ctakes 4.0.1 - Install experience
First off, thank you all and congratulations on this major release milestone. I went through the dev install on a clean Ubuntu 17.04 desktop VM and while things went pretty smoothly I had a few comments to add in case others are going through the same process: 1. I received an error message when trying to runCVD due to a missing dependency version in the clinical pipeline pom.xml file * Error message: "The project org.apache.ctakes:ctakes-clinical-pipeline:4.0.1-SNAPSHOT () has 1 error 'build.plugins.plugin[org.codehaus.mojo:exec-maven-plugin].dependencies.dependency.version' for gov.nih.nlm.nls.lvg:lvgdist:jar is missing. @ org.apache.ctakes:ctakes-clinical-pipeline:[unknown-version], /ctakes-clinical-pipeline/pom.xml, line 174, column 20 * Fix: Add "2016.0" for that dependency in pom.xml (line 177) 2. I had to uncheck the box "Use command line client: [svn]" at File>Settings>Version Control>Subversion before I could checkout the project. I don't know Subversion or IntelliJ well enough to know whether this is idiosyncratic to my environment. Consider adding is as Step 2 for IntelliJ install instructions under "Perform a fresh checkout from SVN" right after "1. Start IntelliJ IDEA" in case it is a more global issue. 3. I decided to switch from Eclipse to IntelliJ for this fresh install and wasn't sure where to add my umlsuser and pw in IntelliJ. I ended up finding File>Settings>Build,Execution,Deployment>Build Tools>Maven>Runner>VM Options and put them there: -Dctakes.umlsuser=<> -Dctakes.umlspw=<> Those were the only minor stumbles along the way. Tom
RE: URI is not hierarchical
Hi Dima, Good news: no uri hierarchical error. All that I did was import your ctakes-misc pom as a maven project in intellij. Then I created a run configuration with my sandbox as the working directory, ctakes-misc as the classpath module, and my umls credentials in the environment. All database files were found, the pipeline ran, and I got output: total 2448 drwxrwx---+ 1 Administrators Domain Users 0 Apr 26 14:19 . drwxrwx---+ 1 Administrators Domain Users 0 Apr 26 14:18 .. -rwxrwx---+ 1 Administrators Domain Users 187675 Apr 26 14:19 GenSurg_UmbilicalHernia_1.rtf.xmi -rwxrwx---+ 1 Administrators Domain Users 238984 Apr 26 14:19 OBGYN_Gen_Abscess_1.rtf.xmi -rwxrwx---+ 1 Administrators Domain Users 82308 Apr 26 14:19 OBGYN_HysterectomyAndBSO_1.rtf.xmi -rwxrwx---+ 1 Administrators Domain Users 115983 Apr 26 14:19 OBGYN_IUD_1.rtf.xmi -rwxrwx---+ 1 Administrators Domain Users 35919 Apr 26 14:19 OBGYN_LaborProgressNote_1.rtf.xmi -rwxrwx---+ 1 Administrators Domain Users 112460 Apr 26 14:19 OBGYN_MVAPrego_1.rtf.xmi -rwxrwx---+ 1 Administrators Domain Users 173695 Apr 26 14:19 OBGYN_PROMCheck_1.rtf.xmi -rwxrwx---+ 1 Administrators Domain Users 340626 Apr 26 14:19 Peds_Dysphagia_1.rtf.xmi -rwxrwx---+ 1 Administrators Domain Users 390928 Apr 26 14:19 Peds_FebrileSez_1.rtf.xmi -rwxrwx---+ 1 Administrators Domain Users 115064 Apr 26 14:19 Peds_RoutBirthNote_1.rtf.xmi -rwxrwx---+ 1 Administrators Domain Users 101234 Apr 26 14:19 VascSurg_AAA_Leak_1.rtf.xmi -rwxrwx---+ 1 Administrators Domain Users 98015 Apr 26 14:19 VascSurg_FollowUp_1.rtf.xmi -rwxrwx---+ 1 Administrators Domain Users 218863 Apr 26 14:19 VascSurg_PVD_1.rtf.xmi -rwxrwx---+ 1 Administrators Domain Users 72902 Apr 26 14:19 VascSurg_RO-AAA_1.rtf.xmi -rwxrwx---+ 1 Administrators Domain Users 90825 Apr 26 14:19 VascSurg_RO-DVT.rtf.xmi -rwxrwx---+ 1 Administrators Domain Users 90827 Apr 26 14:19 VascSurg_RO-DVT_1.rtf.xmi Bad news: it looks like the hsqldb port of lvg2008 is not working. Caused by: org.hsqldb.HsqlException: error in script file line: 61 C:\Spiffy\prj_ctakes_401\dev\ctakes-lvg-res\target\classes\org\apache\ctakes\lvg\data\HSqlDb\lvg2008 java.io.IOException: read beyond end of file in statement [SET TABLE PUBLIC.INFLECTION INDEX '3198299 60557 3198299 53129 0 0 0 0 1210464'] Caused by: org.hsqldb.HsqlException: Caused by: java.io.IOException: read beyond end of file At any rate, the code works but I think that something is missing from your configuration. Since it is a personal development environment you are kind of on your own. Good luck, Sean P.S. Get rid of the ctakes-resources-umls2011ab artifact in your pom. It isn't doing anything. -Original Message- From: Dligach, Dmitriy [mailto:ddlig...@luc.edu] Sent: Wednesday, April 26, 2017 12:46 PM To: dev@ctakes.apache.org Subject: Re: URI is not hierarchical I am definitely still seeing the “URI is not hierarchical” issue. Here’s a piece of information that might help you figure out what the problem is: It only happens if the pipeline includes dictionary lookup. For instance, this one fails: https://urldefense.proofpoint.com/v2/url?u=https-3A__github.com_dmitriydligach_ctakes-2Dmisc_blob_master_src_main_java_org_apache_ctakes_pipelines_UmlsLookupPipeline.java=DwIGaQ=qS4goWBT7poplM69zy_3xhKwEW14JZMSdioCoppxeFU=fs67GvlGZstTpyIisCYNYmQCP6r0bcpKGd4f7d4gTao=Cv-_Ac9CLY7wNnflevRAn-a92vTj9x79u08OjKfQ9U4=MKa09-UzlhjAB_1jVyGOpVo1UHWrMDqjWiVAnk0jfPE= But this one succeeds: https://urldefense.proofpoint.com/v2/url?u=https-3A__github.com_dmitriydligach_ctakes-2Dmisc_blob_master_src_main_java_org_apache_ctakes_pipelines_BasicPipeline.java=DwIGaQ=qS4goWBT7poplM69zy_3xhKwEW14JZMSdioCoppxeFU=fs67GvlGZstTpyIisCYNYmQCP6r0bcpKGd4f7d4gTao=Cv-_Ac9CLY7wNnflevRAn-a92vTj9x79u08OjKfQ9U4=U0zGx-soIiYY64nHK_tIAIpeind9vdIrAuDQIxh-4ek= (it’s the same as the first one, but the dictionary lookup part is removed). Dima > On Apr 26, 2017, at 11:37, Finan, Sean> wrote: > > Hi again Dima, > > The piper files are not meant to replace uimafit. Uimafit is great for many > purposes. > > As for that annoying old "URI is not hierarchical" bug, a while back I > checked in a fix that worked for me. Since then I cannot duplicate it. > > Sean > > -Original Message- > From: Dligach, Dmitriy [mailto:ddlig...@luc.edu] > Sent: Wednesday, April 26, 2017 12:18 PM > To: dev@ctakes.apache.org > Subject: Re: URI is not hierarchical > > As I said in my previous email, the piper approach looks very promising. > However many of us probably still have lots of existing uimaFIT pipelines and > it would be nice to be able to run them from command line. > > So, are there any plans to finally fix this old “URI is not hierarchical” > problem? Do we at least know what’s causing it? > > Dima > > > >> On Apr 14, 2017, at 12:14, Finan, Sean
RE: URI is not hierarchical
Hi Dima, I don't know what to say. "Stop using Eclipse"? Just kidding. When I run from command line I use scripts that set my classpath. When I run from Intellij I specify the "working directory: ..." and "use classpath of module: ..." that encompass everything that I need. I understand what you are saying: "dictionary doesn't work". What I am saying is "dictionary doesn't work as you are running it, and I don't know how you are running it. Code isn't everything." Does that make sense? I am going to pull your github code and see if it runs for me. Sean Just in case there is confusion on pipers and uimafit ... you can mix piper files (e.g. default clinical) with uimafit: PiperFileReader r = new PiperFileReader( piperFilePath ); AnalysisEngineDescription aed = r.getBuilder().getAnalysisEngineDesc(); Your github class with the ~70 line method building a duplicate of the clinical pipeline could be reduced to: protected static AggregateBuilder getXMIWritingPreprocessorAggregateBuilder() throws Exception { AggregateBuilder aggregateBuilder = new AggregateBuilder(); aggregateBuilder.add( AnalysisEngineFactory.createEngineDescription( UriToDocumentTextAnnotatorCtakes.class ) ); PiperFileReader reader = new PiperFileReader( clinicalPiperPath ); AnalysisEngineDescription clinical = r.getBuilder().getAnalysisEngineDesc(); aggregateBuilder.add( clinical ); aggregateBuilder.add(AnalysisEngineFactory.createEngineDescription( XMIWriter.class, XMIWriter.PARAM_XMI_DIRECTORY, outputDirectory)); return aggregateBuilder; } Just to really confuse you, your entire custom pipeline can be run with something like: PiperFileReader reader = new PiperFileReader(); PipelineBuilder builder = reader.getBuilder() .reader( UriCollectionReader.class, "files", files ) .add( UriToDocumentTextAnnotatorCtakes.class ); reader.loadPipelineFile( piperFilePath ); builder.writeXmis( outputDirectory ) .run(); -Original Message- From: Dligach, Dmitriy [mailto:ddlig...@luc.edu] Sent: Wednesday, April 26, 2017 12:18 PM To: dev@ctakes.apache.org Subject: Re: URI is not hierarchical As I said in my previous email, the piper approach looks very promising. However many of us probably still have lots of existing uimaFIT pipelines and it would be nice to be able to run them from command line. So, are there any plans to finally fix this old “URI is not hierarchical” problem? Do we at least know what’s causing it? Dima > On Apr 14, 2017, at 12:14, Finan, Sean> wrote: > > Ok, thanks. For your original question: > >> it fails with “URI is not hierarchical” when the dictionary lookup is >> enabled. >> I believe this is an old issue, so are there any plans for fixing it in the >> new release? > > I thought that I had already fixed it. So much for my thorough testing. > > Let me know what happens with the piper approach. > Sean > > > -Original Message- > From: Dligach, Dmitriy [mailto:ddlig...@luc.edu] > Sent: Friday, April 14, 2017 12:47 PM > To: dev@ctakes.apache.org > Subject: Re: URI is not hierarchical > > Hi Sean, > > > > The pipeline I am trying to run is this: > > > > https://urldefense.proofpoint.com/v2/url?u=https-3A__github.com_dmitriydligach_ctakes-2Dmisc_blob_master_src_main_java_org_apache_ctakes_pipelines_UmlsLookupPipeline.java=DwIGaQ=qS4goWBT7poplM69zy_3xhKwEW14JZMSdioCoppxeFU=fs67GvlGZstTpyIisCYNYmQCP6r0bcpKGd4f7d4gTao=xe6RaWm66aSyUeXjU20x8dEc3xfYTHbadlwhgSdrIsw=ixO9vX_WJ7p-dpJc0RdnzRfjwQqbNw4gkLwf0SSpp0I= > > > > > (This is the UmlsLookupPipeline class). > > > > It runs fine in Eclipse but fails when I run from command line. > > > > I will look into the solution you are suggesting (thanks!). > > > > Dima > > > > > > > >> On Apr 14, 2017, at 11:35, Finan, Sean >> wrote: > >> > >> Hi Dima, > >> > >> Where did you get that class? I don't have UmlsLookupPipeline or the >> package org.apache.ctakes.pipelines. > >> > >> If you want to run from command-line I highly recommend that you use the >> PiperFileRunner class in core.pipeline. > >> > >> To run the clinical pipeline use cli parameters: > >> -p DefaultFastPipeline.piper > >> -i {inputDir} > >> --xmiOut {outputDir} > >> --user {umlsUsername} > >> --pass {umlsPassword} > >> > >> If you have the binary installation there is a runClinicalPipeline script in >> bin/ > >> > >> PiperFileRunner can run other piper files and take other parameters > >> # Runs the pipeline in the piper file specified by -p (piperfile) > >> # with any other provided parameters. Standard parameters are: > >> # -i , --inputDir {inputDirectory} > >> # -o , --outputDir {outputDirectory} > >> # -s , --subDir {subDirectory} (for i/o) > >> # --xmiOut {xmiOutputDirectory} (if different from -o) > >> # -l , --lookupXml
Re: URI is not hierarchical
I am definitely still seeing the “URI is not hierarchical” issue. Here’s a piece of information that might help you figure out what the problem is: It only happens if the pipeline includes dictionary lookup. For instance, this one fails: https://github.com/dmitriydligach/ctakes-misc/blob/master/src/main/java/org/apache/ctakes/pipelines/UmlsLookupPipeline.java But this one succeeds: https://github.com/dmitriydligach/ctakes-misc/blob/master/src/main/java/org/apache/ctakes/pipelines/BasicPipeline.java (it’s the same as the first one, but the dictionary lookup part is removed). Dima > On Apr 26, 2017, at 11:37, Finan, Sean> wrote: > > Hi again Dima, > > The piper files are not meant to replace uimafit. Uimafit is great for many > purposes. > > As for that annoying old "URI is not hierarchical" bug, a while back I > checked in a fix that worked for me. Since then I cannot duplicate it. > > Sean > > -Original Message- > From: Dligach, Dmitriy [mailto:ddlig...@luc.edu] > Sent: Wednesday, April 26, 2017 12:18 PM > To: dev@ctakes.apache.org > Subject: Re: URI is not hierarchical > > As I said in my previous email, the piper approach looks very promising. > However many of us probably still have lots of existing uimaFIT pipelines and > it would be nice to be able to run them from command line. > > So, are there any plans to finally fix this old “URI is not hierarchical” > problem? Do we at least know what’s causing it? > > Dima > > > >> On Apr 14, 2017, at 12:14, Finan, Sean >> wrote: >> >> Ok, thanks. For your original question: >> >>> it fails with “URI is not hierarchical” when the dictionary lookup is >>> enabled. >>> I believe this is an old issue, so are there any plans for fixing it in the >>> new release? >> >> I thought that I had already fixed it. So much for my thorough testing. >> >> Let me know what happens with the piper approach. >> Sean >> >> >> -Original Message- >> From: Dligach, Dmitriy [mailto:ddlig...@luc.edu] >> Sent: Friday, April 14, 2017 12:47 PM >> To: dev@ctakes.apache.org >> Subject: Re: URI is not hierarchical >> >> Hi Sean, >> >> >> >> The pipeline I am trying to run is this: >> >> >> >> https://urldefense.proofpoint.com/v2/url?u=https-3A__github.com_dmitriydligach_ctakes-2Dmisc_blob_master_src_main_java_org_apache_ctakes_pipelines_UmlsLookupPipeline.java=DwIGaQ=qS4goWBT7poplM69zy_3xhKwEW14JZMSdioCoppxeFU=fs67GvlGZstTpyIisCYNYmQCP6r0bcpKGd4f7d4gTao=xe6RaWm66aSyUeXjU20x8dEc3xfYTHbadlwhgSdrIsw=ixO9vX_WJ7p-dpJc0RdnzRfjwQqbNw4gkLwf0SSpp0I= >> >> >> >> >> (This is the UmlsLookupPipeline class). >> >> >> >> It runs fine in Eclipse but fails when I run from command line. >> >> >> >> I will look into the solution you are suggesting (thanks!). >> >> >> >> Dima >> >> >> >> >> >> >> >>> On Apr 14, 2017, at 11:35, Finan, Sean >>> wrote: >> >>> >> >>> Hi Dima, >> >>> >> >>> Where did you get that class? I don't have UmlsLookupPipeline or the >>> package org.apache.ctakes.pipelines. >> >>> >> >>> If you want to run from command-line I highly recommend that you use the >>> PiperFileRunner class in core.pipeline. >> >>> >> >>> To run the clinical pipeline use cli parameters: >> >>> -p DefaultFastPipeline.piper >> >>> -i {inputDir} >> >>> --xmiOut {outputDir} >> >>> --user {umlsUsername} >> >>> --pass {umlsPassword} >> >>> >> >>> If you have the binary installation there is a runClinicalPipeline script >>> in bin/ >> >>> >> >>> PiperFileRunner can run other piper files and take other parameters >> >>> # Runs the pipeline in the piper file specified by -p (piperfile) >> >>> # with any other provided parameters. Standard parameters are: >> >>> # -i , --inputDir {inputDirectory} >> >>> # -o , --outputDir {outputDirectory} >> >>> # -s , --subDir {subDirectory} (for i/o) >> >>> # --xmiOut {xmiOutputDirectory} (if different from -o) >> >>> # -l , --lookupXml {dictionaryConfigFile} (fast only) >> >>> # --user {umlsUsername} >> >>> # --pass {umlsPassword} >> >>> # -? , --help >> >>> # >> >>> # Other parameters may be declared in the piper file using the cli >>> command: >> >>> # cli {parameterName}={singleCharacter} >> >>> # For instance, for declaration of ParagraphAnnotator path to regex file >>> optional parameter PARAGRAPH_TYPES_PATH, >> >>> # in the custom piper file add the line: >> >>> # cli PARAGRAPH_TYPES_PATH=t >> >>> # and when executing this script use: >> >>> # runPiperFile -p path/to/my/custom.piper -t path/to/my/custom.bsv ... >> >>> >> >>> >> >>> The above is a snippet from the runPiperFile script in the bin/ directory. >> >>> >> >>> I am in the process of writing documentation on piper files in the wiki. >> >>> >> >>>
Re: URI is not hierarchical
Sean, thanks for getting back to me on this. I am now trying to run PiperFileRunner in Eclipse (ultimately I want to run it from command line), so I believe the working directory now is ctakes-core. I am specifying the full path to DefaultFastPipeline.piper because I couldn’t get it to work any other way. Dima > On Apr 26, 2017, at 11:27, Finan, Sean> wrote: > > Hi Dima, > > The error messages is telling you that ContextDependentTokenizerAnnotator is > not found. That is the first ae outside of core. It is in > ctakes-contexttokenizer. > > It also looks like you are specifying a full path to > DefaultFastPipeline.piper. > > So I have to ask: what is your working directory and what is your classpath? > > Thanks, > Sean > > -Original Message- > From: Dligach, Dmitriy [mailto:ddlig...@luc.edu] > Sent: Wednesday, April 26, 2017 12:16 PM > To: dev@ctakes.apache.org > Subject: Re: URI is not hierarchical > > Hi Sean, > > > > Thanks again for providing this information — the piper approach looks very > promising. > > > > So I gave it a try, but it didn’t quite work. As you suggested, I am trying > to run the PiperFileRunner class in core.pipelines. I give it the following > parameters: > > > > -p > /Users/Dima/Loyola/Workspaces/cTakes/ctakes/ctakes-clinical-pipeline-res/src/main/resources/org/apache/ctakes/clinical/pipeline/DefaultFastPipeline.piper > > -i > /Users/Dima/Loyola/Workspaces/cTakes/ctakes/ctakes-examples-res/src/main/resources/org/apache/ctakes/examples/notes/rtf/ > > --xmiOut /Users/Dima/Temp/ > > --user > > --pass > > > > I get this error: > > > > 26 Apr 2017 11:11:40 ERROR PiperFileRunner - MESSAGE LOCALIZATION FAILED: > Can't find resource for bundle java.util.PropertyResourceBundle, key No > Analysis Component found for ContextDependentTokenizerAnnotator > > > > Any thoughts? > > > > Best, > > > > Dima > > > > > > > >> On Apr 14, 2017, at 11:35, Finan, Sean >> wrote: > >> > >> Hi Dima, > >> > >> Where did you get that class? I don't have UmlsLookupPipeline or the >> package org.apache.ctakes.pipelines. > >> > >> If you want to run from command-line I highly recommend that you use the >> PiperFileRunner class in core.pipeline. > >> > >> To run the clinical pipeline use cli parameters: > >> -p DefaultFastPipeline.piper > >> -i {inputDir} > >> --xmiOut {outputDir} > >> --user {umlsUsername} > >> --pass {umlsPassword} > >> > >> If you have the binary installation there is a runClinicalPipeline script in >> bin/ > >> > >> PiperFileRunner can run other piper files and take other parameters > >> # Runs the pipeline in the piper file specified by -p (piperfile) > >> # with any other provided parameters. Standard parameters are: > >> # -i , --inputDir {inputDirectory} > >> # -o , --outputDir {outputDirectory} > >> # -s , --subDir {subDirectory} (for i/o) > >> # --xmiOut {xmiOutputDirectory} (if different from -o) > >> # -l , --lookupXml {dictionaryConfigFile} (fast only) > >> # --user {umlsUsername} > >> # --pass {umlsPassword} > >> # -? , --help > >> # > >> # Other parameters may be declared in the piper file using the cli command: > >> # cli {parameterName}={singleCharacter} > >> # For instance, for declaration of ParagraphAnnotator path to regex file >> optional parameter PARAGRAPH_TYPES_PATH, > >> # in the custom piper file add the line: > >> # cli PARAGRAPH_TYPES_PATH=t > >> # and when executing this script use: > >> # runPiperFile -p path/to/my/custom.piper -t path/to/my/custom.bsv ... > >> > >> > >> The above is a snippet from the runPiperFile script in the bin/ directory. > >> > >> I am in the process of writing documentation on piper files in the wiki. > >> > >> https://urldefense.proofpoint.com/v2/url?u=https-3A__cwiki.apache.org_confluence_display_CTAKES_Piper-2BFiles=DwIGaQ=qS4goWBT7poplM69zy_3xhKwEW14JZMSdioCoppxeFU=fs67GvlGZstTpyIisCYNYmQCP6r0bcpKGd4f7d4gTao=QvJHfwMRwE-eH8zLpe6-qM3SFrDndefi0oRgnOIDowI=jfzeOzImM6Wyvbi7yAj5D6CfSaqoIspmc0NCbrt4Fcs= >> > >> > >> > >> -Original Message- > >> From: Dligach, Dmitriy [mailto:ddlig...@luc.edu] > >> Sent: Friday, April 14, 2017 12:17 PM > >> To: cTAKES Developer list > >> Subject: URI is not hierarchical > >> > >> Dear cTAKES developers, > >> > >> > >> > >> I am trying to run a simple pipeline that involves dictionary lookup: > >> > >> > >> > >> https://urldefense.proofpoint.com/v2/url?u=https-3A__github.com_dmitriydligach_ctakes-2Dmisc_blob_master_src_main_java_org_apache_ctakes_pipelines_UmlsLookupPipeline.java=DwIGaQ=qS4goWBT7poplM69zy_3xhKwEW14JZMSdioCoppxeFU=fs67GvlGZstTpyIisCYNYmQCP6r0bcpKGd4f7d4gTao=-N-wYwlTFXbedV0kkcf_qgdQj3HYIeLCeKr9Q303b0Q=NWoDwPNChJkDhFseM3j8Gi7KNDdVqzi1BFt0onalf9I= >> > >> > >> > >> > >> from command line as
RE: URI is not hierarchical
Hi again Dima, The piper files are not meant to replace uimafit. Uimafit is great for many purposes. As for that annoying old "URI is not hierarchical" bug, a while back I checked in a fix that worked for me. Since then I cannot duplicate it. Sean -Original Message- From: Dligach, Dmitriy [mailto:ddlig...@luc.edu] Sent: Wednesday, April 26, 2017 12:18 PM To: dev@ctakes.apache.org Subject: Re: URI is not hierarchical As I said in my previous email, the piper approach looks very promising. However many of us probably still have lots of existing uimaFIT pipelines and it would be nice to be able to run them from command line. So, are there any plans to finally fix this old “URI is not hierarchical” problem? Do we at least know what’s causing it? Dima > On Apr 14, 2017, at 12:14, Finan, Sean> wrote: > > Ok, thanks. For your original question: > >> it fails with “URI is not hierarchical” when the dictionary lookup is >> enabled. >> I believe this is an old issue, so are there any plans for fixing it in the >> new release? > > I thought that I had already fixed it. So much for my thorough testing. > > Let me know what happens with the piper approach. > Sean > > > -Original Message- > From: Dligach, Dmitriy [mailto:ddlig...@luc.edu] > Sent: Friday, April 14, 2017 12:47 PM > To: dev@ctakes.apache.org > Subject: Re: URI is not hierarchical > > Hi Sean, > > > > The pipeline I am trying to run is this: > > > > https://urldefense.proofpoint.com/v2/url?u=https-3A__github.com_dmitriydligach_ctakes-2Dmisc_blob_master_src_main_java_org_apache_ctakes_pipelines_UmlsLookupPipeline.java=DwIGaQ=qS4goWBT7poplM69zy_3xhKwEW14JZMSdioCoppxeFU=fs67GvlGZstTpyIisCYNYmQCP6r0bcpKGd4f7d4gTao=xe6RaWm66aSyUeXjU20x8dEc3xfYTHbadlwhgSdrIsw=ixO9vX_WJ7p-dpJc0RdnzRfjwQqbNw4gkLwf0SSpp0I= > > > > > (This is the UmlsLookupPipeline class). > > > > It runs fine in Eclipse but fails when I run from command line. > > > > I will look into the solution you are suggesting (thanks!). > > > > Dima > > > > > > > >> On Apr 14, 2017, at 11:35, Finan, Sean >> wrote: > >> > >> Hi Dima, > >> > >> Where did you get that class? I don't have UmlsLookupPipeline or the >> package org.apache.ctakes.pipelines. > >> > >> If you want to run from command-line I highly recommend that you use the >> PiperFileRunner class in core.pipeline. > >> > >> To run the clinical pipeline use cli parameters: > >> -p DefaultFastPipeline.piper > >> -i {inputDir} > >> --xmiOut {outputDir} > >> --user {umlsUsername} > >> --pass {umlsPassword} > >> > >> If you have the binary installation there is a runClinicalPipeline script in >> bin/ > >> > >> PiperFileRunner can run other piper files and take other parameters > >> # Runs the pipeline in the piper file specified by -p (piperfile) > >> # with any other provided parameters. Standard parameters are: > >> # -i , --inputDir {inputDirectory} > >> # -o , --outputDir {outputDirectory} > >> # -s , --subDir {subDirectory} (for i/o) > >> # --xmiOut {xmiOutputDirectory} (if different from -o) > >> # -l , --lookupXml {dictionaryConfigFile} (fast only) > >> # --user {umlsUsername} > >> # --pass {umlsPassword} > >> # -? , --help > >> # > >> # Other parameters may be declared in the piper file using the cli command: > >> # cli {parameterName}={singleCharacter} > >> # For instance, for declaration of ParagraphAnnotator path to regex file >> optional parameter PARAGRAPH_TYPES_PATH, > >> # in the custom piper file add the line: > >> # cli PARAGRAPH_TYPES_PATH=t > >> # and when executing this script use: > >> # runPiperFile -p path/to/my/custom.piper -t path/to/my/custom.bsv ... > >> > >> > >> The above is a snippet from the runPiperFile script in the bin/ directory. > >> > >> I am in the process of writing documentation on piper files in the wiki. > >> > >> https://urldefense.proofpoint.com/v2/url?u=https-3A__cwiki.apache.org_confluence_display_CTAKES_Piper-2BFiles=DwIGaQ=qS4goWBT7poplM69zy_3xhKwEW14JZMSdioCoppxeFU=fs67GvlGZstTpyIisCYNYmQCP6r0bcpKGd4f7d4gTao=xe6RaWm66aSyUeXjU20x8dEc3xfYTHbadlwhgSdrIsw=pNVLJSOBMk5orJZ4Oy4ZwtpkPp0tU48-g1qxtpIBucw= >> > >> > >> > >> -Original Message- > >> From: Dligach, Dmitriy [mailto:ddlig...@luc.edu] > >> Sent: Friday, April 14, 2017 12:17 PM > >> To: cTAKES Developer list > >> Subject: URI is not hierarchical > >> > >> Dear cTAKES developers, > >> > >> > >> > >> I am trying to run a simple pipeline that involves dictionary lookup: > >> > >> > >> > >>
Re: URI is not hierarchical
As I said in my previous email, the piper approach looks very promising. However many of us probably still have lots of existing uimaFIT pipelines and it would be nice to be able to run them from command line. So, are there any plans to finally fix this old “URI is not hierarchical” problem? Do we at least know what’s causing it? Dima > On Apr 14, 2017, at 12:14, Finan, Sean> wrote: > > Ok, thanks. For your original question: > >> it fails with “URI is not hierarchical” when the dictionary lookup is >> enabled. >> I believe this is an old issue, so are there any plans for fixing it in the >> new release? > > I thought that I had already fixed it. So much for my thorough testing. > > Let me know what happens with the piper approach. > Sean > > > -Original Message- > From: Dligach, Dmitriy [mailto:ddlig...@luc.edu] > Sent: Friday, April 14, 2017 12:47 PM > To: dev@ctakes.apache.org > Subject: Re: URI is not hierarchical > > Hi Sean, > > > > The pipeline I am trying to run is this: > > > > https://urldefense.proofpoint.com/v2/url?u=https-3A__github.com_dmitriydligach_ctakes-2Dmisc_blob_master_src_main_java_org_apache_ctakes_pipelines_UmlsLookupPipeline.java=DwIGaQ=qS4goWBT7poplM69zy_3xhKwEW14JZMSdioCoppxeFU=fs67GvlGZstTpyIisCYNYmQCP6r0bcpKGd4f7d4gTao=xe6RaWm66aSyUeXjU20x8dEc3xfYTHbadlwhgSdrIsw=ixO9vX_WJ7p-dpJc0RdnzRfjwQqbNw4gkLwf0SSpp0I= > > > > > (This is the UmlsLookupPipeline class). > > > > It runs fine in Eclipse but fails when I run from command line. > > > > I will look into the solution you are suggesting (thanks!). > > > > Dima > > > > > > > >> On Apr 14, 2017, at 11:35, Finan, Sean >> wrote: > >> > >> Hi Dima, > >> > >> Where did you get that class? I don't have UmlsLookupPipeline or the >> package org.apache.ctakes.pipelines. > >> > >> If you want to run from command-line I highly recommend that you use the >> PiperFileRunner class in core.pipeline. > >> > >> To run the clinical pipeline use cli parameters: > >> -p DefaultFastPipeline.piper > >> -i {inputDir} > >> --xmiOut {outputDir} > >> --user {umlsUsername} > >> --pass {umlsPassword} > >> > >> If you have the binary installation there is a runClinicalPipeline script in >> bin/ > >> > >> PiperFileRunner can run other piper files and take other parameters > >> # Runs the pipeline in the piper file specified by -p (piperfile) > >> # with any other provided parameters. Standard parameters are: > >> # -i , --inputDir {inputDirectory} > >> # -o , --outputDir {outputDirectory} > >> # -s , --subDir {subDirectory} (for i/o) > >> # --xmiOut {xmiOutputDirectory} (if different from -o) > >> # -l , --lookupXml {dictionaryConfigFile} (fast only) > >> # --user {umlsUsername} > >> # --pass {umlsPassword} > >> # -? , --help > >> # > >> # Other parameters may be declared in the piper file using the cli command: > >> # cli {parameterName}={singleCharacter} > >> # For instance, for declaration of ParagraphAnnotator path to regex file >> optional parameter PARAGRAPH_TYPES_PATH, > >> # in the custom piper file add the line: > >> # cli PARAGRAPH_TYPES_PATH=t > >> # and when executing this script use: > >> # runPiperFile -p path/to/my/custom.piper -t path/to/my/custom.bsv ... > >> > >> > >> The above is a snippet from the runPiperFile script in the bin/ directory. > >> > >> I am in the process of writing documentation on piper files in the wiki. > >> > >> https://urldefense.proofpoint.com/v2/url?u=https-3A__cwiki.apache.org_confluence_display_CTAKES_Piper-2BFiles=DwIGaQ=qS4goWBT7poplM69zy_3xhKwEW14JZMSdioCoppxeFU=fs67GvlGZstTpyIisCYNYmQCP6r0bcpKGd4f7d4gTao=xe6RaWm66aSyUeXjU20x8dEc3xfYTHbadlwhgSdrIsw=pNVLJSOBMk5orJZ4Oy4ZwtpkPp0tU48-g1qxtpIBucw= >> > >> > >> > >> -Original Message- > >> From: Dligach, Dmitriy [mailto:ddlig...@luc.edu] > >> Sent: Friday, April 14, 2017 12:17 PM > >> To: cTAKES Developer list > >> Subject: URI is not hierarchical > >> > >> Dear cTAKES developers, > >> > >> > >> > >> I am trying to run a simple pipeline that involves dictionary lookup: > >> > >> > >> > >> https://urldefense.proofpoint.com/v2/url?u=https-3A__github.com_dmitriydligach_ctakes-2Dmisc_blob_master_src_main_java_org_apache_ctakes_pipelines_UmlsLookupPipeline.java=DwIGaQ=qS4goWBT7poplM69zy_3xhKwEW14JZMSdioCoppxeFU=fs67GvlGZstTpyIisCYNYmQCP6r0bcpKGd4f7d4gTao=-N-wYwlTFXbedV0kkcf_qgdQj3HYIeLCeKr9Q303b0Q=NWoDwPNChJkDhFseM3j8Gi7KNDdVqzi1BFt0onalf9I= >> > >> > >> > >> > >> from command line as follows: > >> > >> > >> > >> mvn exec:java >> -Dexec.mainClass=“org.apache.ctakes.pipelines.UmlsLookupPipeline” > >> > >> > >> > >> It runs fine if the dictionary lookup related fragmented is commented out, >> but it fails with “URI is not hierarchical” when the dictionary lookup is >>
Re: URI is not hierarchical
Hi Sean, Thanks again for providing this information — the piper approach looks very promising. So I gave it a try, but it didn’t quite work. As you suggested, I am trying to run the PiperFileRunner class in core.pipelines. I give it the following parameters: -p /Users/Dima/Loyola/Workspaces/cTakes/ctakes/ctakes-clinical-pipeline-res/src/main/resources/org/apache/ctakes/clinical/pipeline/DefaultFastPipeline.piper -i /Users/Dima/Loyola/Workspaces/cTakes/ctakes/ctakes-examples-res/src/main/resources/org/apache/ctakes/examples/notes/rtf/ --xmiOut /Users/Dima/Temp/ --user --pass I get this error: 26 Apr 2017 11:11:40 ERROR PiperFileRunner - MESSAGE LOCALIZATION FAILED: Can't find resource for bundle java.util.PropertyResourceBundle, key No Analysis Component found for ContextDependentTokenizerAnnotator Any thoughts? Best, Dima > On Apr 14, 2017, at 11:35, Finan, Sean> wrote: > > Hi Dima, > > Where did you get that class? I don't have UmlsLookupPipeline or the package > org.apache.ctakes.pipelines. > > If you want to run from command-line I highly recommend that you use the > PiperFileRunner class in core.pipeline. > > To run the clinical pipeline use cli parameters: > -p DefaultFastPipeline.piper > -i {inputDir} > --xmiOut {outputDir} > --user {umlsUsername} > --pass {umlsPassword} > > If you have the binary installation there is a runClinicalPipeline script in > bin/ > > PiperFileRunner can run other piper files and take other parameters > # Runs the pipeline in the piper file specified by -p (piperfile) > # with any other provided parameters. Standard parameters are: > # -i , --inputDir {inputDirectory} > # -o , --outputDir {outputDirectory} > # -s , --subDir {subDirectory} (for i/o) > # --xmiOut {xmiOutputDirectory} (if different from -o) > # -l , --lookupXml {dictionaryConfigFile} (fast only) > # --user {umlsUsername} > # --pass {umlsPassword} > # -? , --help > # > # Other parameters may be declared in the piper file using the cli command: > # cli {parameterName}={singleCharacter} > # For instance, for declaration of ParagraphAnnotator path to regex file > optional parameter PARAGRAPH_TYPES_PATH, > # in the custom piper file add the line: > # cli PARAGRAPH_TYPES_PATH=t > # and when executing this script use: > # runPiperFile -p path/to/my/custom.piper -t path/to/my/custom.bsv ... > > > The above is a snippet from the runPiperFile script in the bin/ directory. > > I am in the process of writing documentation on piper files in the wiki. > > https://cwiki.apache.org/confluence/display/CTAKES/Piper+Files > > > -Original Message- > From: Dligach, Dmitriy [mailto:ddlig...@luc.edu] > Sent: Friday, April 14, 2017 12:17 PM > To: cTAKES Developer list > Subject: URI is not hierarchical > > Dear cTAKES developers, > > > > I am trying to run a simple pipeline that involves dictionary lookup: > > > > https://urldefense.proofpoint.com/v2/url?u=https-3A__github.com_dmitriydligach_ctakes-2Dmisc_blob_master_src_main_java_org_apache_ctakes_pipelines_UmlsLookupPipeline.java=DwIGaQ=qS4goWBT7poplM69zy_3xhKwEW14JZMSdioCoppxeFU=fs67GvlGZstTpyIisCYNYmQCP6r0bcpKGd4f7d4gTao=-N-wYwlTFXbedV0kkcf_qgdQj3HYIeLCeKr9Q303b0Q=NWoDwPNChJkDhFseM3j8Gi7KNDdVqzi1BFt0onalf9I= > > > > > from command line as follows: > > > > mvn exec:java > -Dexec.mainClass=“org.apache.ctakes.pipelines.UmlsLookupPipeline” > > > > It runs fine if the dictionary lookup related fragmented is commented out, > but it fails with “URI is not hierarchical” when the dictionary lookup is > enabled. > > > > I believe this is an old issue, so are there any plans for fixing it in the > new release? In the meantime, are there any workarounds? > > > > Many thanks! > > > > The full error is below. > > > > Dima > > > > > > > > 14 Apr 2017 11:04:24 INFO LvgAnnotator - URL for lvg.properties > =file:/home/dima/.m2/repository/net/sourceforge/ctakesresources/ctakes-resources-lvg2008/4.0.0/ctakes-resources-lvg2008-4.0.0.jar!/org/apache/ctakes/lvg/data/config/lvg.properties > > 14 Apr 2017 11:04:24 INFO SentenceDetector - Sentence detector model file: > org/apache/ctakes/core/sentdetect/sd-med-model.zip > > 14 Apr 2017 11:04:24 INFO TokenizerAnnotatorPTB - Initializing > org.apache.ctakes.core.ae.TokenizerAnnotatorPTB > > 14 Apr 2017 11:04:24 INFO ContextDependentTokenizerAnnotator - Finite state > machines loaded. > > 14 Apr 2017 11:04:24 INFO POSTagger - POS tagger model file: > org/apache/ctakes/postagger/models/mayo-pos.zip > > 14 Apr 2017 11:04:24 INFO Chunker - Chunker model file: > /home/dima/cTakes/trunk/ctakes-chunker-res/src/main/resources/org/apache/ctakes/chunker/models/chunker-model.zip > > 14 Apr 2017 11:04:26 INFO AbstractJCasTermAnnotator - Using dictionary > lookup window type: org.apache.ctakes.typesystem.type.textspan.Sentence > > 14 Apr 2017 11:04:26 INFO
Re: Demos menu option on cTAKES homepage
Hi James, The demos were being upgraded to use 4.0.0 last night. They should be up and running now. Let me know if you encounter any issues. --Pei On Tue, Apr 25, 2017 at 10:59 PM, James Masanzwrote: > Pei and others with access to update http://healthnlp.github.io/examples/ > > Following the Get Started -> Demo menu on http://ctakes.apache.org/ leads > to a page with demos that aren't currently working. > Will you have a chance to fix those soon or should the Demo menu be removed > until they get fixed? > > -- James