Add flow-through test case generator

This commit is contained in:
Chris Smowton
2021-06-03 15:14:03 +01:00
parent d4bb8a70c2
commit 6fee40cfde
4 changed files with 491 additions and 0 deletions

View File

@@ -0,0 +1,144 @@
import errno
import json
import os
import os.path
import re
import shlex
import shutil
import subprocess
import sys
import tempfile
if len(sys.argv) != 4:
print("Usage: GenerateFlowTestCase.py specsToTest.ssv projectPom.xml outdir", file=sys.stderr)
print("specsToTest.ssv should contain SSV rows describing method taint-propagation specifications to test", file=sys.stderr)
print("projectPom.xml should import dependencies sufficient to resolve the types used in specsToTest.ssv", file=sys.stderr)
sys.exit(1)
try:
os.makedirs(sys.argv[3])
except Exception as e:
if e.errno != errno.EEXIST:
print("Failed to create output directory %s: %s" % (sys.argv[3], e))
sys.exit(1)
resultJava = os.path.join(sys.argv[3], "Test.java")
resultQl = os.path.join(sys.argv[3], "test.ql")
if os.path.exists(resultJava) or os.path.exists(resultQl):
print("Won't overwrite existing files '%s' or '%s'" % (resultJava, resultQl), file = sys.stderr)
sys.exit(1)
workDir = tempfile.mkdtemp()
# Step 1: make a database that touches all types whose methods we want to test:
print("Creating Maven project")
projectDir = os.path.join(workDir, "mavenProject")
os.makedirs(projectDir)
try:
shutil.copyfile(sys.argv[2], os.path.join(projectDir, "pom.xml"))
except Exception as e:
print("Failed to read project POM %s: %s" % (sys.argv[2], e), file = sys.stderr)
sys.exit(1)
commentRegex = re.compile("^\s*(//|#)")
def isComment(s):
return commentRegex.match(s) is not None
try:
with open(sys.argv[1], "r") as f:
specs = [l for l in f if not isComment(l)]
except Exception as e:
print("Failed to open %s: %s\n" % (sys.argv[1], e))
sys.exit(1)
projectTestPkgDir = os.path.join(projectDir, "src", "main", "java", "test")
projectTestFile = os.path.join(projectTestPkgDir, "Test.java")
os.makedirs(projectTestPkgDir)
def qualifiedOuterNameFromSsvRow(row):
cells = row.split(";")
if len(cells) < 2:
return None
return cells[0] + "." + cells[1].replace("$", ".")
with open(projectTestFile, "w") as testJava:
testJava.write("package test;\n\npublic class Test {\n\n")
for i, spec in enumerate(specs):
outerName = qualifiedOuterNameFromSsvRow(spec)
if outerName is None:
print("A taint specification has the wrong format: should be 'package;classname;methodname....'", file = sys.stderr)
print("Mis-formatted row: " + spec, file = sys.stderr)
sys.exit(1)
testJava.write("\t%s obj%d = null;\n" % (outerName, i))
testJava.write("}")
print("Creating project database")
cmd = ["codeql", "database", "create", "--language=java", "db"]
ret = subprocess.call(cmd, cwd = projectDir)
if ret != 0:
print("Failed to create project database. Check that '%s' is a valid POM that pulls in all necessary dependencies, and '%s' specifies valid classes and methods." % (sys.argv[2], sys.argv[1]), file = sys.stderr)
print("Failed command was: %s (cwd: %s)" % (shlex.join(cmd), projectDir), file = sys.stderr)
sys.exit(1)
print("Creating test-generation query")
queryDir = os.path.join(workDir, "query")
os.makedirs(queryDir)
qlFile = os.path.join(queryDir, "gen.ql")
with open(os.path.join(queryDir, "qlpack.yml"), "w") as f:
f.write("name: test-generation-query\nversion: 0.0.0\nlibraryPathDependencies: codeql-java")
with open(qlFile, "w") as f:
f.write("import java\nimport utils.GenerateFlowTestCase\n\nclass GenRow extends CsvRow {\n\n\tGenRow() {\n\t\tthis = [\n")
f.write(",\n".join('\t\t\t"%s"' % spec.strip() for spec in specs))
f.write("\n\t\t]\n\t}\n}\n")
print("Generating tests")
generatedBqrs = os.path.join(queryDir, "out.bqrs")
cmd = ['codeql', 'query', 'run', qlFile, '--database', os.path.join(projectDir, "db"), '--output', generatedBqrs]
ret = subprocess.call(cmd)
if ret != 0:
print("Failed to generate tests. Failed command was: " + shlex.join(cmd))
sys.exit(1)
generatedJson = os.path.join(queryDir, "out.json")
cmd = ['codeql', 'bqrs', 'decode', generatedBqrs, '--format=json', '--output', generatedJson]
ret = subprocess.call(cmd)
if ret != 0:
print("Failed to decode BQRS. Failed command was: " + shlex.join(cmd))
sys.exit(1)
def getTuples(queryName, jsonResult, fname):
if queryName not in jsonResult or "tuples" not in jsonResult[queryName]:
print("Failed to read generated tests: expected key '%s' with a 'tuples' subkey in file '%s'" % (queryName, fname), file = sys.stderr)
sys.exit(1)
return jsonResult[queryName]["tuples"]
with open(generatedJson, "r") as f:
generateOutput = json.load(f)
testCaseRows = getTuples("getTestCase", generateOutput, generatedJson)
supportModelRows = getTuples("getASupportMethodModel", generateOutput, generatedJson)
if len(testCaseRows) != 1 or len(testCaseRows[0]) != 1:
print("Expected exactly one getTestCase result with one column (got: %s)" % json.dumps(testCaseRows), file = sys.stderr)
if any(len(row) != 1 for row in supportModelRows):
print("Expected exactly one column in getASupportMethodModel relation (got: %s)" % json.dumps(supportModelRows), file = sys.stderr)
with open(resultJava, "w") as f:
f.write(generateOutput["getTestCase"]["tuples"][0][0])
scriptPath = os.path.dirname(sys.argv[0])
with open(resultQl, "w") as f:
with open(os.path.join(scriptPath, "testHeader.qlfrag"), "r") as header:
shutil.copyfileobj(header, f)
f.write(", ".join('"%s"' % modelSpecRow[0].strip() for modelSpecRow in supportModelRows))
with open(os.path.join(scriptPath, "testFooter.qlfrag"), "r") as header:
shutil.copyfileobj(header, f)
cmd = ['codeql', 'query', 'format', '-qq', '-i', resultQl]
subprocess.call(cmd)
shutil.rmtree(workDir)

View File

@@ -0,0 +1,295 @@
import java
import semmle.code.java.dataflow.internal.DataFlowPrivate
import semmle.code.java.dataflow.ExternalFlow
import semmle.code.java.dataflow.FlowSummary
bindingset[this]
abstract class CsvRow extends string { }
Type getParameterType(SummarizedCallableExternal callable, int i) {
if i = -1 then result = callable.getDeclaringType() else result = callable.getParameterType(i)
}
string getZero(PrimitiveType t) {
t.hasName("float") and result = "0.0f"
or
t.hasName("double") and result = "0.0"
or
t.hasName("int") and result = "0"
or
t.hasName("boolean") and result = "false"
or
t.hasName("short") and result = "(short)0"
or
t.hasName("byte") and result = "(byte)0"
or
t.hasName("char") and result = "'a'"
or
t.hasName("long") and result = "0L"
}
string getFiller(Type t) {
t instanceof RefType and result = "null"
or
result = getZero(t)
}
Content getContent(SummaryComponent component) { component = SummaryComponent::content(result) }
string getFieldToken(FieldContent fc) {
result =
fc.getField().getDeclaringType().getSourceDeclaration().getName() + "_" +
fc.getField().getName()
}
string contentToken(Content c) {
c instanceof ArrayContent and result = "ArrayElement"
or
c instanceof CollectionContent and result = "Element"
or
c instanceof MapKeyContent and result = "MapKey"
or
c instanceof MapValueContent and result = "MapValue"
or
result = getFieldToken(c)
}
RefType getRootType(RefType t) {
if t instanceof NestedType
then result = getRootType(t.(NestedType).getEnclosingType())
else result = t
}
Type getRootSourceDeclaration(Type t) {
if t instanceof RefType then result = getRootType(t).getSourceDeclaration() else result = t
}
newtype TRowTestSnippet =
MkSnippet(
CsvRow row, SummarizedCallableExternal callable, SummaryComponentStack input,
SummaryComponentStack output, boolean preservesValue
) {
callable.propagatesFlowForRow(input, output, preservesValue, row)
}
class RowTestSnippet extends TRowTestSnippet {
string row;
SummarizedCallableExternal callable;
SummaryComponentStack input;
SummaryComponentStack output;
SummaryComponentStack baseInput;
SummaryComponentStack baseOutput;
boolean preservesValue;
RowTestSnippet() {
this = MkSnippet(row, callable, input, output, preservesValue) and
baseInput = input.drop(input.length() - 1) and
baseOutput = output.drop(output.length() - 1)
}
string toString() {
result =
row + " / " + callable + " / " + input + " / " + output + " / " + baseInput + " / " +
baseOutput + " / " + preservesValue
}
string getArgument(int i) {
(i = -1 or exists(callable.getParameter(i))) and
if baseInput = SummaryComponentStack::argument(i)
then result = "in"
else (
if baseOutput = SummaryComponentStack::argument(i)
then result = "out"
else (
if i = -1 then result = "instance" else result = getFiller(getParameterType(callable, i))
)
)
}
string makeCall() {
// out = in.method(filler);
// or
// out = filler.method(filler, in, filler);
// or
// out = Type.method(filler, in, filler);
// or
// filler.method(filler, in, out, filler);
// or
// Type.method(filler, in, out, filler);
// or
// out = new Type(filler, in, filler);
// or
// new Type(filler, in, out, filler);
exists(string storePrefix, string invokePrefix, string args |
(
if baseOutput = SummaryComponentStack::return()
then storePrefix = "out = "
else storePrefix = ""
) and
(
if callable instanceof Constructor
then invokePrefix = "new "
else
if callable.(Method).isStatic()
then
invokePrefix =
getShortNameIfPossible(callable.getDeclaringType().getSourceDeclaration()) + "."
else invokePrefix = this.getArgument(-1) + "."
) and
args = concat(int i | i >= 0 | this.getArgument(i), ", " order by i) and
result = storePrefix + invokePrefix + callable.getName() + "(" + args + ")"
)
}
string getExpectation() {
preservesValue = true and result = "// $hasValueFlow"
or
preservesValue = false and result = "// $hasTaintFlow"
}
string getInstancePrefix() {
if
callable instanceof Method and
not callable.(Method).isStatic() and
baseOutput != SummaryComponentStack::argument(-1) and
baseInput != SummaryComponentStack::argument(-1)
then
// In this case `out` is the instance.
result = getShortNameIfPossible(callable.getDeclaringType()) + " instance = null;\n"
else result = ""
}
Type getOutputType() {
if baseOutput = SummaryComponentStack::return()
then result = callable.getReturnType()
else
exists(int i |
baseOutput = SummaryComponentStack::argument(i) and
result = getParameterType(callable, i)
)
}
Type getInputType() {
exists(int i |
baseInput = SummaryComponentStack::argument(i) and
result = getParameterType(callable, i)
)
}
string getInputTypeString() { result = getShortNameIfPossible(this.getInputType()) }
string getInput(SummaryComponentStack componentStack) {
componentStack = input.drop(_) and
(
if componentStack = baseInput
then result = "source()"
else
result =
"newWith" + contentToken(getContent(componentStack.head())) + "(" +
this.getInput(componentStack.tail()) + ")"
)
}
string getOutput(SummaryComponentStack componentStack) {
componentStack = output.drop(_) and
(
if componentStack = baseOutput
then result = "out"
else
result =
"get" + contentToken(getContent(componentStack.head())) + "(" +
this.getOutput(componentStack.tail()) + ")"
)
}
string getASupportMethod() {
result =
"Object newWith" + contentToken(getContent(input.drop(_).head())) +
"(Object element) { return null; }" or
result =
"Object get" + contentToken(getContent(output.drop(_).head())) +
"(Object container) { return null; }"
}
string getASupportMethodModel() {
exists(SummaryComponent c, string contentSsvDescription |
c = input.drop(_).head() and c = interpretComponent(contentSsvDescription)
|
result =
"generatedtest;Test;false;newWith" + contentToken(getContent(c)) + ";;;Argument[0];" +
contentSsvDescription + " of ReturnValue;value"
)
or
exists(SummaryComponent c, string contentSsvDescription |
c = output.drop(_).head() and c = interpretComponent(contentSsvDescription)
|
result =
"generatedtest;Test;false;get" + contentToken(getContent(c)) + ";;;" + contentSsvDescription
+ " of Argument[0];ReturnValue;value"
)
}
Type getADesiredImport() {
result =
getRootSourceDeclaration([
this.getOutputType(), this.getInputType(), callable.getDeclaringType()
])
}
string getATestSnippetForRow(string row_) {
row_ = row and
result =
"\t\t{\n\t\t\t// \"" + row + "\"\n\t\t\t" + getShortNameIfPossible(this.getOutputType()) +
" out = null;\n\t\t\t" + this.getInputTypeString() + " in = (" + this.getInputTypeString() +
")" + this.getInput(input) + ";\n\t\t\t" + this.getInstancePrefix() + this.makeCall() +
";\n\t\t\t" + "sink(" + this.getOutput(output) + "); " + this.getExpectation() + "\n\t\t}\n"
}
}
predicate isImportable(Type t) {
t = any(RowTestSnippet r).getADesiredImport() and
t =
unique(Type sharesBaseName |
sharesBaseName = any(RowTestSnippet r).getADesiredImport() and
sharesBaseName.getName() = t.getName()
|
sharesBaseName
)
}
string getShortNameIfPossible(Type t) {
getRootSourceDeclaration(t) = any(RowTestSnippet r).getADesiredImport() and
if t instanceof RefType and not isImportable(getRootSourceDeclaration(t))
then result = t.(RefType).getPackage().getName() + "." + t.getName()
else result = t.getName()
}
string getAnImportStatement() {
exists(RefType t |
t = any(RowTestSnippet r).getADesiredImport() and
isImportable(t) and
t.getPackage().getName() != "java.lang"
|
result = "import " + t.getPackage().getName() + "." + t.getName() + ";"
)
}
string getASupportMethod() {
result = "Object source() { return null; }" or
result = "void sink(Object o) { }" or
result = any(RowTestSnippet r).getASupportMethod()
}
query string getASupportMethodModel() { result = any(RowTestSnippet r).getASupportMethodModel() }
query string getTestCase() {
result =
"package generatedtest;\n\n" + concat(getAnImportStatement() + "\n") +
"\n//Test case generated by GenerateFlowTestCase.ql\npublic class Test {\n\n" +
concat("\t" + getASupportMethod() + "\n") + "\n\tpublic void test() {\n\n" +
concat(string row, string snippet |
snippet = any(RowTestSnippet r).getATestSnippetForRow(row)
|
snippet order by row
) + "\n\t}\n\n}"
}

View File

@@ -0,0 +1,41 @@
]
}
}
class ValueFlowConf extends DataFlow::Configuration {
ValueFlowConf() { this = "qltest:valueFlowConf" }
override predicate isSource(DataFlow::Node n) { n.asExpr().(MethodAccess).getMethod().hasName("source") }
override predicate isSink(DataFlow::Node n) { n.asExpr().(Argument).getCall().getCallee().hasName("sink") }
}
class TaintFlowConf extends TaintTracking::Configuration {
TaintFlowConf() { this = "qltest:taintFlowConf" }
override predicate isSource(DataFlow::Node n) { n.asExpr().(MethodAccess).getMethod().hasName("source") }
override predicate isSink(DataFlow::Node n) { n.asExpr().(Argument).getCall().getCallee().hasName("sink") }
}
class HasFlowTest extends InlineExpectationsTest {
HasFlowTest() { this = "HasFlowTest" }
override string getARelevantTag() { result = ["hasValueFlow", "hasTaintFlow"] }
override predicate hasActualResult(Location location, string element, string tag, string value) {
tag = "hasValueFlow" and
exists(DataFlow::Node src, DataFlow::Node sink, ValueFlowConf conf | conf.hasFlow(src, sink) |
sink.getLocation() = location and
element = sink.toString() and
value = ""
)
or
tag = "hasTaintFlow" and
exists(DataFlow::Node src, DataFlow::Node sink, TaintFlowConf conf | conf.hasFlow(src, sink) |
sink.getLocation() = location and
element = sink.toString() and
value = ""
)
}
}

View File

@@ -0,0 +1,11 @@
import java
import semmle.code.java.dataflow.DataFlow
import semmle.code.java.dataflow.ExternalFlow
import semmle.code.java.dataflow.TaintTracking
import TestUtilities.InlineExpectationsTest
class SummaryModelTest extends SummaryModelCsv {
override predicate row(string row) {
row =
[
//"package;type;overrides;name;signature;ext;inputspec;outputspec;kind",