update: bump polars to 0.44.2

dwpeng · Nov 7, 2024 · 845a952 · 845a952
1 parent 47d922e
commit 845a952
Show file tree

Hide file tree

Showing 22 changed files with 1,982 additions and 231 deletions.
diff --git a/Cargo.lock b/Cargo.lock
diff --git a/Cargo.toml b/Cargo.toml
@@ -9,13 +9,12 @@ homepage = "https://github.com/dwpeng/filterx"
 exclude = [
     "spec/",
     "docs/",
-    "python/"
 ]
 
 [dependencies]
 anyhow = "1.0.89"
 clap = { version = "4.5.17", features = ["derive"] }
-polars = { version = "0.43.1", default-features = false, features = [
+polars = { version = "0.44.2", default-features = false, features = [
     "lazy",
     "csv",
     "strings",
@@ -31,10 +30,12 @@ polars = { version = "0.43.1", default-features = false, features = [
     "dtype-i16",
     "dtype-i8",
     "abs",
-    "decompress"
+    "decompress",
+    "bitwise",
+    "describe",
 ] }
 rustpython-parser = "0.4.0"
-thiserror = "1.0.63"
+thiserror = "2.0.0"
 flate2 = { version = "1.0.34", features = ["zlib-rs"] }
 regex = "1.11.1"
 colored = "2.1.0"

diff --git a/src/engine/eval/assign.rs b/src/engine/eval/assign.rs
@@ -44,7 +44,7 @@ impl<'a> Eval<'a> for ast::StmtAssign {
         let new_col = eval!(vm, target, Call);
 
         let new_col = match new_col {
-            Value::Column(col) => col.col_name,
+            Value::Item(col) => col.col_name,
             Value::Name(n) => n.name,
             _ => {
                 let h = &mut vm.hint;

diff --git a/src/engine/eval/call/builtin/col.rs b/src/engine/eval/call/builtin/col.rs
@@ -25,14 +25,14 @@ pub fn col(vm: &mut Vm, args: &Vec<ast::Expr>) -> FilterxResult<value::Value> {
         }
         value::Value::Str(s) => s,
         value::Value::Name(c) => c.name,
-        value::Value::Column(c) => c.col_name,
+        value::Value::Item(c) => c.col_name,
         _ => {
             let h = &mut vm.hint;
             h.white("col only support column index, column name, or function which return a column name.").print_and_exit();
         }
     };
 
-    Ok(value::Value::Column(value::Column {
+    Ok(value::Value::Item(value::Item {
         col_name: c,
         data_type: None,
     }))

diff --git a/src/engine/eval/call/builtin/dup.rs b/src/engine/eval/call/builtin/dup.rs
@@ -20,7 +20,7 @@ pub fn dup<'a>(
         let col = eval!(vm, arg, Name, Call);
         let col = match col {
             value::Value::Name(c) => c.name,
-            value::Value::Column(c) => c.col_name,
+            value::Value::Item(c) => c.col_name,
             _ => {
                 let h = &mut vm.hint;
                 h.white("dup only support column name").print_and_exit();

diff --git a/src/engine/eval/call/builtin/gc.rs b/src/engine/eval/call/builtin/gc.rs
@@ -4,7 +4,7 @@ use polars::prelude::*;
 
 use polars::prelude::col;
 
-fn compute_gc(s: Series) -> PolarsResult<Option<Series>> {
+fn compute_gc(s: Column) -> PolarsResult<Option<Column>> {
     if !s.dtype().is_string() {
         return Err(PolarsError::InvalidOperation(
             format!(
@@ -15,6 +15,9 @@ fn compute_gc(s: Series) -> PolarsResult<Option<Series>> {
             .into(),
         ));
     }
+
+    let s = s.as_series().unwrap();
+
     let v = s
         .iter()
         .map(|seq| {
@@ -32,7 +35,7 @@ fn compute_gc(s: Series) -> PolarsResult<Option<Series>> {
             return gc_base as f32 / seq.len() as f32;
         })
         .collect::<Vec<f32>>();
-    Ok(Some(Series::new("gc".into(), v)))
+    Ok(Some(Column::new("gc".into(), v)))
 }
 
 pub fn gc<'a>(vm: &'a mut Vm, args: &Vec<ast::Expr>) -> FilterxResult<value::Value> {
@@ -45,7 +48,7 @@ pub fn gc<'a>(vm: &'a mut Vm, args: &Vec<ast::Expr>) -> FilterxResult<value::Val
     }
     let col_name = eval!(vm, &args[0], Name, Call);
     let col_name = match col_name {
-        value::Value::Column(c) => c.col_name,
+        value::Value::Item(c) => c.col_name,
         value::Value::Name(n) => n.name,
         _ => {
             let h = &mut vm.hint;

diff --git a/src/engine/eval/call/builtin/lower.rs b/src/engine/eval/call/builtin/lower.rs
@@ -18,7 +18,7 @@ pub fn lower<'a>(
 
     let col_name = eval!(vm, &args[0], Name, Call);
     let col_name = match col_name {
-        value::Value::Column(c) => c.col_name,
+        value::Value::Item(c) => c.col_name,
         value::Value::Name(n) => n.name,
         _ => {
             let h = &mut vm.hint;

diff --git a/src/engine/eval/call/builtin/rename.rs b/src/engine/eval/call/builtin/rename.rs
@@ -22,7 +22,7 @@ pub fn rename(vm: &mut Vm, args: &Vec<ast::Expr>) -> FilterxResult<value::Value>
             }
         }
         value::Value::Str(s) => s,
-        value::Value::Column(c) => c.col_name,
+        value::Value::Item(c) => c.col_name,
         value::Value::Name(c) => c.name,
         _ => {
             let h = &mut vm.hint;
@@ -41,7 +41,7 @@ pub fn rename(vm: &mut Vm, args: &Vec<ast::Expr>) -> FilterxResult<value::Value>
 
     let new_col = match new_col_value {
         value::Value::Str(s) => s,
-        value::Value::Column(c) => c.col_name,
+        value::Value::Item(c) => c.col_name,
         value::Value::Name(n) => n.name,
         _ => {
             let h = &mut vm.hint;

diff --git a/src/engine/eval/call/builtin/replace.rs b/src/engine/eval/call/builtin/replace.rs
@@ -19,7 +19,7 @@ pub fn replace<'a>(
 
     let col_name = eval!(vm, &args[0], Name, Call);
     let col_name = match col_name {
-        value::Value::Column(c) => c.col_name,
+        value::Value::Item(c) => c.col_name,
         value::Value::Name(n) => n.name,
         _ => {
             let h = &mut vm.hint;

diff --git a/src/engine/eval/call/builtin/rev.rs b/src/engine/eval/call/builtin/rev.rs
@@ -25,7 +25,7 @@ pub fn rev<'a>(
         Call
     );
     let col_name = match col_name {
-        value::Value::Column(c) => c.col_name.to_string(),
+        value::Value::Item(c) => c.col_name.to_string(),
         value::Value::Name(n) => n.name,
         _ => {
             let h = &mut vm.hint;

diff --git a/src/engine/eval/call/builtin/revcomp.rs b/src/engine/eval/call/builtin/revcomp.rs
@@ -7,10 +7,9 @@ use polars::error::PolarsResult;
 use polars::prelude::col;
 use polars::prelude::ChunkApply;
 use polars::prelude::GetOutput;
-use polars::series::IntoSeries;
-use polars::series::Series;
+use polars::prelude::*;
 
-fn compute_revcomp(s: Series) -> PolarsResult<Option<Series>> {
+fn compute_revcomp(s: Column) -> PolarsResult<Option<Column>> {
     let ca = s.str()?;
     let ca = ca.apply_values(|s| {
         let s = s.chars().rev().collect::<String>();
@@ -30,7 +29,7 @@ fn compute_revcomp(s: Series) -> PolarsResult<Option<Series>> {
             .collect::<String>();
         Cow::Owned(s)
     });
-    Ok(Some(ca.into_series()))
+    Ok(Some(ca.into_column()))
 }
 
 pub fn revcomp<'a>(
@@ -49,7 +48,7 @@ pub fn revcomp<'a>(
 
     let col_name = eval!(vm, &args[0], Name, Call);
     let col_name = match col_name {
-        value::Value::Column(c) => c.col_name,
+        value::Value::Item(c) => c.col_name,
         value::Value::Name(n) => n.name,
         _ => {
             let h = &mut vm.hint;

diff --git a/src/engine/eval/call/builtin/select.rs b/src/engine/eval/call/builtin/select.rs
@@ -18,7 +18,7 @@ pub fn select<'a>(vm: &'a mut Vm, args: &Vec<ast::Expr>) -> FilterxResult<value:
     for arg in args {
         let col = eval!(vm, arg, Name, Call);
         let col = match col {
-            value::Value::Column(c) => c.col_name,
+            value::Value::Item(c) => c.col_name,
             value::Value::Name(c) => c.name,
             _ => {
                 let h = &mut vm.hint;

diff --git a/src/engine/eval/call/builtin/slice.rs b/src/engine/eval/call/builtin/slice.rs
@@ -44,7 +44,7 @@ pub fn slice<'a>(
     let col_name = eval!(vm, &args[0], Name, Call, UnaryOp);
     let col_name = match col_name {
         value::Value::Name(n) => n.name,
-        value::Value::Column(c) => c.col_name,
+        value::Value::Item(c) => c.col_name,
         _ => {
             let h = &mut vm.hint;
             h.white("slice: expected a column name as first argument")

diff --git a/src/engine/eval/call/builtin/sort.rs b/src/engine/eval/call/builtin/sort.rs
@@ -28,7 +28,7 @@ pub fn sort(vm: &mut Vm, args: &Vec<ast::Expr>, incr: bool) -> FilterxResult<val
                 }
             }
             value::Value::Str(s) => s,
-            value::Value::Column(c) => c.col_name,
+            value::Value::Item(c) => c.col_name,
             value::Value::Name(c) => c.name,
             _ => {
                 let h = &mut vm.hint;

diff --git a/src/engine/eval/call/builtin/strip.rs b/src/engine/eval/call/builtin/strip.rs
@@ -27,7 +27,7 @@ pub fn strip<'a>(
     );
 
     let col_name = match col_name {
-        value::Value::Column(c) => c.col_name,
+        value::Value::Item(c) => c.col_name,
         value::Value::Name(n) => n.name,
         _ => {
             let h = &mut vm.hint;

diff --git a/src/engine/eval/call/builtin/upper.rs b/src/engine/eval/call/builtin/upper.rs
@@ -19,7 +19,7 @@ pub fn upper<'a>(
     let col_name = eval!(vm, &args[0], Name, Call, UnaryOp);
 
     let col_name = match col_name {
-        value::Value::Column(c) => c.col_name,
+        value::Value::Item(c) => c.col_name,
         value::Value::Name(n) => n.name,
         _ => {
             let h = &mut vm.hint;

diff --git a/src/engine/eval/ops.rs b/src/engine/eval/ops.rs
@@ -55,8 +55,7 @@ impl<'a> Eval<'a> for ast::ExprUnaryOp {
                 let r = unary(v, self.op)?;
                 return Ok(r);
             }
-            Value::Name(_) | Value::Column(_) => {
-                println!("{:?}", -v.expr()?);
+            Value::Name(_) | Value::Item(_) => {
                 return Ok(Value::Expr(-(v.expr()?)));
             }
             Value::Expr(e) => {
@@ -363,7 +362,7 @@ fn str_in_col<'a>(vm: &'a mut Vm, left: Value, right: Value, op: &CmpOp) -> Filt
         _ => unreachable!(),
     };
     let right_col = match &right {
-        Value::Column(c) => c.col_name.clone(),
+        Value::Item(c) => c.col_name.clone(),
         _ => unreachable!(),
     };
 
@@ -436,7 +435,7 @@ fn compare_in_and_not_in_dataframe<'a>(
                         }
                     }
                 }
-                DataFrame::new(vec![Series::new(right_col.as_str().into(), v)])?
+                DataFrame::new(vec![Column::new(right_col.as_str().into(), v)])?
             }
             DataType::Int32 | DataType::Int64 => {
                 let mut v = Vec::new();
@@ -451,7 +450,7 @@ fn compare_in_and_not_in_dataframe<'a>(
                         }
                     }
                 }
-                DataFrame::new(vec![Series::new(right_col.as_str().into(), v)])?
+                DataFrame::new(vec![Column::new(right_col.as_str().into(), v)])?
             }
             DataType::String => {
                 let mut v = Vec::new();
@@ -467,7 +466,7 @@ fn compare_in_and_not_in_dataframe<'a>(
                         }
                     }
                 }
-                DataFrame::new(vec![Series::new(right_col.as_str().into(), v)])?
+                DataFrame::new(vec![Column::new(right_col.as_str().into(), v)])?
             }
             _ => {
                 return Err(FilterxError::RuntimeError(

diff --git a/src/engine/value.rs b/src/engine/value.rs
@@ -12,7 +12,7 @@ pub enum Value {
     Str(String),
     Name(Name),
     List(Vec<Value>),
-    Column(Column),
+    Item(Item),
     Ident((String, Box<Value>)),
     AttrMethod(AttrMethod),
     File(File),
@@ -117,7 +117,7 @@ impl Value {
             Value::Float(f) => f.lit(),
             Value::Int(i) => i.lit(),
             Value::Str(s) => s.clone().lit(),
-            Value::Column(c) => col(c.col_name.clone()),
+            Value::Item(c) => col(c.col_name.clone()),
             Value::Name(n) => col(n.name.clone()),
             Value::Expr(e) => e.clone(),
             Value::Null => Expr::Literal(LiteralValue::Null),
@@ -130,7 +130,7 @@ impl Value {
     pub fn text(&self) -> FilterxResult<String> {
         match self {
             Value::Str(s) => Ok(s.to_owned()),
-            Value::Column(c) => Ok(c.col_name.to_owned()),
+            Value::Item(c) => Ok(c.col_name.to_owned()),
             Value::Name(n) => Ok(n.name.to_owned()),
             _ => {
                 return Err(FilterxError::RuntimeError(
@@ -160,7 +160,7 @@ impl Value {
 
     pub fn is_column(&self) -> bool {
         match self {
-            Value::Column(_) => true,
+            Value::Item(_) => true,
             Value::Name(_) => true,
             _ => false,
         }
@@ -212,14 +212,14 @@ impl Default for Slice {
 }
 
 #[derive(Debug, Clone, PartialEq)]
-pub struct Column {
+pub struct Item {
     pub col_name: String,
     pub data_type: Option<DataType>,
 }
 
-impl Column {
+impl Item {
     pub fn new(col_name: String) -> Self {
-        Column {
+        Item {
             col_name,
             data_type: None,
         }
@@ -230,9 +230,9 @@ impl Column {
     }
 }
 
-impl Default for Column {
+impl Default for Item {
     fn default() -> Self {
-        Column {
+        Item {
             col_name: String::new(),
             data_type: None,
         }
@@ -241,7 +241,7 @@ impl Default for Column {
 
 #[derive(Debug, PartialEq, Clone)]
 pub struct AttrMethod {
-    pub col: Column,
+    pub col: Item,
     pub method: String,
     pub value: Vec<Value>,
 }
@@ -324,7 +324,7 @@ impl Value {
                 s
             }
             Value::File(f) => f.file_name.clone(),
-            Value::Column(c) => c.col_name.clone(),
+            Value::Item(c) => c.col_name.clone(),
             Value::Ident(i) => {
                 let mut s = String::from("(");
                 s.push_str(&i.0);

diff --git a/src/source/block/fastx/fasta.rs b/src/source/block/fastx/fasta.rs
@@ -314,11 +314,11 @@ impl<'a> TableLike<'a> for Fasta {
         }
         let mut cols = Vec::with_capacity(3);
         cols.append(&mut vec![
-            polars::prelude::Series::new("name".into(), names),
-            polars::prelude::Series::new("seq".into(), sequences),
+            polars::prelude::Column::new("name".into(), names),
+            polars::prelude::Column::new("seq".into(), sequences),
         ]);
         if !comments.is_empty() {
-            cols.push(polars::prelude::Series::new("comment".into(), comments));
+            cols.push(polars::prelude::Column::new("comment".into(), comments));
         }
         let df = polars::prelude::DataFrame::new(cols)?;
 
@@ -344,11 +344,11 @@ impl<'a> TableLike<'a> for Fasta {
             }
         }
         let mut cols = vec![
-            polars::prelude::Series::new("name".into(), headers),
-            polars::prelude::Series::new("seq".into(), sequences),
+            polars::prelude::Column::new("name".into(), headers),
+            polars::prelude::Column::new("seq".into(), sequences),
         ];
         if comments.len() > 0 {
-            cols.push(polars::prelude::Series::new("comment".into(), comments));
+            cols.push(polars::prelude::Column::new("comment".into(), comments));
         }
         let df = polars::prelude::DataFrame::new(cols)?;
         Ok(df)