Further testing shows that the fact that the AEs are non-primitive seems to not be a factor. Eg. when trying to run the AggregatePlaintextFastUMLSProcessor as a piper file with
> // Load a simple token processing pipeline from another pipeline file load DefaultTokenizerPipeline.piper > // adding the individual components of the > AggregatePlaintextFastUMLSProcessor non-primitive AE add TokenizerAnnotator add ContextDependentTokenizerAnnotator //add UmlsLookupAnnotator //add UmlsOverlapLookupAnnotator add StatusAnnotator add NegationAnnotator add ExtractionPrepAnnotator add SentenceDetectorAnnotator add AdjustNounPhraseToIncludeFollowingNP add AdjustNounPhraseToIncludeFollowingPPNP add SimpleSegmentAnnotator add POSTagger add LvgAnnotator add DrugMentionAnnotator add GenericCleartkAnalysisEngine add HistoryCleartkAnalysisEngine add PolarityCleartkAnalysisEngine add SubjectCleartkAnalysisEngine add UncertaintyCleartkAnalysisEngine add ClearNLPDependencyParserAE add ClearNLPSemanticRoleLabelerAE add ConstituencyParserAnnotator > where I just tried to ADD all of the individual AEs that are listed as dependencies of the AggregatePlaintextFastUMLSProcessor (which from quick inspection seem to all be primitive=true), I am still getting similar error such as > log4j: Adding appender named [consoleAppender] to category [root]. 24 Mar 2018 20:15:55 ERROR PiperFileRunner - MESSAGE LOCALIZATION FAILED: > Can't find resource for bundle java.util.PropertyResourceBundle, key No > Analysis Component found for UmlsLookupAnnotator > > At this point I am at a bit of a loss and would appreciate any extra input as to what is going on here. Am I simply misusing something the piper syntax? Misunderstanding how AEs are used in piper files? On Sat, Mar 24, 2018 at 7:05 PM, Reed Villanueva <[email protected]> wrote: > Forgot to add an example of the error message that I am seeing when I try > to use these AEs: > > Here is the error message when trying to use bin/runPiperFile.sh -p > mypipes/AggPlaintextFastPipeline.piper -i testdata --xmiOut > testdata-output --user myuser --pass mypass > where mypipes/AggPlaintextFastPipeline.piper looks like > > >> > load DefaultTokenizerPipeline.piper > > >> add AggregatePlaintextFastUMLSProcessor > > >> >> > Error: > >> .... > > log4j: Setting property [conversionPattern] to [%d{dd MMM yyyy HH:mm:ss} >> %5p %c{1} - %m%n]. > > log4j: Adding appender named [consoleAppender] to category [root]. > > 24 Mar 2018 18:24:37 ERROR PiperFileRunner - MESSAGE LOCALIZATION FAILED: >> Can't find resource for bundle java.util.PropertyResourceBundle, key No >> Analysis Component found for AggregatePlaintextFastUMLSProcessor > > .... > > > > On Sat, Mar 24, 2018 at 6:57 PM, Reed Villanueva <[email protected] > > wrote: > >> Having problem where seem to not be able to ADD analysis engines (AEs) in >> piper files (as described in the docs: https://cwiki.apache.org/confl >> uence/display/CTAKES/Piper+Files). >> >> It seems like the main difference, looking at the first few lines of the >> raw xmi of these AEs (in say google chrome) as compared to existing piper >> files (eg. those in >> $CTAKES_HOME/resources/org/apache/ctakes/examples/pipeline), >> is that the AEs that can be added to a piper file have the <primitive> tag >> set to "true" (while those AEs that throw an error wen added to a piper >> have these tags as "false"). >> >> My question then is: what does this mean? How are these AEs ever to be >> used in a piper file, then? Is there functionality expected to be subsumed >> in an existing AE xmi file in the same directory or something (eg. how >> could I use the desc/ctakes-clinical-pipeline/desc/analysis_engine/Aggre >> gatePlaintextFastUMLSProcessor.xml in a piper file)? Is there >> documentation that could explain this a bit more? Thank you. >> > >
