| 45 | // TODO(annaveronika): more efficient apply. |
| 46 | template <typename TStringVector, typename TUnitToTokenId> |
| 47 | static void ApplyImpl( |
| 48 | TStringVector tokens, |
| 49 | TVector<TTokenId>* tokenIds, |
| 50 | const IDictionary* alphabet, |
| 51 | const TUnitToTokenId& unitToTokenId, |
| 52 | EUnknownTokenPolicy unknownTokenPolicy |
| 53 | ) { |
| 54 | tokenIds->clear(); |
| 55 | alphabet->Apply(tokens, tokenIds, unknownTokenPolicy); |
| 56 | |
| 57 | if (tokenIds->size() <= 1) { |
| 58 | return; |
| 59 | } |
| 60 | |
| 61 | TUnitQueue minIds; |
| 62 | TEraseList<TTokenId> tokenIdsList(*tokenIds); |
| 63 | TPositionsMap positions(Max<TTokenId>(), tokenIds->size()); |
| 64 | |
| 65 | for (size_t i = 0; i + 1 < tokenIds->size(); ++i) { |
| 66 | auto unit = TUnit((*tokenIds)[i], (*tokenIds)[i + 1]); |
| 67 | auto maybeTokenId = unitToTokenId(unit); |
| 68 | if (!maybeTokenId) { |
| 69 | continue; |
| 70 | } |
| 71 | auto unitId = *maybeTokenId; |
| 72 | minIds.push({unitId, unit}); |
| 73 | positions[unitId].push_back(i); |
| 74 | } |
| 75 | |
| 76 | while (!minIds.empty()) { |
| 77 | auto bestIdWithUnit = minIds.top(); |
| 78 | auto bestId = bestIdWithUnit.first; |
| 79 | auto bestUnit = bestIdWithUnit.second; |
| 80 | |
| 81 | while (!minIds.empty() && minIds.top() == bestIdWithUnit) { |
| 82 | minIds.pop(); |
| 83 | } |
| 84 | |
| 85 | for (int firstPosition : positions[bestId]) { |
| 86 | if ( |
| 87 | !tokenIdsList.IsValidElement(firstPosition) || |
| 88 | tokenIdsList.IsLastElement(firstPosition) || |
| 89 | tokenIdsList.GetPair(firstPosition) != bestUnit |
| 90 | ) { |
| 91 | continue; |
| 92 | } |
| 93 | tokenIdsList.Erase(tokenIdsList.GetNextPosition(firstPosition)); |
| 94 | tokenIdsList.UpdateToken(firstPosition, bestId); |
| 95 | if (!tokenIdsList.IsFirstElement(firstPosition)) { |
| 96 | int prevPosition = tokenIdsList.GetPrevPosition(firstPosition); |
| 97 | AddPair(prevPosition, unitToTokenId, &tokenIdsList, &positions, &minIds); |
| 98 | } |
| 99 | AddPair(firstPosition, unitToTokenId, &tokenIdsList, &positions, &minIds); |
| 100 | } |
| 101 | } |
| 102 | |
| 103 | (*tokenIds) = tokenIdsList.GetValidElements(); |
| 104 | } |
no test coverage detected