Add file names and line numbers to error messages.
[pspp] / rust / src / lib.rs
index 0b8c541e54612fcee67b45596a6370ec128f5c5b..8bf72fe02c064c8dcb5f30ff44c8cd6208ec4ba2 100644 (file)
@@ -1,11 +1,16 @@
-#![allow(unused_variables)]
-use endian::{Endian, Parse};
+use endian::{Endian, Parse, ToBytes};
+use flate2::read::ZlibDecoder;
 use num::Integer;
 use num_derive::FromPrimitive;
-use std::io::{BufReader, Error as IoError, ErrorKind, Read, Seek};
+use std::{
+    collections::VecDeque,
+    io::{Error as IoError, Read, Seek, SeekFrom},
+    iter::FusedIterator,
+};
 use thiserror::Error;
 
 pub mod endian;
+pub mod sack;
 
 #[derive(Error, Debug)]
 pub enum Error {
@@ -16,10 +21,7 @@ pub enum Error {
     BadMagic([u8; 4]),
 
     #[error("I/O error ({0})")]
-    Io(
-        #[from]
-        IoError,
-    ),
+    Io(#[from] IoError),
 
     #[error("Invalid SAV compression code {0}")]
     InvalidSavCompression(u32),
@@ -75,6 +77,35 @@ pub enum Error {
 
     #[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)]
@@ -100,6 +131,8 @@ pub enum Record {
     VarIndexes(VarIndexes),
     Extension(Extension),
     EndOfHeaders,
+    ZHeader(ZHeader),
+    ZTrailer(ZTrailer),
     Case(Vec<Value>),
 }
 
@@ -141,7 +174,7 @@ pub struct Header {
     pub file_label: [u8; 64],
 
     /// Endianness of the data in the file header.
-    pub endianness: Endian,
+    pub endian: Endian,
 }
 
 #[derive(Copy, Clone, PartialEq, Eq, Hash)]
@@ -186,17 +219,253 @@ impl VarType {
     }
 }
 
-pub struct Reader<R: Read> {
-    r: BufReader<R>,
+trait State {
+    #[allow(clippy::type_complexity)]
+    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>,
-    state: ReaderState,
 }
 
-enum ReaderState {
-    Start,
-    Headers(Endian, Option<Compression>),
-    Data(Endian),
-    End,
+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))))
+    }
+}
+
+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(ZlibHeader(self.0)),
+                };
+                return Ok(Some((Record::EndOfHeaders, next_state)));
+            }
+            _ => {
+                return Err(Error::BadRecordType {
+                    offset: self.0.reader.stream_position()?,
+                    rec_type,
+                })
+            }
+        };
+        Ok(Some((record, self)))
+    }
+}
+
+struct ZlibHeader<R: Read + Seek>(CommonState<R>);
+
+impl<R: Read + Seek + 'static> State for ZlibHeader<R> {
+    fn read(mut self: Box<Self>) -> Result<Option<(Record, Box<dyn State>)>, Error> {
+        let zheader = read_zheader(&mut self.0.reader, self.0.endian)?;
+        Ok(Some((Record::ZHeader(zheader), self)))
+    }
+}
+
+struct ZlibTrailer<R: Read + Seek>(CommonState<R>, ZHeader);
+
+impl<R: Read + Seek + 'static> State for ZlibTrailer<R> {
+    fn read(mut self: Box<Self>) -> Result<Option<(Record, Box<dyn State>)>, Error> {
+        let retval = read_ztrailer(&mut self.0.reader, self.0.endian, self.1.ztrailer_offset, self.1.ztrailer_len)?;
+        let next_state = 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
+        }));
+        match retval {
+            None => next_state.read(),
+            Some(ztrailer) => Ok(Some((Record::ZTrailer(ztrailer), next_state)))
+        }        
+    }
+}
+
+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> {
+        self.reader.as_mut().unwrap().get_mut().seek(pos)
+    }
 }
 
 #[derive(Copy, Clone)]
@@ -217,110 +486,51 @@ impl Value {
     }
 }
 
-impl<R: Read + Seek> Reader<R> {
-    pub fn new(r: R) -> Result<Reader<R>, Error> {
+pub struct Reader {
+    state: Option<Box<dyn State>>,
+}
+
+impl Reader {
+    pub fn new<R: Read + Seek + 'static>(reader: R) -> Result<Reader, Error> {
         Ok(Reader {
-            r: BufReader::new(r),
-            var_types: Vec::new(),
-            state: ReaderState::Start,
+            state: Some(Box::new(Start { reader })),
         })
     }
-    fn _next(&mut self) -> Result<Option<(Record, ReaderState)>, Error> {
-        match self.state {
-            ReaderState::Start => {
-                let header = read_header(&mut self.r)?;
-                let next_state = ReaderState::Headers(header.endianness, header.compression);
-                Ok(Some((Record::Header(header), next_state)))
-            }
-            ReaderState::Headers(endian, compression) => {
-                let rec_type: u32 = endian.parse(read_bytes(&mut self.r)?);
-                let record = match rec_type {
-                    2 => {
-                        let variable = read_variable_record(&mut self.r, endian)?;
-                        self.var_types.push(VarType::from_width(variable.width));
-                        Record::Variable(variable)
-                    }
-                    3 => Record::ValueLabel(read_value_label_record(&mut self.r, endian)?),
-                    4 => Record::VarIndexes(read_var_indexes_record(&mut self.r, endian)?),
-                    6 => Record::Document(read_document_record(&mut self.r, endian)?),
-                    7 => Record::Extension(read_extension_record(&mut self.r, endian)?),
-                    999 => {
-                        let _: [u8; 4] = read_bytes(&mut self.r)?;
-                        let next_state = match compression {
-                            None => ReaderState::Data(endian),
-                            _ => ReaderState::End,
-                        };
-                        return Ok(Some((Record::EndOfHeaders, next_state)));
-                    }
-                    _ => {
-                        return Err(Error::BadRecordType {
-                            offset: self.r.stream_position()?,
-                            rec_type,
-                        })
-                    }
-                };
-                Ok(Some((record, ReaderState::Headers(endian, compression))))
-            }
-            ReaderState::Data(endian) => {
-                let mut values = Vec::with_capacity(self.var_types.len());
-                for (i, &var_type) in self.var_types.iter().enumerate() {
-                    let raw = match read_bytes(&mut self.r) {
-                        Ok(raw) => raw,
-                        Err(err) => {
-                            if i == 0 && err.kind() == ErrorKind::UnexpectedEof {
-                                return Ok(None);
-                            } else {
-                                return Err(Error::Io(err));
-                            }
-                        }
-                    };
-                    values.push(Value::from_raw(var_type, raw, endian));
-                }
-                Ok(Some((Record::Case(values), ReaderState::Data(endian))))
-            }
-            ReaderState::End => Ok(None),
-        }
-    }
 }
 
-impl<R: Read + Seek> Iterator for Reader<R> {
+impl Iterator for Reader {
     type Item = Result<Record, Error>;
 
     fn next(&mut self) -> Option<Self::Item> {
-        let retval = self._next();
-        match retval {
-            Ok(None) => {
-                self.state = ReaderState::End;
-                None
-            }
+        match self.state.take()?.read() {
             Ok(Some((record, next_state))) => {
-                self.state = next_state;
+                self.state = Some(next_state);
                 Some(Ok(record))
             }
-            Err(error) => {
-                self.state = ReaderState::End;
-                Some(Err(error))
-            }
+            Ok(None) => None,
+            Err(error) => 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 endianness = Endian::identify_u32(2, layout_code)
+    let endian = Endian::identify_u32(2, layout_code)
         .or_else(|| Endian::identify_u32(2, layout_code))
         .ok_or_else(|| Error::NotASystemFile)?;
-    let layout_code = endianness.parse(layout_code);
+    let layout_code = endian.parse(layout_code);
 
-    let nominal_case_size: u32 = endianness.parse(read_bytes(r)?);
+    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 = endianness.parse(read_bytes(r)?);
+    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)),
@@ -329,13 +539,13 @@ fn read_header<R: Read>(r: &mut R) -> Result<Header, Error> {
         (_, code) => return Err(Error::InvalidSavCompression(code)),
     };
 
-    let weight_index: u32 = endianness.parse(read_bytes(r)?);
+    let weight_index: u32 = endian.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: u32 = endian.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)?);
+    let bias: f64 = endian.parse(read_bytes(r)?);
 
     let creation_date: [u8; 9] = read_bytes(r)?;
     let creation_time: [u8; 8] = read_bytes(r)?;
@@ -354,7 +564,7 @@ fn read_header<R: Read>(r: &mut R) -> Result<Header, Error> {
         creation_time,
         eye_catcher,
         file_label,
-        endianness,
+        endian,
     })
 }
 
@@ -384,22 +594,19 @@ pub struct Variable {
     pub label: Option<Vec<u8>>,
 }
 
-fn read_variable_record<R: Read + Seek>(
-    r: &mut BufReader<R>,
-    e: Endian,
-) -> Result<Variable, Error> {
+fn read_variable_record<R: Read + Seek>(r: &mut R, endian: Endian) -> Result<Variable, 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 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 = e.parse(read_bytes(r)?);
+            let len: u32 = endian.parse(read_bytes(r)?);
             let read_len = len.min(65535) as usize;
             let label = Some(read_vec(r, read_len)?);
 
@@ -465,12 +672,9 @@ impl ValueLabel {
     pub const MAX: u32 = u32::MAX / 8;
 }
 
-fn read_value_label_record<R: Read + Seek>(
-    r: &mut BufReader<R>,
-    e: Endian,
-) -> Result<ValueLabel, Error> {
+fn read_value_label_record<R: Read + Seek>(r: &mut R, endian: Endian) -> Result<ValueLabel, Error> {
     let offset = r.stream_position()?;
-    let n: u32 = e.parse(read_bytes(r)?);
+    let n: u32 = endian.parse(read_bytes(r)?);
     if n > ValueLabel::MAX {
         return Err(Error::BadNumberOfValueLabels {
             offset,
@@ -482,7 +686,7 @@ fn read_value_label_record<R: Read + Seek>(
     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: u8 = endian.parse(read_bytes(r)?);
         let label_len = label_len as usize;
         let padded_len = Integer::next_multiple_of(&(label_len + 1), &8);
 
@@ -506,12 +710,9 @@ impl VarIndexes {
     pub const MAX: u32 = u32::MAX / 8;
 }
 
-fn read_var_indexes_record<R: Read + Seek>(
-    r: &mut BufReader<R>,
-    e: Endian,
-) -> Result<VarIndexes, Error> {
+fn read_var_indexes_record<R: Read + Seek>(r: &mut R, endian: Endian) -> Result<VarIndexes, Error> {
     let offset = r.stream_position()?;
-    let n: u32 = e.parse(read_bytes(r)?);
+    let n: u32 = endian.parse(read_bytes(r)?);
     if n > VarIndexes::MAX {
         return Err(Error::BadNumberOfVarIndexes {
             offset,
@@ -521,7 +722,7 @@ fn read_var_indexes_record<R: Read + Seek>(
     }
     let mut var_indexes = Vec::with_capacity(n as usize);
     for _ in 0..n {
-        var_indexes.push(e.parse(read_bytes(r)?));
+        var_indexes.push(endian.parse(read_bytes(r)?));
     }
 
     Ok(VarIndexes {
@@ -541,12 +742,9 @@ pub struct Document {
     pub lines: Vec<[u8; DOC_LINE_LEN as usize]>,
 }
 
-fn read_document_record<R: Read + Seek>(
-    r: &mut BufReader<R>,
-    e: Endian,
-) -> Result<Document, Error> {
+fn read_document_record<R: Read + Seek>(r: &mut R, endian: Endian) -> Result<Document, Error> {
     let offset = r.stream_position()?;
-    let n: u32 = e.parse(read_bytes(r)?);
+    let n: u32 = endian.parse(read_bytes(r)?);
     match n {
         0..=DOC_MAX_LINES => {
             let pos = r.stream_position()?;
@@ -648,14 +846,11 @@ fn extension_record_size_requirements(extension: ExtensionType) -> (u32, u32) {
     }
 }
 
-fn read_extension_record<R: Read + Seek>(
-    r: &mut BufReader<R>,
-    e: Endian,
-) -> Result<Extension, Error> {
-    let subtype = e.parse(read_bytes(r)?);
+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 = e.parse(read_bytes(r)?);
-    let count = e.parse(read_bytes(r)?);
+    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,
@@ -675,25 +870,25 @@ fn read_extension_record<R: Read + Seek>(
     })
 }
 
-struct ZHeader {
+pub struct ZHeader {
     /// File offset to the start of the record.
-    offset: u64,
+    pub offset: u64,
 
     /// File offset to the ZLIB data header.
-    zheader_offset: u64,
+    pub zheader_offset: u64,
 
     /// File offset to the ZLIB trailer.
-    ztrailer_offset: u64,
+    pub ztrailer_offset: u64,
 
     /// Length of the ZLIB trailer in bytes.
-    ztrailer_len: u64,
+    pub ztrailer_len: u64,
 }
 
-fn read_zheader<R: Read + Seek>(r: &mut BufReader<R>, e: Endian) -> Result<ZHeader, Error> {
+fn read_zheader<R: Read + Seek>(r: &mut R, endian: 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)?);
+    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,
@@ -703,13 +898,106 @@ fn read_zheader<R: Read + Seek>(r: &mut BufReader<R>, e: Endian) -> Result<ZHead
     })
 }
 
+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)
+    }
+}
+
 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)
 }
 
-fn read_vec<R: Read>(r: &mut BufReader<R>, n: usize) -> Result<Vec<u8>, IoError> {
+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)