Add initial purl map

Signed-off-by: Jono Yang <jyang@nexb.com>

Author: JonoYang

src/purl_validator/__init__.py

@@ -26,28 +26,30 @@
 import ducer
 
 
-PURLS = """
-pkg:maven/be.sweetmustard.vinegar/vinegar-pattern-matcher@0.1.1
-pkg:maven/be.sweetmustard.vinegar/vinegar-pattern-matcher@0.1.1?classifier=sources
-pkg:maven/be.sweetmustard.vinegar/vinegar-pattern-matcher@0.1.0
-pkg:maven/be.sweetmustard.vinegar/vinegar-pattern-matcher@0.1.0?classifier=sources
-""".split()
+PURL_MAP_LOCATION = Path(__file__).parent / "purls.map"
 
 
 def create_purl_map(purls):
-    # Ensure elements of `purls` are strings:
-    purl_strs = []
+    # Ensure elements of `purls` are PackageURLs:
+    purl_objs = []
     for purl in purls:
         if not isinstance(purl, (PackageURL, str)):
             raise ValueError(f"invalid purl in `purls`: {purl}")
-        if isinstance(purl, PackageURL):
-            purl_str = purl.to_string()
+        if isinstance(purl, str):
+            p = PackageURL.from_string(purl)
         else:
-            purl_str = purl
-        purl_strs.append(purl_str)
-
-    # purl strs must be sorted and converted to bytes before going into the Map
-    prepared_purl_strs = [(bytes(purl_str, "utf-8"), 1) for purl_str in sorted(purl_strs)]
+            p = purl
+        purl_objs.append(p)
+
+    # purl strs must be unique, sorted, and converted to bytes before going into the Map
+    purl_strs = set()
+    for purl_obj in purl_objs:
+        if purl_obj.namespace:
+            purl_str = f"{purl_obj.type}/{purl_obj.namespace}/{purl_obj.name}"
+        else:
+            purl_str = f"{purl_obj.type}/{purl_obj.name}"
+        purl_strs.add(purl_str)
+    prepared_purl_strs = sorted((bytes(purl_str, "utf-8"), 1) for purl_str in purl_strs)
 
     # create map
     temp_dir = fileutils.get_temp_dir()
@@ -58,18 +60,18 @@ def create_purl_map(purls):
 
 
 class PurlValidator:
-    def __init__(self, purl_map_loc=None):
+    def __init__(self, purls=None, purl_map_loc=PURL_MAP_LOCATION):
         self.created_maps = []
 
-        if purl_map_loc:
+        if purls:
+            # Create purl map from list of purls
+            purl_map_loc = self.create_purl_map(purls=purls)
+        elif purl_map_loc:
             if not isinstance(purl_map_loc, (Path, str)):
                 raise ValueError("`purl_map_loc` must be a Path or path string")
             if not isinstance(purl_map_loc, Path):
                 # Ensure purl_map_loc is a Path
                 purl_map_loc = Path(purl_map_loc)
-        else:
-            # Create purl map from PURLS
-            purl_map_loc = self.create_purl_map(purls=PURLS)
 
         self.purl_map = self.load_purl_map(location=purl_map_loc)
 
@@ -97,15 +99,12 @@ def validate_purl(self, purl):
             purl = PackageURL.from_string(purl)
 
         # Convert purl to bytes
-        purl_bytes = bytes(purl.to_string(), "utf-8")
-
-        return bool(self.purl_map.get(purl_bytes))
+        if purl.namespace:
+            purl_str = f"{purl.type}/{purl.namespace}/{purl.name}"
+        else:
+            purl_str = f"{purl.type}/{purl.name}"
 
+        purl_bytes = bytes(purl_str, "utf-8")
+        in_purl_map = bool(self.purl_map.get(purl_bytes))
 
-if __name__ == "__main__":
-    purl_validator = PurlValidator()
-    print(
-        purl_validator.validate_purl(
-            "pkg:maven/be.sweetmustard.vinegar/vinegar-pattern-matcher@0.1.1"
-        )
-    )
+        return in_purl_map