Also fuzz None values

pdfminer · Jun 27, 2024 · 9b55c8d · 9b55c8d
1 parent cd4d715
commit 9b55c8d
Show file tree

Hide file tree

Showing 5 changed files with 34 additions and 30 deletions.
diff --git a/fuzzing/extract_text_fuzzer.py b/fuzzing/extract_text_fuzzer.py
@@ -19,14 +19,12 @@ def fuzz_one_input(data: bytes) -> None:
     fdp = EnhancedFuzzedDataProvider(data)
 
     try:
-        with fdp.ConsumeMemoryFile() as f:
-            max_pages = fdp.ConsumeIntInRange(0, 1000)
-            extract_text(
-                f,
-                maxpages=max_pages,
-                page_numbers=fdp.ConsumeIntList(fdp.ConsumeIntInRange(0, max_pages), 2),
-                laparams=PDFValidator.generate_layout_parameters(fdp),
-            )
+        extract_text(
+            fdp.ConsumeMemoryFile(),
+            maxpages=fdp.ConsumeIntInRange(0, 10),
+            page_numbers=fdp.ConsumeOptionalIntList(10, 0, 10),
+            laparams=PDFValidator.generate_layout_parameters(fdp),
+        )
     except (AssertionError, PSException):
         return
     except Exception as e:

diff --git a/fuzzing/extract_text_to_fp_fuzzer.py b/fuzzing/extract_text_to_fp_fuzzer.py
@@ -23,14 +23,13 @@ def fuzz_one_input(data: bytes) -> None:
 
     try:
         with fdp.ConsumeMemoryFile(all_data=False) as f_in, io.BytesIO() as f_out:
-            max_pages = fdp.ConsumeIntInRange(0, 1000)
             extract_text_to_fp(
                 f_in,
                 f_out,
                 output_type=fdp.PickValueInList(available_output_formats),
                 laparams=PDFValidator.generate_layout_parameters(fdp),
-                maxpages=max_pages,
-                page_numbers=fdp.ConsumeIntList(fdp.ConsumeIntInRange(0, max_pages), 2),
+                maxpages=fdp.ConsumeIntInRange(0, 10),
+                page_numbers=fdp.ConsumeOptionalIntList(10, 0, 10),
                 scale=fdp.ConsumeFloatInRange(0.0, 2.0),
                 rotation=fdp.ConsumeIntInRange(0, 360),
                 layoutmode=fdp.PickValueInList(available_layout_modes),

diff --git a/fuzzing/fuzz_helpers.py b/fuzzing/fuzz_helpers.py
@@ -1,4 +1,5 @@
 import io
+from typing import List, Optional
 
 from atheris import FuzzedDataProvider
 
@@ -23,3 +24,11 @@ def ConsumeMemoryFile(self, all_data: bool = False) -> io.BytesIO:
             return io.BytesIO(self.ConsumeRemainingBytes())
         else:
             return io.BytesIO(self.ConsumeRandomBytes())
+
+    def ConsumeOptionalIntList(
+        self, max_count: int, min: int, max: int
+    ) -> Optional[List[int]]:
+        if self.ConsumeBool():
+            count = self.ConsumeIntInRange(0, max_count)
+            return [int(i) for i in self.ConsumeIntListInRange(count, min, max)]
+        return None
diff --git a/fuzzing/page_extraction_fuzzer.py b/fuzzing/page_extraction_fuzzer.py
@@ -19,14 +19,11 @@ def fuzz_one_input(data: bytes) -> None:
 
     try:
         with fdp.ConsumeMemoryFile() as f:
-            max_pages = fdp.ConsumeIntInRange(0, 1000)
             list(
                 extract_pages(
                     f,
-                    maxpages=max_pages,
-                    page_numbers=fdp.ConsumeIntList(
-                        fdp.ConsumeIntInRange(0, max_pages), 2
-                    ),
+                    maxpages=fdp.ConsumeIntInRange(0, 10),
+                    page_numbers=fdp.ConsumeOptionalIntList(10, 0, 10),
                     laparams=PDFValidator.generate_layout_parameters(fdp),
                 )
             )

diff --git a/fuzzing/pdf_utils.py b/fuzzing/pdf_utils.py
@@ -52,20 +52,21 @@ def is_valid_byte_stream(data: bytes) -> bool:
     def generate_layout_parameters(
         fdp: atheris.FuzzedDataProvider,
     ) -> Optional[LAParams]:
-        return (
-            LAParams(
-                line_overlap=fdp.ConsumeFloat(),
-                char_margin=fdp.ConsumeFloat(),
-                line_margin=fdp.ConsumeFloat(),
-                word_margin=fdp.ConsumeFloat(),
-                boxes_flow=fdp.ConsumeFloatInRange(-1.0, 1.0)
-                if fdp.ConsumeBool()
-                else None,
-                detect_vertical=fdp.ConsumeBool(),
-                all_texts=fdp.ConsumeBool(),
-            )
-            if fdp.ConsumeBool()
-            else None
+        if fdp.ConsumeBool():
+            return None
+
+        boxes_flow: Optional[float] = None
+        if fdp.ConsumeBool():
+            boxes_flow = fdp.ConsumeFloatInRange(-1.0, 1.0)
+
+        return LAParams(
+            line_overlap=fdp.ConsumeFloat(),
+            char_margin=fdp.ConsumeFloat(),
+            line_margin=fdp.ConsumeFloat(),
+            word_margin=fdp.ConsumeFloat(),
+            boxes_flow=boxes_flow,
+            detect_vertical=fdp.ConsumeBool(),
+            all_texts=fdp.ConsumeBool(),
         )
 
     @staticmethod