update pipeline

Signed-off-by: Bhumika Rao <[email protected]>

Data/Soil/.gitignore

@@ -1,6 +1,11 @@
-/GlobalSoilOrganicCarbonDensityinkgCarbon_m2to1meterdepth.tif
+
 /data
 /metadata
 /Global Soil Organic Carbon Density in kg Carbon_m2 to 1 meter depth.zip
 /CONTENTS.txt
 /TERMS OF USE.txt
+/soilcarbon.ovr
+/processed
+/GlobalSoilOrganicCarbonDensityinkgCm_1mDepth.tif
+/.botmKQAeq3934dqKuUWyps.tmp
+/.kfLrvZdTQqnxpJRSCrHh96.tmp

Data/Soil/soilcarbon.ovr.dvc

@@ -0,0 +1,4 @@
+outs:
+- md5: 33956e4dd24c1caa1dcea956e85e1f5f
+  size: 131473
+  path: soilcarbon.ovr

dvc.lock

@@ -0,0 +1,23 @@
+schema: '2.0'
+stages:
+  clean:
+    cmd: rm -rf tmp_unzip_path
+  load:
+    cmd:
+    - dvc push
+  extract:
+    cmd:
+    - python scripts/extract.py
+    - python scripts/rename_files.py
+  transform:
+    cmd:
+    - "python scripts/transform.py --input tmp_unzip_path/data/commonData_Data0_soilcarbon.ovr\
+      \ \\\n    --output Data/Soil/processed/GlobalSoilOrganicCarbonDensityinkgCm_1mDepth.tif"
+    deps:
+    - path: tmp_unzip_path/data/commonData_Data0_soilcarbon.ovr
+      md5: 33956e4dd24c1caa1dcea956e85e1f5f
+      size: 131473
+    outs:
+    - path: Data/Soil/processed/GlobalSoilOrganicCarbonDensityinkgCm_1mDepth.tif
+      md5: 96f78155b79a835f56d019586d4c1f14
+      size: 1038282

dvc.yaml

@@ -0,0 +1,21 @@
+stages:
+  extract:
+    cmd:
+      - python scripts/extract.py
+      - python scripts/rename_files.py
+  transform:
+    cmd:
+      - >-
+        python scripts/transform.py --input tmp_unzip_path/data/commonData_Data0_soilcarbon.ovr \
+            --output Data/Soil/processed/GlobalSoilOrganicCarbonDensityinkgCm_1mDepth.tif
+    deps:
+    - tmp_unzip_path/data/commonData_Data0_soilcarbon.ovr
+    outs:
+    - Data/Soil/processed/GlobalSoilOrganicCarbonDensityinkgCm_1mDepth.tif
+  load:
+    cmd:
+      - dvc push
+  clean:
+    cmd: rm -rf tmp_unzip_path
+
+

scripts/extract.py

@@ -0,0 +1,32 @@
+import argparse
+import requests
+import zipfile
+import io
+import os
+
+URL = "https://databasin2-filestore.s3.amazonaws.com/a4cb6d367eae4e52a08902874f8bfedf/download/a4cb6d367eae4e52a08902874f8bfedf_1_zip_en.zip?Signature=148HP0SJFI49y7HmTmOaNMAlUDw%3D&Expires=1680448953&AWSAccessKeyId=AKIAI4RK5BEPK3FCQPUQ"
+
+def ensure_url_is_accessible(URL):
+    r = requests.get(URL)
+    if not r.ok:
+        print("Download link expired. Please update download link")
+    else:
+        download_and_unzip_files(r.content)
+
+def download_and_unzip_files(content):
+    current_directory = os.getcwd()
+    target_parent_dir = os.path.join(current_directory, r'tmp_unzip_path')
+    if not os.path.exists(target_parent_dir):
+        os.mkdir(target_parent_dir)
+    try:
+        z = zipfile.ZipFile(io.BytesIO(content))
+        z.extractall(target_parent_dir)
+    except Exception as e:
+        print(e)
+    else:
+        print("unzipped successfully")
+
+ensure_url_is_accessible(URL)
+
+
+

scripts/rename_files.py

@@ -0,0 +1,12 @@
+import os
+
+current_directory = os.getcwd()
+target_parent_dir = os.path.join(current_directory, r'tmp_unzip_path/data')
+if os.path.exists(target_parent_dir):
+    for file_name in os.listdir(target_parent_dir):
+        if '\\' in file_name:
+            old_file_name = os.path.join(target_parent_dir, file_name)
+            filename = os.fsdecode(file_name)
+            changed_name = filename.replace("\\", "_")
+            new_file_name = os.path.join(target_parent_dir, changed_name)
+            os.rename(old_file_name,new_file_name)

scripts/transform.py

@@ -0,0 +1,22 @@
+import subprocess
+import argparse
+import os
+
+parser = argparse.ArgumentParser()
+parser.add_argument('--input', help="Directory of file to transform")
+parser.add_argument('--output', help="Directory for transformed files")
+args = vars(parser.parse_args())
+
+
+def run_shell_cmd(cmd):
+    try:
+        p = subprocess.Popen(cmd.split(), stdout=subprocess.PIPE)
+        last_stdout_bytes, last_stderr_bytes = p.communicate()
+        if last_stdout_bytes:
+            return last_stdout_bytes.decode('utf-8', 'replace')
+        else:
+            return last_stderr_bytes
+    except Exception as e:
+        print(e)
+
+run_shell_cmd("gdalwarp -s_srs EPSG:4326 -t_srs EPSG:4326 -to SRC_METHOD=NO_GEOTRANSFORM -tr 0.5 0.5 -r near -te -180.0 -90.0 180.0 90.0 -te_srs EPSG:4326 -of GTiff " + args.get('input') + " " + args.get('output'))