void testCallingSequence1(uima::util::ConsoleUI& rclConsole, const TCHAR* cpszConfigFilename)
/* ----------------------------------------------------------------------- */
{
ErrorInfo errInfo;
uima::TextAnalysisEngine* pEngine = TextAnalysisEngine::createTextAnalysisEngine(cpszConfigFilename, errInfo);
failIfNotTrue(errInfo.getErrorId() == UIMA_ERR_NONE);
failIfNotTrue(pEngine != NULL);
CAS* cas = pEngine->newCAS();
failIfNotTrue(cas != NULL);
uima::UnicodeStringRef us(UnicodeString("a"));
// UnicodeStringRef uRef(us);
rclConsole.formatHeader(_TEXT("testing Engine CallingSequence1"));
cas->setDocumentText(us.getBuffer(), us.length());
cas->getDocumentAnnotation().setLanguage("en");
failIfNotTrue(pEngine->process(*cas) == UIMA_ERR_NONE);
failIfNotTrue(cas->reset() == UIMA_ERR_NONE);
failIfNotTrue(pEngine->destroy() == UIMA_ERR_NONE);
cas->setDocumentText(us.getBuffer(), us.length());
cas->getDocumentAnnotation().setLanguage("en");
failIfNotTrue(pEngine->process(*cas) == UIMA_ERR_ENGINE_INVALID_CALLING_SEQUENCE);
TyErrorId deInitRC = pEngine->destroy();
rclConsole.format("RC of deInit()", deInitRC);
failIfNotTrue(deInitRC == UIMA_ERR_ENGINE_INVALID_CALLING_SEQUENCE);
rclConsole.formatBool(_TEXT("testing Engine CallingSequence1 OK"),
true); //lint !e944: argument for operator '!' always evaluates to False
delete cas;
delete pEngine;
} //lint !e715: cpszConfigFilename (line 99) not referenced
void testCallingSequence3(uima::util::ConsoleUI& rclConsole, const TCHAR* cpszConfigFilename)
/* ----------------------------------------------------------------------- */
{
uima::TextAnalysisEngine* pEngine = NULL;
uima::Language clLanguage(MAIN_DEFAULT_LANG);
const char* clCCSID = MAIN_DEFAULT_CCSID_STR;
TyErrorId utErrorId;
UnicodeString us("a");
UnicodeStringRef uref(us);
rclConsole.formatHeader(_TEXT("testing Engine CallingSequence3"));
ErrorInfo errInfo;
pEngine = TextAnalysisEngine::createTextAnalysisEngine(cpszConfigFilename, errInfo);
failIfNotTrue(errInfo.getErrorId() == UIMA_ERR_NONE);
failIfNotTrue(pEngine != NULL);
CAS* cas = pEngine->newCAS();
failIfNotTrue(cas != NULL);
/* test for NULL ptrs */
UnicodeStringRef uref2(NULL);
cas->setDocumentText(uref2.getBuffer(), uref2.length());
cas->getDocumentAnnotation().setLanguage("en");
failIfNotTrue(pEngine->process(*cas) == UIMA_ERR_NONE);
failIfNotTrue(cas->reset() == UIMA_ERR_NONE);
/* test for subsequent processes */
cas->setDocumentText(uref2.getBuffer(), uref2.length());
cas->getDocumentAnnotation().setLanguage("en");
failIfNotTrue(pEngine->process(*cas) == UIMA_ERR_NONE);
failIfNotTrue(pEngine->process(*cas) == UIMA_ERR_NONE);
utErrorId = pEngine->destroy();
failIfNotTrue(utErrorId == UIMA_ERR_NONE);
delete cas;
delete pEngine;
rclConsole.formatBool(_TEXT("testing Engine CallingSequence3 OK"),
true); //lint !e944: argument for operator '!' always evaluates to False
}
void testCasMultiplier(uima::util::ConsoleUI& rclConsole)
/* ----------------------------------------------------------------------- */
{
rclConsole.info("testCasMultiplier start.");
uima::TextAnalysisEngine* pEngine;
ErrorInfo errInfo;
UnicodeString filename("SimpleTextSegmenter.xml");
UnicodeString fn = ResourceManager::resolveFilename(filename, filename);
pEngine = TextAnalysisEngine::createTextAnalysisEngine(UnicodeStringRef(fn).asUTF8().c_str(), errInfo);
failIfNotTrue(errInfo.getErrorId() == UIMA_ERR_NONE);
failIfNotTrue(pEngine != NULL);
//test operational properties settings
failIfNotTrue(pEngine->getAnalysisEngineMetaData().getOperationalProperties()->getOutputsNewCASes() == true);
failIfNotTrue(pEngine->getAnalysisEngineMetaData().getOperationalProperties()->getModifiesCas() == false);
failIfNotTrue(
pEngine->getAnalysisEngineMetaData().getOperationalProperties()->isMultipleDeploymentAllowed() == true);
CAS* cas = pEngine->newCAS();
cas->setDocumentText(
UnicodeString("This is the first sentence. This is the second sentence. This is the third sentence."));
CASIterator iter = pEngine->processAndOutputNewCASes(*cas);
int num = 0;
while (iter.hasNext()) {
num++;
CAS& seg = iter.next();
failIfNotTrue(seg.getDocumentText().length() > 0);
pEngine->getAnnotatorContext().releaseCAS(seg);
}
failIfNotTrue(num == 3);
delete pEngine;
delete cas;
rclConsole.info("testCasMultiplier finished.");
}
void testProcessTerm(uima::util::ConsoleUI& rclConsole,
uima::TextAnalysisEngine& rclEngine,
///const uima::CCSID & crclCCSID,
const char* crclCCSID,
const uima::Language& crclLanguage,
const TCHAR* cpszInpTerm)
/* ----------------------------------------------------------------------- */
{
TyErrorId utErrorId;
failIfNotTrue(EXISTS(cpszInpTerm));
rclConsole.format(_TEXT("Input term"), cpszInpTerm);
DocBuffer docBuffer;
docBuffer.addDocPart(cpszInpTerm, strlen(cpszInpTerm), crclCCSID);
//? assert(false);
CAS* cas = rclEngine.newCAS();
failIfNotTrue(cas != NULL);
// For terms we always add a term annotation for the whole "document"
/* since we already added a complete doc, we may not add anything else */
cas->setDocumentText(docBuffer.getDocBuffer(), docBuffer.getLength());
cas->getDocumentAnnotation().setLanguage(crclLanguage);
utErrorId = rclEngine.process(*cas);
uimaToolHandleErrorId(rclConsole, utErrorId, rclEngine.getAnnotatorContext().getLogger().getLastErrorAsCStr(),
_TEXT("uima::Engine::processDocument"), gs_lExpectedProcessDocumentRc);
if (utErrorId == UIMA_ERR_NONE && gs_bDoIterTest) {
failIfNotTrue(false);
// iteratorTest(rclConsole, rclEngine);
}
utErrorId = cas->reset();
uimaToolHandleErrorId(rclConsole, utErrorId, rclEngine.getAnnotatorContext().getLogger().getLastErrorAsCStr(),
_TEXT("uima::Engine::resetDocument"));
delete cas;
}
void testProcessDocu(uima::util::ConsoleUI& rclConsole,
uima::TextAnalysisEngine& rclEngine,
const char* crclCCSID,
const uima::Language& crclLanguage)
/* ----------------------------------------------------------------------- */
{
TyErrorId utErrorId;
string clstrInputFileContent;
size_t uiNumOfInputDocs = 0;
uima::DocBuffer docBuffer;
CAS* cas = rclEngine.newCAS();
failIfNotTrue(cas != NULL);
/* iterate through all doc specs on command line */
for (rclConsole.setToFirst(); rclConsole.isValid(); rclConsole.setToNext()) {
////uima::util::Filename clInputFilename(rclConsole.getAsCString());
//replaced with a hard wired data file
UnicodeString filename("tdoc_001_enus_850.asc");
UnicodeString fn = ResourceManager::resolveFilename(filename, filename);
uima::util::Filename clInputFilename(UnicodeStringRef(fn).asUTF8().c_str());
size_t uiSize;
if (!clInputFilename.isExistent()) {
rclConsole.fatal(1, _TEXT("Input file not found"), clInputFilename.getAsCString());
}
if (crclCCSID == NULL) /**** (!crclCCSID.isValid()) ***/
{
rclConsole.fatal(1, _TEXT("Invalid CCSID specified - cannot load document"),
crclCCSID /**crclCCSID.getName() **/);
}
rclConsole.format(_TEXT("Adding Document"), clInputFilename.getAsCString());
uiSize = ftool_ReadFileToString(clInputFilename, clstrInputFileContent);
docBuffer.addDocPart(clstrInputFileContent.data(), uiSize, crclCCSID);
// For real file based documents we only add a term annotation for the
// whole "document" if the appropriate switch is set
if (gs_bDocIsTerm) {
assert(false);
}
UnicodeString ustrInputFileContent(clstrInputFileContent.data(), uiSize, crclCCSID);
/* since we already added a complete doc, we may not add anything else */
/// failIfNotTrue(rclEngine.addDocPart(ustrInputFileContent) == UIMA_ERR_ENGINE_INVALID_CALLING_SEQUENCE);
/// failIfNotTrue(rclEngine.addDoc(ustrInputFileContent) == UIMA_ERR_ENGINE_INVALID_CALLING_SEQUENCE);
cas->setDocumentText(docBuffer.getDocBuffer(), docBuffer.getLength());
cas->getDocumentAnnotation().setLanguage(crclLanguage);
utErrorId = rclEngine.process(*cas);
uimaToolHandleErrorId(rclConsole, utErrorId, rclEngine.getAnnotatorContext().getLogger().getLastErrorAsCStr(),
_TEXT("uima::Engine::processDocument"), gs_lExpectedProcessDocumentRc);
if (utErrorId == UIMA_ERR_NONE && gs_bDoIterTest) {
failIfNotTrue(false);
// iteratorTest(rclConsole, rclEngine);
}
utErrorId = cas->reset();
uimaToolHandleErrorId(rclConsole, utErrorId, rclEngine.getAnnotatorContext().getLogger().getLastErrorAsCStr(),
_TEXT("uima::Engine::resetDocument"));
++uiNumOfInputDocs;
}
if (uiNumOfInputDocs == 0) {
rclConsole.warning(_TEXT("No input file(s) specified"));
}
delete cas;
}
