ferray_strings/
lib.rs

1// ferray-strings: Vectorized string operations on arrays of strings
2//
3// Implements `numpy.strings` (NumPy 2.0+): vectorized elementwise string
4// operations on arrays of strings with broadcasting. Covers case manipulation,
5// alignment/padding, stripping, find/replace, splitting/joining, and regex
6// support. Operates on `StringArray` — a separate array type backed by
7// `Vec<String>`.
8
9//! # ferray-strings
10//!
11//! Vectorized string operations on arrays of strings, analogous to
12//! `numpy.strings` in NumPy 2.0+.
13//!
14//! The primary type is [`StringArray`], a specialized N-dimensional array
15//! backed by `Vec<String>`. Since `String` does not implement
16//! [`ferray_core::Element`], this type is separate from `NdArray<T, D>`.
17//!
18//! # Quick Start
19//!
20//! ```ignore
21//! use ferray_strings::*;
22//!
23//! let a = array(&["hello", "world"]).unwrap();
24//! let b = upper(&a).unwrap();
25//! assert_eq!(b.as_slice(), &["HELLO", "WORLD"]);
26//! ```
27
28pub mod align;
29pub mod case;
30pub mod classify;
31pub mod concat;
32pub mod regex_ops;
33pub mod search;
34pub mod split_join;
35pub mod str_ops;
36pub mod string_array;
37pub mod strip;
38
39// Re-export types
40pub use string_array::{StringArray, StringArray1, StringArray2, array};
41
42// Re-export operations for flat namespace (like numpy.strings.upper etc.)
43pub use align::{center, ljust, ljust_with, rjust, rjust_with, zfill};
44pub use case::{capitalize, lower, title, upper};
45pub use classify::{isalnum, isalpha, isdigit, islower, isnumeric, isspace, istitle, isupper};
46pub use concat::{add, add_same, multiply};
47pub use regex_ops::{extract, extract_compiled, match_, match_compiled};
48// Re-export `regex::Regex` so callers of `match_compiled`/`extract_compiled`
49// don't have to add a direct `regex` dependency to construct one.
50pub use regex::Regex;
51pub use search::{count, endswith, find, replace, startswith};
52pub use split_join::{join, join_array, split};
53pub use str_ops::{equal, greater, greater_equal, less, less_equal, not_equal, str_len, swapcase};
54pub use strip::{lstrip, rstrip, strip};
55
56#[cfg(test)]
57mod integration_tests {
58    use super::*;
59
60    #[test]
61    fn ac1_upper() {
62        // AC-1: strings::upper(&["hello", "world"]) produces ["HELLO", "WORLD"]
63        let a = array(&["hello", "world"]).unwrap();
64        let b = upper(&a).unwrap();
65        assert_eq!(b.as_slice(), &["HELLO", "WORLD"]);
66    }
67
68    #[test]
69    fn ac2_add_broadcast_scalar() {
70        // AC-2: strings::add broadcasts a scalar string against an array correctly
71        let a = array(&["hello", "world"]).unwrap();
72        let b = array(&["!"]).unwrap();
73        let c = add(&a, &b).unwrap();
74        assert_eq!(c.as_slice(), &["hello!", "world!"]);
75    }
76
77    #[test]
78    fn ac3_find_indices() {
79        // AC-3: strings::find(&a, "ll") returns correct indices
80        let a = array(&["hello", "world"]).unwrap();
81        let b = find(&a, "ll").unwrap();
82        let data = b.as_slice().unwrap();
83        assert_eq!(data, &[2_i64, -1_i64]);
84    }
85
86    #[test]
87    fn ac4_split() {
88        // AC-4: strings::split(&["a-b", "c-d"], "-") returns [vec!["a","b"], vec!["c","d"]]
89        let a = array(&["a-b", "c-d"]).unwrap();
90        let result = split(&a, "-").unwrap();
91        assert_eq!(
92            result,
93            vec![
94                vec!["a".to_string(), "b".to_string()],
95                vec!["c".to_string(), "d".to_string()],
96            ]
97        );
98    }
99
100    #[test]
101    fn ac5_regex() {
102        // AC-5: Regex match_ and extract work correctly with capture groups
103        let a = array(&["abc123", "def", "ghi456"]).unwrap();
104
105        let matched = match_(&a, r"\d+").unwrap();
106        let matched_data = matched.as_slice().unwrap();
107        assert_eq!(matched_data, &[true, false, true]);
108
109        let extracted = extract(&a, r"(\d+)").unwrap();
110        assert_eq!(extracted.as_slice(), &["123", "", "456"]);
111    }
112
113    #[test]
114    fn full_pipeline() {
115        // End-to-end: strip, upper, add suffix, search
116        let raw = array(&["  Hello  ", " World "]).unwrap();
117        let stripped = strip(&raw, None).unwrap();
118        let uppered = upper(&stripped).unwrap();
119        let suffix = array(&["!"]).unwrap();
120        let result = add(&uppered, &suffix).unwrap();
121        assert_eq!(result.as_slice(), &["HELLO!", "WORLD!"]);
122
123        let has_excl = endswith(&result, "!").unwrap();
124        let data = has_excl.as_slice().unwrap();
125        assert_eq!(data, &[true, true]);
126    }
127
128    #[test]
129    fn case_round_trip() {
130        let a = array(&["Hello World"]).unwrap();
131        let low = lower(&a).unwrap();
132        let titled = title(&low).unwrap();
133        assert_eq!(titled.as_slice(), &["Hello World"]);
134    }
135
136    #[test]
137    fn alignment_operations() {
138        let a = array(&["hi"]).unwrap();
139        let c = center(&a, 6, '-').unwrap();
140        assert_eq!(c.as_slice(), &["--hi--"]);
141
142        let l = ljust(&a, 6).unwrap();
143        assert_eq!(l.as_slice(), &["hi    "]);
144
145        let r = rjust(&a, 6).unwrap();
146        assert_eq!(r.as_slice(), &["    hi"]);
147
148        let z = zfill(&array(&["42"]).unwrap(), 5).unwrap();
149        assert_eq!(z.as_slice(), &["00042"]);
150    }
151
152    #[test]
153    fn strip_operations() {
154        let a = array(&["  hello  "]).unwrap();
155        assert_eq!(strip(&a, None).unwrap().as_slice(), &["hello"]);
156        assert_eq!(lstrip(&a, None).unwrap().as_slice(), &["hello  "]);
157        assert_eq!(rstrip(&a, None).unwrap().as_slice(), &["  hello"]);
158    }
159
160    #[test]
161    fn search_operations() {
162        let a = array(&["hello world", "foo bar"]).unwrap();
163        let c = count(&a, "o").unwrap();
164        let data = c.as_slice().unwrap();
165        // "hello world" has 2 'o's, "foo bar" has 2 'o's
166        assert_eq!(data, &[2_u64, 2]);
167    }
168
169    #[test]
170    fn replace_operation() {
171        let a = array(&["hello world"]).unwrap();
172        let b = replace(&a, "world", "rust", None).unwrap();
173        assert_eq!(b.as_slice(), &["hello rust"]);
174    }
175
176    #[test]
177    fn multiply_operation() {
178        let a = array(&["ab"]).unwrap();
179        let b = multiply(&a, 3).unwrap();
180        assert_eq!(b.as_slice(), &["ababab"]);
181    }
182
183    #[test]
184    fn join_operation() {
185        let parts = vec![
186            vec!["a".to_string(), "b".to_string()],
187            vec!["c".to_string(), "d".to_string()],
188        ];
189        let result = join("-", &parts).unwrap();
190        assert_eq!(result.as_slice(), &["a-b", "c-d"]);
191    }
192
193    #[test]
194    fn capitalize_operation() {
195        let a = array(&["hello world", "RUST"]).unwrap();
196        let b = capitalize(&a).unwrap();
197        assert_eq!(b.as_slice(), &["Hello world", "Rust"]);
198    }
199
200    #[test]
201    fn string_array_2d() {
202        let a = StringArray2::from_rows(&[&["a", "b"], &["c", "d"]]).unwrap();
203        assert_eq!(a.shape(), &[2, 2]);
204        let b = upper(&a).unwrap();
205        assert_eq!(b.as_slice(), &["A", "B", "C", "D"]);
206        assert_eq!(b.shape(), &[2, 2]);
207    }
208
209    // -----------------------------------------------------------------
210    // #520 — shape preservation for ND StringArrays across every op.
211    //
212    // Every operation that takes `StringArray<D>` must thread `D`
213    // through the output. These tests pin that contract on a 2x2
214    // input so regressions (e.g. an accidental `Ix1::new([len])` in
215    // the output constructor) are caught immediately.
216    // -----------------------------------------------------------------
217
218    fn two_by_two(vals: &[&str; 4]) -> crate::StringArray2 {
219        crate::StringArray2::from_rows(&[&[vals[0], vals[1]], &[vals[2], vals[3]]]).unwrap()
220    }
221
222    #[test]
223    fn shape_preserved_case_ops_2d() {
224        let a = two_by_two(&["Hello", "World", "foo", "Bar"]);
225        assert_eq!(upper(&a).unwrap().shape(), &[2, 2]);
226        assert_eq!(lower(&a).unwrap().shape(), &[2, 2]);
227        assert_eq!(capitalize(&a).unwrap().shape(), &[2, 2]);
228        assert_eq!(title(&a).unwrap().shape(), &[2, 2]);
229    }
230
231    #[test]
232    fn shape_preserved_align_ops_2d() {
233        let a = two_by_two(&["a", "bb", "ccc", "dddd"]);
234        assert_eq!(center(&a, 6, ' ').unwrap().shape(), &[2, 2]);
235        assert_eq!(ljust(&a, 6).unwrap().shape(), &[2, 2]);
236        assert_eq!(rjust(&a, 6).unwrap().shape(), &[2, 2]);
237        assert_eq!(zfill(&a, 6).unwrap().shape(), &[2, 2]);
238    }
239
240    #[test]
241    fn shape_preserved_strip_ops_2d() {
242        let a = two_by_two(&["  a  ", "  b  ", "  c  ", "  d  "]);
243        assert_eq!(strip(&a, None).unwrap().shape(), &[2, 2]);
244        assert_eq!(lstrip(&a, None).unwrap().shape(), &[2, 2]);
245        assert_eq!(rstrip(&a, None).unwrap().shape(), &[2, 2]);
246    }
247
248    #[test]
249    fn shape_preserved_concat_ops_2d() {
250        let a = two_by_two(&["ab", "cd", "ef", "gh"]);
251        // `add` currently flattens to IxDyn regardless of input rank
252        // (it supports cross-rank broadcasting), so it's not expected
253        // to preserve D. Just verify the total element count.
254        let b = two_by_two(&["!", "!", "!", "!"]);
255        let ab = add(&a, &b).unwrap();
256        assert_eq!(ab.shape(), &[2, 2]);
257        // multiply preserves D
258        assert_eq!(multiply(&a, 2).unwrap().shape(), &[2, 2]);
259    }
260
261    #[test]
262    fn shape_preserved_search_ops_2d() {
263        let a = two_by_two(&["hello", "help", "world", "word"]);
264        // Each of these returns Array<T, D> — verify D is preserved.
265        assert_eq!(find(&a, "ell").unwrap().shape(), &[2, 2]);
266        assert_eq!(count(&a, "l").unwrap().shape(), &[2, 2]);
267        assert_eq!(startswith(&a, "he").unwrap().shape(), &[2, 2]);
268        assert_eq!(endswith(&a, "d").unwrap().shape(), &[2, 2]);
269        assert_eq!(replace(&a, "l", "L", None).unwrap().shape(), &[2, 2]);
270    }
271
272    #[test]
273    fn shape_preserved_regex_ops_2d() {
274        let a = two_by_two(&["abc123", "x", "y42", "zzz"]);
275        // match_ preserves D; extract flattens by design (ragged results).
276        assert_eq!(match_(&a, r"\d+").unwrap().shape(), &[2, 2]);
277    }
278
279    #[test]
280    fn shape_preserved_case_ops_3d() {
281        // Just to confirm we aren't special-casing Ix2 somewhere — bump to Ix3.
282        use ferray_core::dimension::Ix3;
283        let data: Vec<String> = (0..8).map(|i| format!("s{i}")).collect();
284        let a = crate::StringArray::<Ix3>::from_vec(Ix3::new([2, 2, 2]), data).unwrap();
285        assert_eq!(upper(&a).unwrap().shape(), &[2, 2, 2]);
286        assert_eq!(lower(&a).unwrap().shape(), &[2, 2, 2]);
287    }
288
289    // --- Unicode / multi-byte character tests ---
290
291    #[test]
292    fn unicode_upper_lower() {
293        let a = array(&["café", "naïve", "über"]).unwrap();
294        let u = upper(&a).unwrap();
295        assert_eq!(u.as_slice(), &["CAFÉ", "NAÏVE", "ÜBER"]);
296        let l = lower(&u).unwrap();
297        assert_eq!(l.as_slice(), &["café", "naïve", "über"]);
298    }
299
300    #[test]
301    fn unicode_capitalize() {
302        let a = array(&["ñoño", "straße"]).unwrap();
303        let c = capitalize(&a).unwrap();
304        assert_eq!(c.as_slice()[0], "Ñoño");
305        // Rust's capitalize of "straße" -> "Straße"
306        assert_eq!(c.as_slice()[1], "Straße");
307    }
308
309    #[test]
310    fn unicode_find() {
311        let a = array(&["日本語テスト", "こんにちは"]).unwrap();
312        let r = find(&a, "テスト").unwrap();
313        let data = r.as_slice().unwrap();
314        assert_eq!(data[0], 3); // "テスト" starts at byte position, but find uses char position...
315        // Actually, find returns the byte index via str::find. Check:
316        // "日本語テスト".find("テスト") returns byte offset 9
317        // But our find should return character index or byte index?
318        // Let's just verify it finds it (>= 0) vs not found (-1)
319        assert!(data[0] >= 0); // found
320        assert_eq!(data[1], -1); // not found
321    }
322
323    #[test]
324    fn unicode_strip() {
325        let a = array(&["  héllo  ", "  wörld  "]).unwrap();
326        let s = strip(&a, None).unwrap();
327        assert_eq!(s.as_slice(), &["héllo", "wörld"]);
328    }
329
330    #[test]
331    fn unicode_replace() {
332        let a = array(&["café latte"]).unwrap();
333        let r = replace(&a, "café", "tea", None).unwrap();
334        assert_eq!(r.as_slice(), &["tea latte"]);
335    }
336
337    #[test]
338    fn emoji_operations() {
339        let a = array(&["hello 🌍", "rust 🦀"]).unwrap();
340        let u = upper(&a).unwrap();
341        assert_eq!(u.as_slice(), &["HELLO 🌍", "RUST 🦀"]);
342        let c = count(&a, "🌍").unwrap();
343        assert_eq!(c.as_slice().unwrap(), &[1, 0]);
344    }
345
346    #[test]
347    fn cjk_characters() {
348        let a = array(&["你好世界", "こんにちは"]).unwrap();
349        let starts = startswith(&a, "你好").unwrap();
350        assert_eq!(starts.as_slice().unwrap(), &[true, false]);
351        let ends = endswith(&a, "世界").unwrap();
352        assert_eq!(ends.as_slice().unwrap(), &[true, false]);
353    }
354
355    // ----- Empty array tests (#282) -----
356
357    #[test]
358    fn empty_array_upper() {
359        let a = StringArray1::from_vec(ferray_core::dimension::Ix1::new([0]), vec![]).unwrap();
360        let u = upper(&a).unwrap();
361        assert_eq!(u.len(), 0);
362    }
363
364    #[test]
365    fn empty_array_str_len() {
366        let a = StringArray1::from_vec(ferray_core::dimension::Ix1::new([0]), vec![]).unwrap();
367        let l = str_len(&a).unwrap();
368        assert_eq!(l.size(), 0);
369    }
370
371    #[test]
372    fn empty_array_find() {
373        let a = StringArray1::from_vec(ferray_core::dimension::Ix1::new([0]), vec![]).unwrap();
374        let f = find(&a, "x").unwrap();
375        assert_eq!(f.size(), 0);
376    }
377}
ferray_strings/lib.rs

ferray_strings/
lib.rs
