polars_ops/series/ops/
strings.rs

1use 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()); // Checked at IR construction
19
20    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 validity = None;
35    let mut num_scalar_inputs = 0;
36    for c in cs.iter_mut() {
37        if let Some(c_validity) = c.rechunk_validity() {
38            // Column with only nulls means output is only nulls.
39            if c.null_count() == c.len() {
40                return Ok(Column::full_null(
41                    output_name,
42                    output_length,
43                    &DataType::String,
44                ));
45            }
46
47            match &mut validity {
48                v @ None => *v = Some(c_validity),
49                Some(v) => *v = arrow::bitmap::and(v, &c_validity),
50            }
51        }
52
53        *c = c.cast(&DataType::String)?;
54        num_scalar_inputs += usize::from(c.len() == 1);
55    }
56
57    let mut format = Cow::Borrowed(format);
58    let mut insertions = Cow::Borrowed(insertions);
59
60    // Fill in any constants into the format string.
61    if num_scalar_inputs > 0 {
62        let mut filled_format = String::new();
63        filled_format.push_str(&format[..*insertions.first().unwrap()]);
64        insertions = Cow::Owned(
65            cs.iter()
66                .enumerate()
67                .filter_map(|(i, c)| {
68                    let v = if c.len() == 1 {
69                        filled_format.push_str(opt_str_to_string(c.str().unwrap().get(0)));
70                        None
71                    } else {
72                        Some(filled_format.len())
73                    };
74
75                    let s = if i == cs.len() - 1 {
76                        &format[insertions[i]..]
77                    } else {
78                        &format[insertions[i]..insertions[i + 1]]
79                    };
80                    filled_format.push_str(s);
81
82                    v
83                })
84                .collect(),
85        );
86        format = filled_format.into();
87    }
88
89    let format = format.as_ref();
90    let insertions = insertions.as_ref();
91
92    // If the format string is constant.
93    if num_scalar_inputs == cs.len() {
94        let sc = Scalar::from(PlSmallStr::from_str(format));
95        return Ok(Column::new_scalar(output_name, sc, output_length));
96    }
97
98    let mut builder = Utf8ViewArrayBuilder::new(ArrowDataType::Utf8View);
99    builder.reserve(output_length);
100
101    let mut arrays = cs
102        .iter()
103        .filter(|c| c.len() != 1)
104        .map(|c| {
105            let ca = c.str().unwrap();
106            let mut iter = ca.downcast_iter();
107            let arr = iter.next().unwrap();
108            (iter, arr, 0)
109        })
110        .collect::<Vec<_>>();
111
112    // @Performance. There is some smarter stuff that can be done with views and stuff. Don't think
113    // it is worth the complexity.
114
115    // Amortize the format string allocation.
116    let mut s = String::new();
117    for i in 0..output_length {
118        if validity
119            .as_ref()
120            .is_some_and(|v| !unsafe { v.get_bit_unchecked(i) })
121        {
122            unsafe { builder.push_inline_view_ignore_validity(Default::default()) };
123
124            for (iter, arr, elem_idx) in arrays.iter_mut() {
125                *elem_idx += 1;
126                if i + 1 != output_length && *elem_idx == arr.len() {
127                    *arr = iter.next().unwrap();
128                    *elem_idx = 0;
129                }
130            }
131
132            continue;
133        }
134
135        s.clear();
136        s.push_str(&format[..insertions[0]]);
137
138        for (j, (iter, arr, elem_idx)) in arrays.iter_mut().enumerate() {
139            s.push_str(opt_str_to_string(arr.get(*elem_idx)));
140            let start = insertions[j];
141            let end = insertions.get(j + 1).copied().unwrap_or(format.len());
142            s.push_str(&format[start..end]);
143
144            *elem_idx += 1;
145            if i + 1 != output_length && *elem_idx == arr.len() {
146                *arr = iter.next().unwrap();
147                *elem_idx = 0;
148            }
149        }
150
151        builder.push_value_ignore_validity(&s);
152    }
153
154    let array = builder.freeze().with_validity(validity).to_boxed();
155    Ok(unsafe { StringChunked::from_chunks(output_name, vec![array]) }.into_column())
156}