add raise and logging of errors during parsing. also added tests to see correct errors are raised

Author: avishaycohen

src/mpd_parser/exceptions.py

@@ -7,6 +7,9 @@ class UnicodeDeclaredError(Exception):
     """ Raised when the XML has an encoding declaration in it's manifest and the parser did not remove it """
     description = "xml has encoding declaration, lxml cannot process it"
 
+class UnknownValueError(Exception):
+    """ Raised when the XML parsing fails on unexpected issue, check error for more information """
+    description = "lxml failed to parse manifest, verify the input"
 
 class UnknownElementTreeParseError(Exception):
     """ Raised after a etree parse operation fails on an unexpected error """

src/mpd_parser/parser.py

@@ -2,15 +2,19 @@
 Main module of the package, Parser class
 """
 
+import logging
 from re import Match, sub
 from urllib.request import urlopen
 
 from lxml import etree
 
-from mpd_parser.exceptions import UnicodeDeclaredError, UnknownElementTreeParseError
+from mpd_parser.exceptions import UnicodeDeclaredError, UnknownElementTreeParseError, UnknownValueError
 from mpd_parser.models.composite_tags import MPD
 
+# module level logger, application will configure formatting and handlers
+logger = logging.getLogger(__name__)
 
+# Regular expression to match encoding declaration in XML
 ENCODING_PATTERN = r"<\?.*?\s(encoding=\"\S*\").*\?>"
 
 
@@ -48,7 +52,10 @@ def cut_and_burn(match: Match) -> str:
         except ValueError as err:
             if "Unicode" in err.args[0]:
                 raise UnicodeDeclaredError() from err
+            logger.exception("Failed to parse manifest string")
+            raise UnknownValueError() from err
         except Exception as err:
+            logger.exception("Failed to parse manifest string")
             raise UnknownElementTreeParseError() from err
         if encoding:
             return MPD(root, encoding=encoding[0].groups()[0])
@@ -69,7 +76,10 @@ def from_file(cls, manifest_file_name: str) -> MPD:
         except ValueError as err:
             if "Unicode" in err.args[0]:
                 raise UnicodeDeclaredError() from err
+            logger.exception("Failed to parse manifest file %s", manifest_file_name)
+            raise UnknownValueError() from err
         except Exception as err:
+            logger.exception("Failed to parse manifest file %s", manifest_file_name)
             raise UnknownElementTreeParseError() from err
         return MPD(tree.getroot())
 
@@ -89,7 +99,10 @@ def from_url(cls, url: str) -> MPD:
         except ValueError as err:
             if "Unicode" in err.args[0]:
                 raise UnicodeDeclaredError() from err
+            logger.exception("Failed to parse manifest from URL %s", url)
+            raise UnknownValueError() from err
         except Exception as err:
+            logger.exception("Failed to parse manifest from URL %s", url)
             raise UnknownElementTreeParseError() from err
         return MPD(tree.getroot())
 

tests/test_manifets.py

@@ -1,13 +1,24 @@
 """
 Test the parsing of full manifests
 """
+import io
 import os
 
-from pytest import mark
+from pytest import mark, raises
+from mpd_parser.exceptions import UnicodeDeclaredError, UnknownElementTreeParseError, UnknownValueError
 from mpd_parser.parser import Parser
 
 from tests.conftest import touch_attributes, MANIFESTS_DIR
 
+class DummyFile(io.BytesIO):
+    def __enter__(self):
+        return self
+    def __exit__(self, exc_type, exc_val, exc_tb):
+        self.close()
+
+def dummy_urlopen(url):
+    return DummyFile(b"<MPD></MPD>")
+
 
 @mark.parametrize("input_file", [
     "./../manifests/bigBuckBunny-onDemend.mpd",
@@ -104,3 +115,49 @@ def test_to_string(input_file):
                    0].range == \
                transformed_mpd.periods[0].adaptation_sets[0].representations[0].segment_bases[0].initializations[
                    0].range
+
+@mark.parametrize(
+    "exception,patch_func",
+    [
+        (UnicodeDeclaredError, lambda: ValueError("Unicode something")),
+        (UnknownValueError, lambda: ValueError("Some other value error")),
+        (UnknownElementTreeParseError, lambda: RuntimeError("Some runtime error")),
+    ]
+)
+def test_from_string_error_handling(monkeypatch, exception, patch_func):
+    def fake_parse(*args, **kwargs):
+        raise patch_func()
+    monkeypatch.setattr("mpd_parser.parser.etree.fromstring", fake_parse)
+    with raises(exception):
+        Parser.from_string("<MPD></MPD>")
+
+@mark.parametrize(
+    "exception,patch_func",
+    [
+        (UnicodeDeclaredError, lambda: ValueError("Unicode something")),
+        (UnknownValueError, lambda: ValueError("Some other value error")),
+        (UnknownElementTreeParseError, lambda: RuntimeError("Some runtime error")),
+    ]
+)
+def test_from_file_error_handling(monkeypatch, exception, patch_func):
+    def fake_parse(*args, **kwargs):
+        raise patch_func()
+    monkeypatch.setattr("mpd_parser.parser.etree.parse", fake_parse)
+    with raises(exception):
+        Parser.from_file("dummy_file.mpd")
+
+@mark.parametrize(
+    "exception,patch_func",
+    [
+        (UnicodeDeclaredError, lambda: ValueError("Unicode something")),
+        (UnknownValueError, lambda: ValueError("Some other value error")),
+        (UnknownElementTreeParseError, lambda: RuntimeError("Some runtime error")),
+    ]
+)
+def test_from_url_error_handling(monkeypatch, exception, patch_func):
+    def fake_parse(*args, **kwargs):
+        raise patch_func()
+    monkeypatch.setattr("mpd_parser.parser.etree.parse", fake_parse)
+    monkeypatch.setattr("mpd_parser.parser.urlopen", dummy_urlopen)
+    with raises(exception):
+        Parser.from_url("http://dummy.url/manifest.mpd")