From 2d6af889d0794b05b83c71c38178f4541a696b4c Mon Sep 17 00:00:00 2001
From: "CR (covid cron)" <cramakri+covid-cron@ethz.ch>
Date: Wed, 25 Mar 2020 08:22:58 +0000
Subject: [PATCH] renku rerun data/covidtracking/states-metadata.json
data/covidtracking/states-daily.json
---
.../349ff7f9cbbd41febc2ee66f691da823.cwl | 120 ++++++++++++++++++
data/covidtracking/states-daily.json | 4 +-
data/covidtracking/states-metadata.json | 4 +-
runs/download-covidtracking-data.runs.ipynb | 112 ++++++++--------
4 files changed, 180 insertions(+), 60 deletions(-)
create mode 100644 .renku/workflow/349ff7f9cbbd41febc2ee66f691da823.cwl
diff --git a/.renku/workflow/349ff7f9cbbd41febc2ee66f691da823.cwl b/.renku/workflow/349ff7f9cbbd41febc2ee66f691da823.cwl
new file mode 100644
index 00000000..453e6e03
--- /dev/null
+++ b/.renku/workflow/349ff7f9cbbd41febc2ee66f691da823.cwl
@@ -0,0 +1,120 @@
+class: Workflow
+cwlVersion: v1.0
+hints: []
+inputs:
+ input_1:
+ default: states-daily.json
+ streamable: false
+ type: string
+ input_2:
+ default: states-metadata.json
+ streamable: false
+ type: string
+ input_3:
+ default: out_folder
+ streamable: false
+ type: string
+ input_4:
+ default: data/covidtracking
+ streamable: false
+ type: string
+ input_5:
+ default:
+ class: File
+ path: ../../notebooks/process/download-covidtracking-data.ipynb
+ streamable: false
+ type: File
+ input_6:
+ default: runs/download-covidtracking-data.runs.ipynb
+ streamable: false
+ type: string
+outputs:
+ output_0:
+ outputSource: step_3/output_1
+ streamable: false
+ type: Directory
+ output_3:
+ outputSource: step_3/output_0
+ streamable: false
+ type: File
+requirements: []
+steps:
+ step_1:
+ in:
+ filename: input_1
+ input_directory: step_3/output_1
+ out:
+ - output_file
+ run:
+ arguments: []
+ baseCommand:
+ - 'true'
+ class: CommandLineTool
+ cwlVersion: v1.0
+ hints: []
+ inputs:
+ filename:
+ default: states-daily.json
+ streamable: false
+ type: string
+ input_directory:
+ streamable: false
+ type: Directory
+ outputs:
+ output_file:
+ outputBinding:
+ glob: $(inputs.filename)
+ streamable: false
+ type: File
+ permanentFailCodes: []
+ requirements:
+ - &id001
+ class: InlineJavascriptRequirement
+ - &id002
+ class: InitialWorkDirRequirement
+ listing: $(inputs.input_directory.listing)
+ successCodes: []
+ temporaryFailCodes: []
+ step_2:
+ in:
+ filename: input_2
+ input_directory: step_3/output_1
+ out:
+ - output_file
+ run:
+ arguments: []
+ baseCommand:
+ - 'true'
+ class: CommandLineTool
+ cwlVersion: v1.0
+ hints: []
+ inputs:
+ filename:
+ default: states-metadata.json
+ streamable: false
+ type: string
+ input_directory:
+ streamable: false
+ type: Directory
+ outputs:
+ output_file:
+ outputBinding:
+ glob: $(inputs.filename)
+ streamable: false
+ type: File
+ permanentFailCodes: []
+ requirements:
+ - *id001
+ - *id002
+ successCodes: []
+ temporaryFailCodes: []
+ step_3:
+ in:
+ input_1: input_3
+ input_2: input_4
+ input_3: input_5
+ input_4: input_6
+ out:
+ - output_1
+ - output_0
+ run: ff5f474d38144c5d882c0036bd1059c2_papermill.cwl
diff --git a/data/covidtracking/states-daily.json b/data/covidtracking/states-daily.json
index a6234d1e..ee2b6bc8 100644
--- a/data/covidtracking/states-daily.json
+++ b/data/covidtracking/states-daily.json
@@ -1,3 +1,3 @@
version https://git-lfs.github.com/spec/v1
-oid sha256:04a4bbec05fc52248512469d41ee013fdce83b6fef928e9562dfe801ad82aff9
-size 151715
+oid sha256:4b0927c74f3dbb27dacfddb47b00d2c543f8ec932625dc1502b3d048fcc2668e
+size 160445
diff --git a/data/covidtracking/states-metadata.json b/data/covidtracking/states-metadata.json
index 55645103..08dcd867 100644
--- a/data/covidtracking/states-metadata.json
+++ b/data/covidtracking/states-metadata.json
@@ -1,3 +1,3 @@
version https://git-lfs.github.com/spec/v1
-oid sha256:7eaff61b1c975bf893a6e7e35a694980140435ce8cb5afc02164ce685cfa5baa
-size 25218
+oid sha256:02330f655200284fbdaef1b3d4272cc84385c648ce7c3b73d910a0cacc3fa529
+size 25509
diff --git a/runs/download-covidtracking-data.runs.ipynb b/runs/download-covidtracking-data.runs.ipynb
index 24fe640f..5c6e420d 100644
--- a/runs/download-covidtracking-data.runs.ipynb
+++ b/runs/download-covidtracking-data.runs.ipynb
@@ -5,10 +5,10 @@
"execution_count": 1,
"metadata": {
"papermill": {
- "duration": 1.484962,
- "end_time": "2020-03-24T11:08:51.882283",
+ "duration": 0.564716,
+ "end_time": "2020-03-25T08:22:56.714459",
"exception": false,
- "start_time": "2020-03-24T11:08:50.397321",
+ "start_time": "2020-03-25T08:22:56.149743",
"status": "completed"
},
"tags": []
@@ -25,10 +25,10 @@
"execution_count": 2,
"metadata": {
"papermill": {
- "duration": 0.034224,
- "end_time": "2020-03-24T11:08:51.927200",
+ "duration": 0.019507,
+ "end_time": "2020-03-25T08:22:56.745116",
"exception": false,
- "start_time": "2020-03-24T11:08:51.892976",
+ "start_time": "2020-03-25T08:22:56.725609",
"status": "completed"
},
"tags": [
@@ -46,10 +46,10 @@
"execution_count": 3,
"metadata": {
"papermill": {
- "duration": 0.024298,
- "end_time": "2020-03-24T11:08:51.965529",
+ "duration": 0.018572,
+ "end_time": "2020-03-25T08:22:56.773076",
"exception": false,
- "start_time": "2020-03-24T11:08:51.941231",
+ "start_time": "2020-03-25T08:22:56.754504",
"status": "completed"
},
"tags": [
@@ -59,7 +59,7 @@
"outputs": [],
"source": [
"# Parameters\n",
- "PAPERMILL_INPUT_PATH = \"/tmp/kq4860jy/notebooks/process/download-covidtracking-data.ipynb\"\n",
+ "PAPERMILL_INPUT_PATH = \"/tmp/e18fw9c9/notebooks/process/download-covidtracking-data.ipynb\"\n",
"PAPERMILL_OUTPUT_PATH = \"runs/download-covidtracking-data.runs.ipynb\"\n",
"out_folder = \"data/covidtracking\"\n"
]
@@ -68,10 +68,10 @@
"cell_type": "markdown",
"metadata": {
"papermill": {
- "duration": 0.007354,
- "end_time": "2020-03-24T11:08:51.981277",
+ "duration": 0.007795,
+ "end_time": "2020-03-25T08:22:56.790849",
"exception": false,
- "start_time": "2020-03-24T11:08:51.973923",
+ "start_time": "2020-03-25T08:22:56.783054",
"status": "completed"
},
"tags": []
@@ -87,10 +87,10 @@
"execution_count": 4,
"metadata": {
"papermill": {
- "duration": 1.994092,
- "end_time": "2020-03-24T11:08:53.981279",
+ "duration": 1.020789,
+ "end_time": "2020-03-25T08:22:57.818863",
"exception": false,
- "start_time": "2020-03-24T11:08:51.987187",
+ "start_time": "2020-03-25T08:22:56.798074",
"status": "completed"
},
"tags": []
@@ -107,10 +107,10 @@
"execution_count": 5,
"metadata": {
"papermill": {
- "duration": 0.028446,
- "end_time": "2020-03-24T11:08:54.020108",
+ "duration": 0.022913,
+ "end_time": "2020-03-25T08:22:57.852810",
"exception": false,
- "start_time": "2020-03-24T11:08:53.991662",
+ "start_time": "2020-03-25T08:22:57.829897",
"status": "completed"
},
"tags": []
@@ -129,10 +129,10 @@
"execution_count": 6,
"metadata": {
"papermill": {
- "duration": 0.080991,
- "end_time": "2020-03-24T11:08:54.110615",
+ "duration": 0.066514,
+ "end_time": "2020-03-25T08:22:57.927763",
"exception": false,
- "start_time": "2020-03-24T11:08:54.029624",
+ "start_time": "2020-03-25T08:22:57.861249",
"status": "completed"
},
"tags": []
@@ -199,7 +199,7 @@
" <td>@alpublichealth</td>\n",
" <td>No data</td>\n",
" <td>False</td>\n",
- " <td>Last negative count from 3/16. Last update tim...</td>\n",
+ " <td>Last update time taken from [main page](http:/...</td>\n",
" <td>Alabama</td>\n",
" </tr>\n",
" </tbody>\n",
@@ -221,7 +221,7 @@
"\n",
" pui pum notes name \n",
"0 All data False We count the reported number as \"persons teste... Alaska \n",
- "1 No data False Last negative count from 3/16. Last update tim... Alabama "
+ "1 No data False Last update time taken from [main page](http:/... Alabama "
]
},
"execution_count": 6,
@@ -239,10 +239,10 @@
"cell_type": "markdown",
"metadata": {
"papermill": {
- "duration": 0.010469,
- "end_time": "2020-03-24T11:08:54.132051",
+ "duration": 0.008089,
+ "end_time": "2020-03-25T08:22:57.947587",
"exception": false,
- "start_time": "2020-03-24T11:08:54.121582",
+ "start_time": "2020-03-25T08:22:57.939498",
"status": "completed"
},
"tags": []
@@ -256,10 +256,10 @@
"execution_count": 7,
"metadata": {
"papermill": {
- "duration": 0.162493,
- "end_time": "2020-03-24T11:08:54.304650",
+ "duration": 0.184993,
+ "end_time": "2020-03-25T08:22:58.140940",
"exception": false,
- "start_time": "2020-03-24T11:08:54.142157",
+ "start_time": "2020-03-25T08:22:57.955947",
"status": "completed"
},
"tags": []
@@ -276,10 +276,10 @@
"execution_count": 8,
"metadata": {
"papermill": {
- "duration": 0.02792,
- "end_time": "2020-03-24T11:08:54.344357",
+ "duration": 0.022977,
+ "end_time": "2020-03-25T08:22:58.177014",
"exception": false,
- "start_time": "2020-03-24T11:08:54.316437",
+ "start_time": "2020-03-25T08:22:58.154037",
"status": "completed"
},
"tags": []
@@ -298,10 +298,10 @@
"execution_count": 9,
"metadata": {
"papermill": {
- "duration": 0.06445,
- "end_time": "2020-03-24T11:08:54.419822",
+ "duration": 0.060602,
+ "end_time": "2020-03-25T08:22:58.251228",
"exception": false,
- "start_time": "2020-03-24T11:08:54.355372",
+ "start_time": "2020-03-25T08:22:58.190626",
"status": "completed"
},
"tags": []
@@ -311,7 +311,7 @@
"name": "stdout",
"output_type": "stream",
"text": [
- "981 data points\n"
+ "1037 data points\n"
]
},
{
@@ -349,27 +349,27 @@
" <tbody>\n",
" <tr>\n",
" <th>0</th>\n",
- " <td>20200323</td>\n",
+ " <td>20200324</td>\n",
" <td>AK</td>\n",
- " <td>22.0</td>\n",
- " <td>946.0</td>\n",
+ " <td>36.0</td>\n",
+ " <td>986.0</td>\n",
" <td>NaN</td>\n",
" <td>0.0</td>\n",
" <td>NaN</td>\n",
- " <td>968</td>\n",
- " <td>2020-03-23T20:00:00Z</td>\n",
+ " <td>1022</td>\n",
+ " <td>2020-03-24T20:00:00Z</td>\n",
" </tr>\n",
" <tr>\n",
" <th>1</th>\n",
- " <td>20200323</td>\n",
+ " <td>20200324</td>\n",
" <td>AL</td>\n",
- " <td>167.0</td>\n",
- " <td>1665.0</td>\n",
+ " <td>215.0</td>\n",
+ " <td>2106.0</td>\n",
" <td>NaN</td>\n",
" <td>NaN</td>\n",
" <td>0.0</td>\n",
- " <td>1832</td>\n",
- " <td>2020-03-23T20:00:00Z</td>\n",
+ " <td>2321</td>\n",
+ " <td>2020-03-24T20:00:00Z</td>\n",
" </tr>\n",
" </tbody>\n",
"</table>\n",
@@ -377,12 +377,12 @@
],
"text/plain": [
" date state positive negative pending hospitalized death total \\\n",
- "0 20200323 AK 22.0 946.0 NaN 0.0 NaN 968 \n",
- "1 20200323 AL 167.0 1665.0 NaN NaN 0.0 1832 \n",
+ "0 20200324 AK 36.0 986.0 NaN 0.0 NaN 1022 \n",
+ "1 20200324 AL 215.0 2106.0 NaN NaN 0.0 2321 \n",
"\n",
" dateChecked \n",
- "0 2020-03-23T20:00:00Z \n",
- "1 2020-03-23T20:00:00Z "
+ "0 2020-03-24T20:00:00Z \n",
+ "1 2020-03-24T20:00:00Z "
]
},
"execution_count": 9,
@@ -416,18 +416,18 @@
"version": "3.7.3"
},
"papermill": {
- "duration": 5.713541,
- "end_time": "2020-03-24T11:08:54.762805",
+ "duration": 3.460914,
+ "end_time": "2020-03-25T08:22:58.575452",
"environment_variables": {},
"exception": null,
- "input_path": "/tmp/kq4860jy/notebooks/process/download-covidtracking-data.ipynb",
+ "input_path": "/tmp/e18fw9c9/notebooks/process/download-covidtracking-data.ipynb",
"output_path": "runs/download-covidtracking-data.runs.ipynb",
"parameters": {
- "PAPERMILL_INPUT_PATH": "/tmp/kq4860jy/notebooks/process/download-covidtracking-data.ipynb",
+ "PAPERMILL_INPUT_PATH": "/tmp/e18fw9c9/notebooks/process/download-covidtracking-data.ipynb",
"PAPERMILL_OUTPUT_PATH": "runs/download-covidtracking-data.runs.ipynb",
"out_folder": "data/covidtracking"
},
- "start_time": "2020-03-24T11:08:49.049264",
+ "start_time": "2020-03-25T08:22:55.114538",
"version": "1.1.0"
}
},
--
GitLab