polars_io/csv/read/
options.rs

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
use std::path::PathBuf;
use std::sync::Arc;

use polars_core::datatypes::{DataType, Field};
use polars_core::schema::{Schema, SchemaRef};
use polars_error::PolarsResult;
use polars_utils::pl_str::PlSmallStr;
#[cfg(feature = "serde")]
use serde::{Deserialize, Serialize};

use crate::RowIndex;

#[derive(Clone, Debug, PartialEq, Eq, Hash)]
#[cfg_attr(feature = "serde", derive(Serialize, Deserialize))]
pub struct CsvReadOptions {
    pub path: Option<PathBuf>,
    // Performance related options
    pub rechunk: bool,
    pub n_threads: Option<usize>,
    pub low_memory: bool,
    // Row-wise options
    pub n_rows: Option<usize>,
    pub row_index: Option<RowIndex>,
    // Column-wise options
    pub columns: Option<Arc<[PlSmallStr]>>,
    pub projection: Option<Arc<Vec<usize>>>,
    pub schema: Option<SchemaRef>,
    pub schema_overwrite: Option<SchemaRef>,
    pub dtype_overwrite: Option<Arc<Vec<DataType>>>,
    // CSV-specific options
    pub parse_options: Arc<CsvParseOptions>,
    pub has_header: bool,
    pub chunk_size: usize,
    pub skip_rows: usize,
    pub skip_rows_after_header: usize,
    pub infer_schema_length: Option<usize>,
    pub raise_if_empty: bool,
    pub ignore_errors: bool,
    pub fields_to_cast: Vec<Field>,
}

#[derive(Clone, Debug, PartialEq, Eq, Hash)]
#[cfg_attr(feature = "serde", derive(Serialize, Deserialize))]
pub struct CsvParseOptions {
    pub separator: u8,
    pub quote_char: Option<u8>,
    pub eol_char: u8,
    pub encoding: CsvEncoding,
    pub null_values: Option<NullValues>,
    pub missing_is_null: bool,
    pub truncate_ragged_lines: bool,
    pub comment_prefix: Option<CommentPrefix>,
    pub try_parse_dates: bool,
    pub decimal_comma: bool,
}

impl Default for CsvReadOptions {
    fn default() -> Self {
        Self {
            path: None,

            rechunk: false,
            n_threads: None,
            low_memory: false,

            n_rows: None,
            row_index: None,

            columns: None,
            projection: None,
            schema: None,
            schema_overwrite: None,
            dtype_overwrite: None,

            parse_options: Default::default(),
            has_header: true,
            chunk_size: 1 << 18,
            skip_rows: 0,
            skip_rows_after_header: 0,
            infer_schema_length: Some(100),
            raise_if_empty: true,
            ignore_errors: false,
            fields_to_cast: vec![],
        }
    }
}

/// Options related to parsing the CSV format.
impl Default for CsvParseOptions {
    fn default() -> Self {
        Self {
            separator: b',',
            quote_char: Some(b'"'),
            eol_char: b'\n',
            encoding: Default::default(),
            null_values: None,
            missing_is_null: true,
            truncate_ragged_lines: false,
            comment_prefix: None,
            try_parse_dates: false,
            decimal_comma: false,
        }
    }
}

impl CsvReadOptions {
    pub fn get_parse_options(&self) -> Arc<CsvParseOptions> {
        self.parse_options.clone()
    }

    pub fn with_path<P: Into<PathBuf>>(mut self, path: Option<P>) -> Self {
        self.path = path.map(|p| p.into());
        self
    }

    /// Whether to makes the columns contiguous in memory.
    pub fn with_rechunk(mut self, rechunk: bool) -> Self {
        self.rechunk = rechunk;
        self
    }

    /// Number of threads to use for reading. Defaults to the size of the polars
    /// thread pool.
    pub fn with_n_threads(mut self, n_threads: Option<usize>) -> Self {
        self.n_threads = n_threads;
        self
    }

    /// Reduce memory consumption at the expense of performance
    pub fn with_low_memory(mut self, low_memory: bool) -> Self {
        self.low_memory = low_memory;
        self
    }

    /// Limits the number of rows to read.
    pub fn with_n_rows(mut self, n_rows: Option<usize>) -> Self {
        self.n_rows = n_rows;
        self
    }

    /// Adds a row index column.
    pub fn with_row_index(mut self, row_index: Option<RowIndex>) -> Self {
        self.row_index = row_index;
        self
    }

    /// Which columns to select.
    pub fn with_columns(mut self, columns: Option<Arc<[PlSmallStr]>>) -> Self {
        self.columns = columns;
        self
    }

    /// Which columns to select denoted by their index. The index starts from 0
    /// (i.e. [0, 4] would select the 1st and 5th column).
    pub fn with_projection(mut self, projection: Option<Arc<Vec<usize>>>) -> Self {
        self.projection = projection;
        self
    }

    /// Set the schema to use for CSV file. The length of the schema must match
    /// the number of columns in the file. If this is [None], the schema is
    /// inferred from the file.
    pub fn with_schema(mut self, schema: Option<SchemaRef>) -> Self {
        self.schema = schema;
        self
    }

    /// Overwrites the data types in the schema by column name.
    pub fn with_schema_overwrite(mut self, schema_overwrite: Option<SchemaRef>) -> Self {
        self.schema_overwrite = schema_overwrite;
        self
    }

    /// Overwrite the dtypes in the schema in the order of the slice that's given.
    /// This is useful if you don't know the column names beforehand
    pub fn with_dtype_overwrite(mut self, dtype_overwrite: Option<Arc<Vec<DataType>>>) -> Self {
        self.dtype_overwrite = dtype_overwrite;
        self
    }

    /// Sets the CSV parsing options. See [map_parse_options][Self::map_parse_options]
    /// for an easier way to mutate them in-place.
    pub fn with_parse_options(mut self, parse_options: CsvParseOptions) -> Self {
        self.parse_options = Arc::new(parse_options);
        self
    }

    /// Sets whether the CSV file has a header row.
    pub fn with_has_header(mut self, has_header: bool) -> Self {
        self.has_header = has_header;
        self
    }

    /// Sets the chunk size used by the parser. This influences performance.
    pub fn with_chunk_size(mut self, chunk_size: usize) -> Self {
        self.chunk_size = chunk_size;
        self
    }

    /// Number of rows to skip before the header row.
    pub fn with_skip_rows(mut self, skip_rows: usize) -> Self {
        self.skip_rows = skip_rows;
        self
    }

    /// Number of rows to skip after the header row.
    pub fn with_skip_rows_after_header(mut self, skip_rows_after_header: usize) -> Self {
        self.skip_rows_after_header = skip_rows_after_header;
        self
    }

    /// Number of rows to use for schema inference. Pass [None] to use all rows.
    pub fn with_infer_schema_length(mut self, infer_schema_length: Option<usize>) -> Self {
        self.infer_schema_length = infer_schema_length;
        self
    }

    /// Whether to raise an error if the frame is empty. By default an empty
    /// DataFrame is returned.
    pub fn with_raise_if_empty(mut self, raise_if_empty: bool) -> Self {
        self.raise_if_empty = raise_if_empty;
        self
    }

    /// Continue with next batch when a ParserError is encountered.
    pub fn with_ignore_errors(mut self, ignore_errors: bool) -> Self {
        self.ignore_errors = ignore_errors;
        self
    }

    /// Apply a function to the parse options.
    pub fn map_parse_options<F: Fn(CsvParseOptions) -> CsvParseOptions>(
        mut self,
        map_func: F,
    ) -> Self {
        let parse_options = Arc::unwrap_or_clone(self.parse_options);
        self.parse_options = Arc::new(map_func(parse_options));
        self
    }
}

impl CsvParseOptions {
    /// The character used to separate fields in the CSV file. This
    /// is most often a comma ','.
    pub fn with_separator(mut self, separator: u8) -> Self {
        self.separator = separator;
        self
    }

    /// Set the character used for field quoting. This is most often double
    /// quotes '"'. Set this to [None] to disable quote parsing.
    pub fn with_quote_char(mut self, quote_char: Option<u8>) -> Self {
        self.quote_char = quote_char;
        self
    }

    /// Set the character used to indicate an end-of-line (eol).
    pub fn with_eol_char(mut self, eol_char: u8) -> Self {
        self.eol_char = eol_char;
        self
    }

    /// Set the encoding used by the file.
    pub fn with_encoding(mut self, encoding: CsvEncoding) -> Self {
        self.encoding = encoding;
        self
    }

    /// Set values that will be interpreted as missing/null.
    ///
    /// Note: These values are matched before quote-parsing, so if the null values
    /// are quoted then those quotes also need to be included here.
    pub fn with_null_values(mut self, null_values: Option<NullValues>) -> Self {
        self.null_values = null_values;
        self
    }

    /// Treat missing fields as null.
    pub fn with_missing_is_null(mut self, missing_is_null: bool) -> Self {
        self.missing_is_null = missing_is_null;
        self
    }

    /// Truncate lines that are longer than the schema.
    pub fn with_truncate_ragged_lines(mut self, truncate_ragged_lines: bool) -> Self {
        self.truncate_ragged_lines = truncate_ragged_lines;
        self
    }

    /// Sets the comment prefix for this instance. Lines starting with this
    /// prefix will be ignored.
    pub fn with_comment_prefix<T: Into<CommentPrefix>>(
        mut self,
        comment_prefix: Option<T>,
    ) -> Self {
        self.comment_prefix = comment_prefix.map(Into::into);
        self
    }

    /// Automatically try to parse dates/datetimes and time. If parsing fails,
    /// columns remain of dtype [`DataType::String`].
    pub fn with_try_parse_dates(mut self, try_parse_dates: bool) -> Self {
        self.try_parse_dates = try_parse_dates;
        self
    }

    /// Parse floats with a comma as decimal separator.
    pub fn with_decimal_comma(mut self, decimal_comma: bool) -> Self {
        self.decimal_comma = decimal_comma;
        self
    }
}

#[derive(Copy, Clone, Debug, Default, Eq, PartialEq, Hash)]
#[cfg_attr(feature = "serde", derive(Serialize, Deserialize))]
pub enum CsvEncoding {
    /// Utf8 encoding.
    #[default]
    Utf8,
    /// Utf8 encoding and unknown bytes are replaced with �.
    LossyUtf8,
}

#[derive(Clone, Debug, Eq, PartialEq, Hash)]
#[cfg_attr(feature = "serde", derive(Serialize, Deserialize))]
pub enum CommentPrefix {
    /// A single byte character that indicates the start of a comment line.
    Single(u8),
    /// A string that indicates the start of a comment line.
    /// This allows for multiple characters to be used as a comment identifier.
    Multi(PlSmallStr),
}

impl CommentPrefix {
    /// Creates a new `CommentPrefix` for the `Single` variant.
    pub fn new_single(prefix: u8) -> Self {
        CommentPrefix::Single(prefix)
    }

    /// Creates a new `CommentPrefix` for the `Multi` variant.
    pub fn new_multi(prefix: PlSmallStr) -> Self {
        CommentPrefix::Multi(prefix)
    }

    /// Creates a new `CommentPrefix` from a `&str`.
    pub fn new_from_str(prefix: &str) -> Self {
        if prefix.len() == 1 && prefix.chars().next().unwrap().is_ascii() {
            let c = prefix.as_bytes()[0];
            CommentPrefix::Single(c)
        } else {
            CommentPrefix::Multi(PlSmallStr::from_str(prefix))
        }
    }
}

impl From<&str> for CommentPrefix {
    fn from(value: &str) -> Self {
        Self::new_from_str(value)
    }
}

#[derive(Clone, Debug, Eq, PartialEq, Hash)]
#[cfg_attr(feature = "serde", derive(Serialize, Deserialize))]
pub enum NullValues {
    /// A single value that's used for all columns
    AllColumnsSingle(PlSmallStr),
    /// Multiple values that are used for all columns
    AllColumns(Vec<PlSmallStr>),
    /// Tuples that map column names to null value of that column
    Named(Vec<(PlSmallStr, PlSmallStr)>),
}

impl NullValues {
    pub fn compile(self, schema: &Schema) -> PolarsResult<NullValuesCompiled> {
        Ok(match self {
            NullValues::AllColumnsSingle(v) => NullValuesCompiled::AllColumnsSingle(v),
            NullValues::AllColumns(v) => NullValuesCompiled::AllColumns(v),
            NullValues::Named(v) => {
                let mut null_values = vec![PlSmallStr::from_static(""); schema.len()];
                for (name, null_value) in v {
                    let i = schema.try_index_of(&name)?;
                    null_values[i] = null_value;
                }
                NullValuesCompiled::Columns(null_values)
            },
        })
    }
}

#[derive(Debug, Clone)]
pub enum NullValuesCompiled {
    /// A single value that's used for all columns
    AllColumnsSingle(PlSmallStr),
    // Multiple null values that are null for all columns
    AllColumns(Vec<PlSmallStr>),
    /// A different null value per column, computed from `NullValues::Named`
    Columns(Vec<PlSmallStr>),
}

impl NullValuesCompiled {
    /// # Safety
    ///
    /// The caller must ensure that `index` is in bounds
    pub(super) unsafe fn is_null(&self, field: &[u8], index: usize) -> bool {
        use NullValuesCompiled::*;
        match self {
            AllColumnsSingle(v) => v.as_bytes() == field,
            AllColumns(v) => v.iter().any(|v| v.as_bytes() == field),
            Columns(v) => {
                debug_assert!(index < v.len());
                v.get_unchecked(index).as_bytes() == field
            },
        }
    }
}