* along with this program. If not, see <http://www.gnu.org/licenses/>. */
use anyhow::Result;
-use clap::Parser;
-use pspp::raw::{Reader, Record, UnencodedStr, Value};
+use clap::{Parser, ValueEnum};
+use encoding_rs::Encoding;
+use pspp::cooked::decode;
+use pspp::raw::{Reader, Record, Magic};
use std::fs::File;
use std::io::BufReader;
use std::path::{Path, PathBuf};
-use std::str::{self, from_utf8};
+use std::str;
+use thiserror::Error as ThisError;
/// A utility to dissect SPSS system files.
#[derive(Parser, Debug)]
/// Files to dissect.
#[arg(required = true)]
files: Vec<PathBuf>,
+
+ /// How to dissect the file.
+ #[arg(short, long, value_enum, default_value_t)]
+ mode: Mode,
+
+ /// The encoding to use.
+ #[arg(long, value_parser = parse_encoding)]
+ encoding: Option<&'static Encoding>,
+}
+
+#[derive(ThisError, Debug)]
+#[error("{0}: unknown encoding")]
+struct UnknownEncodingError(String);
+
+fn parse_encoding(arg: &str) -> Result<&'static Encoding, UnknownEncodingError> {
+ match Encoding::for_label_no_replacement(arg.as_bytes()) {
+ Some(encoding) => Ok(encoding),
+ None => Err(UnknownEncodingError(arg.to_string())),
+ }
+}
+
+#[derive(Clone, Copy, Debug, Default, ValueEnum)]
+enum Mode {
+ Identify,
+ Raw,
+ #[default]
+ Cooked,
}
fn main() -> Result<()> {
- let Args { max_cases, files } = Args::parse();
+ let Args {
+ max_cases,
+ files,
+ mode,
+ encoding,
+ } = Args::parse();
for file in files {
- if let Err(error) = dissect(&file, max_cases) {
- println!("{}: {error}", file.display());
- }
+ dissect(&file, max_cases, mode, encoding)?;
}
Ok(())
}
-fn dissect(file_name: &Path, _max_cases: u64) -> Result<()> {
+fn dissect(file_name: &Path, max_cases: u64, mode: Mode, encoding: Option<&'static Encoding>) -> Result<()> {
let reader = File::open(file_name)?;
let reader = BufReader::new(reader);
let mut reader = Reader::new(reader)?;
- let records: Vec<Record> = reader.collect_headers()?;
-
- let mut character_code = None;
- for record in records {
- //println!("{record:?}");
- if let Record::IntegerInfo(ref info) = record {
- character_code = Some(info.character_code);
- }
- if let Record::EndOfHeaders(_) = record {
- break;
- };
- }
- if character_code != Some(65001) {
- return Ok(());
- }
- let mut n = 0;
- while let Some(Ok(Record::Case(data))) = reader.next() {
- n += 1;
- let mut strings = Vec::new();
- for value in data.iter() {
- if let Value::String(UnencodedStr(s)) = value {
- strings.extend_from_slice(&s[..]);
+ match mode {
+ Mode::Identify => {
+ let Record::Header(header) = reader.next().unwrap()? else { unreachable!() };
+ match header.magic {
+ Magic::Sav => println!("SPSS System File"),
+ Magic::Zsav => println!("SPSS System File with Zlib compression"),
+ Magic::Ebcdic => println!("EBCDIC-encoded SPSS System File"),
}
+ return Ok(())
}
-
- let mut rest = &strings[..];
- let mut any_errors = false;
- while let Err(error) = from_utf8(&rest) {
- if !any_errors {
- print!("{}: UTF-8 error", file_name.display());
- any_errors = true;
- }
- let start = error.valid_up_to();
- let len = match error.error_len() {
- Some(len) => len,
- None => rest.len() - start
- };
-// print!(" {}", (start + len) % 8);
- print!("[");
- for i in 0..len {
- print!("{:02x}", rest[i + start]);
+ Mode::Raw => {
+ let headers: Vec<Record> = reader.collect_headers()?;
+ for header in headers {
+ println!("{header:?}");
}
- print!("]");
- rest = &rest[start + len..];
}
- if any_errors {
- println!();
- println!("Lossy: {}", String::from_utf8_lossy(&strings[..]).replace(char::REPLACEMENT_CHARACTER, "??????").replace(&[' ', '\0'], ""));
- return Ok(())
+ Mode::Cooked => {
+ let headers: Vec<Record> = reader.collect_headers()?;
+ let headers = decode(headers, encoding, &|e| panic!("{e}"))?;
+ for header in headers {
+ println!("{header:?}");
+ }
}
- //println!("{:?}", data);
}
- println!("{}: read {n} records", file_name.display());
+
+ for _ in 0..max_cases {
+ let Some(Ok(record)) = reader.next() else {
+ break;
+ };
+ println!("{:?}", record);
+ }
Ok(())
}