polars-df 0.1.1 → 0.1.2
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +4 -4
- data/CHANGELOG.md +4 -0
- data/Cargo.lock +1 -1
- data/ext/polars/Cargo.toml +1 -1
- data/ext/polars/src/batched_csv.rs +120 -0
- data/ext/polars/src/conversion.rs +105 -5
- data/ext/polars/src/dataframe.rs +132 -4
- data/ext/polars/src/error.rs +9 -0
- data/ext/polars/src/file.rs +8 -7
- data/ext/polars/src/lazy/apply.rs +7 -0
- data/ext/polars/src/lazy/dataframe.rs +132 -0
- data/ext/polars/src/lazy/dsl.rs +38 -0
- data/ext/polars/src/lazy/meta.rs +1 -1
- data/ext/polars/src/lazy/mod.rs +1 -0
- data/ext/polars/src/lib.rs +77 -3
- data/ext/polars/src/series.rs +8 -9
- data/lib/polars/batched_csv_reader.rb +95 -0
- data/lib/polars/data_frame.rb +585 -19
- data/lib/polars/expr.rb +17 -2
- data/lib/polars/io.rb +342 -2
- data/lib/polars/lazy_frame.rb +156 -2
- data/lib/polars/lazy_functions.rb +154 -11
- data/lib/polars/series.rb +806 -18
- data/lib/polars/utils.rb +33 -0
- data/lib/polars/version.rb +1 -1
- data/lib/polars.rb +9 -0
- metadata +5 -2
@@ -1,4 +1,5 @@
|
|
1
1
|
use magnus::{RArray, RHash, Value};
|
2
|
+
use polars::io::RowCount;
|
2
3
|
use polars::lazy::frame::{LazyFrame, LazyGroupBy};
|
3
4
|
use polars::prelude::*;
|
4
5
|
use std::cell::RefCell;
|
@@ -52,6 +53,137 @@ impl From<LazyFrame> for RbLazyFrame {
|
|
52
53
|
}
|
53
54
|
|
54
55
|
impl RbLazyFrame {
|
56
|
+
pub fn new_from_ndjson(
|
57
|
+
path: String,
|
58
|
+
infer_schema_length: Option<usize>,
|
59
|
+
batch_size: Option<usize>,
|
60
|
+
n_rows: Option<usize>,
|
61
|
+
low_memory: bool,
|
62
|
+
rechunk: bool,
|
63
|
+
row_count: Option<(String, IdxSize)>,
|
64
|
+
) -> RbResult<Self> {
|
65
|
+
let row_count = row_count.map(|(name, offset)| RowCount { name, offset });
|
66
|
+
|
67
|
+
let lf = LazyJsonLineReader::new(path)
|
68
|
+
.with_infer_schema_length(infer_schema_length)
|
69
|
+
.with_batch_size(batch_size)
|
70
|
+
.with_n_rows(n_rows)
|
71
|
+
.low_memory(low_memory)
|
72
|
+
.with_rechunk(rechunk)
|
73
|
+
.with_row_count(row_count)
|
74
|
+
.finish()
|
75
|
+
.map_err(RbPolarsErr::from)?;
|
76
|
+
Ok(lf.into())
|
77
|
+
}
|
78
|
+
|
79
|
+
pub fn new_from_csv(arguments: &[Value]) -> RbResult<Self> {
|
80
|
+
// start arguments
|
81
|
+
// this pattern is needed for more than 16
|
82
|
+
let path: String = arguments[0].try_convert()?;
|
83
|
+
let sep: String = arguments[1].try_convert()?;
|
84
|
+
let has_header: bool = arguments[2].try_convert()?;
|
85
|
+
let ignore_errors: bool = arguments[3].try_convert()?;
|
86
|
+
let skip_rows: usize = arguments[4].try_convert()?;
|
87
|
+
let n_rows: Option<usize> = arguments[5].try_convert()?;
|
88
|
+
let cache: bool = arguments[6].try_convert()?;
|
89
|
+
let overwrite_dtype: Option<Vec<(String, Wrap<DataType>)>> = arguments[7].try_convert()?;
|
90
|
+
let low_memory: bool = arguments[8].try_convert()?;
|
91
|
+
let comment_char: Option<String> = arguments[9].try_convert()?;
|
92
|
+
let quote_char: Option<String> = arguments[10].try_convert()?;
|
93
|
+
let null_values: Option<Wrap<NullValues>> = arguments[11].try_convert()?;
|
94
|
+
let infer_schema_length: Option<usize> = arguments[12].try_convert()?;
|
95
|
+
let with_schema_modify: Option<Value> = arguments[13].try_convert()?;
|
96
|
+
let rechunk: bool = arguments[14].try_convert()?;
|
97
|
+
let skip_rows_after_header: usize = arguments[15].try_convert()?;
|
98
|
+
let encoding: Wrap<CsvEncoding> = arguments[16].try_convert()?;
|
99
|
+
let row_count: Option<(String, IdxSize)> = arguments[17].try_convert()?;
|
100
|
+
let parse_dates: bool = arguments[18].try_convert()?;
|
101
|
+
let eol_char: String = arguments[19].try_convert()?;
|
102
|
+
// end arguments
|
103
|
+
|
104
|
+
let null_values = null_values.map(|w| w.0);
|
105
|
+
let comment_char = comment_char.map(|s| s.as_bytes()[0]);
|
106
|
+
let quote_char = quote_char.map(|s| s.as_bytes()[0]);
|
107
|
+
let delimiter = sep.as_bytes()[0];
|
108
|
+
let eol_char = eol_char.as_bytes()[0];
|
109
|
+
|
110
|
+
let row_count = row_count.map(|(name, offset)| RowCount { name, offset });
|
111
|
+
|
112
|
+
let overwrite_dtype = overwrite_dtype.map(|overwrite_dtype| {
|
113
|
+
let fields = overwrite_dtype
|
114
|
+
.into_iter()
|
115
|
+
.map(|(name, dtype)| Field::new(&name, dtype.0));
|
116
|
+
Schema::from(fields)
|
117
|
+
});
|
118
|
+
let r = LazyCsvReader::new(path)
|
119
|
+
.with_infer_schema_length(infer_schema_length)
|
120
|
+
.with_delimiter(delimiter)
|
121
|
+
.has_header(has_header)
|
122
|
+
.with_ignore_parser_errors(ignore_errors)
|
123
|
+
.with_skip_rows(skip_rows)
|
124
|
+
.with_n_rows(n_rows)
|
125
|
+
.with_cache(cache)
|
126
|
+
.with_dtype_overwrite(overwrite_dtype.as_ref())
|
127
|
+
.low_memory(low_memory)
|
128
|
+
.with_comment_char(comment_char)
|
129
|
+
.with_quote_char(quote_char)
|
130
|
+
.with_end_of_line_char(eol_char)
|
131
|
+
.with_rechunk(rechunk)
|
132
|
+
.with_skip_rows_after_header(skip_rows_after_header)
|
133
|
+
.with_encoding(encoding.0)
|
134
|
+
.with_row_count(row_count)
|
135
|
+
.with_parse_dates(parse_dates)
|
136
|
+
.with_null_values(null_values);
|
137
|
+
|
138
|
+
if let Some(_lambda) = with_schema_modify {
|
139
|
+
todo!();
|
140
|
+
}
|
141
|
+
|
142
|
+
Ok(r.finish().map_err(RbPolarsErr::from)?.into())
|
143
|
+
}
|
144
|
+
|
145
|
+
pub fn new_from_parquet(
|
146
|
+
path: String,
|
147
|
+
n_rows: Option<usize>,
|
148
|
+
cache: bool,
|
149
|
+
parallel: Wrap<ParallelStrategy>,
|
150
|
+
rechunk: bool,
|
151
|
+
row_count: Option<(String, IdxSize)>,
|
152
|
+
low_memory: bool,
|
153
|
+
) -> RbResult<Self> {
|
154
|
+
let row_count = row_count.map(|(name, offset)| RowCount { name, offset });
|
155
|
+
let args = ScanArgsParquet {
|
156
|
+
n_rows,
|
157
|
+
cache,
|
158
|
+
parallel: parallel.0,
|
159
|
+
rechunk,
|
160
|
+
row_count,
|
161
|
+
low_memory,
|
162
|
+
};
|
163
|
+
let lf = LazyFrame::scan_parquet(path, args).map_err(RbPolarsErr::from)?;
|
164
|
+
Ok(lf.into())
|
165
|
+
}
|
166
|
+
|
167
|
+
pub fn new_from_ipc(
|
168
|
+
path: String,
|
169
|
+
n_rows: Option<usize>,
|
170
|
+
cache: bool,
|
171
|
+
rechunk: bool,
|
172
|
+
row_count: Option<(String, IdxSize)>,
|
173
|
+
memory_map: bool,
|
174
|
+
) -> RbResult<Self> {
|
175
|
+
let row_count = row_count.map(|(name, offset)| RowCount { name, offset });
|
176
|
+
let args = ScanArgsIpc {
|
177
|
+
n_rows,
|
178
|
+
cache,
|
179
|
+
rechunk,
|
180
|
+
row_count,
|
181
|
+
memmap: memory_map,
|
182
|
+
};
|
183
|
+
let lf = LazyFrame::scan_ipc(path, args).map_err(RbPolarsErr::from)?;
|
184
|
+
Ok(lf.into())
|
185
|
+
}
|
186
|
+
|
55
187
|
pub fn write_json(&self, rb_f: Value) -> RbResult<()> {
|
56
188
|
let file = BufWriter::new(get_file_like(rb_f, true)?);
|
57
189
|
serde_json::to_writer(file, &self.ldf.logical_plan)
|
data/ext/polars/src/lazy/dsl.rs
CHANGED
@@ -6,6 +6,7 @@ use polars::prelude::*;
|
|
6
6
|
use polars::series::ops::NullBehavior;
|
7
7
|
|
8
8
|
use crate::conversion::*;
|
9
|
+
use crate::lazy::apply::*;
|
9
10
|
use crate::lazy::utils::rb_exprs_to_exprs;
|
10
11
|
use crate::RbResult;
|
11
12
|
|
@@ -901,6 +902,10 @@ impl RbExpr {
|
|
901
902
|
self.inner.clone().suffix(&suffix).into()
|
902
903
|
}
|
903
904
|
|
905
|
+
pub fn exclude(&self, columns: Vec<String>) -> Self {
|
906
|
+
self.inner.clone().exclude(columns).into()
|
907
|
+
}
|
908
|
+
|
904
909
|
pub fn interpolate(&self) -> Self {
|
905
910
|
self.inner.clone().interpolate().into()
|
906
911
|
}
|
@@ -1333,6 +1338,29 @@ pub fn col(name: String) -> RbExpr {
|
|
1333
1338
|
dsl::col(&name).into()
|
1334
1339
|
}
|
1335
1340
|
|
1341
|
+
pub fn count() -> RbExpr {
|
1342
|
+
dsl::count().into()
|
1343
|
+
}
|
1344
|
+
|
1345
|
+
pub fn first() -> RbExpr {
|
1346
|
+
dsl::first().into()
|
1347
|
+
}
|
1348
|
+
|
1349
|
+
pub fn last() -> RbExpr {
|
1350
|
+
dsl::last().into()
|
1351
|
+
}
|
1352
|
+
|
1353
|
+
pub fn cols(names: Vec<String>) -> RbExpr {
|
1354
|
+
dsl::cols(names).into()
|
1355
|
+
}
|
1356
|
+
|
1357
|
+
pub fn fold(acc: &RbExpr, lambda: Value, exprs: RArray) -> RbResult<RbExpr> {
|
1358
|
+
let exprs = rb_exprs_to_exprs(exprs)?;
|
1359
|
+
|
1360
|
+
let func = move |a: Series, b: Series| binary_lambda(lambda, a, b);
|
1361
|
+
Ok(polars::lazy::dsl::fold_exprs(acc.inner.clone(), func, exprs).into())
|
1362
|
+
}
|
1363
|
+
|
1336
1364
|
// TODO improve
|
1337
1365
|
pub fn lit(value: Value) -> RbResult<RbExpr> {
|
1338
1366
|
if value.is_nil() {
|
@@ -1387,3 +1415,13 @@ impl RbWhenThen {
|
|
1387
1415
|
pub fn when(predicate: &RbExpr) -> RbWhen {
|
1388
1416
|
dsl::when(predicate.inner.clone()).into()
|
1389
1417
|
}
|
1418
|
+
|
1419
|
+
pub fn concat_str(s: RArray, sep: String) -> RbResult<RbExpr> {
|
1420
|
+
let s = rb_exprs_to_exprs(s)?;
|
1421
|
+
Ok(dsl::concat_str(s, &sep).into())
|
1422
|
+
}
|
1423
|
+
|
1424
|
+
pub fn concat_lst(s: RArray) -> RbResult<RbExpr> {
|
1425
|
+
let s = rb_exprs_to_exprs(s)?;
|
1426
|
+
Ok(dsl::concat_lst(s).into())
|
1427
|
+
}
|
data/ext/polars/src/lazy/meta.rs
CHANGED
data/ext/polars/src/lazy/mod.rs
CHANGED
data/ext/polars/src/lib.rs
CHANGED
@@ -1,3 +1,4 @@
|
|
1
|
+
mod batched_csv;
|
1
2
|
mod conversion;
|
2
3
|
mod dataframe;
|
3
4
|
mod error;
|
@@ -5,14 +6,18 @@ mod file;
|
|
5
6
|
mod lazy;
|
6
7
|
mod series;
|
7
8
|
|
8
|
-
use
|
9
|
+
use batched_csv::RbBatchedCsv;
|
10
|
+
use conversion::*;
|
9
11
|
use dataframe::RbDataFrame;
|
10
12
|
use error::{RbPolarsErr, RbValueError};
|
13
|
+
use file::get_file_like;
|
11
14
|
use lazy::dataframe::{RbLazyFrame, RbLazyGroupBy};
|
12
15
|
use lazy::dsl::{RbExpr, RbWhen, RbWhenThen};
|
13
16
|
use magnus::{
|
14
|
-
define_module, function, memoize, method, prelude::*, Error, RArray, RClass, RModule,
|
17
|
+
define_module, function, memoize, method, prelude::*, Error, RArray, RClass, RHash, RModule,
|
18
|
+
Value,
|
15
19
|
};
|
20
|
+
use polars::datatypes::DataType;
|
16
21
|
use polars::error::PolarsResult;
|
17
22
|
use polars::frame::DataFrame;
|
18
23
|
use polars::functions::{diag_concat_df, hor_concat_df};
|
@@ -34,11 +39,19 @@ fn init() -> RbResult<()> {
|
|
34
39
|
module.define_singleton_method("_concat_df", function!(concat_df, 1))?;
|
35
40
|
module.define_singleton_method("_diag_concat_df", function!(rb_diag_concat_df, 1))?;
|
36
41
|
module.define_singleton_method("_hor_concat_df", function!(rb_hor_concat_df, 1))?;
|
42
|
+
module.define_singleton_method("_concat_series", function!(concat_series, 1))?;
|
43
|
+
module.define_singleton_method("_ipc_schema", function!(ipc_schema, 1))?;
|
44
|
+
module.define_singleton_method("_parquet_schema", function!(parquet_schema, 1))?;
|
45
|
+
|
46
|
+
let class = module.define_class("RbBatchedCsv", Default::default())?;
|
47
|
+
class.define_singleton_method("new", function!(RbBatchedCsv::new, -1))?;
|
48
|
+
class.define_method("next_batches", method!(RbBatchedCsv::next_batches, 1))?;
|
37
49
|
|
38
50
|
let class = module.define_class("RbDataFrame", Default::default())?;
|
39
51
|
class.define_singleton_method("new", function!(RbDataFrame::init, 1))?;
|
40
|
-
class.define_singleton_method("read_csv", function!(RbDataFrame::read_csv,
|
52
|
+
class.define_singleton_method("read_csv", function!(RbDataFrame::read_csv, -1))?;
|
41
53
|
class.define_singleton_method("read_parquet", function!(RbDataFrame::read_parquet, 1))?;
|
54
|
+
class.define_singleton_method("read_ipc", function!(RbDataFrame::read_ipc, 6))?;
|
42
55
|
class.define_singleton_method("read_hash", function!(RbDataFrame::read_hash, 1))?;
|
43
56
|
class.define_singleton_method("read_json", function!(RbDataFrame::read_json, 1))?;
|
44
57
|
class.define_singleton_method("read_ndjson", function!(RbDataFrame::read_ndjson, 1))?;
|
@@ -46,6 +59,7 @@ fn init() -> RbResult<()> {
|
|
46
59
|
class.define_method("write_json", method!(RbDataFrame::write_json, 3))?;
|
47
60
|
class.define_method("write_ndjson", method!(RbDataFrame::write_ndjson, 1))?;
|
48
61
|
class.define_method("write_csv", method!(RbDataFrame::write_csv, 10))?;
|
62
|
+
class.define_method("write_ipc", method!(RbDataFrame::write_ipc, 2))?;
|
49
63
|
class.define_method("write_parquet", method!(RbDataFrame::write_parquet, 5))?;
|
50
64
|
class.define_method("rechunk", method!(RbDataFrame::rechunk, 0))?;
|
51
65
|
class.define_method("to_s", method!(RbDataFrame::to_s, 0))?;
|
@@ -294,6 +308,7 @@ fn init() -> RbResult<()> {
|
|
294
308
|
class.define_method("keep_name", method!(RbExpr::keep_name, 0))?;
|
295
309
|
class.define_method("prefix", method!(RbExpr::prefix, 1))?;
|
296
310
|
class.define_method("suffix", method!(RbExpr::suffix, 1))?;
|
311
|
+
class.define_method("exclude", method!(RbExpr::exclude, 1))?;
|
297
312
|
class.define_method("interpolate", method!(RbExpr::interpolate, 0))?;
|
298
313
|
class.define_method("rolling_sum", method!(RbExpr::rolling_sum, 6))?;
|
299
314
|
class.define_method("rolling_min", method!(RbExpr::rolling_min, 6))?;
|
@@ -364,11 +379,28 @@ fn init() -> RbResult<()> {
|
|
364
379
|
|
365
380
|
// maybe add to different class
|
366
381
|
class.define_singleton_method("col", function!(crate::lazy::dsl::col, 1))?;
|
382
|
+
class.define_singleton_method("count", function!(crate::lazy::dsl::count, 0))?;
|
383
|
+
class.define_singleton_method("first", function!(crate::lazy::dsl::first, 0))?;
|
384
|
+
class.define_singleton_method("last", function!(crate::lazy::dsl::last, 0))?;
|
385
|
+
class.define_singleton_method("cols", function!(crate::lazy::dsl::cols, 1))?;
|
386
|
+
class.define_singleton_method("fold", function!(crate::lazy::dsl::fold, 3))?;
|
367
387
|
class.define_singleton_method("lit", function!(crate::lazy::dsl::lit, 1))?;
|
368
388
|
class.define_singleton_method("arange", function!(crate::lazy::dsl::arange, 3))?;
|
369
389
|
class.define_singleton_method("when", function!(crate::lazy::dsl::when, 1))?;
|
390
|
+
class.define_singleton_method("concat_str", function!(crate::lazy::dsl::concat_str, 2))?;
|
391
|
+
class.define_singleton_method("concat_lst", function!(crate::lazy::dsl::concat_lst, 1))?;
|
370
392
|
|
371
393
|
let class = module.define_class("RbLazyFrame", Default::default())?;
|
394
|
+
class.define_singleton_method(
|
395
|
+
"new_from_ndjson",
|
396
|
+
function!(RbLazyFrame::new_from_ndjson, 7),
|
397
|
+
)?;
|
398
|
+
class.define_singleton_method("new_from_csv", function!(RbLazyFrame::new_from_csv, -1))?;
|
399
|
+
class.define_singleton_method(
|
400
|
+
"new_from_parquet",
|
401
|
+
function!(RbLazyFrame::new_from_parquet, 7),
|
402
|
+
)?;
|
403
|
+
class.define_singleton_method("new_from_ipc", function!(RbLazyFrame::new_from_ipc, 6))?;
|
372
404
|
class.define_method("write_json", method!(RbLazyFrame::write_json, 1))?;
|
373
405
|
class.define_method("describe_plan", method!(RbLazyFrame::describe_plan, 0))?;
|
374
406
|
class.define_method(
|
@@ -567,3 +599,45 @@ fn rb_hor_concat_df(seq: RArray) -> RbResult<RbDataFrame> {
|
|
567
599
|
let df = hor_concat_df(&dfs).map_err(RbPolarsErr::from)?;
|
568
600
|
Ok(df.into())
|
569
601
|
}
|
602
|
+
|
603
|
+
fn concat_series(seq: RArray) -> RbResult<RbSeries> {
|
604
|
+
let mut iter = seq.each();
|
605
|
+
let first = iter.next().unwrap()?;
|
606
|
+
|
607
|
+
let mut s = get_series(first)?;
|
608
|
+
|
609
|
+
for res in iter {
|
610
|
+
let item = res?;
|
611
|
+
let item = get_series(item)?;
|
612
|
+
s.append(&item).map_err(RbPolarsErr::from)?;
|
613
|
+
}
|
614
|
+
Ok(s.into())
|
615
|
+
}
|
616
|
+
|
617
|
+
fn ipc_schema(rb_f: Value) -> RbResult<Value> {
|
618
|
+
use polars::export::arrow::io::ipc::read::read_file_metadata;
|
619
|
+
let mut r = get_file_like(rb_f, false)?;
|
620
|
+
let metadata = read_file_metadata(&mut r).map_err(RbPolarsErr::arrow)?;
|
621
|
+
|
622
|
+
let dict = RHash::new();
|
623
|
+
for field in metadata.schema.fields {
|
624
|
+
let dt: Wrap<DataType> = Wrap((&field.data_type).into());
|
625
|
+
dict.aset(field.name, dt)?;
|
626
|
+
}
|
627
|
+
Ok(dict.into())
|
628
|
+
}
|
629
|
+
|
630
|
+
fn parquet_schema(rb_f: Value) -> RbResult<Value> {
|
631
|
+
use polars::export::arrow::io::parquet::read::{infer_schema, read_metadata};
|
632
|
+
|
633
|
+
let mut r = get_file_like(rb_f, false)?;
|
634
|
+
let metadata = read_metadata(&mut r).map_err(RbPolarsErr::arrow)?;
|
635
|
+
let arrow_schema = infer_schema(&metadata).map_err(RbPolarsErr::arrow)?;
|
636
|
+
|
637
|
+
let dict = RHash::new();
|
638
|
+
for field in arrow_schema.fields {
|
639
|
+
let dt: Wrap<DataType> = Wrap((&field.data_type).into());
|
640
|
+
dict.aset(field.name, dt)?;
|
641
|
+
}
|
642
|
+
Ok(dict.into())
|
643
|
+
}
|
data/ext/polars/src/series.rs
CHANGED
@@ -116,11 +116,10 @@ init_method_opt!(new_opt_f32, Float32Type, f32);
|
|
116
116
|
init_method_opt!(new_opt_f64, Float64Type, f64);
|
117
117
|
|
118
118
|
impl RbSeries {
|
119
|
-
pub fn new_str(name: String, val:
|
120
|
-
let
|
121
|
-
let mut s = Utf8Chunked::new(&name, v).into_series();
|
119
|
+
pub fn new_str(name: String, val: Wrap<Utf8Chunked>, _strict: bool) -> Self {
|
120
|
+
let mut s = val.0.into_series();
|
122
121
|
s.rename(&name);
|
123
|
-
|
122
|
+
RbSeries::new(s)
|
124
123
|
}
|
125
124
|
|
126
125
|
pub fn estimated_size(&self) -> usize {
|
@@ -199,16 +198,16 @@ impl RbSeries {
|
|
199
198
|
self.series.borrow_mut().rename(&name);
|
200
199
|
}
|
201
200
|
|
202
|
-
pub fn dtype(&self) ->
|
203
|
-
self.series.borrow().dtype().
|
201
|
+
pub fn dtype(&self) -> Value {
|
202
|
+
Wrap(self.series.borrow().dtype().clone()).into()
|
204
203
|
}
|
205
204
|
|
206
|
-
pub fn inner_dtype(&self) -> Option<
|
205
|
+
pub fn inner_dtype(&self) -> Option<Value> {
|
207
206
|
self.series
|
208
207
|
.borrow()
|
209
208
|
.dtype()
|
210
209
|
.inner_dtype()
|
211
|
-
.map(|dt| dt.
|
210
|
+
.map(|dt| Wrap(dt.clone()).into())
|
212
211
|
}
|
213
212
|
|
214
213
|
pub fn set_sorted(&self, reverse: bool) -> Self {
|
@@ -634,5 +633,5 @@ impl RbSeries {
|
|
634
633
|
}
|
635
634
|
|
636
635
|
pub fn to_rbseries_collection(s: Vec<Series>) -> Vec<RbSeries> {
|
637
|
-
s.into_iter().map(
|
636
|
+
s.into_iter().map(RbSeries::new).collect()
|
638
637
|
}
|
@@ -0,0 +1,95 @@
|
|
1
|
+
module Polars
|
2
|
+
class BatchedCsvReader
|
3
|
+
attr_accessor :_reader, :new_columns
|
4
|
+
|
5
|
+
def initialize(
|
6
|
+
file,
|
7
|
+
has_header: true,
|
8
|
+
columns: nil,
|
9
|
+
sep: ",",
|
10
|
+
comment_char: nil,
|
11
|
+
quote_char: '"',
|
12
|
+
skip_rows: 0,
|
13
|
+
dtypes: nil,
|
14
|
+
null_values: nil,
|
15
|
+
ignore_errors: false,
|
16
|
+
parse_dates: false,
|
17
|
+
n_threads: nil,
|
18
|
+
infer_schema_length: 100,
|
19
|
+
batch_size: 50_000,
|
20
|
+
n_rows: nil,
|
21
|
+
encoding: "utf8",
|
22
|
+
low_memory: false,
|
23
|
+
rechunk: true,
|
24
|
+
skip_rows_after_header: 0,
|
25
|
+
row_count_name: nil,
|
26
|
+
row_count_offset: 0,
|
27
|
+
sample_size: 1024,
|
28
|
+
eol_char: "\n",
|
29
|
+
new_columns: nil
|
30
|
+
)
|
31
|
+
if file.is_a?(String) || (defined?(Pathname) && file.is_a?(Pathname))
|
32
|
+
path = Utils.format_path(file)
|
33
|
+
end
|
34
|
+
|
35
|
+
dtype_list = nil
|
36
|
+
dtype_slice = nil
|
37
|
+
if !dtypes.nil?
|
38
|
+
if dtypes.is_a?(Hash)
|
39
|
+
dtype_list = []
|
40
|
+
dtypes.each do|k, v|
|
41
|
+
dtype_list << [k, Utils.rb_type_to_dtype(v)]
|
42
|
+
end
|
43
|
+
elsif dtypes.is_a?(Array)
|
44
|
+
dtype_slice = dtypes
|
45
|
+
else
|
46
|
+
raise ArgumentError, "dtype arg should be list or dict"
|
47
|
+
end
|
48
|
+
end
|
49
|
+
|
50
|
+
processed_null_values = Utils._process_null_values(null_values)
|
51
|
+
projection, columns = Utils.handle_projection_columns(columns)
|
52
|
+
|
53
|
+
self._reader = RbBatchedCsv.new(
|
54
|
+
infer_schema_length,
|
55
|
+
batch_size,
|
56
|
+
has_header,
|
57
|
+
ignore_errors,
|
58
|
+
n_rows,
|
59
|
+
skip_rows,
|
60
|
+
projection,
|
61
|
+
sep,
|
62
|
+
rechunk,
|
63
|
+
columns,
|
64
|
+
encoding,
|
65
|
+
n_threads,
|
66
|
+
path,
|
67
|
+
dtype_list,
|
68
|
+
dtype_slice,
|
69
|
+
low_memory,
|
70
|
+
comment_char,
|
71
|
+
quote_char,
|
72
|
+
processed_null_values,
|
73
|
+
parse_dates,
|
74
|
+
skip_rows_after_header,
|
75
|
+
Utils._prepare_row_count_args(row_count_name, row_count_offset),
|
76
|
+
sample_size,
|
77
|
+
eol_char
|
78
|
+
)
|
79
|
+
self.new_columns = new_columns
|
80
|
+
end
|
81
|
+
|
82
|
+
def next_batches(n)
|
83
|
+
batches = _reader.next_batches(n)
|
84
|
+
if !batches.nil?
|
85
|
+
if new_columns
|
86
|
+
batches.map { |df| Utils._update_columns(Utils.wrap_df(df), new_columns) }
|
87
|
+
else
|
88
|
+
batches.map { |df| Utils.wrap_df(df) }
|
89
|
+
end
|
90
|
+
else
|
91
|
+
nil
|
92
|
+
end
|
93
|
+
end
|
94
|
+
end
|
95
|
+
end
|