Merge pull request #549 from gwenn/non-utf-8

Do not assume `sqlite3_column_text` is valid UTF-8.

src/inner_connection.rs

@@ -216,7 +216,7 @@ impl InnerConnection {
 
     pub fn prepare<'a>(&mut self, conn: &'a Connection, sql: &str) -> Result<Statement<'a>> {
         let mut c_stmt = MaybeUninit::uninit();
-        let (c_sql, len, _) = str_for_sqlite(sql)?;
+        let (c_sql, len, _) = str_for_sqlite(sql.as_bytes())?;
         let r = unsafe {
             if cfg!(feature = "unlock_notify") {
                 let mut rc;

src/lib.rs

@@ -248,9 +248,9 @@ fn str_to_cstring(s: &str) -> Result<CString> {
 /// The `sqlite3_destructor_type` item is always `SQLITE_TRANSIENT` unless
 /// the string was empty (in which case it's `SQLITE_STATIC`, and the ptr is
 /// static).
-fn str_for_sqlite(s: &str) -> Result<(*const c_char, c_int, ffi::sqlite3_destructor_type)> {
+fn str_for_sqlite(s: &[u8]) -> Result<(*const c_char, c_int, ffi::sqlite3_destructor_type)> {
     let len = len_as_c_int(s.len())?;
-    if memchr::memchr(0, s.as_bytes()).is_none() {
+    if memchr::memchr(0, s).is_none() {
         let (ptr, dtor_info) = if len != 0 {
             (s.as_ptr() as *const c_char, ffi::SQLITE_TRANSIENT())
         } else {

src/pragma.rs

@@ -86,6 +86,7 @@ impl Sql {
                 self.push_real(r);
             }
             ValueRef::Text(s) => {
+                let s = std::str::from_utf8(s)?;
                 self.push_string_literal(s);
             }
             _ => {

src/statement.rs

@@ -622,10 +622,7 @@ impl Statement<'_> {
                     CStr::from_ptr(text as *const c_char)
                 };
 
-                // sqlite3_column_text returns UTF8 data, so our unwrap here should be fine.
-                let s = s
-                    .to_str()
-                    .expect("sqlite3_column_text returned invalid UTF-8");
+                let s = s.to_bytes();
                 ValueRef::Text(s)
             }
             ffi::SQLITE_BLOB => {

src/types/serde_json.rs

@@ -17,7 +17,7 @@ impl ToSql for Value {
 impl FromSql for Value {
     fn column_result(value: ValueRef<'_>) -> FromSqlResult<Self> {
         match value {
-            ValueRef::Text(s) => serde_json::from_str(s),
+            ValueRef::Text(s) => serde_json::from_slice(s),
             ValueRef::Blob(b) => serde_json::from_slice(b),
             _ => return Err(FromSqlError::InvalidType),
         }

src/types/url.rs

@@ -14,10 +14,12 @@ impl ToSql for Url {
 impl FromSql for Url {
     fn column_result(value: ValueRef<'_>) -> FromSqlResult<Self> {
         match value {
-            ValueRef::Text(s) => Url::parse(s),
-            _ => return Err(FromSqlError::InvalidType),
+            ValueRef::Text(s) => {
+                let s = std::str::from_utf8(s).map_err(|e| FromSqlError::Other(Box::new(e)))?;
+                Url::parse(s).map_err(|e| FromSqlError::Other(Box::new(e)))
+            }
+            _ => Err(FromSqlError::InvalidType),
         }
-        .map_err(|err| FromSqlError::Other(Box::new(err)))
     }
 }
 

src/types/value_ref.rs

@@ -14,7 +14,7 @@ pub enum ValueRef<'a> {
     /// The value is a floating point number.
     Real(f64),
     /// The value is a text string.
-    Text(&'a str),
+    Text(&'a [u8]),
     /// The value is a blob of data
     Blob(&'a [u8]),
 }
@@ -54,7 +54,9 @@ impl<'a> ValueRef<'a> {
     /// `Err(Error::InvalidColumnType)`.
     pub fn as_str(&self) -> FromSqlResult<&'a str> {
         match *self {
-            ValueRef::Text(t) => Ok(t),
+            ValueRef::Text(t) => {
+                std::str::from_utf8(t).map_err(|e| FromSqlError::Other(Box::new(e)))
+            }
             _ => Err(FromSqlError::InvalidType),
         }
     }
@@ -75,7 +77,10 @@ impl From<ValueRef<'_>> for Value {
             ValueRef::Null => Value::Null,
             ValueRef::Integer(i) => Value::Integer(i),
             ValueRef::Real(r) => Value::Real(r),
-            ValueRef::Text(s) => Value::Text(s.to_string()),
+            ValueRef::Text(s) => {
+                let s = std::str::from_utf8(s).expect("invalid UTF-8");
+                Value::Text(s.to_string())
+            }
             ValueRef::Blob(b) => Value::Blob(b.to_vec()),
         }
     }
@@ -83,7 +88,7 @@ impl From<ValueRef<'_>> for Value {
 
 impl<'a> From<&'a str> for ValueRef<'a> {
     fn from(s: &str) -> ValueRef<'_> {
-        ValueRef::Text(s)
+        ValueRef::Text(s.as_bytes())
     }
 }
 
@@ -99,7 +104,7 @@ impl<'a> From<&'a Value> for ValueRef<'a> {
             Value::Null => ValueRef::Null,
             Value::Integer(i) => ValueRef::Integer(i),
             Value::Real(r) => ValueRef::Real(r),
-            Value::Text(ref s) => ValueRef::Text(s),
+            Value::Text(ref s) => ValueRef::Text(s.as_bytes()),
             Value::Blob(ref b) => ValueRef::Blob(b),
         }
     }
@@ -125,10 +130,7 @@ impl<'a> ValueRef<'a> {
                 );
                 let s = CStr::from_ptr(text as *const c_char);
 
-                // sqlite3_value_text returns UTF8 data, so our unwrap here should be fine.
-                let s = s
-                    .to_str()
-                    .expect("sqlite3_value_text returned invalid UTF-8");
+                let s = s.to_bytes();
                 ValueRef::Text(s)
             }
             ffi::SQLITE_BLOB => {