diff --git a/.gitattributes b/.gitattributes index 014db9806979d9b624ba970fbc788bd20bfa233a..a4c86d29f0d894cc3bb02a586d30494182d369b7 100644 --- a/.gitattributes +++ b/.gitattributes @@ -278,3 +278,4 @@ data/AB/1906/05_annotatedxml.tar.gz filter=lfs diff=lfs merge=lfs -text data/AB/1907/05_annotatedxml.tar.gz filter=lfs diff=lfs merge=lfs -text data/AB/1908/05_annotatedxml.tar.gz filter=lfs diff=lfs merge=lfs -text data/AB/1909/05_annotatedxml.tar.gz filter=lfs diff=lfs merge=lfs -text +data/AB/1910/05_annotatedxml.tar.gz filter=lfs diff=lfs merge=lfs -text diff --git a/.renku/workflow/74545142cd9e4813ab933fc9b5941dfd_python.cwl b/.renku/workflow/74545142cd9e4813ab933fc9b5941dfd_python.cwl new file mode 100644 index 0000000000000000000000000000000000000000..004fa4709e83a95eb90fe39ec34097b02cf60e22 --- /dev/null +++ b/.renku/workflow/74545142cd9e4813ab933fc9b5941dfd_python.cwl @@ -0,0 +1,91 @@ +arguments: [] +baseCommand: +- python +class: CommandLineTool +cwlVersion: v1.0 +hints: [] +inputs: + input_1: + default: + class: File + path: ../../src/python/run_extract_discussions.py + inputBinding: + position: 1 + separate: true + shellQuote: true + streamable: false + type: File + input_2: + default: + class: File + path: ../../data/politicians/lastnames/1910_MPs.pickle + inputBinding: + position: 2 + separate: true + shellQuote: true + streamable: false + type: File + input_3: + default: + class: File + path: ../../data/AB/1910/04_correctedxml.tar.gz + inputBinding: + position: 3 + separate: true + shellQuote: true + streamable: false + type: File + input_4: + default: + class: File + path: ../../data/AB/1910/03_correctedmeta.tar.gz + inputBinding: + position: 4 + separate: true + shellQuote: true + streamable: false + type: File + input_5: + default: + class: File + path: ../../data/lists/not_names.txt + inputBinding: + position: 5 + separate: true + shellQuote: true + streamable: false + type: File + input_6: + default: + class: File + path: ../../data/lists/dict_overlaps.pickle + inputBinding: + position: 6 + separate: true + shellQuote: true + streamable: false + type: File + input_7: + default: data/AB/1910/05_annotatedxml.tar.gz + inputBinding: + position: 7 + separate: true + shellQuote: true + streamable: false + type: string +outputs: + output_0: + outputBinding: + glob: $(inputs.input_7) + streamable: false + type: File +permanentFailCodes: [] +requirements: +- class: InlineJavascriptRequirement +- class: InitialWorkDirRequirement + listing: + - entry: '$({"listing": [], "class": "Directory"})' + entryname: data/AB/1910 + writable: true +successCodes: [] +temporaryFailCodes: [] diff --git a/data/AB/1910/05_annotatedxml.tar.gz b/data/AB/1910/05_annotatedxml.tar.gz new file mode 100644 index 0000000000000000000000000000000000000000..4361152ddff986b40f698cf25002984e0bdda2bc --- /dev/null +++ b/data/AB/1910/05_annotatedxml.tar.gz @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:cc4c008d59cdd149b2e624973fb1be9f136fddc0580ffbc99d01974eecf7eccf +size 4685054