polars_lazy/scan/
csv.rs

1use polars_core::prelude::*;
2use polars_io::cloud::CloudOptions;
3use polars_io::csv::read::{
4    CommentPrefix, CsvEncoding, CsvParseOptions, CsvReadOptions, NullValues, infer_file_schema,
5};
6use polars_io::path_utils::expand_paths;
7use polars_io::utils::compression::maybe_decompress_bytes;
8use polars_io::utils::get_reader_bytes;
9use polars_io::{HiveOptions, RowIndex};
10use polars_utils::mmap::MemSlice;
11use polars_utils::plpath::PlPath;
12use polars_utils::slice_enum::Slice;
13
14use crate::prelude::*;
15
16#[derive(Clone)]
17#[cfg(feature = "csv")]
18pub struct LazyCsvReader {
19    sources: ScanSources,
20    glob: bool,
21    cache: bool,
22    read_options: CsvReadOptions,
23    cloud_options: Option<CloudOptions>,
24    include_file_paths: Option<PlSmallStr>,
25}
26
27#[cfg(feature = "csv")]
28impl LazyCsvReader {
29    /// Re-export to shorten code.
30    pub fn map_parse_options<F: Fn(CsvParseOptions) -> CsvParseOptions>(
31        mut self,
32        map_func: F,
33    ) -> Self {
34        self.read_options = self.read_options.map_parse_options(map_func);
35        self
36    }
37
38    pub fn new_paths(paths: Arc<[PlPath]>) -> Self {
39        Self::new_with_sources(ScanSources::Paths(paths))
40    }
41
42    pub fn new_with_sources(sources: ScanSources) -> Self {
43        LazyCsvReader {
44            sources,
45            glob: true,
46            cache: true,
47            read_options: Default::default(),
48            cloud_options: Default::default(),
49            include_file_paths: None,
50        }
51    }
52
53    pub fn new(path: PlPath) -> Self {
54        Self::new_with_sources(ScanSources::Paths([path].into()))
55    }
56
57    /// Skip this number of rows after the header location.
58    #[must_use]
59    pub fn with_skip_rows_after_header(mut self, offset: usize) -> Self {
60        self.read_options.skip_rows_after_header = offset;
61        self
62    }
63
64    /// Add a row index column.
65    #[must_use]
66    pub fn with_row_index(mut self, row_index: Option<RowIndex>) -> Self {
67        self.read_options.row_index = row_index;
68        self
69    }
70
71    /// Try to stop parsing when `n` rows are parsed. During multithreaded parsing the upper bound `n` cannot
72    /// be guaranteed.
73    #[must_use]
74    pub fn with_n_rows(mut self, num_rows: Option<usize>) -> Self {
75        self.read_options.n_rows = num_rows;
76        self
77    }
78
79    /// Set the number of rows to use when inferring the csv schema.
80    /// The default is 100 rows.
81    /// Setting to [None] will do a full table scan, which is very slow.
82    #[must_use]
83    pub fn with_infer_schema_length(mut self, num_rows: Option<usize>) -> Self {
84        self.read_options.infer_schema_length = num_rows;
85        self
86    }
87
88    /// Continue with next batch when a ParserError is encountered.
89    #[must_use]
90    pub fn with_ignore_errors(mut self, ignore: bool) -> Self {
91        self.read_options.ignore_errors = ignore;
92        self
93    }
94
95    /// Set the CSV file's schema
96    #[must_use]
97    pub fn with_schema(mut self, schema: Option<SchemaRef>) -> Self {
98        self.read_options.schema = schema;
99        self
100    }
101
102    /// Skip the first `n` rows during parsing. The header will be parsed at row `n`.
103    /// Note that by row we mean valid CSV, encoding and comments are respected.
104    #[must_use]
105    pub fn with_skip_rows(mut self, skip_rows: usize) -> Self {
106        self.read_options.skip_rows = skip_rows;
107        self
108    }
109
110    /// Skip the first `n` lines during parsing. The header will be parsed at line `n`.
111    /// We don't respect CSV escaping when skipping lines.
112    #[must_use]
113    pub fn with_skip_lines(mut self, skip_lines: usize) -> Self {
114        self.read_options.skip_lines = skip_lines;
115        self
116    }
117
118    /// Overwrite the schema with the dtypes in this given Schema. The given schema may be a subset
119    /// of the total schema.
120    #[must_use]
121    pub fn with_dtype_overwrite(mut self, schema: Option<SchemaRef>) -> Self {
122        self.read_options.schema_overwrite = schema;
123        self
124    }
125
126    /// Set whether the CSV file has headers
127    #[must_use]
128    pub fn with_has_header(mut self, has_header: bool) -> Self {
129        self.read_options.has_header = has_header;
130        self
131    }
132
133    /// Sets the chunk size used by the parser. This influences performance.
134    /// This can be used as a way to reduce memory usage during the parsing at the cost of performance.
135    pub fn with_chunk_size(mut self, chunk_size: usize) -> Self {
136        self.read_options.chunk_size = chunk_size;
137        self
138    }
139
140    /// Set the CSV file's column separator as a byte character
141    #[must_use]
142    pub fn with_separator(self, separator: u8) -> Self {
143        self.map_parse_options(|opts| opts.with_separator(separator))
144    }
145
146    /// Set the comment prefix for this instance. Lines starting with this prefix will be ignored.
147    #[must_use]
148    pub fn with_comment_prefix(self, comment_prefix: Option<PlSmallStr>) -> Self {
149        self.map_parse_options(|opts| {
150            opts.with_comment_prefix(comment_prefix.clone().map(|s| {
151                if s.len() == 1 && s.chars().next().unwrap().is_ascii() {
152                    CommentPrefix::Single(s.as_bytes()[0])
153                } else {
154                    CommentPrefix::Multi(s)
155                }
156            }))
157        })
158    }
159
160    /// Set the `char` used as quote char. The default is `b'"'`. If set to [`None`] quoting is disabled.
161    #[must_use]
162    pub fn with_quote_char(self, quote_char: Option<u8>) -> Self {
163        self.map_parse_options(|opts| opts.with_quote_char(quote_char))
164    }
165
166    /// Set the `char` used as end of line. The default is `b'\n'`.
167    #[must_use]
168    pub fn with_eol_char(self, eol_char: u8) -> Self {
169        self.map_parse_options(|opts| opts.with_eol_char(eol_char))
170    }
171
172    /// Set values that will be interpreted as missing/ null.
173    #[must_use]
174    pub fn with_null_values(self, null_values: Option<NullValues>) -> Self {
175        self.map_parse_options(|opts| opts.with_null_values(null_values.clone()))
176    }
177
178    /// Treat missing fields as null.
179    pub fn with_missing_is_null(self, missing_is_null: bool) -> Self {
180        self.map_parse_options(|opts| opts.with_missing_is_null(missing_is_null))
181    }
182
183    /// Cache the DataFrame after reading.
184    #[must_use]
185    pub fn with_cache(mut self, cache: bool) -> Self {
186        self.cache = cache;
187        self
188    }
189
190    /// Reduce memory usage at the expense of performance
191    #[must_use]
192    pub fn with_low_memory(mut self, low_memory: bool) -> Self {
193        self.read_options.low_memory = low_memory;
194        self
195    }
196
197    /// Set  [`CsvEncoding`]
198    #[must_use]
199    pub fn with_encoding(self, encoding: CsvEncoding) -> Self {
200        self.map_parse_options(|opts| opts.with_encoding(encoding))
201    }
202
203    /// Automatically try to parse dates/datetimes and time.
204    /// If parsing fails, columns remain of dtype [`DataType::String`].
205    #[cfg(feature = "temporal")]
206    pub fn with_try_parse_dates(self, try_parse_dates: bool) -> Self {
207        self.map_parse_options(|opts| opts.with_try_parse_dates(try_parse_dates))
208    }
209
210    /// Raise an error if CSV is empty (otherwise return an empty frame)
211    #[must_use]
212    pub fn with_raise_if_empty(mut self, raise_if_empty: bool) -> Self {
213        self.read_options.raise_if_empty = raise_if_empty;
214        self
215    }
216
217    /// Truncate lines that are longer than the schema.
218    #[must_use]
219    pub fn with_truncate_ragged_lines(self, truncate_ragged_lines: bool) -> Self {
220        self.map_parse_options(|opts| opts.with_truncate_ragged_lines(truncate_ragged_lines))
221    }
222
223    #[must_use]
224    pub fn with_decimal_comma(self, decimal_comma: bool) -> Self {
225        self.map_parse_options(|opts| opts.with_decimal_comma(decimal_comma))
226    }
227
228    #[must_use]
229    /// Expand path given via globbing rules.
230    pub fn with_glob(mut self, toggle: bool) -> Self {
231        self.glob = toggle;
232        self
233    }
234
235    pub fn with_cloud_options(mut self, cloud_options: Option<CloudOptions>) -> Self {
236        self.cloud_options = cloud_options;
237        self
238    }
239
240    /// Modify a schema before we run the lazy scanning.
241    ///
242    /// Important! Run this function latest in the builder!
243    pub fn with_schema_modify<F>(mut self, f: F) -> PolarsResult<Self>
244    where
245        F: Fn(Schema) -> PolarsResult<Schema>,
246    {
247        let n_threads = self.read_options.n_threads;
248
249        let infer_schema = |bytes: MemSlice| {
250            let skip_rows = self.read_options.skip_rows;
251            let skip_lines = self.read_options.skip_lines;
252            let parse_options = self.read_options.get_parse_options();
253
254            let mut owned = vec![];
255            let bytes = maybe_decompress_bytes(bytes.as_ref(), &mut owned)?;
256
257            PolarsResult::Ok(
258                infer_file_schema(
259                    &get_reader_bytes(&mut std::io::Cursor::new(bytes))?,
260                    &parse_options,
261                    self.read_options.infer_schema_length,
262                    self.read_options.has_header,
263                    // we set it to None and modify them after the schema is updated
264                    None,
265                    skip_rows,
266                    skip_lines,
267                    self.read_options.skip_rows_after_header,
268                    self.read_options.raise_if_empty,
269                )?
270                .0,
271            )
272        };
273
274        let schema = match self.sources.clone() {
275            ScanSources::Paths(paths) => {
276                // TODO: Path expansion should happen when converting to the IR
277                // https://github.com/pola-rs/polars/issues/17634
278                let paths = expand_paths(&paths[..], self.glob(), self.cloud_options())?;
279
280                let Some(path) = paths.first() else {
281                    polars_bail!(ComputeError: "no paths specified for this reader");
282                };
283
284                infer_schema(MemSlice::from_file(&polars_utils::open_file(
285                    path.as_ref().as_local_path().unwrap(),
286                )?)?)?
287            },
288            ScanSources::Files(files) => {
289                let Some(file) = files.first() else {
290                    polars_bail!(ComputeError: "no buffers specified for this reader");
291                };
292
293                infer_schema(MemSlice::from_file(file)?)?
294            },
295            ScanSources::Buffers(buffers) => {
296                let Some(buffer) = buffers.first() else {
297                    polars_bail!(ComputeError: "no buffers specified for this reader");
298                };
299
300                infer_schema(buffer.clone())?
301            },
302        };
303
304        self.read_options.n_threads = n_threads;
305        let mut schema = f(schema)?;
306
307        // the dtypes set may be for the new names, so update again
308        if let Some(overwrite_schema) = &self.read_options.schema_overwrite {
309            for (name, dtype) in overwrite_schema.iter() {
310                schema.with_column(name.clone(), dtype.clone());
311            }
312        }
313
314        Ok(self.with_schema(Some(Arc::new(schema))))
315    }
316
317    pub fn with_include_file_paths(mut self, include_file_paths: Option<PlSmallStr>) -> Self {
318        self.include_file_paths = include_file_paths;
319        self
320    }
321}
322
323impl LazyFileListReader for LazyCsvReader {
324    /// Get the final [LazyFrame].
325    fn finish(self) -> PolarsResult<LazyFrame> {
326        let rechunk = self.rechunk();
327        let row_index = self.row_index().cloned();
328        let pre_slice = self.n_rows().map(|len| Slice::Positive { offset: 0, len });
329
330        let lf: LazyFrame = DslBuilder::scan_csv(
331            self.sources,
332            self.read_options,
333            UnifiedScanArgs {
334                schema: None,
335                cloud_options: self.cloud_options,
336                hive_options: HiveOptions::new_disabled(),
337                rechunk,
338                cache: self.cache,
339                glob: self.glob,
340                projection: None,
341                row_index,
342                pre_slice,
343                cast_columns_policy: CastColumnsPolicy::ERROR_ON_MISMATCH,
344                missing_columns_policy: MissingColumnsPolicy::Raise,
345                extra_columns_policy: ExtraColumnsPolicy::Raise,
346                include_file_paths: self.include_file_paths,
347                column_mapping: None,
348                deletion_files: None,
349            },
350        )?
351        .build()
352        .into();
353        Ok(lf)
354    }
355
356    fn finish_no_glob(self) -> PolarsResult<LazyFrame> {
357        unreachable!();
358    }
359
360    fn glob(&self) -> bool {
361        self.glob
362    }
363
364    fn sources(&self) -> &ScanSources {
365        &self.sources
366    }
367
368    fn with_sources(mut self, sources: ScanSources) -> Self {
369        self.sources = sources;
370        self
371    }
372
373    fn with_n_rows(mut self, n_rows: impl Into<Option<usize>>) -> Self {
374        self.read_options.n_rows = n_rows.into();
375        self
376    }
377
378    fn with_row_index(mut self, row_index: impl Into<Option<RowIndex>>) -> Self {
379        self.read_options.row_index = row_index.into();
380        self
381    }
382
383    fn rechunk(&self) -> bool {
384        self.read_options.rechunk
385    }
386
387    /// Rechunk the memory to contiguous chunks when parsing is done.
388    fn with_rechunk(mut self, rechunk: bool) -> Self {
389        self.read_options.rechunk = rechunk;
390        self
391    }
392
393    /// Try to stop parsing when `n` rows are parsed. During multithreaded parsing the upper bound `n` cannot
394    /// be guaranteed.
395    fn n_rows(&self) -> Option<usize> {
396        self.read_options.n_rows
397    }
398
399    /// Return the row index settings.
400    fn row_index(&self) -> Option<&RowIndex> {
401        self.read_options.row_index.as_ref()
402    }
403
404    fn concat_impl(&self, lfs: Vec<LazyFrame>) -> PolarsResult<LazyFrame> {
405        // set to false, as the csv parser has full thread utilization
406        let args = UnionArgs {
407            rechunk: self.rechunk(),
408            parallel: false,
409            to_supertypes: false,
410            from_partitioned_ds: true,
411            ..Default::default()
412        };
413        concat_impl(&lfs, args)
414    }
415
416    /// [CloudOptions] used to list files.
417    fn cloud_options(&self) -> Option<&CloudOptions> {
418        self.cloud_options.as_ref()
419    }
420}
polars_lazy/scan/csv.rs

polars_lazy/scan/
csv.rs
