From d1a7348ea5d482a598ed6df647682bb73e1865d4 Mon Sep 17 00:00:00 2001
From: Ben Pfaff <blp@cs.stanford.edu>
Date: Sun, 13 Apr 2025 10:05:23 -0700
Subject: [PATCH] Use RawString instead of Box<[u8]> in Value.

---
 rust/pspp/src/cooked.rs         |  4 +---
 rust/pspp/src/dictionary.rs     |  6 +++---
 rust/pspp/src/format/display.rs |  8 ++++++--
 rust/pspp/src/format/mod.rs     |  4 ++--
 rust/pspp/src/raw.rs            | 10 ++++++++--
 5 files changed, 20 insertions(+), 12 deletions(-)

diff --git a/rust/pspp/src/cooked.rs b/rust/pspp/src/cooked.rs
index 96d78ff1f2..7c68d47c8b 100644
--- a/rust/pspp/src/cooked.rs
+++ b/rust/pspp/src/cooked.rs
@@ -714,9 +714,7 @@ pub fn decode(
             // XXX warn about too-long value?
             value.0.resize(width, b' ');
             // XXX warn abouat duplicate value labels?
-            variable
-                .value_labels
-                .insert(Value::String(value.0.into_boxed_slice()), label);
+            variable.value_labels.insert(Value::String(value), label);
         }
     }
 
diff --git a/rust/pspp/src/dictionary.rs b/rust/pspp/src/dictionary.rs
index 67d5539ba7..d6f0980c22 100644
--- a/rust/pspp/src/dictionary.rs
+++ b/rust/pspp/src/dictionary.rs
@@ -16,7 +16,7 @@ use unicase::UniCase;
 use crate::{
     format::Format,
     identifier::{ByIdentifier, HasIdentifier, Identifier},
-    raw::{Alignment, CategoryLabels, Measure, MissingValues, VarType},
+    raw::{Alignment, CategoryLabels, Measure, MissingValues, RawString, VarType},
 };
 
 pub type DictIndex = usize;
@@ -157,7 +157,7 @@ impl From<VarWidth> for VarType {
 }
 
 #[derive(Clone)]
-pub enum Value<S = Box<[u8]>> {
+pub enum Value<S = RawString> {
     Number(Option<f64>),
     String(S),
 }
@@ -233,7 +233,7 @@ impl Value {
         }
     }
 
-    pub fn as_string(&self) -> Option<&[u8]> {
+    pub fn as_string(&self) -> Option<&RawString> {
         match self {
             Value::Number(_) => None,
             Value::String(s) => Some(s),
diff --git a/rust/pspp/src/format/display.rs b/rust/pspp/src/format/display.rs
index 8955d6b7b9..6edf3dc9f3 100644
--- a/rust/pspp/src/format/display.rs
+++ b/rust/pspp/src/format/display.rs
@@ -44,11 +44,15 @@ impl Display for DisplayValue<'_, '_> {
             Value::Number(number) => *number,
             Value::String(string) => {
                 if self.format.type_() == Type::AHex {
-                    for byte in string {
+                    for byte in &string.0 {
                         write!(f, "{byte:02x}")?;
                     }
                 } else {
-                    write!(f, "{}", self.encoding.decode_without_bom_handling(string).0)?;
+                    write!(
+                        f,
+                        "{}",
+                        self.encoding.decode_without_bom_handling(&string.0).0
+                    )?;
                 }
                 return Ok(());
             }
diff --git a/rust/pspp/src/format/mod.rs b/rust/pspp/src/format/mod.rs
index 3b87aa4664..8f520266a6 100644
--- a/rust/pspp/src/format/mod.rs
+++ b/rust/pspp/src/format/mod.rs
@@ -13,7 +13,7 @@ use unicode_width::UnicodeWidthStr;
 
 use crate::{
     dictionary::{Value, VarWidth},
-    raw::{self, VarType},
+    raw::{self, RawString, VarType},
 };
 
 mod display;
@@ -592,7 +592,7 @@ impl Format {
     pub fn default_value(&self) -> Value {
         match self.var_width() {
             VarWidth::Numeric => Value::sysmis(),
-            VarWidth::String(width) => Value::String((0..width).map(|_| 0u8).collect()),
+            VarWidth::String(width) => Value::String(RawString::spaces(width as usize)),
         }
     }
 }
diff --git a/rust/pspp/src/raw.rs b/rust/pspp/src/raw.rs
index b8d0ea2cb9..b9f25230ad 100644
--- a/rust/pspp/src/raw.rs
+++ b/rust/pspp/src/raw.rs
@@ -770,7 +770,7 @@ impl RawValue {
             Self::Number(x) => Value::Number(*x),
             Self::String(s) => {
                 let width = width.as_string_width().unwrap();
-                Value::String(Box::from(&s.0[..width]))
+                Value::String(RawString::from(&s.0[..width]))
             }
         }
     }
@@ -1370,9 +1370,15 @@ impl Debug for UntypedValue {
     }
 }
 
-#[derive(Clone)]
+#[derive(Clone, PartialEq, Eq, PartialOrd, Ord, Hash)]
 pub struct RawString(pub Vec<u8>);
 
+impl RawString {
+    pub fn spaces(n: usize) -> Self {
+        Self(std::iter::repeat_n(b' ', n).collect())
+    }
+}
+
 impl From<Vec<u8>> for RawString {
     fn from(source: Vec<u8>) -> Self {
         Self(source)
-- 
2.30.2