OSSFuzz Initial Integration (#1198)

fuzzing/build.sh

@@ -0,0 +1,8 @@
+cd "$SRC"/dateparser
+pip3 install .
+
+# Build fuzzers in $OUT
+for fuzzer in $(find fuzzing -name '*_fuzzer.py');do
+  compile_python_fuzzer "$fuzzer"
+done
+zip -q $OUT/dateparser_fuzzer_seed_corpus.zip $SRC/corpus/*

fuzzing/corpus/current

@@ -0,0 +1 @@
+now EST

fuzzing/corpus/date_time

@@ -0,0 +1 @@
+January 12, 2012 10:00 PM

fuzzing/corpus/french

@@ -0,0 +1 @@
+Le 11 Décembre 2014 à 09:00

fuzzing/corpus/minutes_offset

@@ -0,0 +1 @@
+2 minutes ago

fuzzing/corpus/russian

@@ -0,0 +1 @@
+13 января 2015 г. в 13:34

fuzzing/corpus/thai

@@ -0,0 +1 @@
+1 เดือนตุลาคม 2005, 1:00 AM

fuzzing/corpus/time

@@ -0,0 +1 @@
+10:00 am

fuzzing/corpus/time_offset

@@ -0,0 +1 @@
+2 hours ago -0500

fuzzing/corpus/turkish

@@ -0,0 +1 @@
+yaklaşık 23 saat önce

fuzzing/dateparser_fuzzer.py

@@ -0,0 +1,114 @@
+import sys
+from typing import List
+
+import atheris
+from fuzz_helpers import EnhancedFuzzedDataProvider
+
+with atheris.instrument_imports():
+    import dateparser
+
+import re
+
+import pytz
+
+import dateparser.data
+import dateparser.parser
+
+language_codes = dateparser.data.languages_info.language_order
+directives = [
+    "%a",
+    "%A",
+    "%w",
+    "%d",
+    "%b",
+    "%B",
+    "%m",
+    "%y",
+    "%Y",
+    "%H",
+    "%I",
+    "%p",
+    "%M",
+    "%S",
+    "%f",
+    "%z",
+    "%Z",
+    "%j",
+    "%U",
+    "%W",
+    "%c",
+    "%x",
+    "%X",
+    "%%",
+    "%G",
+    "%u",
+    "%V",
+    "%:Z",
+]
+locale_codes = ["fr-PF", "qu-EC", "af-NA"]
+date_order = list(dateparser.parser.date_order_chart.keys())
+timezone = list(pytz.all_timezones)
+preferred_date = ["last", "first", "current"]
+preferred_dates_from = ["past", "future", "current_period"]
+parsers = [
+    "timestamp",
+    "negative-timestamp",
+    "relative-time",
+    "custom-formats",
+    "absolute-time",
+    "no-spaces-time",
+]
+
+
+def _get_format_strings(fdp: EnhancedFuzzedDataProvider) -> List[str]:
+    format_strings = []
+    for _ in range(fdp.ConsumeIntInRange(0, 5)):
+        format_strings.append(fdp.ConsumeString(1).join(fdp.ConsumeSublist(directives)))
+    return format_strings
+
+
+def TestOneInput(data):
+    fdp = EnhancedFuzzedDataProvider(data)
+
+    settings = {
+        "DATE_ORDER": fdp.PickValueInList(date_order),
+        "PREFER_LOCALE_DATE_ORDER": fdp.ConsumeBool(),
+        "TIMEZONE": fdp.PickValueInList(timezone),
+        "TO_TIMEZONE": fdp.PickValueInList(timezone),
+        "RETURN_AS_TIMEZONE_AWARE": fdp.ConsumeBool(),
+        "PREFER_MONTH_OF_YEAR": fdp.PickValueInList(preferred_date),
+        "PREFER_DAY_OF_MONTH": fdp.PickValueInList(preferred_date),
+        "PREFER_DATES_FROM": fdp.PickValueInList(preferred_dates_from),
+        "RELATIVE_BASE": fdp.ConsumeDate(),
+        "STRICT_PARSING": fdp.ConsumeBool(),
+        "REQUIRE_PARTS": [],
+        "SKIP_TOKENS": [
+            fdp.ConsumeRandomString() for _ in range(fdp.ConsumeIntInRange(0, 3))
+        ],
+        "NORMALIZE": fdp.ConsumeBool(),
+        "RETURN_TIME_AS_PERIOD": fdp.ConsumeBool(),
+        "PARSERS": fdp.ConsumeSublist(parsers),
+        "DEFAULT_LANGUAGES": fdp.ConsumeSublist(language_codes),
+        "LANGUAGE_DETECTION_CONFIDENCE_THRESHOLD": fdp.ConsumeProbability(),
+    }
+
+    try:
+        dateparser.parse(
+            fdp.ConsumeRandomString(),
+            date_formats=_get_format_strings(fdp),
+            languages=fdp.ConsumeSublist(language_codes),
+            locales=fdp.ConsumeSublist(locale_codes),
+            region=fdp.ConsumeString(2),
+            settings=settings,
+        )
+    except re.error:
+        return -1
+
+
+def main():
+    atheris.Setup(sys.argv, TestOneInput)
+    atheris.Fuzz()
+
+
+if __name__ == "__main__":
+    main()

fuzzing/fuzz_helpers.py

@@ -0,0 +1,86 @@
+import contextlib
+import datetime
+import io
+import tempfile
+from typing import List, TypeVar
+
+import atheris
+
+T = TypeVar("T")
+
+
+class EnhancedFuzzedDataProvider(atheris.FuzzedDataProvider):
+    def ConsumeRandomBytes(self) -> bytes:
+        return self.ConsumeBytes(self.ConsumeIntInRange(0, self.remaining_bytes()))
+
+    def ConsumeRandomString(self) -> str:
+        return self.ConsumeUnicodeNoSurrogates(
+            self.ConsumeIntInRange(0, self.remaining_bytes())
+        )
+
+    def ConsumeRemainingString(self) -> str:
+        return self.ConsumeUnicodeNoSurrogates(self.remaining_bytes())
+
+    def ConsumeRemainingBytes(self) -> bytes:
+        return self.ConsumeBytes(self.remaining_bytes())
+
+    def ConsumeSublist(self, source: List[T]) -> List[T]:
+        """
+        Returns a shuffled sub-list of the given list of len [1, len(source)]
+        """
+        chosen = [elem for elem in source if self.ConsumeBool()]
+
+        # Shuffle
+        for i in range(len(chosen) - 1, 1, -1):
+            j = self.ConsumeIntInRange(0, i)
+            chosen[i], chosen[j] = chosen[j], chosen[i]
+
+        return chosen or [self.PickValueInList(source)]
+
+    def ConsumeDate(self) -> datetime.datetime:
+        try:
+            return datetime.datetime.fromtimestamp(self.ConsumeFloat())
+        except (OverflowError, OSError, ValueError):
+            return datetime.datetime(year=1970, month=1, day=1)
+
+    @contextlib.contextmanager
+    def ConsumeMemoryFile(
+        self, all_data: bool = False, as_bytes: bool = True
+    ) -> io.BytesIO:
+        if all_data:
+            file_data = (
+                self.ConsumeRemainingBytes()
+                if as_bytes
+                else self.ConsumeRemainingString()
+            )
+        else:
+            file_data = (
+                self.ConsumeRandomBytes() if as_bytes else self.ConsumeRandomString()
+            )
+
+        file = io.BytesIO(file_data) if as_bytes else io.StringIO(file_data)
+        yield file
+        file.close()
+
+    @contextlib.contextmanager
+    def ConsumeTemporaryFile(
+        self, suffix: str, all_data: bool = False, as_bytes: bool = True
+    ) -> str:
+        if all_data:
+            file_data = (
+                self.ConsumeRemainingBytes()
+                if as_bytes
+                else self.ConsumeRemainingString()
+            )
+        else:
+            file_data = (
+                self.ConsumeRandomBytes() if as_bytes else self.ConsumeRandomString()
+            )
+
+        mode = "w+b" if as_bytes else "w+"
+        tfile = tempfile.NamedTemporaryFile(mode=mode, suffix=suffix)
+        tfile.write(file_data)
+        tfile.seek(0)
+        tfile.flush()
+        yield tfile.name
+        tfile.close()

fuzzing/requirements.txt

@@ -0,0 +1 @@
+atheris

tox.ini

@@ -5,6 +5,7 @@ envlist = flake8, py3
 deps =
     -rdateparser_scripts/requirements.txt
     -rtests/requirements.txt
+    -rfuzzing/requirements.txt
 
 [testenv]
 deps =