void ImplicitTagRawRulesDeriver::deriveRawRules(const QStringList inputs,
const QStringList translationScripts,
const QString output)
{
_validateInputs(inputs, translationScripts, output);
LOG_INFO(
"Generating implicit tag rules raw file for inputs: " << inputs <<
", translation scripts: " << translationScripts << ". Writing to output: " << output << "...");
LOG_VARD(_sortParallelCount);
LOG_VARD(_skipFiltering);
LOG_VARD(_translateNamesToEnglish);
_init();
long eligibleFeatureCount = 0;
long totalFeatureCount = 0;
for (int i = 0; i < inputs.size(); i++)
{
boost::shared_ptr<ElementInputStream> inputStream =
_getInputStream(inputs.at(i), translationScripts.at(i));
while (inputStream->hasMoreElements())
{
ElementPtr element = inputStream->readNextElement();
LOG_VART(element);
totalFeatureCount++;
assert(_elementCriterion.get());
if (_skipFiltering || _elementCriterion->isSatisfied(element))
{
QStringList names = element->getTags().getNames();
assert(!names.isEmpty());
//old_name/former_name generally indicates that an element formerly went by the name, so
//not really useful here.
if (names.removeAll("old_name") > 0)
{
LOG_VART("Removed old name tag.");
}
if (names.removeAll("former_name") > 0)
{
LOG_VART("Removed former name tag.");
}
assert(!names.isEmpty());
if (_translateNamesToEnglish)
{
names = ImplicitTagUtils::translateNamesToEnglish(names, element->getTags(), _translator);
}
LOG_VART(names);
//get back only the tags that we'd be interested in applying to future elements implicitly
//based on name
const QStringList kvps = _elementCriterion->getEligibleKvps(element->getTags());
assert(!kvps.isEmpty());
if (kvps.isEmpty())
{
throw HootException("Kvps empty.");
}
//parse whole names and token groups
_parseNames(names, kvps);
eligibleFeatureCount++;
if (eligibleFeatureCount % _statusUpdateInterval == 0)
{
PROGRESS_INFO(
"Parsed " << StringUtils::formatLargeNumber(eligibleFeatureCount) <<
" eligible features / " << StringUtils::formatLargeNumber(totalFeatureCount) <<
" total features.");
}
}
}
_inputReader->finalizePartial();
}
_countFile->close();
LOG_INFO(
"Parsed " << StringUtils::formatLargeNumber(eligibleFeatureCount) <<
" eligible features from " << StringUtils::formatLargeNumber(totalFeatureCount) <<
" total features.");
LOG_INFO(
"Wrote " << StringUtils::formatLargeNumber(_countFileLineCtr) << " lines to count file.");
_sortByTagOccurrence(); //sort in descending count order
_removeDuplicatedKeyTypes();
bool tieCountsNeededResolved = false;
if (_duplicatedWordTagKeyCountsToValues.size() > 0)
{
_resolveCountTies();
tieCountsNeededResolved = true;
}
LOG_INFO(
"Extracted " << StringUtils::formatLargeNumber(_wordKeysToCountsValues.size()) <<
" word/tag associations.");
LOG_INFO("Clearing word/tag associations...");
_wordKeysToCountsValues.clear();
if (tieCountsNeededResolved)
{
_sortByWord(_tieResolvedCountFile);
}
else
{
_sortByWord(_dedupedCountFile);
}
}
void ImplicitTagRulesDatabaseDeriver::deriveRulesDatabase(const QString input, const QString output)
{
_validateInputs(input, output);
LOG_INFO(
"Deriving implicit tag rules for input: " << input << ". Writing to output: " <<
output << "...");
LOG_VARD(_minTagOccurrencesPerWord);
LOG_VARD(_minWordLength);
LOG_VARD(_customRules.getWordIgnoreFile());
LOG_VARD(_customRules.getTagIgnoreFile());
LOG_VARD(_customRules.getCustomRuleFile());
LOG_VARD(_useSchemaTagValuesForWordsOnly);
_customRules.init();
LOG_VARD(_customRules.getWordIgnoreList().size());
LOG_VARD(_customRules.getWordIgnoreList());
LOG_VARD(_customRules.getTagIgnoreList().size());
LOG_VARD(_customRules.getTagIgnoreList());
LOG_VARD(_customRules.getCustomRulesList().size());
LOG_VARD(_customRules.getCustomRulesList());
if (_minTagOccurrencesPerWord == 1 && _minWordLength == 1 &&
_customRules.getWordIgnoreList().size() == 0 && _customRules.getTagIgnoreList().size() == 0 &&
_customRules.getCustomRulesList().size() == 0 && !_useSchemaTagValuesForWordsOnly)
{
LOG_INFO("Skipping filtering, as no filtering criteria were specified...");
if (_minTagOccurrencesPerWord >= 2)
{
_removeKvpsBelowOccurrenceThreshold(input, _minTagOccurrencesPerWord);
_writeRules(_thresholdedCountFile->fileName(), output);
}
else
{
LOG_INFO("Skipping count thresholding since threshold = 1...");
_writeRules(input, output);
}
}
else
{
if (_useSchemaTagValuesForWordsOnly)
{
_populateSchemaTagValues();
}
if (_minTagOccurrencesPerWord >= 2)
{
_removeKvpsBelowOccurrenceThreshold(input, _minTagOccurrencesPerWord);
_applyFiltering(_thresholdedCountFile->fileName());
}
else
{
LOG_INFO("Skipping count thresholding since min occurrence threshold = 1...");
_applyFiltering(input);
}
_writeRules(_filteredCountFile->fileName(), output);
}
}
