Enable the huge_tree option for the lxml parser (#3365)

teemtee · Nov 28, 2024 · dfb5761 · dfb5761
1 parent 36362a1
commit dfb5761
Show file tree

Hide file tree

Showing 4 changed files with 59 additions and 22 deletions.
diff --git a/tests/report/junit/data/custom-deep-tree.xml.j2 b/tests/report/junit/data/custom-deep-tree.xml.j2
@@ -0,0 +1,20 @@
+<?xml version="1.0" ?>
+
+{#
+Test that very deep trees are correctly handled by junit plugin.
+
+Refs.:
+- [PR#3365](https://github.com/teemtee/tmt/pull/3365)
+- [`huge_tree` option](https://lxml.de/api/lxml.etree.XMLParser-class.html)
+#}
+
+{% set DEPTH = 257 %}
+<root>
+{% for i in range(1, DEPTH + 1) %}
+    {{ "  " * i }}<tag{{ i }}>
+{% endfor %}
+
+{% for i in range(DEPTH, 0, -1) %}
+    {{ "  " * i }}</tag{{ i }}>
+{% endfor %}
+</root>
diff --git a/tests/report/junit/data/main.fmf b/tests/report/junit/data/main.fmf
@@ -32,6 +32,14 @@
         /escape"<speci&l>_chars:
             test: ./special_chars.sh
 
+        /big-output:
+            summary: Generate huge text data into <system-out> JUnit tag
+            link:
+              - verifies: https://github.com/teemtee/tmt/issues/3363
+            require:
+              - python
+            test: python -c "print((('a' * 1023) + '\n') * 1024 * 10)"
+
         /subresults/pass:
             summary: Basic pass test of shell subresults
             test: |

diff --git a/tests/report/junit/test.sh b/tests/report/junit/test.sh
@@ -5,62 +5,71 @@ rlJournalStart
     rlPhaseStartSetup
         rlRun "pushd data"
         rlRun "set -o pipefail"
+        rlRun "tmp=\$(mktemp -d)" 0 "Creating tmp directory"
+        rlRun "run_dir=$tmp/run-subresults"
     rlPhaseEnd
 
     for method in tmt; do
         rlPhaseStartTest "[$method] Basic format checks"
-            rlRun "tmt run -avr execute -h $method report -h junit --file junit.xml 2>&1 >/dev/null | tee output" 2
-            rlAssertGrep "5 tests passed, 5 tests failed and 2 errors" "output"
-            rlAssertGrep '00:00:00 pass /test/shell/escape"<speci&l>_chars (on default-0)' "output"
-            rlAssertGrep '<testsuite name="/plan" disabled="0" errors="2" failures="5" skipped="0" tests="12"' "junit.xml"
+            rlRun -s "tmt run -av --id $run_dir execute -h $method report -h junit --file junit.xml 2>&1 >/dev/null" 2
+            rlAssertGrep "6 tests passed, 5 tests failed and 2 errors" "$rlRun_LOG"
+            rlAssertGrep '00:00:00 pass /test/shell/escape"<speci&l>_chars (on default-0)' "$rlRun_LOG"
+            rlAssertGrep '<testsuite name="/plan" disabled="0" errors="2" failures="5" skipped="0" tests="13"' "junit.xml"
             rlAssertGrep 'fail</failure>' "junit.xml"
 
             # Test the escape of special characters
             rlAssertGrep '<testcase name="/test/shell/escape&quot;&lt;speci&amp;l&gt;_chars">' "junit.xml"
             rlAssertGrep '<system-out>&lt;speci&amp;l&gt;"chars and control chars</system-out>' "junit.xml"
 
             # Check there is no schema problem reported
-            rlAssertNotGrep 'The generated XML output is not a valid XML file or it is not valid against the XSD schema\.' "output"
+            rlAssertNotGrep 'The generated XML output is not a valid XML file or it is not valid against the XSD schema\.' "$rlRun_LOG"
         rlPhaseEnd
 
         rlPhaseStartTest "[$method] Check the flavor argument is working"
-            rlRun "tmt run -avr execute -h $method report -h junit --file junit.xml --flavor default 2>&1 >/dev/null | tee output" 2
-            rlAssertGrep "5 tests passed, 5 tests failed and 2 errors" "output"
+            rlRun -s "tmt run --last -v --id $run_dir execute -h $method report -h junit --file junit.xml --flavor default --force 2>&1 >/dev/null" 2
+            rlAssertGrep "6 tests passed, 5 tests failed and 2 errors" "$rlRun_LOG"
 
             # Check there is no schema problem reported
-            rlAssertNotGrep 'The generated XML output is not a valid XML file or it is not valid against the XSD schema\.' "output"
+            rlAssertNotGrep 'The generated XML output is not a valid XML file or it is not valid against the XSD schema\.' "$rlRun_LOG"
         rlPhaseEnd
 
         rlPhaseStartTest "[$method] Check the mutually exclusive arguments"
-            rlRun "tmt run -avr execute -h $method report -h junit --file junit.xml --flavor custom 2>&1 >/dev/null | tee output" 2
-            rlAssertGrep "The 'custom' flavor requires the '--template-path' argument." "output"
+            rlRun -s "tmt run --last -v --id $run_dir execute -h $method report -h junit --file junit.xml --flavor custom --force 2>&1 >/dev/null" 2
+            rlAssertGrep "The 'custom' flavor requires the '--template-path' argument." "$rlRun_LOG"
 
-            rlRun "tmt run -avr execute -h $method report -h junit --file junit.xml --template-path custom.xml.j2 2>&1 >/dev/null | tee output" 2
-            rlAssertGrep "The '--template-path' can be used only with '--flavor=custom'." "output"
+            rlRun -s "tmt run --last -v execute -h $method report -h junit --file junit.xml --template-path custom.xml.j2 --force 2>&1 >/dev/null" 2
+            rlAssertGrep "The '--template-path' can be used only with '--flavor=custom'." "$rlRun_LOG"
 
         rlPhaseEnd
 
         rlPhaseStartTest "[$method] Check the 'custom' flavor with a custom XML template"
-            rlRun "tmt run -avr execute -h $method report -h junit --file custom-template-out.xml --template-path custom.xml.j2 --flavor custom 2>&1 >/dev/null | tee output" 2
+            rlRun -s "tmt run --last -v --id $run_dir execute -h $method report -h junit --file custom-template-out.xml --template-path custom.xml.j2 --flavor custom --force 2>&1 >/dev/null" 2
 
             # There must not be a schema check when using a custom flavor
-            rlAssertGrep "The 'custom' JUnit flavor is used, you are solely responsible for the validity of the XML schema\." "output"
+            rlAssertGrep "The 'custom' JUnit flavor is used, you are solely responsible for the validity of the XML schema\." "$rlRun_LOG"
 
             rlAssertGrep '<test name="/test/beakerlib/fail" value="fail"/>' "custom-template-out.xml"
             rlAssertGrep '<test name="/test/beakerlib/pass" value="pass"/>' "custom-template-out.xml"
             rlAssertGrep '<test name="/test/shell/pass" value="pass"/>' "custom-template-out.xml"
             rlAssertGrep '<test name="/test/shell/timeout" value="error"/>' "custom-template-out.xml"
             rlAssertGrep '<test name="/test/shell/escape&quot;&lt;speci&amp;l&gt;_chars" value="pass"/>' "custom-template-out.xml"
+            rlAssertGrep '<test name="/test/shell/big-output" value="pass"/>' "custom-template-out.xml"
+        rlPhaseEnd
+
+        rlPhaseStartTest "[$method] Check the 'custom' flavor with very deep XML trees"
+            rlRun -s "tmt run --last -v --id $run_dir execute -h $method report -h junit --file custom-deep-tree-template-out.xml --template-path custom-deep-tree.xml.j2 --flavor custom --force 2>&1 >/dev/null" 2
+
+            rlAssertNotGrep 'Excessive depth in document' "$rlRun_LOG"
         rlPhaseEnd
 
         rlPhaseStartTest "[$method] The 'custom' flavor with a custom **non-XML** template must not work"
-            rlRun "tmt run -avr execute -h $method report -h junit --file custom-template-out.xml --template-path non-xml-custom.j2 --flavor custom 2>&1 >/dev/null | tee output" 2
+            rlRun -s "tmt run --last -v execute -h $method report -h junit --file custom-template-out.xml --template-path non-xml-custom.j2 --flavor custom --force 2>&1 >/dev/null" 2
 
-            rlAssertGrep 'The generated XML output is not a valid XML file.' "output"
+            rlAssertGrep 'The generated XML output is not a valid XML file.' "$rlRun_LOG"
         rlPhaseEnd
 
         rlPhaseStartTest "[$method] Check the 'custom' flavor and context for subresults"
-            rlRun "tmt run -avr execute -h $method report -h junit --file custom-subresults-template-out.xml --template-path custom-subresults.xml.j2 --flavor custom 2>&1 >/dev/null | tee output" 2
+            rlRun "tmt run --last -v execute -h $method report -h junit --file custom-subresults-template-out.xml --template-path custom-subresults.xml.j2 --flavor custom --force 2>&1 >/dev/null" 2
 
             # Beakerlib subresults
             rlAssertGrep '<subresult name="/Test" outcome="fail"/>' "custom-subresults-template-out.xml"
@@ -85,7 +94,8 @@ rlJournalStart
     done
 
     rlPhaseStartCleanup
-        rlRun "rm output junit.xml custom-template-out.xml custom-subresults-template-out.xml"
+        rlRun "rm *.xml"
         rlRun "popd"
+        rlRun "rm -rf $tmp"
     rlPhaseEnd
 rlJournalEnd
diff --git a/tmt/steps/report/junit.py b/tmt/steps/report/junit.py
@@ -311,8 +311,8 @@ def _read_log_filter(log: Path) -> str:
 
     xml_parser_kwargs: dict[str, Any] = {
         'remove_blank_text': prettify,
-        'schema': None,
-        }
+        'huge_tree': True,
+        'schema': None}
 
     # The schema check must be done only for a non-custom JUnit flavors
     if flavor != CUSTOM_FLAVOR_NAME:
@@ -334,8 +334,7 @@ def _read_log_filter(log: Path) -> str:
 
     xml_parser = etree.XMLParser(**xml_parser_kwargs)
     try:
-        # S320: Parsing of untrusted data is known to be vulnerable to XML
-        # attacks.
+        # S320: Parsing of untrusted data is known to be vulnerable to XML attacks.
         tree_root: XMLElement = etree.fromstring(xml_data, xml_parser)  # noqa: S320
 
     except etree.XMLSyntaxError as e: