1use sofie_core::{Cast, Error, Result, SqlType, SqlValue};
2use sofie_values::{LATIN_COMBINING_ACCENTS, is_slug_character};
3use unicode_normalization::UnicodeNormalization as _;
4
5use crate::behavior::stored;
6use crate::{
7 Aggregate, Behavior, Column, Dialect, Field, Plan, Query, Read, Source, Step, Uses, Write,
8};
9
10const SLUG_LENGTH: usize = 100;
11const CODE_ALPHABET: &[u8; 32] = b"0123456789ABCDEFGHJKMNPQRSTVWXYZ";
13const CODE_LENGTH: usize = 10;
14
15pub struct Slugged {
16 pub from: &'static str,
17}
18
19impl Behavior for Slugged {
20 fn columns(&self) -> &'static [Field] {
21 const { &[Field::new("slug", String::COLUMNS, false)] }
22 }
23
24 fn written(&self, write: &Write<'_>, values: &mut [SqlValue]) {
26 let Some(SqlValue::Text(text)) = write.value(self.from) else {
27 return;
28 };
29 let kept = matches!(values[0], SqlValue::Text(_))
30 && write.loaded(self.from) == write.value(self.from);
31 if !kept {
32 values[0] = SqlValue::Text(slug(text));
33 }
34 }
35
36 fn planned(&self, write: &Write<'_>, plan: &mut Plan<'_>) -> Result<()> {
38 let index = format!("{}_slug", write.table.name);
39 plan.unique_fields.push(crate::outcome::UniqueField::column(
40 write.table.schema,
41 index,
42 self.from,
43 ));
44 let table = record_table(write);
45 if write.deleted() == (false, true) {
46 let params = vec![table, SqlValue::Uuid(write.id)];
47 plan.push(previous_slugs(plan.dialect(), FORGET_RECORD, params));
48 return Ok(());
49 }
50 let Some((previous, current)) = renamed(write) else {
51 return Ok(());
52 };
53 if current.is_empty() {
54 return Err(Error::invalid(
55 self.from,
56 "needs a letter or digit to name it by",
57 ));
58 }
59 let params = vec![table.clone(), SqlValue::Text(current.to_owned())];
60 plan.push(previous_slugs(plan.dialect(), FORGET_SLUG, params));
61 if let Some(previous) = previous {
62 let params = vec![
63 table,
64 SqlValue::Text(previous.to_owned()),
65 SqlValue::Uuid(write.id),
66 SqlValue::Timestamp(write.stamp.now.timestamp()),
67 ];
68 plan.push(previous_slugs(plan.dialect(), KEEP, params));
69 }
70 Ok(())
71 }
72
73 #[cfg(feature = "retention")]
74 fn pruned(&self, schema: &str, name: &str, ids: &[sofie_core::Uuid], plan: &mut Plan<'_>) {
75 let table = SqlValue::Text(format!("{schema}.{name}"));
76 plan.push(forget_records(plan.dialect(), table, ids));
77 }
78}
79
80#[cfg(feature = "retention")]
81fn forget_records(dialect: &dyn Dialect, table: SqlValue, ids: &[sofie_core::Uuid]) -> Step {
82 let previous_slug = dialect.qualified("sofie", "previous_slug");
83 let ids = ids.iter().map(|id| SqlValue::Uuid(*id));
84 if dialect.binds_arrays() {
85 let sql = format!(
86 "DELETE FROM {previous_slug} WHERE record_table = {} AND record = ANY({}::uuid[])",
87 dialect.placeholder(1),
88 dialect.placeholder(2)
89 );
90 return Step::statement(sql, vec![table, SqlValue::Array(ids.collect())], None);
91 }
92 let mut params = vec![table];
93 params.extend(ids);
94 let records: Vec<String> = (2..=params.len())
95 .map(|index| dialect.placeholder(index))
96 .collect();
97 let sql = format!(
98 "DELETE FROM {previous_slug} WHERE record_table = {} AND record IN ({})",
99 dialect.placeholder(1),
100 records.join(", ")
101 );
102 Step::statement(sql, params, None)
103}
104
105const FORGET_RECORD: &str = "DELETE FROM {table} WHERE record_table = {1} AND record = {2}";
106const FORGET_SLUG: &str = "DELETE FROM {table} WHERE record_table = {1} AND slug = {2}";
107const KEEP: &str = "INSERT INTO {table} (record_table, slug, record, changed_at) \
108 VALUES ({1}, {2}, {3}, {4}) ON CONFLICT (record_table, slug) \
109 DO UPDATE SET record = excluded.record, changed_at = excluded.changed_at";
110
111fn renamed<'w>(write: &'w Write<'_>) -> Option<(Option<&'w str>, &'w str)> {
113 let Some(SqlValue::Text(current)) = write.value("slug") else {
114 return None;
115 };
116 match write.loaded("slug") {
117 Some(SqlValue::Text(previous)) if previous == current => None,
118 Some(SqlValue::Text(previous)) if !previous.is_empty() => Some((Some(previous), current)),
119 _ => Some((None, current)),
120 }
121}
122
123fn record_table(write: &Write<'_>) -> SqlValue {
124 SqlValue::Text(format!("{}.{}", write.table.schema, write.table.name))
125}
126
127fn previous_slugs(dialect: &dyn Dialect, template: &str, params: Vec<SqlValue>) -> Step {
128 let mut sql = template.replace("{table}", &dialect.qualified("sofie", "previous_slug"));
129 for index in 1..=params.len() {
130 sql = sql.replace(&format!("{{{index}}}"), &dialect.placeholder(index));
131 }
132 Step::statement(sql, params, None)
133}
134
135fn slug(text: &str) -> String {
136 let mut slug = String::new();
137 for word in words(text).split('-').filter(|word| !word.is_empty()) {
138 if slug.len() + word.len() + 1 > SLUG_LENGTH {
139 if slug.is_empty() {
140 slug.push_str(cut(word, SLUG_LENGTH));
141 }
142 break;
143 }
144 if !slug.is_empty() {
145 slug.push('-');
146 }
147 slug.push_str(word);
148 }
149 slug
150}
151
152fn words(text: &str) -> String {
153 let mut words = String::with_capacity(text.len());
154 for character in text.nfc() {
155 if let Some(latin) = latin(character) {
156 words.push_str(latin);
157 continue;
158 }
159 for lowered in character.to_lowercase() {
160 match latin(lowered) {
161 Some(latin) => words.push_str(latin),
162 None if LATIN_COMBINING_ACCENTS.contains(&lowered) => {}
164 None if is_slug_character(lowered) => words.push(lowered),
165 None => words.push('-'),
166 }
167 }
168 }
169 words
170}
171
172fn cut(word: &str, length: usize) -> &str {
173 let mut end = length;
174 while !word.is_char_boundary(end) {
175 end -= 1;
176 }
177 &word[..end]
178}
179
180fn latin(character: char) -> Option<&'static str> {
182 Some(match character {
183 'à' | 'á' | 'â' | 'ã' | 'ä' | 'å' | 'ā' | 'ă' | 'ą' => "a",
184 'æ' => "ae",
185 'ç' | 'ć' | 'č' => "c",
186 'ď' | 'đ' | 'ð' => "d",
187 'è' | 'é' | 'ê' | 'ë' | 'ē' | 'ę' | 'ě' => "e",
188 'ğ' => "g",
189 'ì' | 'í' | 'î' | 'ï' | 'ī' | 'ı' | 'İ' => "i",
190 'ł' => "l",
191 'ñ' | 'ń' | 'ň' => "n",
192 'ò' | 'ó' | 'ô' | 'õ' | 'ö' | 'ø' | 'ō' | 'ő' => "o",
193 'œ' => "oe",
194 'ř' => "r",
195 'ß' => "ss",
196 'ś' | 'ş' | 'š' | 'ș' => "s",
197 'ț' | 'ţ' | 'ť' => "t",
198 'þ' => "th",
199 'ù' | 'ú' | 'û' | 'ü' | 'ū' | 'ů' | 'ű' => "u",
200 'ý' | 'ÿ' => "y",
201 'ź' | 'ż' | 'ž' => "z",
202 _ => return cyrillic(character),
203 })
204}
205
206fn cyrillic(character: char) -> Option<&'static str> {
207 Some(match character {
208 'а' => "a",
209 'б' => "b",
210 'в' => "v",
211 'г' | 'ґ' => "g",
212 'д' => "d",
213 'е' | 'ё' | 'э' => "e",
214 'є' => "ye",
215 'ж' => "zh",
216 'з' => "z",
217 'и' | 'і' => "i",
218 'ї' => "yi",
219 'й' | 'ы' => "y",
220 'к' => "k",
221 'л' => "l",
222 'м' => "m",
223 'н' => "n",
224 'о' => "o",
225 'п' => "p",
226 'р' => "r",
227 'с' => "s",
228 'т' => "t",
229 'у' => "u",
230 'ф' => "f",
231 'х' => "kh",
232 'ц' => "ts",
233 'ч' => "ch",
234 'ш' => "sh",
235 'щ' => "shch",
236 'ъ' | 'ь' => "",
237 'ю' => "yu",
238 'я' => "ya",
239 _ => return None,
240 })
241}
242
243pub trait Slugs: Aggregate {
244 const SLUG: Column<Self, String> = Column::new("slug");
245
246 fn slug(&self) -> &str {
247 stored_text(self, "slug")
248 }
249}
250
251impl<A: Uses<Slugged>> Slugs for A {}
252
253impl<T: Source> Query<T>
254where
255 T::Base: Slugs,
256{
257 pub async fn by_slug(self, read: &Read, slug: &str) -> Result<T> {
259 let named = self
260 .clone()
261 .where_(T::Base::SLUG.eq(slug.to_owned()))
262 .first(read)
263 .await?;
264 if let Some(record) = named {
265 return Ok(record);
266 }
267 let Some(current) = current_slug::<T::Base>(read, slug).await? else {
268 return Err(Error::NotFound);
269 };
270 if self
271 .where_(T::Base::SLUG.eq(current.clone()))
272 .exists(read)
273 .await?
274 {
275 return Err(Error::Moved {
276 from: slug.to_owned(),
277 to: current,
278 });
279 }
280 Err(Error::NotFound)
281 }
282}
283
284async fn current_slug<A: Aggregate>(read: &Read, previous: &str) -> Result<Option<String>> {
285 let dialect = read.dialect();
286 let sql = format!(
287 "SELECT named.slug FROM {} AS moved JOIN {} AS named ON named.id = moved.record \
288 WHERE moved.record_table = {} AND moved.slug = {}",
289 dialect.qualified("sofie", "previous_slug"),
290 dialect.qualified(A::MODULE, A::NAME),
291 dialect.placeholder(1),
292 dialect.placeholder(2)
293 );
294 let params = [
295 SqlValue::Text(format!("{}.{}", A::MODULE, A::NAME)),
296 SqlValue::Text(previous.to_owned()),
297 ];
298 let slugs = read
299 .decoded(&sql, ¶ms, 1, vec![Some(SqlType::Text)], String::decode)
300 .await?;
301 Ok(slugs.into_iter().next())
302}
303
304pub struct UniqueCode;
306
307impl Behavior for UniqueCode {
308 fn columns(&self) -> &'static [Field] {
309 const { &[Field::new("code", String::COLUMNS, false)] }
310 }
311
312 fn written(&self, write: &Write<'_>, values: &mut [SqlValue]) {
313 if values[0] == SqlValue::Null {
314 values[0] = SqlValue::Text(code(write.id.as_bytes()));
315 }
316 }
317}
318
319fn code(id: &[u8; 16]) -> String {
321 let random = u64::from_be_bytes(id[8..].try_into().expect("eight bytes"));
322 (0..CODE_LENGTH)
323 .map(|index| {
324 let digit = (random >> (5 * index)) & 0b1_1111;
325 char::from(CODE_ALPHABET[usize::try_from(digit).expect("a five-bit digit")])
326 })
327 .collect()
328}
329
330pub trait Coded: Aggregate {
331 const CODE: Column<Self, String> = Column::new("code");
332
333 fn code(&self) -> &str {
334 stored_text(self, "code")
335 }
336}
337
338impl<A: Uses<UniqueCode>> Coded for A {}
339
340fn stored_text<'a, A: Aggregate>(aggregate: &'a A, column: &str) -> &'a str {
341 match stored(aggregate, column) {
342 Some(SqlValue::Text(text)) => text,
343 _ => "",
344 }
345}
346
347#[cfg(test)]
348mod tests {
349 use super::*;
350
351 #[test]
352 fn a_slug_joins_lowercase_words_by_hyphens() {
353 assert_eq!(slug(" Blue Wool -- Scarf (2026)!"), "blue-wool-scarf-2026");
354 assert_eq!(slug("a/b_c.d,e:f"), "a-b-c-d-e-f");
355 assert_eq!(slug("!?"), "");
356 }
357
358 #[test]
359 fn a_slug_spells_turkish_german_and_accented_latin_letters_in_ascii() {
360 assert_eq!(slug("Öğle Yemeği"), "ogle-yemegi");
361 assert_eq!(
362 slug("İstanbul'da Çay Şöleni ılık"),
363 "istanbul-da-cay-soleni-ilik"
364 );
365 assert_eq!(slug("Große Straße für Äpfel"), "grosse-strasse-fur-apfel");
366 assert_eq!(
367 slug("Crème brûlée à la française"),
368 "creme-brulee-a-la-francaise"
369 );
370 assert_eq!(slug("Cre\u{300}me"), "creme");
371 assert_eq!(
372 slug("Smørrebrød Æble Œuvre Łódź Đak Ñandú Åre"),
373 "smorrebrod-aeble-oeuvre-lodz-dak-nandu-are"
374 );
375 }
376
377 #[test]
378 fn a_slug_spells_cyrillic_in_latin_and_keeps_other_scripts() {
379 assert_eq!(slug("Привет, мир!"), "privet-mir");
380 assert_eq!(slug("Щедрий вечір, Україно"), "shchedriy-vechir-ukrayino");
381 assert_eq!(slug("关于我们 2026"), "关于我们-2026");
382 assert_eq!(slug("Ελλάδα"), "ελλάδα");
383 }
384
385 #[test]
386 fn a_decomposed_title_slugs_as_its_composed_form() {
387 assert_eq!(slug("Мой край"), slug("Мои\u{306} край"));
388 assert_eq!(slug("Ελλάδα"), slug("Ελλα\u{301}δα"));
389 }
390
391 #[test]
392 fn a_slug_keeps_the_marks_of_other_scripts_in_their_word() {
393 assert_eq!(slug("नमस्ते दुनिया"), "नमस्ते-दुनिया");
394 assert_eq!(slug("ไม่ ใช่"), "ไม่-ใช่");
395 assert_eq!(slug("关于 — 我们 🎉 «x»"), "关于-我们-x");
396 }
397
398 #[test]
399 fn a_slug_stops_before_the_word_that_passes_its_length() {
400 assert_eq!(slug(&"word ".repeat(30)).len(), 99);
401 let long = slug(&"ğüş ".repeat(40));
402 assert!(long.len() <= SLUG_LENGTH && long.ends_with("gus"), "{long}");
403 let chinese = slug(&"关".repeat(40));
404 assert_eq!(chinese, "关".repeat(33));
405 }
406
407 #[test]
408 fn a_code_spells_the_random_bits_of_the_id() {
409 let first = code(&[0; 16]);
410 let other = code(&[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1]);
411
412 assert_eq!(first, "0000000000");
413 assert_eq!(other, "1000000000");
414 }
415}