Move installed_files to Package model #2058

And add size to PackageFile model

Signed-off-by: Philippe Ombredanne <pombredanne@nexb.com>

src/packagedcode/alpine.py

@@ -29,12 +29,6 @@
 class AlpinePackage(models.Package):
     extensions = ('.apk',)
     default_type = 'alpine'
-
-    installed_files = List(
-        item_type=models.PackageFile,
-        label='installed files',
-        help='List of files installed by this package.')
-
     def compute_normalized_license(self):
         _declared, detected = detect_declared_license(self.declared_license)
         return detected

src/packagedcode/debian.py

@@ -15,12 +15,9 @@
 from packageurl import PackageURL
 
 from commoncode import filetype
-from commoncode import fileutils
-from commoncode.datautils import List
 from commoncode.datautils import String
 from packagedcode import models
 
-
 """
 Handle Debian packages.
 """
@@ -47,24 +44,17 @@ class DebianPackage(models.Package):
         label='Multi-Arch',
         help='Multi-Arch value from status file')
 
-    installed_files = List(
-        item_type=models.PackageFile,
-        label='installed files',
-        help='List of files installed by this package.')
-
     def to_dict(self, _detailed=False, **kwargs):
-        data = models.Package.to_dict(self, **kwargs)
+        data = super().to_dict(_detailed=_detailed, **kwargs)
         if _detailed:
             #################################################
-            # remove temporary fields
+            # populate temporary fields
             data['multi_arch'] = self.multi_arch
-            data['installed_files'] = [istf.to_dict() for istf in (self.installed_files or [])]
             #################################################
         else:
             #################################################
             # remove temporary fields
             data.pop('multi_arch', None)
-            data.pop('installed_files', None)
             #################################################
 
         return data
@@ -178,8 +168,7 @@ def get_copyright_file_path(self, root_dir):
 
 def get_installed_packages(root_dir, distro='debian', detect_licenses=False, **kwargs):
     """
-    Given a directory to a rootfs, yield a DebianPackage and a list of `installed_files`
-    (path, md5sum) tuples.
+    Yield installed Package objects given a ``root_dir`` rootfs directory.
     """
 
     base_status_file_loc = os.path.join(root_dir, 'var/lib/dpkg/status')
@@ -199,8 +188,7 @@ def get_installed_packages(root_dir, distro='debian', detect_licenses=False, **k
 
 
 def is_debian_status_file(location):
-    return (filetype.is_file(location)
-            and fileutils.file_name(location).lower() == 'status')
+    return filetype.is_file(location) and location.endswith('/status')
 
 
 def parse_status_file(location, distro='debian'):

src/packagedcode/models.py

@@ -336,6 +336,41 @@ class DependentPackage(BaseModel):
              'exact version.')
 
 
+@attr.s()
+class PackageFile(BaseModel):
+    """
+    A file that belongs to a package.
+    """
+
+    path = String(
+        label='Path of this installed file',
+        help='The path of this installed file either relative to a rootfs '
+            '(typical for system packages) or a path in this scan (typical for '
+             'application packages).',
+        repr=True,
+    )
+
+    size = Integer(
+        label='file size',
+        help='size of the file in bytes')
+
+    sha1 = String(
+        label='SHA1 checksum',
+        help='SHA1 checksum for this file in hexadecimal')
+
+    md5 = String(
+        label='MD5 checksum',
+        help='MD5 checksum for this file in hexadecimal')
+
+    sha256 = String(
+        label='SHA256 checksum',
+        help='SHA256 checksum for this file in hexadecimal')
+
+    sha512 = String(
+        label='SHA512 checksum',
+        help='SHA512 checksum for this file in hexadecimal')
+
+
 @attr.s()
 class Package(BasePackage):
     """
@@ -451,6 +486,11 @@ class Package(BasePackage):
              'this package. For instance an SRPM is the "source package" for a '
              'binary RPM.')
 
+    installed_files = List(
+        item_type=PackageFile,
+        label='installed files',
+        help='List of files installed by this package.')
+
     def __attrs_post_init__(self, *args, **kwargs):
         if not self.type and hasattr(self, 'default_type'):
             self.type = self.default_type
@@ -552,6 +592,14 @@ def extra_root_dirs(cls):
         """
         return []
 
+    def to_dict(self, _detailed=False, **kwargs):
+        data = super().to_dict(**kwargs)
+        if _detailed:
+            data['installed_files'] = [istf.to_dict() for istf in (self.installed_files or [])]
+        else:
+            data.pop('installed_files', None)
+        return data
+
 
 def compute_normalized_license(declared_license):
     """
@@ -573,37 +621,6 @@ def compute_normalized_license(declared_license):
         return 'unknown'
 
 
-@attr.s()
-class PackageFile(BaseModel):
-    """
-    A file that belongs to a package.
-    """
-
-    path = String(
-        label='Path of this installed file',
-        help='The path of this installed file either relative to a rootfs '
-            '(typical for system packages) or a path in this scan (typical for '
-             'application packages).',
-        repr=True,
-    )
-
-    sha1 = String(
-        label='SHA1 checksum',
-        help='SHA1 checksum for this file in hexadecimal')
-
-    md5 = String(
-        label='MD5 checksum',
-        help='MD5 checksum for this file in hexadecimal')
-
-    sha256 = String(
-        label='SHA256 checksum',
-        help='SHA256 checksum for this file in hexadecimal')
-
-    sha512 = String(
-        label='SHA512 checksum',
-        help='SHA512 checksum for this file in hexadecimal')
-
-
 # Package types
 # NOTE: this is somewhat redundant with extractcode archive handlers
 # yet the purpose and semantics are rather different here