fix: by comments

Signed-off-by: cutecutecat <junyuchen@tensorchord.ai>

src/datatype/text_svecf32.rs

@@ -5,18 +5,25 @@ use base::scalar::*;
 use base::vector::*;
 use pgrx::pg_sys::Oid;
 use std::ffi::{CStr, CString};
+use std::fmt::Write;
 
 #[pgrx::pg_extern(immutable, strict, parallel_safe)]
 fn _vectors_svecf32_in(input: &CStr, _oid: Oid, _typmod: i32) -> SVecf32Output {
-    use crate::utils::parse::parse_pgvector_svector;
+    use crate::utils::parse::{parse_pgvector_svector, svector_filter_nonzero};
     let v = parse_pgvector_svector(input.to_bytes(), |s| s.parse::<F32>().ok());
     match v {
         Err(e) => {
             bad_literal(&e.to_string());
         }
         Ok((indexes, values, dims)) => {
             check_value_dims_1048575(dims);
-            SVecf32Output::new(SVecf32Borrowed::new(dims as u32, &indexes, &values))
+            check_index_in_bound(&indexes, dims);
+            let (non_zero_indexes, non_zero_values) = svector_filter_nonzero(&indexes, &values);
+            SVecf32Output::new(SVecf32Borrowed::new(
+                dims as u32,
+                &non_zero_indexes,
+                &non_zero_values,
+            ))
         }
     }
 }
@@ -27,16 +34,16 @@ fn _vectors_svecf32_out(vector: SVecf32Input<'_>) -> CString {
     let mut buffer = String::new();
     buffer.push('{');
     let svec = vector.for_borrow();
-    let mut need_splitter = true;
+    let mut need_splitter = false;
     for (&index, &value) in svec.indexes().iter().zip(svec.values().iter()) {
         match need_splitter {
-            true => {
-                buffer.push_str(format!("{}:{}", index, value).as_str());
-                need_splitter = false;
+            false => {
+                write!(buffer, "{}:{}", index, value).unwrap();
+                need_splitter = true;
             }
-            false => buffer.push_str(format!(", {}:{}", index, value).as_str()),
+            true => write!(buffer, ", {}:{}", index, value).unwrap(),
         }
     }
-    buffer.push_str(format!("}}/{}", dims).as_str());
+    write!(buffer, "}}/{}", dims).unwrap();
     CString::new(buffer).unwrap()
 }

src/error.rs

@@ -69,6 +69,20 @@ ADVICE: Check if dimensions of the vector are among 1 and 1_048_575."
     NonZeroU32::new(dims as u32).unwrap()
 }
 
+pub fn check_index_in_bound(indexes: &[u32], dims: usize) -> NonZeroU32 {
+    let mut last: u32 = 0;
+    for (i, index) in indexes.iter().enumerate() {
+        if i > 0 && last == *index {
+            error!("Indexes need to be unique, but there are more than one same index {index}")
+        }
+        if *index >= dims as u32 {
+            error!("Index out of bounds: the dim is {dims} but the index is {index}");
+        }
+        last = *index;
+    }
+    NonZeroU32::new(dims as u32).unwrap()
+}
+
 pub fn bad_literal(hint: &str) -> ! {
     error!(
         "\