Fix dependency:licenses to correctly parse exceptional names (#464)

* Fix dependency:licenses to correctly parse exceptional names

* Soften restrictions as handled later with the class Package which uses Version to validate the text

Author: ArBridgeman

doc/changes/unreleased.md

@@ -5,3 +5,7 @@
 ## ✨ Features
 
 * #426: Allowed configuring the python version used for coverage
+
+## Bugfixes
+
+* #463: Fixed dependency:licenses to correctly parse exceptional names

exasol/toolbox/util/dependencies/poetry_dependencies.py

@@ -2,13 +2,13 @@
 
 import re
 import subprocess
+import sys
 from pathlib import Path
 from typing import Optional
 
 import tomlkit
 from pydantic import (
     BaseModel,
-    model_validator,
 )
 from tomlkit import TOMLDocument
 
@@ -84,19 +84,33 @@ class PoetryDependencies(BaseModel):
     working_directory: Path
 
     @staticmethod
-    def _extract_from_line(line: str) -> Package:
-        pattern = r"\s+(\d+(?:\.\d+)*)\s+"
-        match = re.split(pattern, line)
-        return Package(name=match[0], version=match[1])  #
+    def _extract_from_line(line: str) -> Optional[Package]:
+        # remove (!) from line as indicates not installed in environment,
+        # which could occur for optional dependencies
+        split_line = line.replace("(!)", "").strip().split(maxsplit=2)
+        if len(split_line) < 2:
+            print(f"Unable to parse dependency={line}")
+            return None
+        return Package(name=split_line[0], version=split_line[1])
 
     def _extract_from_poetry_show(self, output_text: str) -> list[Package]:
-        return [self._extract_from_line(line) for line in output_text.splitlines()]
+        return [
+            package
+            for line in output_text.splitlines()
+            if (package := self._extract_from_line(line))
+        ]
 
     @property
     def direct_dependencies(self) -> dict[str, list[Package]]:
         dependencies = {}
         for group in self.groups:
-            command = ("poetry", "show", "--top-level", f"--only={group.name}")
+            command = (
+                "poetry",
+                "show",
+                "--top-level",
+                f"--only={group.name}",
+                "--no-truncate",
+            )
             output = subprocess.run(
                 command,
                 capture_output=True,
@@ -110,7 +124,7 @@ def direct_dependencies(self) -> dict[str, list[Package]]:
 
     @property
     def all_dependencies(self) -> dict[str, list[Package]]:
-        command = ("poetry", "show")
+        command = ("poetry", "show", "--no-truncate")
         output = subprocess.run(
             command,
             capture_output=True,
@@ -128,7 +142,7 @@ def all_dependencies(self) -> dict[str, list[Package]]:
         }
         for line in output.stdout.splitlines():
             dep = self._extract_from_line(line=line)
-            if dep.name not in names_direct_dependencies:
+            if dep and dep.name not in names_direct_dependencies:
                 transitive_dependencies.append(dep)
 
         return direct_dependencies | {TRANSITIVE_GROUP.name: transitive_dependencies}

test/unit/util/dependencies/poetry_dependencies_test.py

@@ -95,6 +95,14 @@ class TestPoetryDependencies:
                 "import-linter                 2.3       Enforces rules for the imports within and between Python packages.",
                 Package(name="import-linter", version="2.3"),
             ),
+            (
+                "python-dateutil               2.9.0.post0     Extensions to the standard Python datetime module",
+                Package(name="python-dateutil", version="2.9.0.post0"),
+            ),
+            (
+                "types-requests                             2.32.0.20250602",
+                Package(name="types-requests", version="2.32.0.20250602"),
+            ),
         ],
     )
     def test_extract_from_line(line, expected):