Don't allow invalid chars in Symbol objects. (#765)

Co-authored-by: Graydon Hoare <graydon@pobox.com>

soroban-env-common/src/symbol.rs

@@ -157,6 +157,25 @@ impl<const N: u32> TryFrom<&StringM<N>> for SymbolSmall {
 }
 
 impl SymbolSmall {
+    #[doc(hidden)]
+    pub const fn validate_char(ch: char) -> Result<(), SymbolError> {
+        match SymbolSmall::encode_char(ch) {
+            Ok(_) => Ok(()),
+            Err(e) => Err(e),
+        }
+    }
+
+    const fn encode_char(ch: char) -> Result<u64, SymbolError> {
+        let v = match ch {
+            '_' => 1,
+            '0'..='9' => 2 + ((ch as u64) - ('0' as u64)),
+            'A'..='Z' => 12 + ((ch as u64) - ('A' as u64)),
+            'a'..='z' => 38 + ((ch as u64) - ('a' as u64)),
+            _ => return Err(SymbolError::BadChar(ch)),
+        };
+        Ok(v)
+    }
+
     pub const fn try_from_bytes(b: &[u8]) -> Result<SymbolSmall, SymbolError> {
         let mut n = 0;
         let mut accum: u64 = 0;
@@ -167,12 +186,9 @@ impl SymbolSmall {
             }
             n += 1;
             accum <<= CODE_BITS;
-            let v = match ch {
-                '_' => 1,
-                '0'..='9' => 2 + ((ch as u64) - ('0' as u64)),
-                'A'..='Z' => 12 + ((ch as u64) - ('A' as u64)),
-                'a'..='z' => 38 + ((ch as u64) - ('a' as u64)),
-                _ => return Err(SymbolError::BadChar(ch)),
+            let v = match SymbolSmall::encode_char(ch) {
+                Ok(v) => v,
+                Err(e) => return Err(e),
             };
             accum |= v;
         }

soroban-env-host/src/host.rs

@@ -1603,6 +1603,9 @@ impl EnvBase for Host {
     }
 
     fn symbol_new_from_slice(&self, s: &str) -> Result<SymbolObject, HostError> {
+        for ch in s.chars() {
+            SymbolSmall::validate_char(ch)?;
+        }
         self.add_host_object(ScSymbol(s.as_bytes().to_vec().try_into()?))
     }
 

soroban-env-host/src/test.rs

@@ -14,6 +14,7 @@ mod num;
 #[cfg(all(feature = "testutils", feature = "vm"))]
 mod storage;
 mod str;
+mod symbol;
 mod vec;
 
 #[cfg(feature = "vm")]

soroban-env-host/src/test/symbol.rs

@@ -0,0 +1,50 @@
+use crate::{Host, HostError};
+use soroban_env_common::{Symbol, TryFromVal};
+
+#[test]
+fn invalid_chars() -> Result<(), HostError> {
+    let host = Host::default();
+
+    let s = "#";
+    let s = Symbol::try_from_val(&host, &s);
+
+    assert!(s.is_err());
+
+    Ok(())
+}
+
+#[test]
+fn overlong() -> Result<(), HostError> {
+    let host = Host::default();
+
+    let s = "123456789012345678901234567890___";
+    let s = Symbol::try_from_val(&host, &s);
+
+    assert!(s.is_err());
+
+    Ok(())
+}
+
+#[test]
+fn max_len() -> Result<(), HostError> {
+    let host = Host::default();
+
+    let s = "123456789012345678901234567890__";
+    let s = Symbol::try_from_val(&host, &s);
+
+    assert!(s.is_ok());
+
+    Ok(())
+}
+
+#[test]
+fn zero_len() -> Result<(), HostError> {
+    let host = Host::default();
+
+    let s = "";
+    let s = Symbol::try_from_val(&host, &s);
+
+    assert!(s.is_ok());
+
+    Ok(())
+}