push filter operator to latest CLAIMED spec

component-library/filter/filter.cwl

@@ -0,0 +1,33 @@
+cwlVersion: v1.2
+class: CommandLineTool
+
+baseCommand: "claimed"
+
+inputs:
+  component:
+    type: string
+    default: romeokienzler/claimed-filter:0.1
+    inputBinding:
+      position: 1
+      prefix: --component
+  log_level:
+    type: string
+    default: "INFO"
+    inputBinding:
+      position: 2
+      prefix: --log_level
+  predicate:
+    type: string
+    default: None
+    inputBinding:
+      position: 3
+      prefix: --predicate
+  file_name:
+    type: string
+    default: None
+    inputBinding:
+      position: 4
+      prefix: --file_name
+
+
+outputs: []

component-library/filter/filter.ipynb

@@ -5,51 +5,51 @@
    "id": "arabic-honey",
    "metadata": {},
    "source": [
-    "# Filters rows based on predicate on pandas data frame\n",
-    "Example \"predicate=~metadata.filename.str.contains('.gz') \" => filters all rows where column \"filename\" contains '.gz\""
+    "# filter"
    ]
   },
   {
-   "cell_type": "code",
-   "execution_count": null,
-   "id": "fabulous-israeli",
+   "cell_type": "markdown",
+   "id": "ebe4a7f0",
    "metadata": {},
-   "outputs": [],
    "source": [
-    "# @param predicate (as described in documentation of the component)\n",
-    "# @param file_name csv file name"
+    "Filters rows based on predicate on pandas data frame\n",
+    "Example \"predicate=~metadata.filename.str.contains('.gz') \" => filters all rows where column \"filename\" contains '.gz\""
    ]
   },
   {
    "cell_type": "code",
    "execution_count": null,
-   "id": "upset-affair",
+   "id": "fabulous-israeli",
    "metadata": {},
    "outputs": [],
    "source": [
-    "import os\n",
-    "predicate = os.environ.get('predicate')\n",
-    "file_name = os.environ.get('file_name', 'metadata.csv')"
+    "!pip install pandas"
    ]
   },
   {
    "cell_type": "code",
    "execution_count": null,
-   "id": "incomplete-monkey",
+   "id": "d92abd54",
    "metadata": {},
    "outputs": [],
    "source": [
-    "!pip3 install pandas==1.2.1"
+    "import pandas as pd\n",
+    "import os"
    ]
   },
   {
    "cell_type": "code",
    "execution_count": null,
-   "id": "auburn-wonder",
+   "id": "upset-affair",
    "metadata": {},
    "outputs": [],
    "source": [
-    "import pandas as pd"
+    "# predicate (as described in documentation of the component)\n",
+    "predicate = os.environ.get('predicate')\n",
+    "\n",
+    "# file name / path of the CSV file to read\n",
+    "file_name = os.environ.get('file_name')"
    ]
   },
   {
@@ -59,7 +59,7 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "metadata = pd.read_csv(file_name)"
+    "df = pd.read_csv(file_name)"
    ]
   },
   {
@@ -69,7 +69,7 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "exec('metadata = metadata[' + predicate + ']')"
+    "exec('df = df[' + predicate + ']')"
    ]
   },
   {
@@ -79,7 +79,7 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "metadata.to_csv(file_name, index=False)"
+    "df.to_csv(file_name, index=False)"
    ]
   }
  ],

component-library/filter/filter.job.yaml

@@ -0,0 +1,22 @@
+apiVersion: batch/v1
+kind: Job
+metadata:
+  name: filter
+spec:
+  template:
+    spec:
+      containers:
+      - name: filter
+        image: romeokienzler/claimed-filter:0.1
+        workingDir: /opt/app-root/src/
+        command: ["/opt/app-root/bin/ipython","filter.py"]
+        env:
+        - name: log_level
+          value: value_of_log_level
+        - name: predicate
+          value: value_of_predicate
+        - name: file_name
+          value: value_of_file_name
+      restartPolicy: OnFailure
+      imagePullSecrets:
+        - name: image_pull_secret

component-library/filter/filter.yaml

@@ -0,0 +1,23 @@
+name: file_name
+description: "# filter Filters rows based on predicate on pandas data frame Example 'predicate=~metadata.filename.str.contains('.gz') ' => filters all rows where column 'filename' contains '.gz' – CLAIMED V0.1"
+
+inputs:
+- {name: log_level, type: String, description: "update log level", default: "INFO"}
+- {name: predicate, type: String, description: "predicate (as described in documentation of the component)"}
+- {name: file_name, type: String, description: "file name / path of the CSV file to read"}
+
+
+outputs:
+
+
+implementation:
+    container:
+        image: romeokienzler/claimed-file_name:0.1
+        command:
+        - sh
+        - -ec
+        - |
+          ipython ./filter.py log_level="${0}" predicate="${1}" file_name="${2}" 
+        - {inputValue: log_level}
+        - {inputValue: predicate}
+        - {inputValue: file_name}