Skip to main content

sofie_db/
codes.rs

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;
11/// Crockford's base 32: no I, L, O or U, so a code read aloud is not misread.
12const 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    /// An unchanged `from` keeps its slug, so a later transliteration does not move the record.
25    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    /// A slug taken now is no longer another record's previous one; a deleted record keeps none.
37    fn planned(&self, write: &Write<'_>, plan: &mut Plan<'_>) -> Result<()> {
38        let index = format!("{}_slug", write.table.name);
39        plan.unique_column(write.table.schema, &index, self.from);
40        let table = record_table(write);
41        if write.deleted() == (false, true) {
42            let params = vec![table, SqlValue::Uuid(write.id)];
43            plan.push(previous_slugs(plan.dialect(), FORGET_RECORD, params));
44            return Ok(());
45        }
46        let Some((previous, current)) = renamed(write) else {
47            return Ok(());
48        };
49        if current.is_empty() {
50            return Err(Error::invalid(
51                self.from,
52                "needs a letter or digit to name it by",
53            ));
54        }
55        let params = vec![table.clone(), SqlValue::Text(current.to_owned())];
56        plan.push(previous_slugs(plan.dialect(), FORGET_SLUG, params));
57        if let Some(previous) = previous {
58            let params = vec![
59                table,
60                SqlValue::Text(previous.to_owned()),
61                SqlValue::Uuid(write.id),
62                SqlValue::Timestamp(write.stamp.now.timestamp()),
63            ];
64            plan.push(previous_slugs(plan.dialect(), KEEP, params));
65        }
66        Ok(())
67    }
68}
69
70const FORGET_RECORD: &str = "DELETE FROM {table} WHERE record_table = {1} AND record = {2}";
71const FORGET_SLUG: &str = "DELETE FROM {table} WHERE record_table = {1} AND slug = {2}";
72const KEEP: &str = "INSERT INTO {table} (record_table, slug, record, changed_at) \
73    VALUES ({1}, {2}, {3}, {4}) ON CONFLICT (record_table, slug) \
74    DO UPDATE SET record = excluded.record, changed_at = excluded.changed_at";
75
76/// The slug the record leaves, if any, and the one it takes.
77fn renamed<'w>(write: &'w Write<'_>) -> Option<(Option<&'w str>, &'w str)> {
78    let Some(SqlValue::Text(current)) = write.value("slug") else {
79        return None;
80    };
81    match write.loaded("slug") {
82        Some(SqlValue::Text(previous)) if previous == current => None,
83        Some(SqlValue::Text(previous)) if !previous.is_empty() => Some((Some(previous), current)),
84        _ => Some((None, current)),
85    }
86}
87
88fn record_table(write: &Write<'_>) -> SqlValue {
89    SqlValue::Text(format!("{}.{}", write.table.schema, write.table.name))
90}
91
92fn previous_slugs(dialect: &dyn Dialect, template: &str, params: Vec<SqlValue>) -> Step {
93    let mut sql = template.replace("{table}", &dialect.qualified("sofie", "previous_slug"));
94    for index in 1..=params.len() {
95        sql = sql.replace(&format!("{{{index}}}"), &dialect.placeholder(index));
96    }
97    Step::statement(sql, params, None)
98}
99
100fn slug(text: &str) -> String {
101    let mut slug = String::new();
102    for word in words(text).split('-').filter(|word| !word.is_empty()) {
103        if slug.len() + word.len() + 1 > SLUG_LENGTH {
104            if slug.is_empty() {
105                slug.push_str(cut(word, SLUG_LENGTH));
106            }
107            break;
108        }
109        if !slug.is_empty() {
110            slug.push('-');
111        }
112        slug.push_str(word);
113    }
114    slug
115}
116
117fn words(text: &str) -> String {
118    let mut words = String::with_capacity(text.len());
119    for character in text.nfc() {
120        if let Some(latin) = latin(character) {
121            words.push_str(latin);
122            continue;
123        }
124        for lowered in character.to_lowercase() {
125            match latin(lowered) {
126                Some(latin) => words.push_str(latin),
127                // A decomposed accent belongs to the Latin letter before it.
128                None if LATIN_COMBINING_ACCENTS.contains(&lowered) => {}
129                None if is_slug_character(lowered) => words.push(lowered),
130                None => words.push('-'),
131            }
132        }
133    }
134    words
135}
136
137fn cut(word: &str, length: usize) -> &str {
138    let mut end = length;
139    while !word.is_char_boundary(end) {
140        end -= 1;
141    }
142    &word[..end]
143}
144
145/// `İ` is matched before lowercasing, which would leave a dot above its `i`.
146fn latin(character: char) -> Option<&'static str> {
147    Some(match character {
148        'à' | 'á' | 'â' | 'ã' | 'ä' | 'å' | 'ā' | 'ă' | 'ą' => "a",
149        'æ' => "ae",
150        'ç' | 'ć' | 'č' => "c",
151        'ď' | 'đ' | 'ð' => "d",
152        'è' | 'é' | 'ê' | 'ë' | 'ē' | 'ę' | 'ě' => "e",
153        'ğ' => "g",
154        'ì' | 'í' | 'î' | 'ï' | 'ī' | 'ı' | 'İ' => "i",
155        'ł' => "l",
156        'ñ' | 'ń' | 'ň' => "n",
157        'ò' | 'ó' | 'ô' | 'õ' | 'ö' | 'ø' | 'ō' | 'ő' => "o",
158        'œ' => "oe",
159        'ř' => "r",
160        'ß' => "ss",
161        'ś' | 'ş' | 'š' | 'ș' => "s",
162        'ț' | 'ţ' | 'ť' => "t",
163        'þ' => "th",
164        'ù' | 'ú' | 'û' | 'ü' | 'ū' | 'ů' | 'ű' => "u",
165        'ý' | 'ÿ' => "y",
166        'ź' | 'ż' | 'ž' => "z",
167        _ => return cyrillic(character),
168    })
169}
170
171fn cyrillic(character: char) -> Option<&'static str> {
172    Some(match character {
173        'а' => "a",
174        'б' => "b",
175        'в' => "v",
176        'г' | 'ґ' => "g",
177        'д' => "d",
178        'е' | 'ё' | 'э' => "e",
179        'є' => "ye",
180        'ж' => "zh",
181        'з' => "z",
182        'и' | 'і' => "i",
183        'ї' => "yi",
184        'й' | 'ы' => "y",
185        'к' => "k",
186        'л' => "l",
187        'м' => "m",
188        'н' => "n",
189        'о' => "o",
190        'п' => "p",
191        'р' => "r",
192        'с' => "s",
193        'т' => "t",
194        'у' => "u",
195        'ф' => "f",
196        'х' => "kh",
197        'ц' => "ts",
198        'ч' => "ch",
199        'ш' => "sh",
200        'щ' => "shch",
201        'ъ' | 'ь' => "",
202        'ю' => "yu",
203        'я' => "ya",
204        _ => return None,
205    })
206}
207
208pub trait Slugs: Aggregate {
209    const SLUG: Column<Self, String> = Column::new("slug");
210
211    fn slug(&self) -> &str {
212        stored_text(self, "slug")
213    }
214}
215
216impl<A: Uses<Slugged>> Slugs for A {}
217
218impl<T: Source> Query<T>
219where
220    T::Base: Slugs,
221{
222    /// A slug that a record this query finds had before is `Error::Moved` to its current one.
223    pub async fn by_slug(self, read: &Read, slug: &str) -> Result<T> {
224        let named = self
225            .clone()
226            .where_(T::Base::SLUG.eq(slug.to_owned()))
227            .first(read)
228            .await?;
229        if let Some(record) = named {
230            return Ok(record);
231        }
232        let Some(current) = current_slug::<T::Base>(read, slug).await? else {
233            return Err(Error::NotFound);
234        };
235        if self
236            .where_(T::Base::SLUG.eq(current.clone()))
237            .exists(read)
238            .await?
239        {
240            return Err(Error::Moved {
241                from: slug.to_owned(),
242                to: current,
243            });
244        }
245        Err(Error::NotFound)
246    }
247}
248
249async fn current_slug<A: Aggregate>(read: &Read, previous: &str) -> Result<Option<String>> {
250    let dialect = read.dialect();
251    let sql = format!(
252        "SELECT named.slug FROM {} AS moved JOIN {} AS named ON named.id = moved.record \
253         WHERE moved.record_table = {} AND moved.slug = {}",
254        dialect.qualified("sofie", "previous_slug"),
255        dialect.qualified(A::MODULE, A::NAME),
256        dialect.placeholder(1),
257        dialect.placeholder(2)
258    );
259    let params = [
260        SqlValue::Text(format!("{}.{}", A::MODULE, A::NAME)),
261        SqlValue::Text(previous.to_owned()),
262    ];
263    let slugs = read
264        .decoded(&sql, &params, 1, vec![Some(SqlType::Text)], String::decode)
265        .await?;
266    Ok(slugs.into_iter().next())
267}
268
269/// From the id's random bits; the migration adds the unique index.
270pub struct UniqueCode;
271
272impl Behavior for UniqueCode {
273    fn columns(&self) -> &'static [Field] {
274        const { &[Field::new("code", String::COLUMNS, false)] }
275    }
276
277    fn written(&self, write: &Write<'_>, values: &mut [SqlValue]) {
278        if values[0] == SqlValue::Null {
279            values[0] = SqlValue::Text(code(write.id.as_bytes()));
280        }
281    }
282}
283
284/// The id's last 50 bits, all random in a `UUIDv7`.
285fn code(id: &[u8; 16]) -> String {
286    let random = u64::from_be_bytes(id[8..].try_into().expect("eight bytes"));
287    (0..CODE_LENGTH)
288        .map(|index| {
289            let digit = (random >> (5 * index)) & 0b1_1111;
290            char::from(CODE_ALPHABET[usize::try_from(digit).expect("a five-bit digit")])
291        })
292        .collect()
293}
294
295pub trait Coded: Aggregate {
296    const CODE: Column<Self, String> = Column::new("code");
297
298    fn code(&self) -> &str {
299        stored_text(self, "code")
300    }
301}
302
303impl<A: Uses<UniqueCode>> Coded for A {}
304
305fn stored_text<'a, A: Aggregate>(aggregate: &'a A, column: &str) -> &'a str {
306    match stored(aggregate, column) {
307        Some(SqlValue::Text(text)) => text,
308        _ => "",
309    }
310}
311
312#[cfg(test)]
313mod tests {
314    use super::*;
315
316    #[test]
317    fn a_slug_joins_lowercase_words_by_hyphens() {
318        assert_eq!(slug("  Blue Wool -- Scarf (2026)!"), "blue-wool-scarf-2026");
319        assert_eq!(slug("a/b_c.d,e:f"), "a-b-c-d-e-f");
320        assert_eq!(slug("!?"), "");
321    }
322
323    #[test]
324    fn a_slug_spells_turkish_german_and_accented_latin_letters_in_ascii() {
325        assert_eq!(slug("Öğle Yemeği"), "ogle-yemegi");
326        assert_eq!(
327            slug("İstanbul'da Çay Şöleni ılık"),
328            "istanbul-da-cay-soleni-ilik"
329        );
330        assert_eq!(slug("Große Straße für Äpfel"), "grosse-strasse-fur-apfel");
331        assert_eq!(
332            slug("Crème brûlée à la française"),
333            "creme-brulee-a-la-francaise"
334        );
335        assert_eq!(slug("Cre\u{300}me"), "creme");
336        assert_eq!(
337            slug("Smørrebrød Æble Œuvre Łódź Đak Ñandú Åre"),
338            "smorrebrod-aeble-oeuvre-lodz-dak-nandu-are"
339        );
340    }
341
342    #[test]
343    fn a_slug_spells_cyrillic_in_latin_and_keeps_other_scripts() {
344        assert_eq!(slug("Привет, мир!"), "privet-mir");
345        assert_eq!(slug("Щедрий вечір, Україно"), "shchedriy-vechir-ukrayino");
346        assert_eq!(slug("关于我们 2026"), "关于我们-2026");
347        assert_eq!(slug("Ελλάδα"), "ελλάδα");
348    }
349
350    #[test]
351    fn a_decomposed_title_slugs_as_its_composed_form() {
352        assert_eq!(slug("Мой край"), slug("Мои\u{306} край"));
353        assert_eq!(slug("Ελλάδα"), slug("Ελλα\u{301}δα"));
354    }
355
356    #[test]
357    fn a_slug_keeps_the_marks_of_other_scripts_in_their_word() {
358        assert_eq!(slug("नमस्ते दुनिया"), "नमस्ते-दुनिया");
359        assert_eq!(slug("ไม่ ใช่"), "ไม่-ใช่");
360        assert_eq!(slug("关于 — 我们 🎉 «x»"), "关于-我们-x");
361    }
362
363    #[test]
364    fn a_slug_stops_before_the_word_that_passes_its_length() {
365        assert_eq!(slug(&"word ".repeat(30)).len(), 99);
366        let long = slug(&"ğüş ".repeat(40));
367        assert!(long.len() <= SLUG_LENGTH && long.ends_with("gus"), "{long}");
368        let chinese = slug(&"关".repeat(40));
369        assert_eq!(chinese, "关".repeat(33));
370    }
371
372    #[test]
373    fn a_code_spells_the_random_bits_of_the_id() {
374        let first = code(&[0; 16]);
375        let other = code(&[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1]);
376
377        assert_eq!(first, "0000000000");
378        assert_eq!(other, "1000000000");
379    }
380}