icydb_core/db/identity/
mod.rs

1//! Module: identity
2//! Responsibility: validated entity/index naming and stable byte ordering contracts.
3//! Does not own: schema metadata, relation policy, or storage-layer persistence.
4//! Boundary: all identity construction/decoding for db data/index key domains.
5//!
6//! Invariants:
7//! - Identities are ASCII, non-empty, and bounded by MAX_* limits.
8//! - All construction paths validate invariants.
9//! - Stored byte representation is canonical and order-preserving.
10//! - Ordering semantics follow the length-prefixed stored-byte layout, not
11//!   lexicographic string ordering.
12
13#![expect(clippy::cast_possible_truncation)]
14
15#[cfg(test)]
16mod tests;
17
18use crate::MAX_INDEX_FIELDS;
19use icydb_utils::to_snake_case;
20use std::{
21    cmp::Ordering,
22    fmt::{self, Display},
23};
24
25///
26/// Constants
27///
28
29pub(super) const MAX_ENTITY_NAME_LEN: usize = 64;
30pub(super) const MAX_INDEX_FIELD_NAME_LEN: usize = 64;
31const MAX_INDEX_NAME_PREFIX_LEN: usize = 5;
32const MAX_ENTITY_NAME_SLUG_LEN: usize = (MAX_ENTITY_NAME_LEN * 3) / 2;
33const MAX_INDEX_FIELD_NAME_SLUG_LEN: usize = (MAX_INDEX_FIELD_NAME_LEN * 3) / 2;
34pub(super) const MAX_INDEX_NAME_LEN: usize = MAX_INDEX_NAME_PREFIX_LEN
35    + MAX_ENTITY_NAME_SLUG_LEN
36    + 2
37    + (MAX_INDEX_FIELDS * MAX_INDEX_FIELD_NAME_SLUG_LEN)
38    + (MAX_INDEX_FIELDS - 1);
39const INDEX_NAME_SEGMENT_DELIMITER: u8 = b'|';
40///
41/// IdentityDecodeError
42/// Decode errors (storage / corruption boundary)
43///
44
45#[derive(Debug)]
46pub enum IdentityDecodeError {
47    InvalidSize,
48
49    InvalidLength,
50
51    NonAscii,
52
53    NonZeroPadding,
54
55    Delimiter,
56}
57
58///
59/// EntityNameError
60///
61
62#[derive(Debug)]
63pub enum EntityNameError {
64    Empty,
65
66    TooLong { len: usize, max: usize },
67
68    NonAscii,
69
70    Delimiter,
71}
72
73///
74/// IndexNameError
75///
76
77#[derive(Debug)]
78pub enum IndexNameError {
79    TooManyFields { len: usize, max: usize },
80
81    NoFields,
82
83    FieldEmpty,
84
85    FieldTooLong { field: String, max: usize },
86
87    FieldNonAscii { field: String },
88
89    FieldDelimiter { field: String },
90
91    TooLong { len: usize, max: usize },
92}
93
94///
95/// EntityName
96///
97
98#[derive(Clone, Copy, Eq, Hash, PartialEq)]
99pub struct EntityName {
100    len: u8,
101    bytes: [u8; MAX_ENTITY_NAME_LEN],
102}
103
104impl EntityName {
105    /// Fixed on-disk size in bytes (stable, protocol-level)
106    pub const STORED_SIZE_BYTES: u64 = 1 + (MAX_ENTITY_NAME_LEN as u64);
107
108    /// Fixed in-memory size (for buffers and arrays)
109    pub const STORED_SIZE_USIZE: usize = Self::STORED_SIZE_BYTES as usize;
110
111    /// Validate and construct an entity name from one ASCII string.
112    pub fn try_from_str(name: &str) -> Result<Self, EntityNameError> {
113        // Phase 1: validate user-visible identity constraints.
114        let bytes = name.as_bytes();
115        let len = bytes.len();
116
117        if len == 0 {
118            return Err(EntityNameError::Empty);
119        }
120        if len > MAX_ENTITY_NAME_LEN {
121            return Err(EntityNameError::TooLong {
122                len,
123                max: MAX_ENTITY_NAME_LEN,
124            });
125        }
126        if !bytes.is_ascii() {
127            return Err(EntityNameError::NonAscii);
128        }
129        if bytes.contains(&INDEX_NAME_SEGMENT_DELIMITER) {
130            return Err(EntityNameError::Delimiter);
131        }
132
133        // Phase 2: write into fixed-size canonical storage.
134        let mut out = [0u8; MAX_ENTITY_NAME_LEN];
135        out[..len].copy_from_slice(bytes);
136
137        Ok(Self {
138            len: len as u8,
139            bytes: out,
140        })
141    }
142
143    /// Return the stored entity-name length.
144    #[must_use]
145    pub const fn len(&self) -> usize {
146        self.len as usize
147    }
148
149    /// Return whether the stored entity-name length is zero.
150    #[must_use]
151    pub const fn is_empty(&self) -> bool {
152        self.len() == 0
153    }
154
155    /// Borrow raw identity bytes excluding trailing fixed-buffer padding.
156    #[must_use]
157    pub fn as_bytes(&self) -> &[u8] {
158        &self.bytes[..self.len()]
159    }
160
161    /// Borrow the entity name as UTF-8 text.
162    #[must_use]
163    pub fn as_str(&self) -> &str {
164        // Invariant: construction and decoding enforce ASCII-only storage,
165        // so UTF-8 decoding cannot fail.
166        std::str::from_utf8(self.as_bytes()).expect("EntityName invariant: ASCII-only storage")
167    }
168
169    /// Encode this identity into its fixed-size persisted representation.
170    #[must_use]
171    pub fn to_bytes(self) -> [u8; Self::STORED_SIZE_USIZE] {
172        let mut out = [0u8; Self::STORED_SIZE_USIZE];
173        out[0] = self.len;
174        out[1..].copy_from_slice(&self.bytes);
175        out
176    }
177
178    /// Decode one fixed-size persisted entity identity payload.
179    pub fn from_bytes(bytes: &[u8]) -> Result<Self, IdentityDecodeError> {
180        // Phase 1: validate layout and payload bounds.
181        if bytes.len() != Self::STORED_SIZE_USIZE {
182            return Err(IdentityDecodeError::InvalidSize);
183        }
184
185        let len = bytes[0] as usize;
186        if len == 0 || len > MAX_ENTITY_NAME_LEN {
187            return Err(IdentityDecodeError::InvalidLength);
188        }
189        if !bytes[1..=len].is_ascii() {
190            return Err(IdentityDecodeError::NonAscii);
191        }
192        if bytes[1..=len].contains(&INDEX_NAME_SEGMENT_DELIMITER) {
193            return Err(IdentityDecodeError::Delimiter);
194        }
195        if bytes[1 + len..].iter().any(|&b| b != 0) {
196            return Err(IdentityDecodeError::NonZeroPadding);
197        }
198
199        // Phase 2: materialize canonical fixed-buffer identity storage.
200        let mut name = [0u8; MAX_ENTITY_NAME_LEN];
201        name.copy_from_slice(&bytes[1..]);
202
203        Ok(Self {
204            len: len as u8,
205            bytes: name,
206        })
207    }
208}
209
210impl Ord for EntityName {
211    fn cmp(&self, other: &Self) -> Ordering {
212        // Keep ordering consistent with `to_bytes()` (length prefix first).
213        // This is deterministic protocol/storage ordering, not lexical string order.
214        self.len.cmp(&other.len).then(self.bytes.cmp(&other.bytes))
215    }
216}
217
218impl PartialOrd for EntityName {
219    fn partial_cmp(&self, other: &Self) -> Option<Ordering> {
220        Some(self.cmp(other))
221    }
222}
223
224impl Display for EntityName {
225    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
226        f.write_str(self.as_str())
227    }
228}
229
230impl fmt::Debug for EntityName {
231    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
232        write!(f, "EntityName({})", self.as_str())
233    }
234}
235
236///
237/// IndexName
238///
239
240#[derive(Clone, Copy, Eq, Hash, PartialEq)]
241pub struct IndexName {
242    len: u16,
243    bytes: [u8; MAX_INDEX_NAME_LEN],
244}
245
246impl IndexName {
247    /// Fixed on-disk size in bytes (stable, protocol-level).
248    pub const STORED_SIZE_BYTES: u64 = 2 + (MAX_INDEX_NAME_LEN as u64);
249    /// Fixed in-memory size (for buffers and arrays).
250    pub const STORED_SIZE_USIZE: usize = Self::STORED_SIZE_BYTES as usize;
251
252    /// Validate and construct one non-unique index identity from an entity +
253    /// field list.
254    pub fn try_from_entity_fields(
255        entity: &EntityName,
256        fields: &[&str],
257    ) -> Result<Self, IndexNameError> {
258        Self::try_from_entity_fields_with_prefix("idx", entity, fields)
259    }
260
261    /// Validate and construct one unique index identity from an entity + field
262    /// list.
263    pub fn try_unique_from_entity_fields(
264        entity: &EntityName,
265        fields: &[&str],
266    ) -> Result<Self, IndexNameError> {
267        Self::try_from_entity_fields_with_prefix("uniq", entity, fields)
268    }
269
270    fn try_from_entity_fields_with_prefix(
271        prefix: &str,
272        entity: &EntityName,
273        fields: &[&str],
274    ) -> Result<Self, IndexNameError> {
275        // Phase 1: validate index-field count and per-field identity constraints.
276        if fields.is_empty() {
277            return Err(IndexNameError::NoFields);
278        }
279        if fields.len() > MAX_INDEX_FIELDS {
280            return Err(IndexNameError::TooManyFields {
281                len: fields.len(),
282                max: MAX_INDEX_FIELDS,
283            });
284        }
285
286        let mut field_slugs = Vec::with_capacity(fields.len());
287        for field in fields {
288            let field_len = field.len();
289            if field_len == 0 {
290                return Err(IndexNameError::FieldEmpty);
291            }
292            if field_len > MAX_INDEX_FIELD_NAME_LEN {
293                return Err(IndexNameError::FieldTooLong {
294                    field: (*field).to_string(),
295                    max: MAX_INDEX_FIELD_NAME_LEN,
296                });
297            }
298            if !field.is_ascii() {
299                return Err(IndexNameError::FieldNonAscii {
300                    field: (*field).to_string(),
301                });
302            }
303            if field.as_bytes().contains(&INDEX_NAME_SEGMENT_DELIMITER) {
304                return Err(IndexNameError::FieldDelimiter {
305                    field: (*field).to_string(),
306                });
307            }
308            let slug = index_name_slug(field);
309            if slug.is_empty() {
310                return Err(IndexNameError::FieldEmpty);
311            }
312            field_slugs.push(slug);
313        }
314
315        let entity_slug = index_name_slug(entity.as_str());
316        let total_len = prefix
317            .len()
318            .saturating_add(1)
319            .saturating_add(entity_slug.len())
320            .saturating_add(2)
321            .saturating_add(field_slugs.iter().map(String::len).sum::<usize>())
322            .saturating_add(field_slugs.len().saturating_sub(1));
323        if total_len > MAX_INDEX_NAME_LEN {
324            return Err(IndexNameError::TooLong {
325                len: total_len,
326                max: MAX_INDEX_NAME_LEN,
327            });
328        }
329
330        // Phase 2: encode canonical `idx_entity__field...` bytes into fixed storage.
331        let mut out = [0u8; MAX_INDEX_NAME_LEN];
332        let mut len = 0usize;
333
334        Self::push_bytes(&mut out, &mut len, prefix.as_bytes());
335        Self::push_bytes(&mut out, &mut len, b"_");
336        Self::push_bytes(&mut out, &mut len, entity_slug.as_bytes());
337        Self::push_bytes(&mut out, &mut len, b"__");
338        for (index, field_slug) in field_slugs.iter().enumerate() {
339            if index > 0 {
340                Self::push_bytes(&mut out, &mut len, b"_");
341            }
342            Self::push_bytes(&mut out, &mut len, field_slug.as_bytes());
343        }
344
345        Ok(Self {
346            len: len as u16,
347            bytes: out,
348        })
349    }
350
351    /// Borrow raw index-identity bytes excluding trailing fixed-buffer padding.
352    #[must_use]
353    pub fn as_bytes(&self) -> &[u8] {
354        &self.bytes[..self.len as usize]
355    }
356
357    /// Borrow the index identity as UTF-8 text.
358    #[must_use]
359    pub fn as_str(&self) -> &str {
360        // Invariant: construction and decoding enforce ASCII-only storage,
361        // so UTF-8 decoding cannot fail.
362        std::str::from_utf8(self.as_bytes()).expect("IndexName invariant: ASCII-only storage")
363    }
364
365    /// Encode this identity into its fixed-size persisted representation.
366    #[must_use]
367    pub fn to_bytes(self) -> [u8; Self::STORED_SIZE_USIZE] {
368        let mut out = [0u8; Self::STORED_SIZE_USIZE];
369        out[..2].copy_from_slice(&self.len.to_be_bytes());
370        out[2..].copy_from_slice(&self.bytes);
371        out
372    }
373
374    /// Decode one fixed-size persisted index identity payload.
375    ///
376    /// This validates the canonical fixed-width byte envelope only. It does not
377    /// reconstruct field segments or prove the bytes were produced by
378    /// `try_from_entity_fields`; callers must ensure persisted bytes originate from a
379    /// previously validated `IndexName`.
380    pub fn from_bytes(bytes: &[u8]) -> Result<Self, IdentityDecodeError> {
381        // Phase 1: validate layout and payload bounds.
382        if bytes.len() != Self::STORED_SIZE_USIZE {
383            return Err(IdentityDecodeError::InvalidSize);
384        }
385
386        let len = u16::from_be_bytes([bytes[0], bytes[1]]) as usize;
387        if len == 0 || len > MAX_INDEX_NAME_LEN {
388            return Err(IdentityDecodeError::InvalidLength);
389        }
390        if !bytes[2..2 + len].is_ascii() {
391            return Err(IdentityDecodeError::NonAscii);
392        }
393        if bytes[2 + len..].iter().any(|&b| b != 0) {
394            return Err(IdentityDecodeError::NonZeroPadding);
395        }
396
397        // Phase 2: materialize canonical fixed-buffer identity storage.
398        let mut name = [0u8; MAX_INDEX_NAME_LEN];
399        name.copy_from_slice(&bytes[2..]);
400
401        Ok(Self {
402            len: len as u16,
403            bytes: name,
404        })
405    }
406
407    // Append bytes into the fixed-size identity buffer while tracking write offset.
408    fn push_bytes(out: &mut [u8; MAX_INDEX_NAME_LEN], len: &mut usize, bytes: &[u8]) {
409        let end = *len + bytes.len();
410        out[*len..end].copy_from_slice(bytes);
411        *len = end;
412    }
413}
414
415fn index_name_slug(value: &str) -> String {
416    let separated = value
417        .chars()
418        .map(|ch| if ch.is_ascii_alphanumeric() { ch } else { '_' })
419        .collect::<String>();
420
421    to_snake_case(separated.as_str())
422}
423
424impl Ord for IndexName {
425    fn cmp(&self, other: &Self) -> Ordering {
426        self.to_bytes().cmp(&other.to_bytes())
427    }
428}
429
430impl PartialOrd for IndexName {
431    fn partial_cmp(&self, other: &Self) -> Option<Ordering> {
432        Some(self.cmp(other))
433    }
434}
435
436impl fmt::Debug for IndexName {
437    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
438        write!(f, "IndexName({})", self.as_str())
439    }
440}
441
442impl Display for IndexName {
443    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
444        f.write_str(self.as_str())
445    }
446}
icydb_core/db/identity/mod.rs

icydb_core/db/identity/
mod.rs
