use either::Either;
use magnus::{
    prelude::*, r_hash::ForEach, typed_data::Obj, IntoValue, RArray, RHash, RString, Value,
};
use polars::frame::row::{rows_to_schema_supertypes, Row};
use polars::frame::NullStrategy;
use polars::io::avro::AvroCompression;
use polars::io::mmap::ReaderBytes;
use polars::io::RowCount;
use polars::prelude::pivot::{pivot, pivot_stable};
use polars::prelude::*;
use polars_core::utils::try_get_supertype;
use std::cell::RefCell;
use std::io::{BufWriter, Cursor};
use std::ops::Deref;

use crate::conversion::*;
use crate::file::{get_file_like, get_mmap_bytes_reader};
use crate::map::dataframe::{
    apply_lambda_unknown, apply_lambda_with_bool_out_type, apply_lambda_with_primitive_out_type,
    apply_lambda_with_utf8_out_type,
};
use crate::rb_modules;
use crate::series::{to_rbseries_collection, to_series_collection};
use crate::{RbExpr, RbLazyFrame, RbPolarsErr, RbResult, RbSeries};

#[magnus::wrap(class = "Polars::RbDataFrame")]
pub struct RbDataFrame {
    pub df: RefCell<DataFrame>,
}

impl From<DataFrame> for RbDataFrame {
    fn from(df: DataFrame) -> Self {
        RbDataFrame::new(df)
    }
}

impl RbDataFrame {
    pub fn new(df: DataFrame) -> Self {
        RbDataFrame {
            df: RefCell::new(df),
        }
    }

    fn finish_from_rows(
        rows: Vec<Row>,
        infer_schema_length: Option<usize>,
        schema_overwrite: Option<Schema>,
    ) -> RbResult<Self> {
        // object builder must be registered.
        crate::on_startup::register_object_builder();

        let schema =
            rows_to_schema_supertypes(&rows, infer_schema_length.map(|n| std::cmp::max(1, n)))
                .map_err(RbPolarsErr::from)?;
        // replace inferred nulls with boolean
        let fields = schema.iter_fields().map(|mut fld| match fld.data_type() {
            DataType::Null => {
                // fld.coerce(DataType::Boolean);
                fld
            }
            DataType::Decimal(_, _) => {
                fld.coerce(DataType::Decimal(None, None));
                fld
            }
            _ => fld,
        });
        let mut schema = Schema::from_iter(fields);

        if let Some(schema_overwrite) = schema_overwrite {
            for (i, (name, dtype)) in schema_overwrite.into_iter().enumerate() {
                if let Some((name_, dtype_)) = schema.get_at_index_mut(i) {
                    *name_ = name;

                    // if user sets dtype unknown, we use the inferred datatype
                    if !matches!(dtype, DataType::Unknown) {
                        *dtype_ = dtype;
                    }
                } else {
                    schema.with_column(name, dtype);
                }
            }
        }

        let df = DataFrame::from_rows_and_schema(&rows, &schema).map_err(RbPolarsErr::from)?;
        Ok(df.into())
    }

    pub fn init(columns: RArray) -> RbResult<Self> {
        let mut cols = Vec::new();
        for i in columns.each() {
            cols.push(<&RbSeries>::try_convert(i?)?.series.borrow().clone());
        }
        let df = DataFrame::new(cols).map_err(RbPolarsErr::from)?;
        Ok(RbDataFrame::new(df))
    }

    pub fn estimated_size(&self) -> usize {
        self.df.borrow().estimated_size()
    }

    pub fn read_csv(arguments: &[Value]) -> RbResult<Self> {
        // start arguments
        // this pattern is needed for more than 16
        let rb_f = arguments[0];
        let infer_schema_length = Option::<usize>::try_convert(arguments[1])?;
        let chunk_size = usize::try_convert(arguments[2])?;
        let has_header = bool::try_convert(arguments[3])?;
        let ignore_errors = bool::try_convert(arguments[4])?;
        let n_rows = Option::<usize>::try_convert(arguments[5])?;
        let skip_rows = usize::try_convert(arguments[6])?;
        let projection = Option::<Vec<usize>>::try_convert(arguments[7])?;
        let separator = String::try_convert(arguments[8])?;
        let rechunk = bool::try_convert(arguments[9])?;
        let columns = Option::<Vec<String>>::try_convert(arguments[10])?;
        let encoding = Wrap::<CsvEncoding>::try_convert(arguments[11])?;
        let n_threads = Option::<usize>::try_convert(arguments[12])?;
        let path = Option::<String>::try_convert(arguments[13])?;
        let overwrite_dtype = Option::<Vec<(String, Wrap<DataType>)>>::try_convert(arguments[14])?;
        // TODO fix
        let overwrite_dtype_slice = Option::<Vec<Wrap<DataType>>>::None; // Option::<Vec<Wrap<DataType>>>::try_convert(arguments[15])?;
        let low_memory = bool::try_convert(arguments[16])?;
        let comment_prefix = Option::<String>::try_convert(arguments[17])?;
        let quote_char = Option::<String>::try_convert(arguments[18])?;
        let null_values = Option::<Wrap<NullValues>>::try_convert(arguments[19])?;
        let try_parse_dates = bool::try_convert(arguments[20])?;
        let skip_rows_after_header = usize::try_convert(arguments[21])?;
        let row_count = Option::<(String, IdxSize)>::try_convert(arguments[22])?;
        let sample_size = usize::try_convert(arguments[23])?;
        let eol_char = String::try_convert(arguments[24])?;
        // end arguments

        let null_values = null_values.map(|w| w.0);
        let eol_char = eol_char.as_bytes()[0];

        let row_count = row_count.map(|(name, offset)| RowCount { name, offset });

        let quote_char = if let Some(s) = quote_char {
            if s.is_empty() {
                None
            } else {
                Some(s.as_bytes()[0])
            }
        } else {
            None
        };

        let overwrite_dtype = overwrite_dtype.map(|overwrite_dtype| {
            overwrite_dtype
                .iter()
                .map(|(name, dtype)| {
                    let dtype = dtype.0.clone();
                    Field::new(name, dtype)
                })
                .collect::<Schema>()
        });

        let overwrite_dtype_slice = overwrite_dtype_slice.map(|overwrite_dtype| {
            overwrite_dtype
                .iter()
                .map(|dt| dt.0.clone())
                .collect::<Vec<_>>()
        });

        let mmap_bytes_r = get_mmap_bytes_reader(rb_f)?;
        let df = CsvReader::new(mmap_bytes_r)
            .infer_schema(infer_schema_length)
            .has_header(has_header)
            .with_n_rows(n_rows)
            .with_separator(separator.as_bytes()[0])
            .with_skip_rows(skip_rows)
            .with_ignore_errors(ignore_errors)
            .with_projection(projection)
            .with_rechunk(rechunk)
            .with_chunk_size(chunk_size)
            .with_encoding(encoding.0)
            .with_columns(columns)
            .with_n_threads(n_threads)
            .with_path(path)
            .with_dtypes(overwrite_dtype.map(Arc::new))
            .with_dtypes_slice(overwrite_dtype_slice.as_deref())
            .low_memory(low_memory)
            .with_comment_prefix(comment_prefix.as_deref())
            .with_null_values(null_values)
            .with_try_parse_dates(try_parse_dates)
            .with_quote_char(quote_char)
            .with_end_of_line_char(eol_char)
            .with_skip_rows_after_header(skip_rows_after_header)
            .with_row_count(row_count)
            .sample_size(sample_size)
            .finish()
            .map_err(RbPolarsErr::from)?;
        Ok(df.into())
    }

    #[allow(clippy::too_many_arguments)]
    pub fn read_parquet(
        rb_f: Value,
        columns: Option<Vec<String>>,
        projection: Option<Vec<usize>>,
        n_rows: Option<usize>,
        parallel: Wrap<ParallelStrategy>,
        row_count: Option<(String, IdxSize)>,
        low_memory: bool,
        use_statistics: bool,
        rechunk: bool,
    ) -> RbResult<Self> {
        let row_count = row_count.map(|(name, offset)| RowCount { name, offset });
        let mmap_bytes_r = get_mmap_bytes_reader(rb_f)?;
        let df = ParquetReader::new(mmap_bytes_r)
            .with_projection(projection)
            .with_columns(columns)
            .read_parallel(parallel.0)
            .with_n_rows(n_rows)
            .with_row_count(row_count)
            .set_low_memory(low_memory)
            .use_statistics(use_statistics)
            .set_rechunk(rechunk)
            .finish()
            .map_err(RbPolarsErr::from)?;
        Ok(RbDataFrame::new(df))
    }

    pub fn read_ipc(
        rb_f: Value,
        columns: Option<Vec<String>>,
        projection: Option<Vec<usize>>,
        n_rows: Option<usize>,
        row_count: Option<(String, IdxSize)>,
        memory_map: bool,
    ) -> RbResult<Self> {
        let row_count = row_count.map(|(name, offset)| RowCount { name, offset });
        let mmap_bytes_r = get_mmap_bytes_reader(rb_f)?;
        let df = IpcReader::new(mmap_bytes_r)
            .with_projection(projection)
            .with_columns(columns)
            .with_n_rows(n_rows)
            .with_row_count(row_count)
            .memory_mapped(memory_map)
            .finish()
            .map_err(RbPolarsErr::from)?;
        Ok(RbDataFrame::new(df))
    }

    pub fn read_avro(
        rb_f: Value,
        columns: Option<Vec<String>>,
        projection: Option<Vec<usize>>,
        n_rows: Option<usize>,
    ) -> RbResult<Self> {
        use polars::io::avro::AvroReader;

        let file = get_file_like(rb_f, false)?;
        let df = AvroReader::new(file)
            .with_projection(projection)
            .with_columns(columns)
            .with_n_rows(n_rows)
            .finish()
            .map_err(RbPolarsErr::from)?;
        Ok(RbDataFrame::new(df))
    }

    pub fn write_avro(
        &self,
        rb_f: Value,
        compression: Wrap<Option<AvroCompression>>,
    ) -> RbResult<()> {
        use polars::io::avro::AvroWriter;

        if let Ok(s) = String::try_convert(rb_f) {
            let f = std::fs::File::create(s).unwrap();
            AvroWriter::new(f)
                .with_compression(compression.0)
                .finish(&mut self.df.borrow_mut())
                .map_err(RbPolarsErr::from)?;
        } else {
            let mut buf = get_file_like(rb_f, true)?;
            AvroWriter::new(&mut buf)
                .with_compression(compression.0)
                .finish(&mut self.df.borrow_mut())
                .map_err(RbPolarsErr::from)?;
        }

        Ok(())
    }

    pub fn read_json(rb_f: Value) -> RbResult<Self> {
        // memmap the file first
        let mmap_bytes_r = get_mmap_bytes_reader(rb_f)?;
        let mmap_read: ReaderBytes = (&mmap_bytes_r).into();
        let bytes = mmap_read.deref();

        // Happy path is our column oriented json as that is most performant
        // on failure we try
        match serde_json::from_slice::<DataFrame>(bytes) {
            Ok(df) => Ok(df.into()),
            // try arrow json reader instead
            // this is row oriented
            Err(e) => {
                let msg = format!("{e}");
                if msg.contains("successful parse invalid data") {
                    let e = RbPolarsErr::from(PolarsError::ComputeError(msg.into()));
                    Err(e)
                } else {
                    let out = JsonReader::new(mmap_bytes_r)
                        .with_json_format(JsonFormat::Json)
                        .finish()
                        .map_err(|e| RbPolarsErr::other(format!("{:?}", e)))?;
                    Ok(out.into())
                }
            }
        }
    }

    pub fn read_ndjson(rb_f: Value) -> RbResult<Self> {
        let mmap_bytes_r = get_mmap_bytes_reader(rb_f)?;

        let out = JsonReader::new(mmap_bytes_r)
            .with_json_format(JsonFormat::JsonLines)
            .finish()
            .map_err(|e| RbPolarsErr::other(format!("{:?}", e)))?;
        Ok(out.into())
    }

    pub fn write_json(&self, rb_f: Value, pretty: bool, row_oriented: bool) -> RbResult<()> {
        let file = BufWriter::new(get_file_like(rb_f, true)?);

        let r = match (pretty, row_oriented) {
            (_, true) => JsonWriter::new(file)
                .with_json_format(JsonFormat::Json)
                .finish(&mut self.df.borrow_mut()),
            (true, _) => serde_json::to_writer_pretty(file, &*self.df.borrow())
                .map_err(|e| PolarsError::ComputeError(format!("{:?}", e).into())),
            (false, _) => serde_json::to_writer(file, &*self.df.borrow())
                .map_err(|e| PolarsError::ComputeError(format!("{:?}", e).into())),
        };
        r.map_err(|e| RbPolarsErr::other(format!("{:?}", e)))?;
        Ok(())
    }

    pub fn write_ndjson(&self, rb_f: Value) -> RbResult<()> {
        let file = BufWriter::new(get_file_like(rb_f, true)?);

        let r = JsonWriter::new(file)
            .with_json_format(JsonFormat::JsonLines)
            .finish(&mut self.df.borrow_mut());

        r.map_err(|e| RbPolarsErr::other(format!("{:?}", e)))?;
        Ok(())
    }

    pub fn read_rows(
        rb_rows: RArray,
        infer_schema_length: Option<usize>,
        schema_overwrite: Option<Wrap<Schema>>,
    ) -> RbResult<Self> {
        let mut rows = Vec::with_capacity(rb_rows.len());
        for v in rb_rows.each() {
            let rb_row = RArray::try_convert(v?)?;
            let mut row = Vec::with_capacity(rb_row.len());
            for val in rb_row.each() {
                row.push(Wrap::<AnyValue>::try_convert(val?)?.0);
            }
            rows.push(Row(row));
        }
        Self::finish_from_rows(
            rows,
            infer_schema_length,
            schema_overwrite.map(|wrap| wrap.0),
        )
    }

    pub fn read_hashes(
        dicts: Value,
        infer_schema_length: Option<usize>,
        schema_overwrite: Option<Wrap<Schema>>,
    ) -> RbResult<Self> {
        let (rows, mut names) = dicts_to_rows(&dicts, infer_schema_length.unwrap_or(50))?;

        // ensure the new names are used
        if let Some(schema) = &schema_overwrite {
            for (new_name, name) in schema.0.iter_names().zip(names.iter_mut()) {
                *name = new_name.to_string();
            }
        }
        let rbdf = Self::finish_from_rows(
            rows,
            infer_schema_length,
            schema_overwrite.map(|wrap| wrap.0),
        )?;

        unsafe {
            rbdf.df
                .borrow_mut()
                .get_columns_mut()
                .iter_mut()
                .zip(&names)
                .for_each(|(s, name)| {
                    s.rename(name);
                });
        }
        let length = names.len();
        if names.into_iter().collect::<PlHashSet<_>>().len() != length {
            let err = PolarsError::SchemaMismatch("duplicate column names found".into());
            Err(RbPolarsErr::from(err))?;
        }

        Ok(rbdf)
    }

    pub fn read_hash(data: RHash) -> RbResult<Self> {
        let mut cols: Vec<Series> = Vec::new();
        data.foreach(|name: String, values: Value| {
            let obj: Value = rb_modules::series().funcall("new", (name, values))?;
            let rbseries = obj.funcall::<_, _, &RbSeries>("_s", ())?;
            cols.push(rbseries.series.borrow().clone());
            Ok(ForEach::Continue)
        })?;
        let df = DataFrame::new(cols).map_err(RbPolarsErr::from)?;
        Ok(df.into())
    }

    #[allow(clippy::too_many_arguments)]
    pub fn write_csv(
        &self,
        rb_f: Value,
        include_header: bool,
        separator: u8,
        quote_char: u8,
        batch_size: usize,
        datetime_format: Option<String>,
        date_format: Option<String>,
        time_format: Option<String>,
        float_precision: Option<usize>,
        null_value: Option<String>,
    ) -> RbResult<()> {
        let null = null_value.unwrap_or_default();

        if let Ok(s) = String::try_convert(rb_f) {
            let f = std::fs::File::create(s).unwrap();
            // no need for a buffered writer, because the csv writer does internal buffering
            CsvWriter::new(f)
                .include_header(include_header)
                .with_separator(separator)
                .with_quote_char(quote_char)
                .with_batch_size(batch_size)
                .with_datetime_format(datetime_format)
                .with_date_format(date_format)
                .with_time_format(time_format)
                .with_float_precision(float_precision)
                .with_null_value(null)
                .finish(&mut self.df.borrow_mut())
                .map_err(RbPolarsErr::from)?;
        } else {
            let mut buf = Cursor::new(Vec::new());
            CsvWriter::new(&mut buf)
                .include_header(include_header)
                .with_separator(separator)
                .with_quote_char(quote_char)
                .with_batch_size(batch_size)
                .with_datetime_format(datetime_format)
                .with_date_format(date_format)
                .with_time_format(time_format)
                .with_float_precision(float_precision)
                .with_null_value(null)
                .finish(&mut self.df.borrow_mut())
                .map_err(RbPolarsErr::from)?;
            // TODO less copying
            let rb_str = RString::from_slice(&buf.into_inner());
            rb_f.funcall::<_, _, Value>("write", (rb_str,))?;
        }

        Ok(())
    }

    pub fn write_ipc(
        &self,
        rb_f: Value,
        compression: Wrap<Option<IpcCompression>>,
    ) -> RbResult<()> {
        if let Ok(s) = String::try_convert(rb_f) {
            let f = std::fs::File::create(s).unwrap();
            IpcWriter::new(f)
                .with_compression(compression.0)
                .finish(&mut self.df.borrow_mut())
                .map_err(RbPolarsErr::from)?;
        } else {
            let mut buf = Cursor::new(Vec::new());
            IpcWriter::new(&mut buf)
                .with_compression(compression.0)
                .finish(&mut self.df.borrow_mut())
                .map_err(RbPolarsErr::from)?;
            // TODO less copying
            let rb_str = RString::from_slice(&buf.into_inner());
            rb_f.funcall::<_, _, Value>("write", (rb_str,))?;
        }
        Ok(())
    }

    pub fn row_tuple(&self, idx: i64) -> Value {
        let idx = if idx < 0 {
            (self.df.borrow().height() as i64 + idx) as usize
        } else {
            idx as usize
        };
        RArray::from_iter(
            self.df
                .borrow()
                .get_columns()
                .iter()
                .map(|s| match s.dtype() {
                    DataType::Object(_, _) => {
                        let obj: Option<&ObjectValue> = s.get_object(idx).map(|any| any.into());
                        obj.unwrap().to_object()
                    }
                    _ => Wrap(s.get(idx).unwrap()).into_value(),
                }),
        )
        .as_value()
    }

    pub fn row_tuples(&self) -> Value {
        let df = &self.df;
        RArray::from_iter((0..df.borrow().height()).map(|idx| {
            RArray::from_iter(
                self.df
                    .borrow()
                    .get_columns()
                    .iter()
                    .map(|s| match s.dtype() {
                        DataType::Object(_, _) => {
                            let obj: Option<&ObjectValue> = s.get_object(idx).map(|any| any.into());
                            obj.unwrap().to_object()
                        }
                        _ => Wrap(s.get(idx).unwrap()).into_value(),
                    }),
            )
        }))
        .as_value()
    }

    pub fn to_numo(&self) -> Option<Value> {
        let mut st = None;
        for s in self.df.borrow().iter() {
            let dt_i = s.dtype();
            match st {
                None => st = Some(dt_i.clone()),
                Some(ref mut st) => {
                    *st = try_get_supertype(st, dt_i).ok()?;
                }
            }
        }
        let _st = st?;

        // TODO
        None
    }

    pub fn write_parquet(
        &self,
        rb_f: Value,
        compression: String,
        compression_level: Option<i32>,
        statistics: bool,
        row_group_size: Option<usize>,
    ) -> RbResult<()> {
        let compression = parse_parquet_compression(&compression, compression_level)?;

        if let Ok(s) = String::try_convert(rb_f) {
            let f = std::fs::File::create(s).unwrap();
            ParquetWriter::new(f)
                .with_compression(compression)
                .with_statistics(statistics)
                .with_row_group_size(row_group_size)
                .finish(&mut self.df.borrow_mut())
                .map_err(RbPolarsErr::from)?;
        } else {
            todo!();
        }

        Ok(())
    }

    pub fn add(&self, s: &RbSeries) -> RbResult<Self> {
        let df = (&*self.df.borrow() + &*s.series.borrow()).map_err(RbPolarsErr::from)?;
        Ok(df.into())
    }

    pub fn sub(&self, s: &RbSeries) -> RbResult<Self> {
        let df = (&*self.df.borrow() - &*s.series.borrow()).map_err(RbPolarsErr::from)?;
        Ok(df.into())
    }

    pub fn div(&self, s: &RbSeries) -> RbResult<Self> {
        let df = (&*self.df.borrow() / &*s.series.borrow()).map_err(RbPolarsErr::from)?;
        Ok(df.into())
    }

    pub fn mul(&self, s: &RbSeries) -> RbResult<Self> {
        let df = (&*self.df.borrow() * &*s.series.borrow()).map_err(RbPolarsErr::from)?;
        Ok(df.into())
    }

    pub fn rem(&self, s: &RbSeries) -> RbResult<Self> {
        let df = (&*self.df.borrow() % &*s.series.borrow()).map_err(RbPolarsErr::from)?;
        Ok(df.into())
    }

    pub fn add_df(&self, s: &Self) -> RbResult<Self> {
        let df = (&*self.df.borrow() + &*s.df.borrow()).map_err(RbPolarsErr::from)?;
        Ok(df.into())
    }

    pub fn sub_df(&self, s: &Self) -> RbResult<Self> {
        let df = (&*self.df.borrow() - &*s.df.borrow()).map_err(RbPolarsErr::from)?;
        Ok(df.into())
    }

    pub fn div_df(&self, s: &Self) -> RbResult<Self> {
        let df = (&*self.df.borrow() / &*s.df.borrow()).map_err(RbPolarsErr::from)?;
        Ok(df.into())
    }

    pub fn mul_df(&self, s: &Self) -> RbResult<Self> {
        let df = (&*self.df.borrow() * &*s.df.borrow()).map_err(RbPolarsErr::from)?;
        Ok(df.into())
    }

    pub fn rem_df(&self, s: &Self) -> RbResult<Self> {
        let df = (&*self.df.borrow() % &*s.df.borrow()).map_err(RbPolarsErr::from)?;
        Ok(df.into())
    }

    pub fn sample_n(
        &self,
        n: &RbSeries,
        with_replacement: bool,
        shuffle: bool,
        seed: Option<u64>,
    ) -> RbResult<Self> {
        let df = self
            .df
            .borrow()
            .sample_n(&n.series.borrow(), with_replacement, shuffle, seed)
            .map_err(RbPolarsErr::from)?;
        Ok(df.into())
    }

    pub fn sample_frac(
        &self,
        frac: &RbSeries,
        with_replacement: bool,
        shuffle: bool,
        seed: Option<u64>,
    ) -> RbResult<Self> {
        let df = self
            .df
            .borrow()
            .sample_frac(&frac.series.borrow(), with_replacement, shuffle, seed)
            .map_err(RbPolarsErr::from)?;
        Ok(df.into())
    }

    pub fn rechunk(&self) -> Self {
        self.df.borrow().agg_chunks().into()
    }

    pub fn to_s(&self) -> String {
        format!("{}", self.df.borrow())
    }

    pub fn get_columns(&self) -> RArray {
        let cols = self.df.borrow().get_columns().to_vec();
        to_rbseries_collection(cols)
    }

    pub fn columns(&self) -> Vec<String> {
        self.df
            .borrow()
            .get_column_names()
            .iter()
            .map(|v| v.to_string())
            .collect()
    }

    pub fn set_column_names(&self, names: Vec<String>) -> RbResult<()> {
        self.df
            .borrow_mut()
            .set_column_names(&names)
            .map_err(RbPolarsErr::from)?;
        Ok(())
    }

    pub fn dtypes(&self) -> RArray {
        RArray::from_iter(
            self.df
                .borrow()
                .iter()
                .map(|s| Wrap(s.dtype().clone()).into_value()),
        )
    }

    pub fn n_chunks(&self) -> usize {
        self.df.borrow().n_chunks()
    }

    pub fn shape(&self) -> (usize, usize) {
        self.df.borrow().shape()
    }

    pub fn height(&self) -> usize {
        self.df.borrow().height()
    }

    pub fn width(&self) -> usize {
        self.df.borrow().width()
    }

    pub fn hstack_mut(&self, columns: RArray) -> RbResult<()> {
        let columns = to_series_collection(columns)?;
        self.df
            .borrow_mut()
            .hstack_mut(&columns)
            .map_err(RbPolarsErr::from)?;
        Ok(())
    }

    pub fn hstack(&self, columns: RArray) -> RbResult<Self> {
        let columns = to_series_collection(columns)?;
        let df = self
            .df
            .borrow()
            .hstack(&columns)
            .map_err(RbPolarsErr::from)?;
        Ok(df.into())
    }

    pub fn extend(&self, df: &RbDataFrame) -> RbResult<()> {
        self.df
            .borrow_mut()
            .extend(&df.df.borrow())
            .map_err(RbPolarsErr::from)?;
        Ok(())
    }

    pub fn vstack_mut(&self, df: &RbDataFrame) -> RbResult<()> {
        self.df
            .borrow_mut()
            .vstack_mut(&df.df.borrow())
            .map_err(RbPolarsErr::from)?;
        Ok(())
    }

    pub fn vstack(&self, df: &RbDataFrame) -> RbResult<Self> {
        let df = self
            .df
            .borrow()
            .vstack(&df.df.borrow())
            .map_err(RbPolarsErr::from)?;
        Ok(df.into())
    }

    pub fn drop_in_place(&self, name: String) -> RbResult<RbSeries> {
        let s = self
            .df
            .borrow_mut()
            .drop_in_place(&name)
            .map_err(RbPolarsErr::from)?;
        Ok(RbSeries::new(s))
    }

    pub fn drop_nulls(&self, subset: Option<Vec<String>>) -> RbResult<Self> {
        let df = self
            .df
            .borrow()
            .drop_nulls(subset.as_ref().map(|s| s.as_ref()))
            .map_err(RbPolarsErr::from)?;
        Ok(df.into())
    }

    pub fn drop(&self, name: String) -> RbResult<Self> {
        let df = self.df.borrow().drop(&name).map_err(RbPolarsErr::from)?;
        Ok(RbDataFrame::new(df))
    }

    pub fn select_at_idx(&self, idx: usize) -> Option<RbSeries> {
        self.df
            .borrow()
            .select_at_idx(idx)
            .map(|s| RbSeries::new(s.clone()))
    }

    pub fn get_column_index(&self, name: String) -> Option<usize> {
        self.df.borrow().get_column_index(&name)
    }

    // TODO remove clone
    pub fn column(&self, name: String) -> RbResult<RbSeries> {
        self.df
            .borrow()
            .column(&name)
            .map(|v| v.clone().into())
            .map_err(RbPolarsErr::from)
    }

    pub fn select(&self, selection: Vec<String>) -> RbResult<Self> {
        let df = self
            .df
            .borrow()
            .select(selection)
            .map_err(RbPolarsErr::from)?;
        Ok(RbDataFrame::new(df))
    }

    pub fn take(&self, indices: Vec<IdxSize>) -> RbResult<Self> {
        let indices = IdxCa::from_vec("", indices);
        let df = self.df.borrow().take(&indices).map_err(RbPolarsErr::from)?;
        Ok(RbDataFrame::new(df))
    }

    pub fn take_with_series(&self, indices: &RbSeries) -> RbResult<Self> {
        let binding = indices.series.borrow();
        let idx = binding.idx().map_err(RbPolarsErr::from)?;
        let df = self.df.borrow().take(idx).map_err(RbPolarsErr::from)?;
        Ok(RbDataFrame::new(df))
    }

    pub fn replace(&self, column: String, new_col: &RbSeries) -> RbResult<()> {
        self.df
            .borrow_mut()
            .replace(&column, new_col.series.borrow().clone())
            .map_err(RbPolarsErr::from)?;
        Ok(())
    }

    pub fn replace_column(&self, index: usize, new_col: &RbSeries) -> RbResult<()> {
        self.df
            .borrow_mut()
            .replace_column(index, new_col.series.borrow().clone())
            .map_err(RbPolarsErr::from)?;
        Ok(())
    }

    pub fn insert_column(&self, index: usize, new_col: &RbSeries) -> RbResult<()> {
        self.df
            .borrow_mut()
            .insert_column(index, new_col.series.borrow().clone())
            .map_err(RbPolarsErr::from)?;
        Ok(())
    }

    pub fn slice(&self, offset: usize, length: Option<usize>) -> Self {
        let df = self.df.borrow().slice(
            offset as i64,
            length.unwrap_or_else(|| self.df.borrow().height()),
        );
        df.into()
    }

    pub fn head(&self, length: Option<usize>) -> Self {
        self.df.borrow().head(length).into()
    }

    pub fn tail(&self, length: Option<usize>) -> Self {
        self.df.borrow().tail(length).into()
    }

    pub fn is_unique(&self) -> RbResult<RbSeries> {
        let mask = self.df.borrow().is_unique().map_err(RbPolarsErr::from)?;
        Ok(mask.into_series().into())
    }

    pub fn is_duplicated(&self) -> RbResult<RbSeries> {
        let mask = self
            .df
            .borrow()
            .is_duplicated()
            .map_err(RbPolarsErr::from)?;
        Ok(mask.into_series().into())
    }

    pub fn equals(&self, other: &RbDataFrame, null_equal: bool) -> bool {
        if null_equal {
            self.df.borrow().equals_missing(&other.df.borrow())
        } else {
            self.df.borrow().equals(&other.df.borrow())
        }
    }

    pub fn with_row_count(&self, name: String, offset: Option<IdxSize>) -> RbResult<Self> {
        let df = self
            .df
            .borrow()
            .with_row_count(&name, offset)
            .map_err(RbPolarsErr::from)?;
        Ok(df.into())
    }

    pub fn clone(&self) -> Self {
        RbDataFrame::new(self.df.borrow().clone())
    }

    pub fn melt(
        &self,
        id_vars: Vec<String>,
        value_vars: Vec<String>,
        value_name: Option<String>,
        variable_name: Option<String>,
    ) -> RbResult<Self> {
        let args = MeltArgs {
            id_vars: strings_to_smartstrings(id_vars),
            value_vars: strings_to_smartstrings(value_vars),
            value_name: value_name.map(|s| s.into()),
            variable_name: variable_name.map(|s| s.into()),
            streamable: false,
        };

        let df = self.df.borrow().melt2(args).map_err(RbPolarsErr::from)?;
        Ok(RbDataFrame::new(df))
    }

    #[allow(clippy::too_many_arguments)]
    pub fn pivot_expr(
        &self,
        values: Vec<String>,
        index: Vec<String>,
        columns: Vec<String>,
        maintain_order: bool,
        sort_columns: bool,
        aggregate_expr: Option<&RbExpr>,
        separator: Option<String>,
    ) -> RbResult<Self> {
        let fun = match maintain_order {
            true => pivot_stable,
            false => pivot,
        };
        let agg_expr = aggregate_expr.map(|aggregate_expr| aggregate_expr.inner.clone());
        let df = fun(
            &self.df.borrow(),
            values,
            index,
            columns,
            sort_columns,
            agg_expr,
            separator.as_deref(),
        )
        .map_err(RbPolarsErr::from)?;
        Ok(RbDataFrame::new(df))
    }

    pub fn partition_by(
        &self,
        by: Vec<String>,
        maintain_order: bool,
        include_key: bool,
    ) -> RbResult<RArray> {
        let out = if maintain_order {
            self.df.borrow().partition_by_stable(by, include_key)
        } else {
            self.df.borrow().partition_by(by, include_key)
        }
        .map_err(RbPolarsErr::from)?;
        Ok(RArray::from_iter(out.into_iter().map(RbDataFrame::new)))
    }

    pub fn shift(&self, periods: i64) -> Self {
        self.df.borrow().shift(periods).into()
    }

    pub fn lazy(&self) -> RbLazyFrame {
        self.df.borrow().clone().lazy().into()
    }

    pub fn max_horizontal(&self) -> RbResult<Option<RbSeries>> {
        let s = self
            .df
            .borrow()
            .max_horizontal()
            .map_err(RbPolarsErr::from)?;
        Ok(s.map(|s| s.into()))
    }

    pub fn min_horizontal(&self) -> RbResult<Option<RbSeries>> {
        let s = self
            .df
            .borrow()
            .min_horizontal()
            .map_err(RbPolarsErr::from)?;
        Ok(s.map(|s| s.into()))
    }

    pub fn sum_horizontal(&self, ignore_nulls: bool) -> RbResult<Option<RbSeries>> {
        let null_strategy = if ignore_nulls {
            NullStrategy::Ignore
        } else {
            NullStrategy::Propagate
        };
        let s = self
            .df
            .borrow()
            .sum_horizontal(null_strategy)
            .map_err(RbPolarsErr::from)?;
        Ok(s.map(|s| s.into()))
    }

    pub fn mean_horizontal(&self, ignore_nulls: bool) -> RbResult<Option<RbSeries>> {
        let null_strategy = if ignore_nulls {
            NullStrategy::Ignore
        } else {
            NullStrategy::Propagate
        };
        let s = self
            .df
            .borrow()
            .mean_horizontal(null_strategy)
            .map_err(RbPolarsErr::from)?;
        Ok(s.map(|s| s.into()))
    }

    pub fn to_dummies(
        &self,
        columns: Option<Vec<String>>,
        separator: Option<String>,
        drop_first: bool,
    ) -> RbResult<Self> {
        let df = match columns {
            Some(cols) => self.df.borrow().columns_to_dummies(
                cols.iter().map(|x| x as &str).collect(),
                separator.as_deref(),
                drop_first,
            ),
            None => self
                .df
                .borrow()
                .to_dummies(separator.as_deref(), drop_first),
        }
        .map_err(RbPolarsErr::from)?;
        Ok(df.into())
    }

    pub fn null_count(&self) -> Self {
        let df = self.df.borrow().null_count();
        df.into()
    }

    pub fn apply(
        &self,
        lambda: Value,
        output_type: Option<Wrap<DataType>>,
        inference_size: usize,
    ) -> RbResult<(Value, bool)> {
        let df = &self.df.borrow();

        let output_type = output_type.map(|dt| dt.0);
        let out = match output_type {
            Some(DataType::Int32) => {
                apply_lambda_with_primitive_out_type::<Int32Type>(df, lambda, 0, None).into_series()
            }
            Some(DataType::Int64) => {
                apply_lambda_with_primitive_out_type::<Int64Type>(df, lambda, 0, None).into_series()
            }
            Some(DataType::UInt32) => {
                apply_lambda_with_primitive_out_type::<UInt32Type>(df, lambda, 0, None)
                    .into_series()
            }
            Some(DataType::UInt64) => {
                apply_lambda_with_primitive_out_type::<UInt64Type>(df, lambda, 0, None)
                    .into_series()
            }
            Some(DataType::Float32) => {
                apply_lambda_with_primitive_out_type::<Float32Type>(df, lambda, 0, None)
                    .into_series()
            }
            Some(DataType::Float64) => {
                apply_lambda_with_primitive_out_type::<Float64Type>(df, lambda, 0, None)
                    .into_series()
            }
            Some(DataType::Boolean) => {
                apply_lambda_with_bool_out_type(df, lambda, 0, None).into_series()
            }
            Some(DataType::Date) => {
                apply_lambda_with_primitive_out_type::<Int32Type>(df, lambda, 0, None)
                    .into_date()
                    .into_series()
            }
            Some(DataType::Datetime(tu, tz)) => {
                apply_lambda_with_primitive_out_type::<Int64Type>(df, lambda, 0, None)
                    .into_datetime(tu, tz)
                    .into_series()
            }
            Some(DataType::String) => {
                apply_lambda_with_utf8_out_type(df, lambda, 0, None).into_series()
            }
            _ => return apply_lambda_unknown(df, lambda, inference_size),
        };

        Ok((Obj::wrap(RbSeries::from(out)).as_value(), false))
    }

    pub fn shrink_to_fit(&self) {
        self.df.borrow_mut().shrink_to_fit();
    }

    pub fn hash_rows(&self, k0: u64, k1: u64, k2: u64, k3: u64) -> RbResult<RbSeries> {
        let hb = ahash::RandomState::with_seeds(k0, k1, k2, k3);
        let hash = self
            .df
            .borrow_mut()
            .hash_rows(Some(hb))
            .map_err(RbPolarsErr::from)?;
        Ok(hash.into_series().into())
    }

    pub fn transpose(&self, keep_names_as: Option<String>, column_names: Value) -> RbResult<Self> {
        let new_col_names = if let Ok(name) = <Vec<String>>::try_convert(column_names) {
            Some(Either::Right(name))
        } else if let Ok(name) = String::try_convert(column_names) {
            Some(Either::Left(name))
        } else {
            None
        };
        Ok(self
            .df
            .borrow()
            .transpose(keep_names_as.as_deref(), new_col_names)
            .map_err(RbPolarsErr::from)?
            .into())
    }

    pub fn upsample(
        &self,
        by: Vec<String>,
        index_column: String,
        every: String,
        offset: String,
        stable: bool,
    ) -> RbResult<Self> {
        let out = if stable {
            self.df.borrow().upsample_stable(
                by,
                &index_column,
                Duration::parse(&every),
                Duration::parse(&offset),
            )
        } else {
            self.df.borrow().upsample(
                by,
                &index_column,
                Duration::parse(&every),
                Duration::parse(&offset),
            )
        };
        let out = out.map_err(RbPolarsErr::from)?;
        Ok(out.into())
    }

    pub fn to_struct(&self, name: String) -> RbSeries {
        let s = self.df.borrow().clone().into_struct(&name);
        s.into_series().into()
    }

    pub fn unnest(&self, names: Vec<String>) -> RbResult<Self> {
        let df = self.df.borrow().unnest(names).map_err(RbPolarsErr::from)?;
        Ok(df.into())
    }
}