ io::{Error as IoError, Read, Seek, SeekFrom},
+ iter::FusedIterator,
+};
use thiserror::Error;
pub mod endian;
#[error("Not an SPSS system file")]
NotASystemFile,
- #[error("I/O error ({source})")]
- Io {
- #[from]
- source: IoError,
- },
+ #[error("Invalid magic number {0:?}")]
+ BadMagic([u8; 4]),
+
+ #[error("I/O error ({0})")]
+ Io(#[from] IoError),
#[error("Invalid SAV compression code {0}")]
InvalidSavCompression(u32),
#[error("Invalid ZSAV compression code {0}")]
InvalidZsavCompression(u32),
+
+ #[error("Variable record at offset {offset:#x} specifies width {width} not in valid range [-1,255).")]
+ BadVariableWidth { offset: u64, width: i32 },
+
+ #[error("Misplaced type 4 record near offset {0:#x}.")]
+ MisplacedType4Record(u64),
+
+ #[error("Document record at offset {offset:#x} has document line count ({n}) greater than the maximum number {max}.")]
+ BadDocumentLength { offset: u64, n: u32, max: u32 },
+
+ #[error("At offset {offset:#x}, Unrecognized record type {rec_type}.")]
+ BadRecordType { offset: u64, rec_type: u32 },
+
+ #[error("At offset {offset:#x}, variable label code ({code}) is not 0 or 1.")]
+ BadVariableLabelCode { offset: u64, code: u32 },
+
+ #[error(
+ "At offset {offset:#x}, numeric missing value code ({code}) is not -3, -2, 0, 1, 2, or 3."
+ )]
+ BadNumericMissingValueCode { offset: u64, code: i32 },
+
+ #[error("At offset {offset:#x}, string missing value code ({code}) is not 0, 1, 2, or 3.")]
+ BadStringMissingValueCode { offset: u64, code: i32 },
+
+ #[error("At offset {offset:#x}, number of value labels ({n}) is greater than the maximum number {max}.")]
+ BadNumberOfValueLabels { offset: u64, n: u32, max: u32 },
+
+ #[error("At offset {offset:#x}, variable index record (type 4) does not immediately follow value label record (type 3) as it should.")]
+ MissingVariableIndexRecord { offset: u64 },
+
+ #[error("At offset {offset:#x}, number of variables indexes ({n}) is greater than the maximum number ({max}).")]
+ BadNumberOfVarIndexes { offset: u64, n: u32, max: u32 },
+
+ #[error("At offset {offset:#x}, record type 7 subtype {subtype} is too large with element size {size} and {count} elements.")]
+ ExtensionRecordTooLarge {
+ offset: u64,
+ subtype: u32,
+ size: u32,
+ count: u32,
+ },
+
+ #[error("Wrong ZLIB data header offset {zheader_offset:#x} (expected {offset:#x}).")]
+ BadZlibHeaderOffset { offset: u64, zheader_offset: u64 },
+
+ #[error("At offset {offset:#x}, impossible ZLIB trailer offset {ztrailer_offset:#x}.")]
+ BadZlibTrailerOffset { offset: u64, ztrailer_offset: u64 },
+
+ #[error("At offset {offset:#x}, impossible ZLIB trailer length {ztrailer_len}.")]
+ BadZlibTrailerLen { offset: u64, ztrailer_len: u64 },
+
+ #[error("Unexpected end of file at offset {offset:#x}, {case_ofs} bytes into a {case_len}-byte case.")]
+ EofInCase {
+ offset: u64,
+ case_ofs: u64,
+ case_len: usize,
+ },
+
+ #[error(
+ "Unexpected end of file at offset {offset:#x}, {case_ofs} bytes into a compressed case."
+ )]
+ EofInCompressedCase { offset: u64, case_ofs: u64 },
+
+ #[error("Data ends at offset {offset:#x}, {case_ofs} bytes into a compressed case.")]
+ PartialCompressedCase { offset: u64, case_ofs: u64 },
+
+ #[error("At {case_ofs} bytes into compressed case starting at offset {offset:#x}, a string was found where a number was expected.")]
+ CompressedNumberExpected { offset: u64, case_ofs: u64 },
+
+ #[error("At {case_ofs} bytes into compressed case starting at offset {offset:#x}, a number was found where a string was expected.")]
+ CompressedStringExpected { offset: u64, case_ofs: u64 },
+
+ #[error("Block count {n_blocks} in ZLIB trailer at offset {offset:#x} differs from expected block count {expected_n_blocks} calculated from trailer length {ztrailer_len}.")]
+ BadZlibTrailerNBlocks {
+ offset: u64,
+ n_blocks: u32,
+ expected_n_blocks: u64,
+ ztrailer_len: u64,
+ },
}
#[derive(Error, Debug)]
pub enum Warning {
#[error("Unexpected floating-point bias {0} or unrecognized floating-point format.")]
UnexpectedBias(f64),
+
+ #[error("Duplicate type 6 (document) record.")]
+ DuplicateDocumentRecord,
}
#[derive(Copy, Clone, Debug)]
ZLib,
}
-pub struct Reader<R: Read> {
- r: BufReader<R>,
+pub enum Record {
+ Header(Header),
+ Document(Document),
+ Variable(Variable),
+ ValueLabel(ValueLabel),
+ VarIndexes(VarIndexes),
+ Extension(Extension),
+ EndOfHeaders,
+ ZHeader(ZHeader),
+ ZTrailer(ZTrailer),
+ Case(Vec<Value>),
}
-pub const ASCII_MAGIC: &[u8; 4] = b"$FL2";
-pub const ASCII_ZMAGIC: &[u8; 4] = b"$FL3";
-pub const EBCDIC_MAGIC: &[u8; 4] = &[0x5b, 0xc6, 0xd3, 0xf2];
+pub struct Header {
+ /// Magic number.
+ pub magic: Magic,
+
+ /// Eye-catcher string, product name, in the file's encoding. Padded
+ /// on the right with spaces.
+ pub eye_catcher: [u8; 60],
-pub struct FileHeader {
- /// First 4 bytes of the file, one of `ASCII_MAGIC`, `ASCII_ZMAGIC`, and
- /// `EBCDIC_MAGIC`.
- pub magic: [u8; 4],
+ /// Layout code, normally either 2 or 3.
+ pub layout_code: u32,
- /// True if `magic` indicates that this file contained zlib-compressed data.
- pub is_zsav: bool,
+ /// Number of variable positions, or `None` if the value in the file is
+ /// questionably trustworthy.
+ pub nominal_case_size: Option<u32>,
- /// True if `magic` indicates that this file contained EBCDIC data.
- pub is_ebcdic: bool,
+ /// Compression type, if any,
+ pub compression: Option<Compression>,
/// 0-based variable index of the weight variable, or `None` if the file is
/// unweighted.
pub weight_index: Option<u32>,
- /// Number of variable positions, or `None` if the value in the file is
- /// questionably trustworthy.
- pub nominal_case_size: Option<u32>,
+ /// Claimed number of cases, if known.
+ pub n_cases: Option<u32>,
+
+ /// Compression bias, usually 100.0.
+ pub bias: f64,
/// `dd mmm yy` in the file's encoding.
pub creation_date: [u8; 9],
/// `HH:MM:SS` in the file's encoding.
pub creation_time: [u8; 8],
- /// Eye-catcher string, then product name, in the file's encoding. Padded
- /// on the right with spaces.
- pub eye_catcher: [u8; 60],
-
/// File label, in the file's encoding. Padded on the right with spaces.
pub file_label: [u8; 64],
+
+ /// Endianness of the data in the file header.
+ pub endian: Endian,
}
-impl<R: Read> Reader<R> {
- pub fn new(r: R, warn: impl Fn(Warning)) -> Result<Reader<R>, Error> {
- let mut r = BufReader::new(r);
+#[derive(Copy, Clone, PartialEq, Eq, Hash)]
+pub struct Magic([u8; 4]);
- let magic: [u8; 4] = read_bytes(&mut r)?;
- let (is_zsav, is_ebcdic) = match &magic {
- ASCII_MAGIC => (false, false),
- ASCII_ZMAGIC => (true, false),
- EBCDIC_MAGIC => (false, true),
- _ => return Err(Error::NotASystemFile),
- };
+impl Magic {
+ /// Magic number for a regular system file.
+ pub const SAV: Magic = Magic(*b"$FL2");
- let eye_catcher: [u8; 60] = read_bytes(&mut r)?;
- let layout_code: [u8; 4] = read_bytes(&mut r)?;
- let endianness = Endian::identify_u32(2, layout_code)
- .or_else(|| Endian::identify_u32(2, layout_code))
- .ok_or_else(|| Error::NotASystemFile)?;
-
- let nominal_case_size: u32 = endianness.parse(read_bytes(&mut r)?);
- let nominal_case_size = (nominal_case_size <= u32::MAX / 32).then_some(nominal_case_size);
-
- let compression_code: u32 = endianness.parse(read_bytes(&mut r)?);
- let compression = match (is_zsav, compression_code) {
- (false, 0) => None,
- (false, 1) => Some(Compression::Simple),
- (true, 2) => Some(Compression::ZLib),
- (false, code) => return Err(Error::InvalidSavCompression(code)),
- (true, code) => return Err(Error::InvalidZsavCompression(code)),
- };
+ /// Magic number for a system file that contains zlib-compressed data.
+ pub const ZSAV: Magic = Magic(*b"$FL3");
- let weight_index: u32 = endianness.parse(read_bytes(&mut r)?);
- let weight_index = (weight_index > 0).then_some(weight_index - 1);
+ /// Magic number for an EBDIC-encoded system file. This is `$FL2` encoded
+ /// in EBCDIC.
+ pub const EBCDIC: Magic = Magic([0x5b, 0xc6, 0xd3, 0xf2]);
+}
+
+impl TryFrom<[u8; 4]> for Magic {
+ type Error = Error;
+
+ fn try_from(value: [u8; 4]) -> Result<Self, Self::Error> {
+ let magic = Magic(value);
+ match magic {
+ Magic::SAV | Magic::ZSAV | Magic::EBCDIC => Ok(magic),
+ _ => Err(Error::BadMagic(value)),
+ }
+ }
+}
- let n_cases: u32 = endianness.parse(read_bytes(&mut r)?);
- let n_cases = (n_cases <= u32::MAX / 4).then_some(n_cases);
+#[derive(Copy, Clone, PartialEq, Eq, Hash)]
+pub enum VarType {
+ Number,
+ String,
+}
- let bias: f64 = endianness.parse(read_bytes(&mut r)?);
- if bias != 100.0 {
- warn(Warning::UnexpectedBias(bias))
+impl VarType {
+ fn from_width(width: i32) -> VarType {
+ match width {
+ 0 => VarType::Number,
+ _ => VarType::String,
}
+ }
+}
+
+trait State {
+ fn read(self: Box<Self>) -> Result<Option<(Record, Box<dyn State>)>, Error>;
+}
+
+struct Start<R: Read + Seek> {
+ reader: R,
+}
+
+struct CommonState<R: Read + Seek> {
+ reader: R,
+ endian: Endian,
+ bias: f64,
+ compression: Option<Compression>,
+ var_types: Vec<VarType>,
+}
+
+impl<R: Read + Seek + 'static> State for Start<R> {
+ fn read(mut self: Box<Self>) -> Result<Option<(Record, Box<dyn State>)>, Error> {
+ let header = read_header(&mut self.reader)?;
+ let next_state = Headers(CommonState {
+ reader: self.reader,
+ endian: header.endian,
+ bias: header.bias,
+ compression: header.compression,
+ var_types: Vec::new(),
+ });
+ Ok(Some((Record::Header(header), Box::new(next_state))))
+ }
+}
- let creation_date: [u8; 9] = read_bytes(&mut r)?;
- let creation_time: [u8; 8] = read_bytes(&mut r)?;
- let file_label: [u8; 64] = read_bytes(&mut r)?;
- let _: [u8; 3] = read_bytes(&mut r)?;
-
- let header = FileHeader {
- magic,
- is_zsav,
- is_ebcdic,
- weight_index,
- nominal_case_size,
- creation_date,
- creation_time,
- eye_catcher,
- file_label,
+struct Headers<R: Read + Seek>(CommonState<R>);
+
+impl<R: Read + Seek + 'static> State for Headers<R> {
+ fn read(mut self: Box<Self>) -> Result<Option<(Record, Box<dyn State>)>, Error> {
+ let endian = self.0.endian;
+ let rec_type: u32 = endian.parse(read_bytes(&mut self.0.reader)?);
+ let record = match rec_type {
+ 2 => {
+ let variable = read_variable_record(&mut self.0.reader, endian)?;
+ self.0.var_types.push(VarType::from_width(variable.width));
+ Record::Variable(variable)
+ }
+ 3 => Record::ValueLabel(read_value_label_record(&mut self.0.reader, endian)?),
+ 4 => Record::VarIndexes(read_var_indexes_record(&mut self.0.reader, endian)?),
+ 6 => Record::Document(read_document_record(&mut self.0.reader, endian)?),
+ 7 => Record::Extension(read_extension_record(&mut self.0.reader, endian)?),
+ 999 => {
+ let _: [u8; 4] = read_bytes(&mut self.0.reader)?;
+ let next_state: Box<dyn State> = match self.0.compression {
+ None => Box::new(Data(self.0)),
+ Some(Compression::Simple) => Box::new(CompressedData::new(self.0)),
+ Some(Compression::ZLib) => Box::new(CompressedData::new(CommonState {
+ reader: ZlibDecodeMultiple::new(self.0.reader),
+ endian: self.0.endian,
+ bias: self.0.bias,
+ compression: self.0.compression,
+ var_types: self.0.var_types
+ })),
+ };
+ return Ok(Some((Record::EndOfHeaders, next_state)));
+ }
+ _ => {
+ return Err(Error::BadRecordType {
+ offset: self.0.reader.stream_position()?,
+ rec_type,
+ })
+ }
};
+ Ok(Some((record, self)))
+ }
+}
+
+struct Data<R: Read + Seek>(CommonState<R>);
+
+impl<R: Read + Seek + 'static> State for Data<R> {
+ fn read(mut self: Box<Self>) -> Result<Option<(Record, Box<dyn State>)>, Error> {
+ let case_start = self.0.reader.stream_position()?;
+ let mut values = Vec::with_capacity(self.0.var_types.len());
+ for (i, &var_type) in self.0.var_types.iter().enumerate() {
+ let Some(raw) = try_read_bytes(&mut self.0.reader)? else {
+ if i == 0 {
+ return Ok(None);
+ } else {
+ let offset = self.0.reader.stream_position()?;
+ return Err(Error::EofInCase {
+ offset,
+ case_ofs: offset - case_start,
+ case_len: self.0.var_types.len() * 8,
+ });
+ }
+ };
+ values.push(Value::from_raw(var_type, raw, self.0.endian));
+ }
+ Ok(Some((Record::Case(values), self)))
+ }
+}
+
+struct CompressedData<R: Read + Seek> {
+ common: CommonState<R>,
+ codes: VecDeque<u8>,
+}
+
+impl<R: Read + Seek + 'static> CompressedData<R> {
+ fn new(common: CommonState<R>) -> CompressedData<R> {
+ CompressedData { common, codes: VecDeque::new() }
+ }
+}
+
+impl<R: Read + Seek + 'static> State for CompressedData<R> {
+ fn read(mut self: Box<Self>) -> Result<Option<(Record, Box<dyn State>)>, Error> {
+ let case_start = self.common.reader.stream_position()?;
+ let mut values = Vec::with_capacity(self.common.var_types.len());
+ for (i, &var_type) in self.common.var_types.iter().enumerate() {
+ let value = loop {
+ let Some(code) = self.codes.pop_front() else {
+ let Some(new_codes): Option<[u8; 8]> = try_read_bytes(&mut self.common.reader)?
+ else {
+ if i == 0 {
+ return Ok(None);
+ } else {
+ let offset = self.common.reader.stream_position()?;
+ return Err(Error::EofInCompressedCase {
+ offset,
+ case_ofs: offset - case_start,
+ });
+ }
+ };
+ self.codes.extend(new_codes.into_iter());
+ continue;
+ };
+ match code {
+ 0 => (),
+ 1..=251 => match var_type {
+ VarType::Number => break Value::Number(Some(code as f64 - self.common.bias)),
+ VarType::String => {
+ break Value::String(self.common.endian.to_bytes(code as f64 - self.common.bias))
+ }
+ },
+ 252 => {
+ if i == 0 {
+ return Ok(None);
+ } else {
+ let offset = self.common.reader.stream_position()?;
+ return Err(Error::PartialCompressedCase {
+ offset,
+ case_ofs: offset - case_start,
+ });
+ }
+ }
+ 253 => {
+ break Value::from_raw(
+ var_type,
+ read_bytes(&mut self.common.reader)?,
+ self.common.endian,
+ )
+ }
+ 254 => match var_type {
+ VarType::String => break Value::String(*b" "), // XXX EBCDIC
+ VarType::Number => {
+ return Err(Error::CompressedStringExpected {
+ offset: case_start,
+ case_ofs: self.common.reader.stream_position()? - case_start,
+ })
+ }
+ },
+ 255 => match var_type {
+ VarType::Number => break Value::Number(None),
+ VarType::String => {
+ return Err(Error::CompressedNumberExpected {
+ offset: case_start,
+ case_ofs: self.common.reader.stream_position()? - case_start,
+ })
+ }
+ },
+ }
+ };
+ values.push(value);
+ }
+ Ok(Some((Record::Case(values), self)))
+ }
+}
+
+struct ZlibDecodeMultiple<R>
+where
+ R: Read + Seek,
+{
+ reader: Option<ZlibDecoder<R>>,
+}
+
+impl<R> ZlibDecodeMultiple<R>
+where
+ R: Read + Seek,
+{
+ fn new(reader: R) -> ZlibDecodeMultiple<R> {
+ ZlibDecodeMultiple {
+ reader: Some(ZlibDecoder::new(reader)),
+ }
+ }
+}
+
+impl<R> Read for ZlibDecodeMultiple<R>
+where
+ R: Read + Seek,
+{
+ fn read(&mut self, buf: &mut [u8]) -> Result<usize, IoError> {
+ loop {
+ match self.reader.as_mut().unwrap().read(buf)? {
+ 0 => {
+ let inner = self.reader.take().unwrap().into_inner();
+ self.reader = Some(ZlibDecoder::new(inner));
+ }
+ n => return Ok(n),
+ };
+ }
+ }
+}
+
+impl<R> Seek for ZlibDecodeMultiple<R>
+where
+ R: Read + Seek,
+{
+ fn seek(&mut self, pos: SeekFrom) -> Result<u64, IoError> {
+ unimplemented!();
+ }
+}
+
+#[derive(Copy, Clone)]
+pub enum Value {
+ Number(Option<f64>),
+ String([u8; 8]),
+}
+
+impl Value {
+ pub fn from_raw(var_type: VarType, raw: [u8; 8], endian: Endian) -> Value {
+ match var_type {
+ VarType::String => Value::String(raw),
+ VarType::Number => {
+ let number: f64 = endian.parse(raw);
+ Value::Number((number != -f64::MAX).then_some(number))
+ }
+ }
+ }
+}
+
+pub struct Reader {
+ state: Option<Box<dyn State>>,
+}
- Ok(Reader { r })
+impl Reader {
+ pub fn new<R: Read + Seek + 'static>(reader: R) -> Result<Reader, Error> {
+ Ok(Reader {
+ state: Some(Box::new(Start { reader })),
+ })
+ }
+}
+
+impl Iterator for Reader {
+ type Item = Result<Record, Error>;
+
+ fn next(&mut self) -> Option<Self::Item> {
+ match self.state.take()?.read() {
+ Ok(Some((record, next_state))) => {
+ self.state = Some(next_state);
+ return Some(Ok(record));
+ }
+ Ok(None) => return None,
+ Err(error) => return Some(Err(error)),
+ }
+ }
+}
+
+impl FusedIterator for Reader {}
+
+fn read_header<R: Read>(r: &mut R) -> Result<Header, Error> {
+ let magic: [u8; 4] = read_bytes(r)?;
+ let magic: Magic = magic.try_into().map_err(|_| Error::NotASystemFile)?;
+
+ let eye_catcher: [u8; 60] = read_bytes(r)?;
+ let layout_code: [u8; 4] = read_bytes(r)?;
+ let endian = Endian::identify_u32(2, layout_code)
+ .or_else(|| Endian::identify_u32(2, layout_code))
+ .ok_or_else(|| Error::NotASystemFile)?;
+ let layout_code = endian.parse(layout_code);
+
+ let nominal_case_size: u32 = endian.parse(read_bytes(r)?);
+ let nominal_case_size =
+ (nominal_case_size <= i32::MAX as u32 / 16).then_some(nominal_case_size);
+
+ let compression_code: u32 = endian.parse(read_bytes(r)?);
+ let compression = match (magic, compression_code) {
+ (Magic::ZSAV, 2) => Some(Compression::ZLib),
+ (Magic::ZSAV, code) => return Err(Error::InvalidZsavCompression(code)),
+ (_, 0) => None,
+ (_, 1) => Some(Compression::Simple),
+ (_, code) => return Err(Error::InvalidSavCompression(code)),
+ };
+
+ let weight_index: u32 = endian.parse(read_bytes(r)?);
+ let weight_index = (weight_index > 0).then_some(weight_index - 1);
+
+ let n_cases: u32 = endian.parse(read_bytes(r)?);
+ let n_cases = (n_cases < i32::MAX as u32 / 2).then_some(n_cases);
+
+ let bias: f64 = endian.parse(read_bytes(r)?);
+
+ let creation_date: [u8; 9] = read_bytes(r)?;
+ let creation_time: [u8; 8] = read_bytes(r)?;
+ let file_label: [u8; 64] = read_bytes(r)?;
+ let _: [u8; 3] = read_bytes(r)?;
+
+ Ok(Header {
+ magic,
+ layout_code,
+ nominal_case_size,
+ compression,
+ weight_index,
+ n_cases,
+ bias,
+ creation_date,
+ creation_time,
+ eye_catcher,
+ file_label,
+ endian,
+ })
+}
+
+pub struct Variable {
+ /// Offset from the start of the file to the start of the record.
+ pub offset: u64,
+
+ /// Variable width, in the range -1..=255.
+ pub width: i32,
+
+ /// Variable name, padded on the right with spaces.
+ pub name: [u8; 8],
+
+ /// Print format.
+ pub print_format: u32,
+
+ /// Write format.
+ pub write_format: u32,
+
+ /// Missing value code, one of -3, -2, 0, 1, 2, or 3.
+ pub missing_value_code: i32,
+
+ /// Raw missing values, up to 3 of them.
+ pub missing: Vec<[u8; 8]>,
+
+ /// Optional variable label.
+ pub label: Option<Vec<u8>>,
+}
+
+fn read_variable_record<R: Read + Seek>(r: &mut R, endian: Endian) -> Result<Variable, Error> {
+ let offset = r.stream_position()?;
+ let width: i32 = endian.parse(read_bytes(r)?);
+ let has_variable_label: u32 = endian.parse(read_bytes(r)?);
+ let missing_value_code: i32 = endian.parse(read_bytes(r)?);
+ let print_format: u32 = endian.parse(read_bytes(r)?);
+ let write_format: u32 = endian.parse(read_bytes(r)?);
+ let name: [u8; 8] = read_bytes(r)?;
+
+ let label = match has_variable_label {
+ 0 => None,
+ 1 => {
+ let len: u32 = endian.parse(read_bytes(r)?);
+ let read_len = len.min(65535) as usize;
+ let label = Some(read_vec(r, read_len)?);
+
+ let padding_bytes = Integer::next_multiple_of(&len, &4) - len;
+ let _ = read_vec(r, padding_bytes as usize)?;
+
+ label
+ }
+ _ => {
+ return Err(Error::BadVariableLabelCode {
+ offset,
+ code: has_variable_label,
+ })
+ }
+ };
+
+ let mut missing = Vec::new();
+ if missing_value_code != 0 {
+ match (width, missing_value_code) {
+ (0, -3 | -2 | 1 | 2 | 3) => (),
+ (0, _) => {
+ return Err(Error::BadNumericMissingValueCode {
+ offset,
+ code: missing_value_code,
+ })
+ }
+ (_, 0..=3) => (),
+ (_, _) => {
+ return Err(Error::BadStringMissingValueCode {
+ offset,
+ code: missing_value_code,
+ })
+ }
+ }
+
+ for _ in 0..missing_value_code.abs() {
+ missing.push(read_bytes(r)?);
+ }
+ }
+
+ Ok(Variable {
+ offset,
+ width,
+ name,
+ print_format,
+ write_format,
+ missing_value_code,
+ missing,
+ label,
+ })
+}
+
+pub struct ValueLabel {
+ /// Offset from the start of the file to the start of the record.
+ pub offset: u64,
+
+ /// The labels.
+ pub labels: Vec<([u8; 8], Vec<u8>)>,
+}
+
+impl ValueLabel {
+ /// Maximum number of value labels in a record.
+ pub const MAX: u32 = u32::MAX / 8;
+}
+
+fn read_value_label_record<R: Read + Seek>(r: &mut R, endian: Endian) -> Result<ValueLabel, Error> {
+ let offset = r.stream_position()?;
+ let n: u32 = endian.parse(read_bytes(r)?);
+ if n > ValueLabel::MAX {
+ return Err(Error::BadNumberOfValueLabels {
+ offset,
+ n,
+ max: ValueLabel::MAX,
+ });
+ }
+
+ let mut labels = Vec::new();
+ for _ in 0..n {
+ let value: [u8; 8] = read_bytes(r)?;
+ let label_len: u8 = endian.parse(read_bytes(r)?);
+ let label_len = label_len as usize;
+ let padded_len = Integer::next_multiple_of(&(label_len + 1), &8);
+
+ let mut label = read_vec(r, padded_len)?;
+ label.truncate(label_len);
+ labels.push((value, label));
+ }
+ Ok(ValueLabel { offset, labels })
+}
+
+pub struct VarIndexes {
+ /// Offset from the start of the file to the start of the record.
+ pub offset: u64,
+
+ /// The 0-based indexes of the variable indexes.
+ pub var_indexes: Vec<u32>,
+}
+
+impl VarIndexes {
+ /// Maximum number of variable indexes in a record.
+ pub const MAX: u32 = u32::MAX / 8;
+}
+
+fn read_var_indexes_record<R: Read + Seek>(r: &mut R, endian: Endian) -> Result<VarIndexes, Error> {
+ let offset = r.stream_position()?;
+ let n: u32 = endian.parse(read_bytes(r)?);
+ if n > VarIndexes::MAX {
+ return Err(Error::BadNumberOfVarIndexes {
+ offset,
+ n,
+ max: VarIndexes::MAX,
+ });
+ }
+ let mut var_indexes = Vec::with_capacity(n as usize);
+ for _ in 0..n {
+ var_indexes.push(endian.parse(read_bytes(r)?));
+ }
+
+ Ok(VarIndexes {
+ offset,
+ var_indexes,
+ })
+}
+
+pub const DOC_LINE_LEN: u32 = 80;
+pub const DOC_MAX_LINES: u32 = i32::MAX as u32 / DOC_LINE_LEN;
+
+pub struct Document {
+ /// Offset from the start of the file to the start of the record.
+ pub pos: u64,
+
+ /// The document, as an array of 80-byte lines.
+ pub lines: Vec<[u8; DOC_LINE_LEN as usize]>,
+}
+
+fn read_document_record<R: Read + Seek>(r: &mut R, endian: Endian) -> Result<Document, Error> {
+ let offset = r.stream_position()?;
+ let n: u32 = endian.parse(read_bytes(r)?);
+ match n {
+ 0..=DOC_MAX_LINES => {
+ let pos = r.stream_position()?;
+ let mut lines = Vec::with_capacity(n as usize);
+ for _ in 0..n {
+ let line: [u8; 80] = read_bytes(r)?;
+ lines.push(line);
+ }
+ Ok(Document { pos, lines })
+ }
+ _ => Err(Error::BadDocumentLength {
+ offset,
+ n,
+ max: DOC_MAX_LINES,
+ }),
+ }
+}
+
+#[derive(FromPrimitive)]
+enum ExtensionType {
+ /// Machine integer info.
+ Integer = 3,
+ /// Machine floating-point info.
+ Float = 4,
+ /// Variable sets.
+ VarSets = 5,
+ /// DATE.
+ Date = 6,
+ /// Multiple response sets.
+ Mrsets = 7,
+ /// SPSS Data Entry.
+ DataEntry = 8,
+ /// Extra product info text.
+ ProductInfo = 10,
+ /// Variable display parameters.
+ Display = 11,
+ /// Long variable names.
+ LongNames = 13,
+ /// Long strings.
+ LongStrings = 14,
+ /// Extended number of cases.
+ Ncases = 16,
+ /// Data file attributes.
+ FileAttrs = 17,
+ /// Variable attributes.
+ VarAttrs = 18,
+ /// Multiple response sets (extended).
+ Mrsets2 = 19,
+ /// Character encoding.
+ Encoding = 20,
+ /// Value labels for long strings.
+ LongLabels = 21,
+ /// Missing values for long strings.
+ LongMissing = 22,
+ /// "Format properties in dataview table".
+ Dataview = 24,
+}
+
+pub struct Extension {
+ /// Offset from the start of the file to the start of the record.
+ pub offset: u64,
+
+ /// Record subtype.
+ pub subtype: u32,
+
+ /// Size of each data element.
+ pub size: u32,
+
+ /// Number of data elements.
+ pub count: u32,
+
+ /// `size * count` bytes of data.
+ pub data: Vec<u8>,
+}
+
+fn extension_record_size_requirements(extension: ExtensionType) -> (u32, u32) {
+ match extension {
+ /* Implemented record types. */
+ ExtensionType::Integer => (4, 8),
+ ExtensionType::Float => (8, 3),
+ ExtensionType::VarSets => (1, 0),
+ ExtensionType::Mrsets => (1, 0),
+ ExtensionType::ProductInfo => (1, 0),
+ ExtensionType::Display => (4, 0),
+ ExtensionType::LongNames => (1, 0),
+ ExtensionType::LongStrings => (1, 0),
+ ExtensionType::Ncases => (8, 2),
+ ExtensionType::FileAttrs => (1, 0),
+ ExtensionType::VarAttrs => (1, 0),
+ ExtensionType::Mrsets2 => (1, 0),
+ ExtensionType::Encoding => (1, 0),
+ ExtensionType::LongLabels => (1, 0),
+ ExtensionType::LongMissing => (1, 0),
+
+ /* Ignored record types. */
+ ExtensionType::Date => (0, 0),
+ ExtensionType::DataEntry => (0, 0),
+ ExtensionType::Dataview => (0, 0),
+ }
+}
+
+fn read_extension_record<R: Read + Seek>(r: &mut R, endian: Endian) -> Result<Extension, Error> {
+ let subtype = endian.parse(read_bytes(r)?);
+ let offset = r.stream_position()?;
+ let size: u32 = endian.parse(read_bytes(r)?);
+ let count = endian.parse(read_bytes(r)?);
+ let Some(product) = size.checked_mul(count) else {
+ return Err(Error::ExtensionRecordTooLarge {
+ offset,
+ subtype,
+ size,
+ count,
+ });
+ };
+ let offset = r.stream_position()?;
+ let data = read_vec(r, product as usize)?;
+ Ok(Extension {
+ offset,
+ subtype,
+ size,
+ count,
+ data,
+ })
+}
+
+pub struct ZHeader {
+ /// File offset to the start of the record.
+ pub offset: u64,
+
+ /// File offset to the ZLIB data header.
+ pub zheader_offset: u64,
+
+ /// File offset to the ZLIB trailer.
+ pub ztrailer_offset: u64,
+
+ /// Length of the ZLIB trailer in bytes.
+ pub ztrailer_len: u64,
+}
+
+fn read_zheader<R: Read + Seek>(r: &mut R, endian: Endian) -> Result<ZHeader, Error> {
+ let offset = r.stream_position()?;
+ let zheader_offset: u64 = endian.parse(read_bytes(r)?);
+ let ztrailer_offset: u64 = endian.parse(read_bytes(r)?);
+ let ztrailer_len: u64 = endian.parse(read_bytes(r)?);
+
+ Ok(ZHeader {
+ offset,
+ zheader_offset,
+ ztrailer_offset,
+ ztrailer_len,
+ })
+}
+
+pub struct ZTrailer {
+ /// File offset to the start of the record.
+ pub offset: u64,
+
+ /// Compression bias as a negative integer, e.g. -100.
+ pub int_bias: i64,
+
+ /// Always observed as zero.
+ pub zero: u64,
+
+ /// Uncompressed size of each block, except possibly the last. Only
+ /// `0x3ff000` has been observed so far.
+ pub block_size: u32,
+
+ /// Block descriptors, always `(ztrailer_len - 24) / 24)` of them.
+ pub blocks: Vec<ZBlock>,
+}
+
+pub struct ZBlock {
+ /// Offset of block of data if simple compression were used.
+ pub uncompressed_ofs: u64,
+
+ /// Actual offset within the file of the compressed data block.
+ pub compressed_ofs: u64,
+
+ /// The number of bytes in this data block after decompression. This is
+ /// `block_size` in every data block but the last, which may be smaller.
+ pub uncompressed_size: u32,
+
+ /// The number of bytes in this data block, as stored compressed in this
+ /// file.
+ pub compressed_size: u32,
+}
+
+fn read_ztrailer<R: Read + Seek>(
+ r: &mut R,
+ endian: Endian,
+ ztrailer_ofs: u64,
+ ztrailer_len: u64,
+) -> Result<Option<ZTrailer>, Error> {
+ let start_offset = r.stream_position()?;
+ if r.seek(SeekFrom::Start(ztrailer_ofs)).is_err() {
+ return Ok(None);
+ }
+ let int_bias = endian.parse(read_bytes(r)?);
+ let zero = endian.parse(read_bytes(r)?);
+ let block_size = endian.parse(read_bytes(r)?);
+ let n_blocks: u32 = endian.parse(read_bytes(r)?);
+ let expected_n_blocks = (ztrailer_len - 24) / 24;
+ if n_blocks as u64 != expected_n_blocks {
+ return Err(Error::BadZlibTrailerNBlocks {
+ offset: ztrailer_ofs,
+ n_blocks,
+ expected_n_blocks,
+ ztrailer_len,
+ });
+ }
+ let mut blocks = Vec::with_capacity(n_blocks as usize);
+ for _ in 0..n_blocks {
+ let uncompressed_ofs = endian.parse(read_bytes(r)?);
+ let compressed_ofs = endian.parse(read_bytes(r)?);
+ let uncompressed_size = endian.parse(read_bytes(r)?);
+ let compressed_size = endian.parse(read_bytes(r)?);
+ blocks.push(ZBlock {
+ uncompressed_ofs,
+ compressed_ofs,
+ uncompressed_size,
+ compressed_size,
+ });
+ }
+ r.seek(SeekFrom::Start(start_offset))?;
+ Ok(Some(ZTrailer {
+ offset: ztrailer_ofs,
+ int_bias,
+ zero,
+ block_size,
+ blocks,
+ }))
+}
+
+fn try_read_bytes<const N: usize, R: Read>(r: &mut R) -> Result<Option<[u8; N]>, IoError> {
+ let mut buf = [0; N];
+ let n = r.read(&mut buf)?;
+ if n > 0 {
+ if n < N {
+ r.read_exact(&mut buf[n..])?;
+ }
+ Ok(Some(buf))
+ } else {
+ Ok(None)
}
}
Ok(buf)
}
+fn read_vec<R: Read>(r: &mut R, n: usize) -> Result<Vec<u8>, IoError> {
+ let mut vec = vec![0; n];
+ r.read_exact(&mut vec)?;
+ Ok(vec)
+}
+
/*
fn trim_end(mut s: Vec<u8>, c: u8) -> Vec<u8> {
while s.last() == Some(&c) {
projects / pspp / blobdiff