Improve unzip_like and remove bash-based tempdirs

Author: sjpfenninger

tests/integration/Snakefile

@@ -19,7 +19,7 @@ rule download_netherlands_shapes:
 
 rule download_netherlands_protected_areas:
     message:
-        "Download and unzip a dummy drop-in dataset for Netherlands protected areas (not based on WDPA)."
+        "Download a dummy drop-in dataset for Netherlands protected areas (not based on WDPA)."
     output:
         "results/integration_test/resources/user/wdpa.gdb.zip",
     log:
@@ -34,7 +34,7 @@ rule download_netherlands_protected_areas:
 
 rule unzip_netherlands_protected_areas:
     message:
-        "Download and unzip a dummy drop-in dataset for Netherlands protected areas (not based on WDPA)."
+        "Download protected areas data."
     input:
         script=workflow.source_path("../../workflow/scripts/unzip_like.py"),
         zipfile=rules.download_netherlands_protected_areas.output,

workflow/rules/automatic.smk

@@ -43,7 +43,9 @@ rule download_globcover:
     conda:
         "../envs/shell.yaml"
     shell:
-        'curl -sSLo {output} "{params.url}"'
+        """
+        curl -sSLo "{output}" "{params.url}"
+        """
 
 
 rule unzip_globcover:
@@ -62,10 +64,7 @@ rule unzip_globcover:
         "../envs/shell.yaml"
     shell:
         """
-        temp_dir=$(mktemp -d)
-        python {input.script} {input.zipfile} -f {params.target_file} -t $temp_dir
-        mv $temp_dir/{params.target_file} {output}
-        rm -R $temp_dir
+        python "{input.script}" "{input.zipfile}" -f "{params.target_file}" -o "{output}" 2> "{log}"
         """
 
 
@@ -81,7 +80,9 @@ rule download_ghsl:
     conda:
         "../envs/shell.yaml"
     shell:
-        'curl -sSLo {output} "{params.url}"'
+        """
+        curl -sSLo "{output}" "{params.url}"
+        """
 
 
 rule unzip_ghsl:
@@ -100,8 +101,5 @@ rule unzip_ghsl:
         "../envs/shell.yaml"
     shell:
         """
-        temp_dir=$(mktemp -d)
-        python {input.script} {input.zipfile} -f {params.target_file} -t $temp_dir
-        mv $temp_dir/{params.target_file} {output}
-        rm -R $temp_dir
+        python "{input.script}" "{input.zipfile}" -f "{params.target_file}" -o "{output}" 2> "{log}"
         """

workflow/scripts/unzip_like.py

@@ -1,6 +1,7 @@
 """Emulates the `unzip` command across platforms."""
 
 import os
+import shutil
 import zipfile
 
 import click
@@ -15,8 +16,18 @@
     default=".",
     help="Target directory to extract to.",
 )
-@click.option("--file", "-f", help="Specific file inside the zip to extract.")
-def unzip(zip_path, target, file):
+@click.option(
+    "--file",
+    "-f",
+    help="Specific file inside the zip to extract.",
+)
+@click.option(
+    "--output",
+    "-o",
+    type=click.Path(),
+    help="Output filename to save the extracted file as (used with -f).",
+)
+def unzip(zip_path, target, file, output):
     """Emulates the `unzip` command across platforms.
 
     ZIP_PATH: Path to the .zip file
@@ -25,12 +36,19 @@ def unzip(zip_path, target, file):
 
     with zipfile.ZipFile(zip_path, "r") as zip_ref:
         if file:
-            # Check if file exists in zip
             if file not in zip_ref.namelist():
                 click.echo(f"Error: '{file}' not found in archive.")
                 return
-            zip_ref.extract(file, target)
-            click.echo(f"Extracted '{file}' to '{target}'.")
+
+            extracted_path = zip_ref.extract(file, target)
+
+            if output:
+                output_path = os.path.join(target, output)
+                os.makedirs(os.path.dirname(output_path), exist_ok=True)
+                shutil.move(extracted_path, output_path)
+                click.echo(f"Extracted '{file}' to '{output_path}'.")
+            else:
+                click.echo(f"Extracted '{file}' to '{extracted_path}'.")
         else:
             zip_ref.extractall(target)
             click.echo(f"Extracted all files to '{target}'.")