![]() |
Tesseract 0.28.4
|
Canonically ordered pair of ids with cached combined hash — the key type for pair-keyed maps (ACM, collision margins, collision coefficients). More...
#include <types.h>
Public Member Functions | |
| OrderedIdPair (const NameId< Tag > &a, const NameId< Tag > &b) | |
| void | assign (const NameId< Tag > &a, const NameId< Tag > &b) |
| In-place equivalent of *this = OrderedIdPair(a, b): re-canonicalizes and reuses the existing ids' string capacity, so reassigning a long-lived pair performs no allocation once capacity suffices. | |
| const NameId< Tag > & | first () const noexcept |
| const NameId< Tag > & | second () const noexcept |
| std::size_t | hash () const noexcept |
| std::pair< const std::string &, const std::string & > | orderedNameView () const &noexcept |
| References to the two names, lexicographically smaller first. | |
| std::pair< const std::string &, const std::string & > | orderedNameView () const &&=delete |
| bool | operator== (const OrderedIdPair &other) const noexcept |
| Hybrid equality: both values are compared first, the names confirm on matches. | |
| bool | operator!= (const OrderedIdPair &other) const noexcept |
| bool | operator< (const OrderedIdPair &other) const noexcept |
| Lexicographic by (first, second) under NameId ordering; consistent with operator==. | |
Canonically ordered pair of ids with cached combined hash — the key type for pair-keyed maps (ACM, collision margins, collision coefficients).
Stores the two full NameIds (values AND names) so pair equality can confirm names whenever the hash values match — the same hybrid scheme as NameId::operator==. This makes every pair-keyed container watertight against hash collisions: colliding pairs compare unequal and coexist as distinct keys, exactly as in a string-keyed hash map.
Canonicalization orders by (value, then name on value ties), so OrderedIdPair(a, b) == OrderedIdPair(b, a) holds even when a and b collide on the hash value.
|
inlinenodiscardnoexcept |
References to the two names, lexicographically smaller first.
first() and second() order by hash value, which is a runtime property and not alphabetical. Anything writing the pair somewhere its order is observable — an archive, a document, a file — must use this instead, or the output depends on the hashing implementation.
The result references this pair's names and does not own them. Besides the usual rule that it must not outlive the pair, assign() re-canonicalizes in place and reuses the ids' string capacity, so a result taken before an assign() is invalidated by it even though the pair itself is still alive. Copy the names if either applies.