-#![allow(unused_variables)]
-use endian::{Endian, Parse};
-use num::Integer;
-use num_derive::FromPrimitive;
-use std::io::{BufReader, Error as IoError, Read, Seek};
-use thiserror::Error;
+use std::io::Error as IoError;
+use thiserror::Error as ThisError;
pub mod endian;
+pub mod raw;
+pub mod cooked;
+pub mod sack;
+pub mod encoding;
+pub mod format;
-#[derive(Error, Debug)]
+#[derive(ThisError, Debug)]
pub enum Error {
#[error("Not an SPSS system file")]
NotASystemFile,
#[error("Invalid magic number {0:?}")]
BadMagic([u8; 4]),
- #[error("I/O error ({source})")]
- Io {
- #[from]
- source: IoError,
- },
+ #[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("Misplaced type 4 record near offset {0:#x}.")]
- MisplacedType4Record(u64),
+ #[error("Variable record at offset {offset:#x} specifies width {width} not in valid range [-1,255).")]
+ BadVariableWidth { offset: u64, width: i32 },
#[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}.")]
+ #[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.")]
#[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 associated with a value label ({n}) is not between 1 and the number of variables ({max}).")]
- BadNumberOfValueLabelVariables { offset: u64, n: u32, max: u32 },
+ #[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 {
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 },
-}
-
-#[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)]
-pub enum Compression {
- Simple,
- ZLib,
-}
-
-pub struct Reader<R: Read> {
- r: BufReader<R>,
- documents: Vec<DocumentRecord>,
- variables: Vec<VariableRecord>,
- value_labels: Vec<ValueLabelRecord>,
- extensions: Vec<ExtensionRecord>,
- zheader: Option<ZHeader>,
-}
-
-pub struct FileHeader {
- /// Magic number.
- pub magic: Magic,
-
- /// Endianness of the data in the file header.
- pub endianness: Endian,
-
- /// 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>,
-
- /// `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],
-}
-
-#[derive(Copy, Clone, PartialEq, Eq, Hash)]
-pub struct Magic([u8; 4]);
-
-impl Magic {
- /// Magic number for a regular system file.
- pub const SAV: Magic = Magic(*b"$FL2");
-
- /// Magic number for a system file that contains zlib-compressed data.
- pub const ZSAV: Magic = Magic(*b"$FL3");
-
- /// 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)),
- }
- }
-}
-
-impl<R: Read + Seek> Reader<R> {
- pub fn new(r: R, warn: impl Fn(Warning)) -> Result<Reader<R>, Error> {
- let mut r = BufReader::new(r);
-
- let header = read_header(&mut r, &warn)?;
- let e = header.endianness;
- let mut documents = Vec::new();
- let mut variables = Vec::new();
- let mut value_labels = Vec::new();
- let mut extensions = Vec::new();
- loop {
- let offset = r.stream_position()?;
- let rec_type: u32 = e.parse(read_bytes(&mut r)?);
- match rec_type {
- 2 => variables.push(read_variable_record(&mut r, e)?),
- 3 => value_labels.push(read_value_label_record(&mut r, e, variables.len())?),
- 4 => return Err(Error::MisplacedType4Record(offset)),
- 6 => documents.push(read_document_record(&mut r, e)?),
- 7 => extensions.push(read_extension_record(&mut r, e)?),
- 999 => break,
- _ => return Err(Error::BadRecordType { offset, rec_type }),
- }
- }
- let _: [u8; 4] = read_bytes(&mut r)?;
- let zheader = match header.magic {
- Magic::ZSAV => Some(read_zheader(&mut r, e)?),
- _ => None,
- };
-
- Ok(Reader {
- r,
- documents,
- variables,
- value_labels,
- extensions,
- zheader,
- })
- }
-}
-
-fn read_header<R: Read>(r: &mut R, warn: impl Fn(Warning)) -> Result<FileHeader, 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 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(r)?);
- let nominal_case_size =
- (nominal_case_size <= i32::MAX as u32 / 16).then_some(nominal_case_size);
-
- let compression_code: u32 = endianness.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 = endianness.parse(read_bytes(r)?);
- let weight_index = (weight_index > 0).then_some(weight_index - 1);
-
- let n_cases: u32 = endianness.parse(read_bytes(r)?);
- let n_cases = (n_cases < i32::MAX as u32 / 2).then_some(n_cases);
-
- let bias: f64 = endianness.parse(read_bytes(r)?);
- if bias != 100.0 {
- warn(Warning::UnexpectedBias(bias))
- }
-
- 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(FileHeader {
- magic,
- endianness,
- weight_index,
- nominal_case_size,
- creation_date,
- creation_time,
- eye_catcher,
- file_label,
- })
-}
-
-pub struct VariableRecord {
- /// 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 BufReader<R>,
- e: Endian,
-) -> Result<VariableRecord, Error> {
- let offset = r.stream_position()?;
- let width: i32 = e.parse(read_bytes(r)?);
- let has_variable_label: u32 = e.parse(read_bytes(r)?);
- let missing_value_code: i32 = e.parse(read_bytes(r)?);
- let print_format: u32 = e.parse(read_bytes(r)?);
- let write_format: u32 = e.parse(read_bytes(r)?);
- let name: [u8; 8] = read_bytes(r)?;
-
- let label = match has_variable_label {
- 0 => None,
- 1 => {
- let len: u32 = e.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(VariableRecord {
- offset,
- width,
- name,
- print_format,
- write_format,
- missing_value_code,
- missing,
- label,
- })
-}
-
-pub struct ValueLabelRecord {
- /// 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>)>,
-
- /// The 0-based indexes of the variables to which the labels are assigned.
- pub var_indexes: Vec<u32>,
-}
-
-pub const MAX_VALUE_LABELS: u32 = u32::MAX / 8;
-
-fn read_value_label_record<R: Read + Seek>(
- r: &mut BufReader<R>,
- e: Endian,
- n_var_records: usize,
-) -> Result<ValueLabelRecord, Error> {
- let offset = r.stream_position()?;
- let n: u32 = e.parse(read_bytes(r)?);
- if n > MAX_VALUE_LABELS {
- return Err(Error::BadNumberOfValueLabels {
- offset,
- n,
- max: MAX_VALUE_LABELS,
- });
- }
-
- let mut labels = Vec::new();
- for _ in 0..n {
- let value: [u8; 8] = read_bytes(r)?;
- let label_len: u8 = e.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));
- }
-
- let rec_type: u32 = e.parse(read_bytes(r)?);
- if rec_type != 4 {
- return Err(Error::MissingVariableIndexRecord {
- offset: r.stream_position()?,
- });
- }
-
- let n_vars: u32 = e.parse(read_bytes(r)?);
- if n_vars < 1 || n_vars as usize > n_var_records {
- return Err(Error::BadNumberOfValueLabelVariables {
- offset: r.stream_position()?,
- n: n_vars,
- max: n_var_records as u32,
- });
- }
- let mut var_indexes = Vec::with_capacity(n_vars as usize);
- for _ in 0..n_vars {
- var_indexes.push(e.parse(read_bytes(r)?));
- }
-
- Ok(ValueLabelRecord {
- offset,
- labels,
- var_indexes,
- })
-}
-
-pub const DOC_LINE_LEN: u32 = 80;
-pub const DOC_MAX_LINES: u32 = i32::MAX as u32 / DOC_LINE_LEN;
-
-pub struct DocumentRecord {
- /// 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 BufReader<R>,
- e: Endian,
-) -> Result<DocumentRecord, Error> {
- let offset = r.stream_position()?;
- let n: u32 = e.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(DocumentRecord { pos, lines })
- }
- _ => Err(Error::BadDocumentLength {
- offset,
- n,
- max: DOC_MAX_LINES,
- }),
- }
-}
-
-#[derive(FromPrimitive)]
-enum Extension {
- /// 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,
-}
-
-struct ExtensionRecord {
- /// Offset from the start of the file to the start of the record.
- offset: u64,
-
- /// Record subtype.
- subtype: u32,
-
- /// Size of each data element.
- size: u32,
-
- /// Number of data elements.
- count: u32,
+ #[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,
+ },
- /// `size * count` bytes of data.
- data: Vec<u8>,
-}
+ #[error(
+ "Unexpected end of file at offset {offset:#x}, {case_ofs} bytes into a compressed case."
+ )]
+ EofInCompressedCase { offset: u64, case_ofs: u64 },
-fn extension_record_size_requirements(extension: Extension) -> (u32, u32) {
- match extension {
- /* Implemented record types. */
- Extension::Integer => (4, 8),
- Extension::Float => (8, 3),
- Extension::VarSets => (1, 0),
- Extension::Mrsets => (1, 0),
- Extension::ProductInfo => (1, 0),
- Extension::Display => (4, 0),
- Extension::LongNames => (1, 0),
- Extension::LongStrings => (1, 0),
- Extension::Ncases => (8, 2),
- Extension::FileAttrs => (1, 0),
- Extension::VarAttrs => (1, 0),
- Extension::Mrsets2 => (1, 0),
- Extension::Encoding => (1, 0),
- Extension::LongLabels => (1, 0),
- Extension::LongMissing => (1, 0),
+ #[error("Data ends at offset {offset:#x}, {case_ofs} bytes into a compressed case.")]
+ PartialCompressedCase { offset: u64, case_ofs: u64 },
- /* Ignored record types. */
- Extension::Date => (0, 0),
- Extension::DataEntry => (0, 0),
- Extension::Dataview => (0, 0),
- }
-}
+ #[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 },
-fn read_extension_record<R: Read + Seek>(
- r: &mut BufReader<R>,
- e: Endian,
-) -> Result<ExtensionRecord, Error> {
- let subtype = e.parse(read_bytes(r)?);
- let offset = r.stream_position()?;
- let size: u32 = e.parse(read_bytes(r)?);
- let count = e.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(ExtensionRecord {
- offset,
- subtype,
- size,
- count,
- data,
- })
-}
+ #[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 },
-struct ZHeader {
- /// File offset to the start of the record.
- offset: 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,
+ },
- /// File offset to the ZLIB data header.
- zheader_offset: u64,
+ #[error("At offset {offset:#x}, {record} has bad size {size} bytes instead of the expected {expected_size}.")]
+ BadRecordSize { offset: u64, record: String, size: u32, expected_size: u32 },
- /// File offset to the ZLIB trailer.
- ztrailer_offset: u64,
+ #[error("At offset {offset:#x}, {record} has bad count {count} instead of the expected {expected_count}.")]
+ BadRecordCount { offset: u64, record: String, count: u32, expected_count: u32 },
- /// Length of the ZLIB trailer in bytes.
- ztrailer_len: u64,
-}
+ #[error("The encoding record at offset {offset:#x} contains an encoding name that is not valid UTF-8.")]
+ BadEncodingName { offset: u64 },
-fn read_zheader<R: Read + Seek>(r: &mut BufReader<R>, e: Endian) -> Result<ZHeader, Error> {
- let offset = r.stream_position()?;
- let zheader_offset: u64 = e.parse(read_bytes(r)?);
- let ztrailer_offset: u64 = e.parse(read_bytes(r)?);
- let ztrailer_len: u64 = e.parse(read_bytes(r)?);
+ #[error("In long string missing values record starting at offset {record_offset:#x}, value length at offset {offset:#x} is {value_len} instead of the expected 8.")]
+ BadLongMissingValueLength { record_offset: u64, offset: u64, value_len: u32 },
- if zheader_offset != offset {
- return Err(Error::BadZlibHeaderOffset {
- offset,
- zheader_offset,
- });
- }
- if ztrailer_offset < offset {
- return Err(Error::BadZlibTrailerOffset {
- offset,
- ztrailer_offset,
- });
- }
- if ztrailer_len < 24 || ztrailer_len % 24 != 0 {
- return Err(Error::BadZlibTrailerLen {
- offset,
- ztrailer_len,
- });
- }
+ #[error("This file has corrupted metadata written by a buggy version of PSPP. To ensure that other software can read it correctly, save a new copy of the file.")]
+ BadLongMissingValueFormat,
- Ok(ZHeader {
- offset,
- zheader_offset,
- ztrailer_offset,
- ztrailer_len,
- })
-}
+ #[error("File creation date {creation_date} is not in the expected format \"DD MMM YY\" format. Using 01 Jan 1970.")]
+ InvalidCreationDate { creation_date: String },
-fn read_bytes<const N: usize, R: Read>(r: &mut R) -> Result<[u8; N], IoError> {
- let mut buf = [0; N];
- r.read_exact(&mut buf)?;
- Ok(buf)
-}
+ #[error("File creation time {creation_time} is not in the expected format \"HH:MM:SS\" format. Using midnight.")]
+ InvalidCreationTime { creation_time: String },
-fn read_vec<R: Read>(r: &mut BufReader<R>, n: usize) -> Result<Vec<u8>, IoError> {
- let mut vec = vec![0; n];
- r.read_exact(&mut vec)?;
- Ok(vec)
+ #[error("Details TBD")]
+ TBD,
}
-/*
-fn trim_end(mut s: Vec<u8>, c: u8) -> Vec<u8> {
- while s.last() == Some(&c) {
- s.pop();
- }
- s
+#[derive(Copy, Clone, Debug)]
+pub enum Compression {
+ Simple,
+ ZLib,
}
-fn skip_bytes<R: Read>(r: &mut R, mut n: u64) -> Result<(), IoError> {
- let mut buf = [0; 1024];
- while n > 0 {
- let chunk = u64::min(n, buf.len() as u64);
- r.read_exact(&mut buf[0..chunk as usize])?;
- n -= chunk;
- }
- Ok(())
+#[derive(Clone, Debug)]
+pub enum CategoryLabels {
+ VarLabels,
+ CountedValues,
}
-
-*/
projects / pspp / blobdiff