From: Ben Pfaff <blp@cs.stanford.edu>
Date: Sat, 4 Nov 2023 00:17:30 +0000 (-0700)
Subject: work
X-Git-Url: https://pintos-os.org/cgi-bin/gitweb.cgi?a=commitdiff_plain;h=6165ed413d9aa818e3246d0a063c646dc4efc7e5;p=pspp

work
---

diff --git a/rust/build.rs b/rust/build.rs
index 48e011b4c5..112f3d3da8 100644
--- a/rust/build.rs
+++ b/rust/build.rs
@@ -102,7 +102,7 @@ use lazy_static::lazy_static;
 use std::collections::HashMap;
 
 lazy_static! {
-    static ref CODEPAGE_NUMBER_TO_NAME: HashMap<u32, &'static str> = {
+    static ref CODEPAGE_NUMBER_TO_NAME: HashMap<i32, &'static str> = {
         let mut map = HashMap::new();
 "
     )?;
diff --git a/rust/src/cooked.rs b/rust/src/cooked.rs
index 3fac184063..ef4b79ccf4 100644
--- a/rust/src/cooked.rs
+++ b/rust/src/cooked.rs
@@ -1,10 +1,11 @@
 use std::{borrow::Cow, cmp::Ordering, collections::HashMap, iter::repeat};
 
 use crate::{
+    encoding::{get_encoding, Error as EncodingError},
     endian::Endian,
     format::{Error as FormatError, Spec, UncheckedSpec},
     identifier::{Error as IdError, Identifier},
-    raw::{self, MissingValues, UnencodedStr, VarType}, encoding::get_encoding,
+    raw::{self, MissingValues, UnencodedStr, VarType},
 };
 use chrono::{NaiveDate, NaiveDateTime, NaiveTime};
 use encoding_rs::{DecoderResult, Encoding};
@@ -16,6 +17,14 @@ pub use crate::raw::{CategoryLabels, Compression};
 
 #[derive(ThisError, Debug)]
 pub enum Error {
+    // XXX this is really an internal error and maybe we should change the
+    // interfaces to make it impossible
+    #[error("Missing header record")]
+    MissingHeaderRecord,
+
+    #[error("{0}")]
+    EncodingError(EncodingError),
+
     #[error("Variable record at offset {offset:#x} specifies width {width} not in valid range [-1,255).")]
     InvalidVariableWidth { offset: u64, width: i32 },
 
@@ -187,7 +196,16 @@ pub struct Decoder {
     n_generated_names: usize,
 }
 
-pub fn decode<T>(headers: Vec<raw::Record>, warn: &impl Fn(Error)) -> Vec<Record> {
+pub fn decode<T>(headers: Vec<raw::Record>, warn: &impl Fn(Error)) -> Result<Vec<Record>, Error> {
+    let Some(header_record) = headers.iter().find_map(|rec| {
+        if let raw::Record::Header(header) = rec {
+            Some(header)
+        } else {
+            None
+        }
+    }) else {
+        return Err(Error::MissingHeaderRecord);
+    };
     let encoding = headers.iter().find_map(|rec| {
         if let raw::Record::Encoding(ref e) = rec {
             Some(e.0.as_str())
@@ -202,12 +220,27 @@ pub fn decode<T>(headers: Vec<raw::Record>, warn: &impl Fn(Error)) -> Vec<Record
             None
         }
     });
-    let encoding = get_encoding(encoding, character_code)
+    let encoding = match get_encoding(encoding, character_code) {
+        Ok(encoding) => encoding,
+        Err(err @ EncodingError::Ebcdic) => return Err(Error::EncodingError(err)),
+        Err(err) => {
+            warn(Error::EncodingError(err));
+            // Warn that we're using the default encoding.
+            
+        }
+    };
 
     let decoder = Decoder {
+        compression: header_record.compression,
+        endian: header_record.endian,
+        encoding,
+        variables: HashMap::new(),
+        var_names: HashMap::new(),
+        n_dict_indexes: 0,
+        n_generated_names: 0,
     };
 
-    Vec::new()
+    unreachable!()
 }
 
 impl Decoder {
diff --git a/rust/src/encoding.rs b/rust/src/encoding.rs
index 3d585a6b63..3509b73954 100644
--- a/rust/src/encoding.rs
+++ b/rust/src/encoding.rs
@@ -9,11 +9,11 @@ pub fn codepage_from_encoding(encoding: &str) -> Option<u32> {
 use thiserror::Error as ThisError;
 #[derive(ThisError, Debug)]
 pub enum Error {
-    #[error("This system file does not indicate its own character encoding.  xFor best results, specify an encoding explicitly.  Use SYSFILE INFO with ENCODING=\"DETECT\" to analyze the possible encodings.")]
+    #[error("This system file does not indicate its own character encoding.  For best results, specify an encoding explicitly.  Use SYSFILE INFO with ENCODING=\"DETECT\" to analyze the possible encodings.")]
     NoEncoding,
 
     #[error("This system file encodes text strings with unknown code page {0}.")]
-    UnknownCodepage(u32),
+    UnknownCodepage(i32),
 
     #[error("This system file is encoded in EBCDIC, which is not supported.")]
     Ebcdic,
@@ -26,7 +26,7 @@ pub fn locale_charset() -> &'static str {
     "UTF-8"
 }
 
-pub fn get_encoding(encoding: Option<&str>, character_code: Option<u32>) -> Result<&str, Error> {
+pub fn get_encoding(encoding: Option<&str>, character_code: Option<i32>) -> Result<&str, Error> {
     if let Some(encoding) = encoding {
         Ok(encoding)
     } else if let Some(codepage) = character_code {