added a string linking stage to ticcl, this adds extra markup informa…

…tion (t-str/t-correction) using the foliatextcontent tool, this is in turn needed by FLAT for proper visualisation. #62
LanguageMachines · Dec 7, 2020 · e67b292 · e67b292
1 parent d423fce
commit e67b292
Show file tree

Hide file tree

Showing 4 changed files with 57 additions and 6 deletions.
diff --git a/codemeta.json b/codemeta.json
@@ -10,7 +10,7 @@
     "@type": "SoftwareSourceCode",
     "identifier": "piccl",
     "name": "PICCL",
-    "version": "0.9.4",
+    "version": "0.9.5",
     "description": "A set of workflows for corpus building through OCR, post-correction, and normalisation.",
     "license": "https://spdx.org/licenses/GPL-3.0",
 	"url": "https://github.com/LanguageMachines/PICCL",

diff --git a/ticcl.nf b/ticcl.nf
@@ -56,6 +56,7 @@ if (params.containsKey('help')) {
     log.info "  --high INT               skip entries from the anagram file longer than 'high' characters. (default=35)"
     log.info "  --chainclean BOOLINT     enable chain clean or not (1 = on, 0 = off, default)"
     log.info "  --nofoliacorrect         skip the FoLiA correct step"
+    log.info "  --nostringlinking        skip the final string linking step"
     exit 2
 }
 
@@ -508,7 +509,7 @@ if (!params.containsKey('nofoliacorrect')) {
             Correct the input documents using the ranked list, produces final output documents with <str>, using FoLiA-correct
         */
 
-        publishDir params.outputdir, mode: 'copy', overwrite: true //publish the output for the end-user to see (this is the final output)
+        publishDir params.outputdir, mode: 'copy', overwrite: true
         label "multicore"
 
         input:
@@ -522,7 +523,7 @@ if (!params.containsKey('nofoliacorrect')) {
         val virtualenv from params.virtualenv
 
         output:
-        file "*.ticcl.folia.xml" into folia_ticcl_documents
+        file "*.foliacorrect.folia.xml" into foliacorrect_documents
 
         script:
         """
@@ -540,13 +541,16 @@ if (!params.containsKey('nofoliacorrect')) {
         FoLiA-correct --inputclass "${inputclass}" --outputclass "${outputclass}" --nums 10 -e ${extension} -O outputdir/ --unk "${unknownfreqlist}" --punct "${punctuationmap}" --rank "${rankedlist}"  -t ${task.cpus} . || exit 1
 
         cd outputdir
+        echo "output files:"
         ls
 
         #rename files so they have *.ticcl.folia.xml as extension (rather than .ticcl.xml which FoLiA-correct produces)
         for f in *.xml; do
             if [[ \$f != "*.xml" ]]; then
                 if [[ \${f%.ticcl.xml} != \$f ]]; then
-                    newf="\${f%.ticcl.xml}.ticcl.folia.xml"
+                    newf="\${f%.ticcl.xml}.foliacorrect.folia.xml" #old folia-correc
+                elif [[ \${f%.ticcl.folia.xml} != \$f ]]; then
+                    newf="\${f%.ticcl.folia.xml}.foliacorrect.folia.xml" #new folia-correct
                 else
                     newf="\$f"
                 fi
@@ -557,6 +561,53 @@ if (!params.containsKey('nofoliacorrect')) {
         """
     }
 
+    if (!params.containsKey('nostringlinking')) {
+        process linkstrings {
+            /*
+             This invokes a tool that adds text markup information (t-str and t-correction) linking to the substrings. It adds a level of redundancy that is needed for proper visualisation in FLAT.
+            */
+
+            publishDir params.outputdir, mode: 'copy', overwrite: true //publish the output for the end-user to see (this is the final output)
+
+            input:
+            file foliadoc from foliacorrect_documents
+            val virtualenv from params.virtualenv
+
+            output:
+            file "*.ticcl.folia.xml" into folia_ticcl_documents
+
+            script:
+            """
+            #!/bin/bash
+            set +u
+            if [ ! -z "${virtualenv}" ]; then
+                source ${virtualenv}/bin/activate
+            fi
+            set -u
+
+            foliatextcontent -M ${foliadoc} > ${foliadoc.simpleName}.ticcl.folia.xml || exit 1
+            """
+        }
+
+    } else {
+        process nolinkstrings {
+            """Simple file rename step"""
+
+            publishDir params.outputdir, mode: 'copy', overwrite: true //publish the output for the end-user to see (this is the final output)
+
+            input:
+            file foliadoc from foliacorrect_documents
+
+            output:
+            file "*.ticcl.folia.xml" into folia_ticcl_documents
+
+            script:
+            """
+            cp ${foliadoc} ${foliadoc.simpleName}.ticcl.folia.xml || exit 1
+            """
+        }
+    }
+
     //explicitly report the final documents created to stdout
     folia_ticcl_documents.subscribe { println "TICCL output document written to " +  params.outputdir + "/" + it.name }
 }
diff --git a/webservice/picclservice/picclservice.py b/webservice/picclservice/picclservice.py
@@ -48,7 +48,7 @@
 #An informative description for this system (this should be fairly short, about one paragraph, and may not contain HTML)
 SYSTEM_DESCRIPTION = "PICCL offers a workflow for corpus building and builds on a variety of tools. The primary component of PICCL is TICCL; a Text-induced Corpus Clean-up system, which performs spelling correction and OCR post-correction (normalisation of spelling variants etc)."
 
-SYSTEM_VERSION = "0.9.4" #also change in codemeta.json and setup.py
+SYSTEM_VERSION = "0.9.5" #also change in codemeta.json and setup.py
 
 SYSTEM_AUTHOR = "Martin Reynaert, Maarten van Gompel, Ko van der Sloot"
 

diff --git a/webservice/setup.py b/webservice/setup.py
@@ -10,7 +10,7 @@
 
 setup(
     name = "PICCL",
-    version = "0.9.4", #also change in codemeta.json and picclservice.py
+    version = "0.9.5", #also change in codemeta.json and picclservice.py
     author = "Martin Reynaert, Maarten van Gompel",
     author_email = "[email protected]",
     description = ("Webservice for PICCL; a set of workflows for corpus building through OCR, post-correction, modernization of historic language and Natural Language Processing"),