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
28// Workspace convention: every public function returns FerrayResult<T> and
29// the FerrayError variants are documented once on the type, not on every
30// returning function.
31#![allow(
32    clippy::missing_errors_doc,
33    clippy::missing_panics_doc,
34    clippy::many_single_char_names,
35    clippy::similar_names,
36    clippy::items_after_statements,
37    clippy::option_if_let_else,
38    clippy::too_long_first_doc_paragraph,
39    clippy::needless_pass_by_value,
40    clippy::match_same_arms
41)]
42
43pub mod align;
44pub mod case;
45pub mod classify;
46pub mod concat;
47pub mod regex_ops;
48pub mod search;
49pub mod split_join;
50pub mod str_ops;
51pub mod string_array;
52pub mod strip;
53
54// Re-export types
55pub use string_array::{StringArray, StringArray1, StringArray2, array};
56
57// Re-export operations for flat namespace (like numpy.strings.upper etc.)
58pub use align::{center, ljust, ljust_with, rjust, rjust_with, zfill};
59pub use case::{capitalize, lower, title, upper};
60pub use classify::{isalnum, isalpha, isdigit, islower, isnumeric, isspace, istitle, isupper};
61pub use concat::{add, add_same, multiply};
62pub use regex_ops::{extract, extract_compiled, match_, match_compiled};
63// Re-export `regex::Regex` so callers of `match_compiled`/`extract_compiled`
64// don't have to add a direct `regex` dependency to construct one.
65pub use regex::Regex;
66pub use search::{count, endswith, find, replace, startswith};
67pub use split_join::{join, join_array, split};
68pub use str_ops::{equal, greater, greater_equal, less, less_equal, not_equal, str_len, swapcase};
69pub use strip::{lstrip, rstrip, strip};
70
71#[cfg(test)]
72mod integration_tests {
73    use super::*;
74
75    #[test]
76    fn ac1_upper() {
77        // AC-1: strings::upper(&["hello", "world"]) produces ["HELLO", "WORLD"]
78        let a = array(&["hello", "world"]).unwrap();
79        let b = upper(&a).unwrap();
80        assert_eq!(b.as_slice(), &["HELLO", "WORLD"]);
81    }
82
83    #[test]
84    fn ac2_add_broadcast_scalar() {
85        // AC-2: strings::add broadcasts a scalar string against an array correctly
86        let a = array(&["hello", "world"]).unwrap();
87        let b = array(&["!"]).unwrap();
88        let c = add(&a, &b).unwrap();
89        assert_eq!(c.as_slice(), &["hello!", "world!"]);
90    }
91
92    #[test]
93    fn ac3_find_indices() {
94        // AC-3: strings::find(&a, "ll") returns correct indices
95        let a = array(&["hello", "world"]).unwrap();
96        let b = find(&a, "ll").unwrap();
97        let data = b.as_slice().unwrap();
98        assert_eq!(data, &[2_i64, -1_i64]);
99    }
100
101    #[test]
102    fn ac4_split() {
103        // AC-4: strings::split(&["a-b", "c-d"], "-") returns [vec!["a","b"], vec!["c","d"]]
104        let a = array(&["a-b", "c-d"]).unwrap();
105        let result = split(&a, "-").unwrap();
106        assert_eq!(
107            result,
108            vec![
109                vec!["a".to_string(), "b".to_string()],
110                vec!["c".to_string(), "d".to_string()],
111            ]
112        );
113    }
114
115    #[test]
116    fn ac5_regex() {
117        // AC-5: Regex match_ and extract work correctly with capture groups
118        let a = array(&["abc123", "def", "ghi456"]).unwrap();
119
120        let matched = match_(&a, r"\d+").unwrap();
121        let matched_data = matched.as_slice().unwrap();
122        assert_eq!(matched_data, &[true, false, true]);
123
124        let extracted = extract(&a, r"(\d+)").unwrap();
125        assert_eq!(extracted.as_slice(), &["123", "", "456"]);
126    }
127
128    #[test]
129    fn full_pipeline() {
130        // End-to-end: strip, upper, add suffix, search
131        let raw = array(&["  Hello  ", " World "]).unwrap();
132        let stripped = strip(&raw, None).unwrap();
133        let uppered = upper(&stripped).unwrap();
134        let suffix = array(&["!"]).unwrap();
135        let result = add(&uppered, &suffix).unwrap();
136        assert_eq!(result.as_slice(), &["HELLO!", "WORLD!"]);
137
138        let has_excl = endswith(&result, "!").unwrap();
139        let data = has_excl.as_slice().unwrap();
140        assert_eq!(data, &[true, true]);
141    }
142
143    #[test]
144    fn case_round_trip() {
145        let a = array(&["Hello World"]).unwrap();
146        let low = lower(&a).unwrap();
147        let titled = title(&low).unwrap();
148        assert_eq!(titled.as_slice(), &["Hello World"]);
149    }
150
151    #[test]
152    fn alignment_operations() {
153        let a = array(&["hi"]).unwrap();
154        let c = center(&a, 6, '-').unwrap();
155        assert_eq!(c.as_slice(), &["--hi--"]);
156
157        let l = ljust(&a, 6).unwrap();
158        assert_eq!(l.as_slice(), &["hi    "]);
159
160        let r = rjust(&a, 6).unwrap();
161        assert_eq!(r.as_slice(), &["    hi"]);
162
163        let z = zfill(&array(&["42"]).unwrap(), 5).unwrap();
164        assert_eq!(z.as_slice(), &["00042"]);
165    }
166
167    #[test]
168    fn strip_operations() {
169        let a = array(&["  hello  "]).unwrap();
170        assert_eq!(strip(&a, None).unwrap().as_slice(), &["hello"]);
171        assert_eq!(lstrip(&a, None).unwrap().as_slice(), &["hello  "]);
172        assert_eq!(rstrip(&a, None).unwrap().as_slice(), &["  hello"]);
173    }
174
175    #[test]
176    fn search_operations() {
177        let a = array(&["hello world", "foo bar"]).unwrap();
178        let c = count(&a, "o").unwrap();
179        let data = c.as_slice().unwrap();
180        // "hello world" has 2 'o's, "foo bar" has 2 'o's
181        assert_eq!(data, &[2_u64, 2]);
182    }
183
184    #[test]
185    fn replace_operation() {
186        let a = array(&["hello world"]).unwrap();
187        let b = replace(&a, "world", "rust", None).unwrap();
188        assert_eq!(b.as_slice(), &["hello rust"]);
189    }
190
191    #[test]
192    fn multiply_operation() {
193        let a = array(&["ab"]).unwrap();
194        let b = multiply(&a, 3).unwrap();
195        assert_eq!(b.as_slice(), &["ababab"]);
196    }
197
198    #[test]
199    fn join_operation() {
200        let parts = vec![
201            vec!["a".to_string(), "b".to_string()],
202            vec!["c".to_string(), "d".to_string()],
203        ];
204        let result = join("-", &parts).unwrap();
205        assert_eq!(result.as_slice(), &["a-b", "c-d"]);
206    }
207
208    #[test]
209    fn capitalize_operation() {
210        let a = array(&["hello world", "RUST"]).unwrap();
211        let b = capitalize(&a).unwrap();
212        assert_eq!(b.as_slice(), &["Hello world", "Rust"]);
213    }
214
215    #[test]
216    fn string_array_2d() {
217        let a = StringArray2::from_rows(&[&["a", "b"], &["c", "d"]]).unwrap();
218        assert_eq!(a.shape(), &[2, 2]);
219        let b = upper(&a).unwrap();
220        assert_eq!(b.as_slice(), &["A", "B", "C", "D"]);
221        assert_eq!(b.shape(), &[2, 2]);
222    }
223
224    // -----------------------------------------------------------------
225    // #520 — shape preservation for ND StringArrays across every op.
226    //
227    // Every operation that takes `StringArray<D>` must thread `D`
228    // through the output. These tests pin that contract on a 2x2
229    // input so regressions (e.g. an accidental `Ix1::new([len])` in
230    // the output constructor) are caught immediately.
231    // -----------------------------------------------------------------
232
233    fn two_by_two(vals: &[&str; 4]) -> crate::StringArray2 {
234        crate::StringArray2::from_rows(&[&[vals[0], vals[1]], &[vals[2], vals[3]]]).unwrap()
235    }
236
237    #[test]
238    fn shape_preserved_case_ops_2d() {
239        let a = two_by_two(&["Hello", "World", "foo", "Bar"]);
240        assert_eq!(upper(&a).unwrap().shape(), &[2, 2]);
241        assert_eq!(lower(&a).unwrap().shape(), &[2, 2]);
242        assert_eq!(capitalize(&a).unwrap().shape(), &[2, 2]);
243        assert_eq!(title(&a).unwrap().shape(), &[2, 2]);
244    }
245
246    #[test]
247    fn shape_preserved_align_ops_2d() {
248        let a = two_by_two(&["a", "bb", "ccc", "dddd"]);
249        assert_eq!(center(&a, 6, ' ').unwrap().shape(), &[2, 2]);
250        assert_eq!(ljust(&a, 6).unwrap().shape(), &[2, 2]);
251        assert_eq!(rjust(&a, 6).unwrap().shape(), &[2, 2]);
252        assert_eq!(zfill(&a, 6).unwrap().shape(), &[2, 2]);
253    }
254
255    #[test]
256    fn shape_preserved_strip_ops_2d() {
257        let a = two_by_two(&["  a  ", "  b  ", "  c  ", "  d  "]);
258        assert_eq!(strip(&a, None).unwrap().shape(), &[2, 2]);
259        assert_eq!(lstrip(&a, None).unwrap().shape(), &[2, 2]);
260        assert_eq!(rstrip(&a, None).unwrap().shape(), &[2, 2]);
261    }
262
263    #[test]
264    fn shape_preserved_concat_ops_2d() {
265        let a = two_by_two(&["ab", "cd", "ef", "gh"]);
266        // `add` currently flattens to IxDyn regardless of input rank
267        // (it supports cross-rank broadcasting), so it's not expected
268        // to preserve D. Just verify the total element count.
269        let b = two_by_two(&["!", "!", "!", "!"]);
270        let ab = add(&a, &b).unwrap();
271        assert_eq!(ab.shape(), &[2, 2]);
272        // multiply preserves D
273        assert_eq!(multiply(&a, 2).unwrap().shape(), &[2, 2]);
274    }
275
276    #[test]
277    fn shape_preserved_search_ops_2d() {
278        let a = two_by_two(&["hello", "help", "world", "word"]);
279        // Each of these returns Array<T, D> — verify D is preserved.
280        assert_eq!(find(&a, "ell").unwrap().shape(), &[2, 2]);
281        assert_eq!(count(&a, "l").unwrap().shape(), &[2, 2]);
282        assert_eq!(startswith(&a, "he").unwrap().shape(), &[2, 2]);
283        assert_eq!(endswith(&a, "d").unwrap().shape(), &[2, 2]);
284        assert_eq!(replace(&a, "l", "L", None).unwrap().shape(), &[2, 2]);
285    }
286
287    #[test]
288    fn shape_preserved_regex_ops_2d() {
289        let a = two_by_two(&["abc123", "x", "y42", "zzz"]);
290        // match_ preserves D; extract flattens by design (ragged results).
291        assert_eq!(match_(&a, r"\d+").unwrap().shape(), &[2, 2]);
292    }
293
294    #[test]
295    fn shape_preserved_case_ops_3d() {
296        // Just to confirm we aren't special-casing Ix2 somewhere — bump to Ix3.
297        use ferray_core::dimension::Ix3;
298        let data: Vec<String> = (0..8).map(|i| format!("s{i}")).collect();
299        let a = crate::StringArray::<Ix3>::from_vec(Ix3::new([2, 2, 2]), data).unwrap();
300        assert_eq!(upper(&a).unwrap().shape(), &[2, 2, 2]);
301        assert_eq!(lower(&a).unwrap().shape(), &[2, 2, 2]);
302    }
303
304    // --- Unicode / multi-byte character tests ---
305
306    #[test]
307    fn unicode_upper_lower() {
308        let a = array(&["café", "naïve", "über"]).unwrap();
309        let u = upper(&a).unwrap();
310        assert_eq!(u.as_slice(), &["CAFÉ", "NAÏVE", "ÜBER"]);
311        let l = lower(&u).unwrap();
312        assert_eq!(l.as_slice(), &["café", "naïve", "über"]);
313    }
314
315    #[test]
316    fn unicode_capitalize() {
317        let a = array(&["ñoño", "straße"]).unwrap();
318        let c = capitalize(&a).unwrap();
319        assert_eq!(c.as_slice()[0], "Ñoño");
320        // Rust's capitalize of "straße" -> "Straße"
321        assert_eq!(c.as_slice()[1], "Straße");
322    }
323
324    #[test]
325    fn unicode_find() {
326        let a = array(&["日本語テスト", "こんにちは"]).unwrap();
327        let r = find(&a, "テスト").unwrap();
328        let data = r.as_slice().unwrap();
329        assert_eq!(data[0], 3); // "テスト" starts at byte position, but find uses char position...
330        // Actually, find returns the byte index via str::find. Check:
331        // "日本語テスト".find("テスト") returns byte offset 9
332        // But our find should return character index or byte index?
333        // Let's just verify it finds it (>= 0) vs not found (-1)
334        assert!(data[0] >= 0); // found
335        assert_eq!(data[1], -1); // not found
336    }
337
338    #[test]
339    fn unicode_strip() {
340        let a = array(&["  héllo  ", "  wörld  "]).unwrap();
341        let s = strip(&a, None).unwrap();
342        assert_eq!(s.as_slice(), &["héllo", "wörld"]);
343    }
344
345    #[test]
346    fn unicode_replace() {
347        let a = array(&["café latte"]).unwrap();
348        let r = replace(&a, "café", "tea", None).unwrap();
349        assert_eq!(r.as_slice(), &["tea latte"]);
350    }
351
352    #[test]
353    fn emoji_operations() {
354        let a = array(&["hello 🌍", "rust 🦀"]).unwrap();
355        let u = upper(&a).unwrap();
356        assert_eq!(u.as_slice(), &["HELLO 🌍", "RUST 🦀"]);
357        let c = count(&a, "🌍").unwrap();
358        assert_eq!(c.as_slice().unwrap(), &[1, 0]);
359    }
360
361    #[test]
362    fn cjk_characters() {
363        let a = array(&["你好世界", "こんにちは"]).unwrap();
364        let starts = startswith(&a, "你好").unwrap();
365        assert_eq!(starts.as_slice().unwrap(), &[true, false]);
366        let ends = endswith(&a, "世界").unwrap();
367        assert_eq!(ends.as_slice().unwrap(), &[true, false]);
368    }
369
370    // ----- Empty array tests (#282) -----
371
372    #[test]
373    fn empty_array_upper() {
374        let a = StringArray1::from_vec(ferray_core::dimension::Ix1::new([0]), vec![]).unwrap();
375        let u = upper(&a).unwrap();
376        assert_eq!(u.len(), 0);
377    }
378
379    #[test]
380    fn empty_array_str_len() {
381        let a = StringArray1::from_vec(ferray_core::dimension::Ix1::new([0]), vec![]).unwrap();
382        let l = str_len(&a).unwrap();
383        assert_eq!(l.size(), 0);
384    }
385
386    #[test]
387    fn empty_array_find() {
388        let a = StringArray1::from_vec(ferray_core::dimension::Ix1::new([0]), vec![]).unwrap();
389        let f = find(&a, "x").unwrap();
390        assert_eq!(f.size(), 0);
391    }
392}
ferray_strings/lib.rs

ferray_strings/
lib.rs
