polars_ops/series/ops/
strings.rs1use std::borrow::Cow;
2
3use arrow::array::builder::StaticArrayBuilder;
4use arrow::array::{Array, Utf8ViewArrayBuilder};
5use arrow::datatypes::ArrowDataType;
6use polars_core::prelude::{Column, DataType, IntoColumn, StringChunked};
7use polars_core::scalar::Scalar;
8use polars_error::{PolarsResult, polars_ensure};
9use polars_utils::pl_str::PlSmallStr;
10
11#[inline(always)]
12fn opt_str_to_string(s: Option<&str>) -> &str {
13 s.unwrap_or("null")
14}
15
16pub fn str_format(cs: &mut [Column], format: &str, insertions: &[usize]) -> PolarsResult<Column> {
17 assert_eq!(cs.len(), insertions.len());
18 assert!(!cs.is_empty()); let output_name = cs[0].name().clone();
21 let mut output_length = 1;
22 for c in cs.iter() {
23 if c.len() != 1 {
24 polars_ensure!(
25 output_length == 1 || output_length == c.len(),
26 length_mismatch = "format",
27 output_length,
28 c.len()
29 );
30 output_length = c.len();
31 }
32 }
33
34 let mut num_scalar_inputs = 0;
35 for c in cs.iter_mut() {
36 *c = c.cast(&DataType::String)?;
37 num_scalar_inputs += usize::from(c.len() == 1);
38 }
39
40 let mut format = Cow::Borrowed(format);
41 let mut insertions = Cow::Borrowed(insertions);
42
43 if num_scalar_inputs > 0 {
45 let mut filled_format = String::new();
46 filled_format.push_str(&format[..*insertions.first().unwrap()]);
47 insertions = Cow::Owned(
48 cs.iter()
49 .enumerate()
50 .filter_map(|(i, c)| {
51 let v = if c.len() == 1 {
52 filled_format.push_str(opt_str_to_string(c.str().unwrap().get(0)));
53 None
54 } else {
55 Some(filled_format.len())
56 };
57
58 let s = if i == cs.len() - 1 {
59 &format[insertions[i]..]
60 } else {
61 &format[insertions[i]..insertions[i + 1]]
62 };
63 filled_format.push_str(s);
64
65 v
66 })
67 .collect(),
68 );
69 format = filled_format.into();
70 }
71
72 let format = format.as_ref();
73 let insertions = insertions.as_ref();
74
75 if num_scalar_inputs == cs.len() {
77 let sc = Scalar::from(PlSmallStr::from_str(format));
78 return Ok(Column::new_scalar(output_name, sc, output_length));
79 }
80
81 let mut builder = Utf8ViewArrayBuilder::new(ArrowDataType::Utf8View);
82 builder.reserve(output_length);
83
84 let mut arrays = cs
85 .iter()
86 .filter(|c| c.len() != 1)
87 .map(|c| {
88 let ca = c.str().unwrap();
89 let mut iter = ca.downcast_iter();
90 let arr = iter.next().unwrap();
91 (iter, arr, 0)
92 })
93 .collect::<Vec<_>>();
94
95 let mut s = String::new();
100 for i in 0..output_length {
101 s.clear();
102 s.push_str(&format[..insertions[0]]);
103
104 for (j, (iter, arr, elem_idx)) in arrays.iter_mut().enumerate() {
105 s.push_str(opt_str_to_string(arr.get(*elem_idx)));
106 let start = insertions[j];
107 let end = insertions.get(j + 1).copied().unwrap_or(format.len());
108 s.push_str(&format[start..end]);
109
110 *elem_idx += 1;
111 if i + 1 != output_length && *elem_idx == arr.len() {
112 *arr = iter.next().unwrap();
113 }
114 }
115
116 builder.push_value_ignore_validity(&s);
117 }
118
119 let array = builder.freeze().to_boxed();
120 Ok(unsafe { StringChunked::from_chunks(output_name, vec![array]) }.into_column())
121}