X-Git-Url: https://pintos-os.org/cgi-bin/gitweb.cgi?a=blobdiff_plain;f=rust%2Fsrc%2Fencoding.rs;h=8fd13f3ea3fb4e9b3f5243cd998037c15ac609d0;hb=a8331d2f67af24ce1f9f5da99641b8d1cdc21300;hp=296e4e65a8e4f32a61e4e415318610e3180513f7;hpb=6225b13e18d6ce5b42aa7112fa44a67b13b3bb41;p=pspp

diff --git a/rust/src/encoding.rs b/rust/src/encoding.rs
index 296e4e65a8..8fd13f3ea3 100644
--- a/rust/src/encoding.rs
+++ b/rust/src/encoding.rs
@@ -1,3 +1,6 @@
+use crate::locale_charset::locale_charset;
+use encoding_rs::{Encoding, UTF_8};
+
 include!(concat!(env!("OUT_DIR"), "/encodings.rs"));
 
 pub fn codepage_from_encoding(encoding: &str) -> Option<u32> {
@@ -6,23 +9,56 @@ pub fn codepage_from_encoding(encoding: &str) -> Option<u32> {
         .copied()
 }
 
-pub fn encoding_from_hints(encoding: Option<&str>, codepage: Option<u32>) -> Option<&str> {
-    if encoding.is_some() {
+use thiserror::Error as ThisError;
+
+#[derive(ThisError, Debug)]
+pub enum Error {
+    #[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(i32),
+
+    #[error("This system file encodes text strings with unknown encoding {0}.")]
+    UnknownEncoding(String),
+
+    #[error("This system file is encoded in EBCDIC, which is not supported.")]
+    Ebcdic,
+}
+
+pub fn default_encoding() -> &'static Encoding {
+    lazy_static! {
+        static ref DEFAULT_ENCODING: &'static Encoding =
+            Encoding::for_label(locale_charset().as_bytes()).unwrap_or(&UTF_8);
+    }
+    &DEFAULT_ENCODING
+}
+
+pub fn get_encoding(
+    encoding: Option<&str>,
+    character_code: Option<i32>,
+) -> Result<&'static Encoding, Error> {
+    let label = if let Some(encoding) = encoding {
         encoding
-    } else if let Some(codepage) = codepage {
+    } else if let Some(codepage) = character_code {
         match codepage {
-            1 => Some("EBCDIC-US"),
+            1 => return Err(Error::Ebcdic),
             2 | 3 => {
                 // These ostensibly mean "7-bit ASCII" and "8-bit ASCII"[sic]
                 // respectively.  However, many files have character code 2 but
                 // data which are clearly not ASCII.  Therefore, ignore these
                 // values.
-                None
-            },
-            4 => Some("MS_KANJI"),
-            _ => CODEPAGE_NUMBER_TO_NAME.get(&codepage).copied()
+                return Err(Error::NoEncoding);
+            }
+            4 => "MS_KANJI",
+            _ => CODEPAGE_NUMBER_TO_NAME
+                .get(&codepage)
+                .copied()
+                .ok_or(Error::UnknownCodepage(codepage))?,
         }
     } else {
-        None
-    }
+        return Err(Error::NoEncoding);
+    };
+
+    Ok(Encoding::for_label(label.as_bytes()).ok_or(Error::UnknownEncoding(label.into()))?)
 }