Split SuggestionsOutputUtils::outputSuggestions.
Bug: 13333066 Change-Id: Ie5e513dacdc5502e7263ddf709824bbd6bc6b74a
This commit is contained in:
Keisuke Kuroyanagi
parent
a0ea92f76c
commit
ff1b3947c6
2 changed files with 86 additions and 74 deletions
|
|
@ -17,11 +17,11 @@
|
|||
#include "suggest/core/result/suggestions_output_utils.h"
|
||||
|
||||
#include <algorithm>
|
||||
#include <vector>
|
||||
|
||||
#include "suggest/core/dicnode/dic_node.h"
|
||||
#include "suggest/core/dicnode/dic_node_utils.h"
|
||||
#include "suggest/core/dictionary/binary_dictionary_shortcut_iterator.h"
|
||||
#include "suggest/core/dictionary/dictionary.h"
|
||||
#include "suggest/core/dictionary/error_type_utils.h"
|
||||
#include "suggest/core/policy/scoring.h"
|
||||
#include "suggest/core/result/suggestion_results.h"
|
||||
|
|
@ -31,40 +31,48 @@ namespace latinime {
|
|||
|
||||
const int SuggestionsOutputUtils::MIN_LEN_FOR_MULTI_WORD_AUTOCORRECT = 16;
|
||||
|
||||
// TODO: Split this method.
|
||||
/* static */ void SuggestionsOutputUtils::outputSuggestions(
|
||||
const Scoring *const scoringPolicy, DicTraverseSession *traverseSession,
|
||||
SuggestionResults *const outSuggestionResults) {
|
||||
#if DEBUG_EVALUATE_MOST_PROBABLE_STRING
|
||||
const int terminalSize = 0;
|
||||
#else
|
||||
const int terminalSize = std::min(MAX_RESULTS,
|
||||
static_cast<int>(traverseSession->getDicTraverseCache()->terminalSize()));
|
||||
const int terminalSize = traverseSession->getDicTraverseCache()->terminalSize();
|
||||
#endif
|
||||
DicNode terminals[MAX_RESULTS]; // Avoiding non-POD variable length array
|
||||
|
||||
std::vector<DicNode> terminals(terminalSize);
|
||||
for (int index = terminalSize - 1; index >= 0; --index) {
|
||||
traverseSession->getDicTraverseCache()->popTerminal(&terminals[index]);
|
||||
}
|
||||
|
||||
const float languageWeight = scoringPolicy->getAdjustedLanguageWeight(
|
||||
traverseSession, terminals, terminalSize);
|
||||
traverseSession, terminals.data(), terminalSize);
|
||||
// Force autocorrection for obvious long multi-word suggestions when the top suggestion is
|
||||
// a long multiple words suggestion.
|
||||
// TODO: Implement a smarter auto-commit method for handling multi-word suggestions.
|
||||
const bool forceCommitMultiWords = scoringPolicy->autoCorrectsToMultiWordSuggestionIfTop()
|
||||
&& (traverseSession->getInputSize() >= MIN_LEN_FOR_MULTI_WORD_AUTOCORRECT
|
||||
&& terminals[0].hasMultipleWords());
|
||||
&& !terminals.empty() && terminals.front().hasMultipleWords());
|
||||
// TODO: have partial commit work even with multiple pointers.
|
||||
const bool outputSecondWordFirstLetterInputIndex =
|
||||
traverseSession->isOnlyOnePointerUsed(0 /* pointerId */);
|
||||
const bool boostExactMatches = traverseSession->getDictionaryStructurePolicy()->
|
||||
getHeaderStructurePolicy()->shouldBoostExactMatches();
|
||||
|
||||
int codePoints[MAX_WORD_LENGTH];
|
||||
// Output suggestion results here
|
||||
for (int terminalIndex = 0; terminalIndex < terminalSize; ++terminalIndex) {
|
||||
DicNode *terminalDicNode = &terminals[terminalIndex];
|
||||
for (auto &terminalDicNode : terminals) {
|
||||
outputSuggestionsOfDicNode(scoringPolicy, traverseSession, &terminalDicNode,
|
||||
languageWeight, boostExactMatches, forceCommitMultiWords,
|
||||
outputSecondWordFirstLetterInputIndex, outSuggestionResults);
|
||||
}
|
||||
scoringPolicy->getMostProbableString(traverseSession, languageWeight, outSuggestionResults);
|
||||
}
|
||||
|
||||
/* static */ void SuggestionsOutputUtils::outputSuggestionsOfDicNode(
|
||||
const Scoring *const scoringPolicy, DicTraverseSession *traverseSession,
|
||||
const DicNode *const terminalDicNode, const float languageWeight,
|
||||
const bool boostExactMatches, const bool forceCommitMultiWords,
|
||||
const bool outputSecondWordFirstLetterInputIndex,
|
||||
SuggestionResults *const outSuggestionResults) {
|
||||
if (DEBUG_GEO_FULL) {
|
||||
terminalDicNode->dump("OUT:");
|
||||
}
|
||||
|
|
@ -100,6 +108,7 @@ const int SuggestionsOutputUtils::MIN_LEN_FOR_MULTI_WORD_AUTOCORRECT = 16;
|
|||
|
||||
// Don't output invalid words. However, we still need to submit their shortcuts if any.
|
||||
if (isValidWord) {
|
||||
int codePoints[MAX_WORD_LENGTH];
|
||||
terminalDicNode->outputResult(codePoints);
|
||||
const int indexToPartialCommit = outputSecondWordFirstLetterInputIndex ?
|
||||
terminalDicNode->getSecondWordFirstInputIndex(
|
||||
|
|
@ -111,13 +120,14 @@ const int SuggestionsOutputUtils::MIN_LEN_FOR_MULTI_WORD_AUTOCORRECT = 16;
|
|||
indexToPartialCommit, computeFirstWordConfidence(terminalDicNode));
|
||||
}
|
||||
|
||||
// Output shortcuts.
|
||||
// Shortcut is not supported for multiple words suggestions.
|
||||
// TODO: Check shortcuts during traversal for multiple words suggestions.
|
||||
if (!terminalDicNode->hasMultipleWords()) {
|
||||
BinaryDictionaryShortcutIterator shortcutIt(
|
||||
traverseSession->getDictionaryStructurePolicy()->getShortcutsStructurePolicy(),
|
||||
traverseSession->getDictionaryStructurePolicy()
|
||||
->getShortcutPositionOfPtNode(terminalDicNode->getPtNodePos()));
|
||||
// Shortcut is not supported for multiple words suggestions.
|
||||
// TODO: Check shortcuts during traversal for multiple words suggestions.
|
||||
const bool sameAsTyped = scoringPolicy->sameAsTyped(traverseSession, terminalDicNode);
|
||||
const int shortcutBaseScore = scoringPolicy->doesAutoCorrectValidWord() ?
|
||||
scoringPolicy->calculateFinalScore(compoundDistance,
|
||||
|
|
@ -127,8 +137,6 @@ const int SuggestionsOutputUtils::MIN_LEN_FOR_MULTI_WORD_AUTOCORRECT = 16;
|
|||
outputShortcuts(&shortcutIt, shortcutBaseScore, sameAsTyped, outSuggestionResults);
|
||||
}
|
||||
}
|
||||
scoringPolicy->getMostProbableString(traverseSession, languageWeight, outSuggestionResults);
|
||||
}
|
||||
|
||||
/* static */ int SuggestionsOutputUtils::computeFirstWordConfidence(
|
||||
const DicNode *const terminalDicNode) {
|
||||
|
|
|
|||
|
|
@ -41,11 +41,15 @@ class SuggestionsOutputUtils {
|
|||
// Inputs longer than this will autocorrect if the suggestion is multi-word
|
||||
static const int MIN_LEN_FOR_MULTI_WORD_AUTOCORRECT;
|
||||
|
||||
static int computeFirstWordConfidence(const DicNode *const terminalDicNode);
|
||||
|
||||
static void outputSuggestionsOfDicNode(const Scoring *const scoringPolicy,
|
||||
DicTraverseSession *traverseSession, const DicNode *const terminalDicNode,
|
||||
const float languageWeight, const bool boostExactMatches,
|
||||
const bool forceCommitMultiWords, const bool outputSecondWordFirstLetterInputIndex,
|
||||
SuggestionResults *const outSuggestionResults);
|
||||
static void outputShortcuts(BinaryDictionaryShortcutIterator *const shortcutIt,
|
||||
const int finalScore, const bool sameAsTyped,
|
||||
SuggestionResults *const outSuggestionResults);
|
||||
static int computeFirstWordConfidence(const DicNode *const terminalDicNode);
|
||||
};
|
||||
} // namespace latinime
|
||||
#endif // LATINIME_SUGGESTIONS_OUTPUT_UTILS
|
||||
|
|
|
|||
Loading…
Reference in a new issue