From 2f806711f8791c45d9ec7a004e78802545f7aa6f Mon Sep 17 00:00:00 2001
From: ebell495 <ericbell495@gmail.com>
Date: Wed, 6 Jul 2022 15:48:43 -0400
Subject: [PATCH] Update Fuzz Harness

---
 fuzz/Dockerfile |  2 +-
 fuzz/fuzz.py    | 35 ++++++++++++++---------------------
 2 files changed, 15 insertions(+), 22 deletions(-)

diff --git a/fuzz/Dockerfile b/fuzz/Dockerfile
index ec2cdde..d8a8ab8 100644
--- a/fuzz/Dockerfile
+++ b/fuzz/Dockerfile
@@ -34,7 +34,7 @@ RUN apt-get update && apt-get upgrade -y && apt-get install -y \
     apt-get clean
 
 # pytest needs to be installed through pip to make sure we have a recent version
-RUN pip3 install pytest atheris
+RUN pip3 install pytest atheris segments dlinfo
 
 # tests expect python to be available as executable 'python' not 'python3'
 RUN ln -s /usr/bin/python3 /usr/bin/python
diff --git a/fuzz/fuzz.py b/fuzz/fuzz.py
index 63775c5..267dfdc 100644
--- a/fuzz/fuzz.py
+++ b/fuzz/fuzz.py
@@ -3,13 +3,8 @@
 import sys
 import os
 import atheris
-# old_stdout = sys.stdout # backup current stdout
-# sys.stdout = open(os.devnull, "w")
-with atheris.instrument_imports():
-    from phonemizer.backend import EspeakBackend, FestivalBackend
-# sys.stdout = old_stdout
-# from phonemizer import phonemize
-# from phonemizer.backend import EspeakBackend, FestivalBackend
+
+from phonemizer.backend import EspeakBackend, FestivalBackend
 
 # Initilize the backends separatly
 # Running phonemizer repeatly increases the memory usage
@@ -18,21 +13,19 @@
 
 @atheris.instrument_func
 def TestOneInput(data):
-    barray = bytearray(data)
-    # espeak.phonemize(str(data).split(" "))
-    if len(barray) > 0:
-        # Choose the backend to use based on the first input byte
-        r = barray[0]
-        if r % 2 == 0:
-            # Make sure to remove the first byte otherwise this will only every test this backend with the first byte being even
-            del barray[0]
-            espeak.phonemize(str(data).split(' '))
-        else:
-            del barray[0]
-            festival.phonemize(str(data).split(' '))
+    fdp = atheris.FuzzedDataProvider(data)
+
+    if len(data) < 1:
+        return
+
+    option = fdp.ConsumeBytes(1)[0]
+    in_string = fdp.ConsumeUnicodeNoSurrogates(len(data))
+
+    if option % 2 == 0:
+        espeak.phonemize(in_string.split(" "))
     else:
-        espeak.phonemize(str(data).split(' '))
-        festival.phonemize(str(data).split(' '))
+        festival.phonemize(in_string.split(" "))
 
+atheris.instrument_all()
 atheris.Setup(sys.argv, TestOneInput)
 atheris.Fuzz()
\ No newline at end of file