Store cran_db in pipeline working_path

Fix mine_cran_packageurls return type
Update minecode-pipelines version to 0.0.1b57

Signed-off-by: ziad hany <[email protected]>

Author: ziadhany

minecode_pipelines/__init__.py

@@ -8,4 +8,4 @@
 #
 
 
-VERSION = "0.0.1b56"
+VERSION = "0.0.1b57"

minecode_pipelines/pipelines/mine_cran.py

@@ -44,7 +44,7 @@ def download_cran_db(self):
         """
         Download the full CRAN package database
         """
-        self.db_path = fetch_cran_db(logger=self.log)
+        self.db_path = fetch_cran_db(working_path=self.working_path, logger=self.log)
 
     def packages_count(self):
         """

minecode_pipelines/pipes/cran.py

@@ -22,19 +22,21 @@
 
 import json
 from pathlib import Path
+from typing import Iterable
+from typing import Tuple
+from typing import List
+
 import requests
 from packageurl import PackageURL
 from aboutcode.hashid import get_core_purl
-import tempfile
 
 
-def fetch_cran_db(logger) -> Path:
+def fetch_cran_db(working_path, logger) -> Path:
     """
     Download the CRAN package database (~250MB JSON) in a memory-efficient way.
     Saves it to a file instead of loading everything into memory.
     """
-    temp_dir = Path(tempfile.mkdtemp())
-    output_path = temp_dir / "cran_db.json"
+    output_path = working_path / "cran_db.json"
     logger(f"Target download path: {output_path}")
 
     url = "https://crandb.r-pkg.org/-/all"
@@ -47,7 +49,7 @@ def fetch_cran_db(logger) -> Path:
     return output_path
 
 
-def mine_cran_packageurls(db_path: Path) -> list:
+def mine_cran_packageurls(db_path: Path) -> Iterable[Tuple[str, List[str]]]:
     """
     Extract package names and their versions from a CRAN DB JSON file.
     Yields a tuple: (base_purl, list_of_purls)

pyproject-minecode_pipelines.toml

@@ -4,7 +4,7 @@ build-backend = "flot.buildapi"
 
 [project]
 name = "minecode_pipelines"
-version = "0.0.1b56"
+version = "0.0.1b57"
 description = "A library for mining packageURLs and package metadata from ecosystem repositories."
 readme = "minecode_pipelines/README.rst"
 license = { text = "Apache-2.0" }