moved version checking and xml tree conversion to package.py

Signed-off-by: Anton Utz <[email protected]>

src/ros_license_toolkit/license_checks/schema_check.py

@@ -21,6 +21,7 @@
 from lxml import etree
 
 from ros_license_toolkit.checks import Check
+from ros_license_toolkit.package import Package
 
 
 class SchemaCheck(Check):
@@ -34,56 +35,46 @@ def __init__(self):
         xml_schema_1_parsed = etree.parse('./schemas/package_format1.xsd')
         self.xml_schema_1 = etree.XMLSchema(xml_schema_1_parsed)
 
-    def _check(self, package):
-        version, status = self.validate(package.abspath + "/package.xml")
+    def _check(self, package: Package):
+        status, message = self.validate(package)
+        version: int = package.package_xml_format_ver
         if status:
             self._success(f"Detected package.xml version {version}, "
                           "validation of scheme successful.")
         else:
-            reason = ''
-            if version == 1:
-                reason = "package.xml contains errors: " +\
-                    f"{self.xml_schema_1.error_log.last_error.message}"
-            elif version == 2:
-                reason = "package.xml contains errors: " +\
-                    f"{self.xml_schema_2.error_log.last_error.message}"
-            elif version == 3:
-                reason = "package.xml contains errors: " +\
-                    f"{self.xml_schema_3.error_log.last_error.message}"
             # Temporary workaround for not implemented version 4
-            elif version == 4:
+            if version == 4:
                 reason = "couldn't check package.xml scheme. Version 4 is " +\
                     "not available right now"
                 self._warning(reason)
                 return
-            elif version == -1:
+
+            if message != '':
+                reason = f"package.xml contains errors: {message}"
+            else:
                 reason = "package.xml does not contain correct package " +\
                     "format number. Please use a real version. " +\
                     "(e.g. <package format=\"3\">)"
             self._failed(reason)
 
-    def validate(self, pck_xml_path: str) -> Tuple[int, bool]:
+    def validate(self, package: Package) -> Tuple[bool, str]:
         """This is validating package.xml file from given path.
         Automatically detects version number and validates
         it with corresponding scheme, e.g. format 3.
         If everything is correct, returns format number, else -1."""
-        xml_doc_parsed = etree.parse(pck_xml_path)
-        for element in xml_doc_parsed.getiterator():
-            if element.tag == 'package' and 'format' in element.attrib:
-                version = element.attrib['format']
-                status_check = False
-                if version == '3':
-                    version = 3
-                    status_check = self.xml_schema_3.validate(xml_doc_parsed)
-                elif version == '2':
-                    version = 2
-                    status_check = self.xml_schema_2.validate(xml_doc_parsed)
-                elif version == '1':
-                    version = 1
-                    status_check = self.xml_schema_1.validate(xml_doc_parsed)
-                elif version == '4':  # Work around: 4 not available right now
-                    version = 4
-                else:
-                    version = -1
-                return (version, status_check)
-        return (-1, False)
+        version = package.package_xml_format_ver
+        xml_doc = package.parsed_package_xml
+        validation_schema: etree.XMLSchema = False
+        if version == 3:
+            validation_schema = self.xml_schema_3
+        elif version == 2:
+            validation_schema = self.xml_schema_2
+        elif version == 1:
+            validation_schema = self.xml_schema_1
+        if validation_schema:
+            status_check = validation_schema.validate(xml_doc)
+            message = ''
+            if not status_check:
+                message = validation_schema.error_log.last_error_message
+            return status_check, message
+        return False, message

src/ros_license_toolkit/package.py

@@ -22,6 +22,7 @@
 import xml.etree.ElementTree as ET
 from typing import Any, Dict, List, Optional
 
+from lxml import etree
 from rospkg import RosPack, list_by_path
 from rospkg.common import PACKAGE_FILE
 from scancode.api import get_licenses
@@ -83,6 +84,12 @@ def __init__(self, path: str, repo: Optional[Repo] = None):
         # All ignored files and folders
         self._ignored_content: List[str] = get_ignored_content(self.abspath)
 
+        # The package.xml file, parsed as etree
+        self._parsed_package_xml: etree = None
+
+        # The package.xml version as set in <package format="x">
+        self._package_xml_format_ver: int = 0
+
     def _get_path_relative_to_pkg(self, path: str) -> str:
         """Get path relative to pkg root"""
         return os.path.relpath(path, self.abspath)
@@ -230,6 +237,32 @@ def license_tags(self) -> Dict[str, LicenseTag]:
 
         return self._license_tags
 
+    @property
+    def package_xml_format_ver(self) -> int:
+        """Returns version of package.xml format as seen in
+        <package format="3">. If Version is not valid, -1 is returned."""
+        if self._package_xml_format_ver is None:
+            root = self.parsed_package_xml.getroot()
+            if root.tag == 'package':
+                if 'format' in root.attrib:
+                    version = root.attrib['format']
+                    try:
+                        self._package_xml_format_ver = int(version)
+                    except ValueError:
+                        self._package_xml_format_ver = -1
+                    return self._package_xml_format_ver
+        self._package_xml_format_ver = -1
+        return self._package_xml_format_ver
+
+    @property
+    def parsed_package_xml(self) -> etree:
+        """Returns the package.xml content parsed as etree."""
+        if self._parsed_package_xml == 0:
+            path = self.abspath + "/package.xml"
+            assert os.path.exists(path), f'Path {path} does not exist.'
+            self._parsed_package_xml = etree.parse(path)
+        return self._parsed_package_xml
+
     @property
     def repo_url(self) -> Optional[str]:
         """Get the url of the repo this package is in if the package is in a