Rust: Initialisation of OFG for Rust projects (#781)

This PR initialise OFG for rust projects. This PR superceeds #762. This PR also fixed the experimental from_scratch approach to work for Rust projects. --------- Signed-off-by: Arthur Chan <[email protected]>
google · Feb 12, 2025 · 248a804 · 248a804
1 parent de2ab8c
commit 248a804
Show file tree

Hide file tree

Showing 22 changed files with 539 additions and 35 deletions.
diff --git a/agent/one_prompt_prototyper.py b/agent/one_prompt_prototyper.py
@@ -55,6 +55,10 @@ def _prompt_builder(self,
       # For Python projects
       return prompt_builder.DefaultPythonTemplateBuilder(
           self.llm, benchmark, self.args.template_directory)
+    if benchmark.language == 'rust':
+      # For Rust projects
+      return prompt_builder.DefaultRustTemplateBuilder(
+          self.llm, benchmark, self.args.template_directory)
 
     if self.args.prompt_builder == 'CSpecific':
       return prompt_builder.CSpecificBuilder(self.llm, benchmark,

diff --git a/benchmark-sets/rust-small/askama.yaml b/benchmark-sets/rust-small/askama.yaml
@@ -0,0 +1,37 @@
+"functions":
+- "name": "derive_template"
+  "params":
+  - "name": "input"
+    "type": "TokenStream"
+  "return_type": "TokenStream"
+  "signature": "pub fn derive_template(input: TokenStream) -> TokenStream"
+- "name": "build_template"
+  "params":
+  - "name": "ast"
+    "type": "&syn::DeriveInput"
+  "return_type": "Result"
+  "signature": "pub(crate) fn build_template(ast: &syn::DeriveInput) -> Result<String, CompileError>"
+- "name": "build_skeleton"
+  "params":
+  - "name": "ast"
+    "type": "&syn::DeriveInput"
+  "return_type": "Result"
+  "signature": "fn build_skeleton(ast: &syn::DeriveInput) -> Result<String, CompileError>"
+- "name": "Generator::build"
+  "params":
+  - "name": "ctx"
+    "type": "&Context<'a>"
+  "return_type": "Result"
+  "signature": "pub(crate) fn build(mut self, ctx: &Context<'a>) -> Result<String, CompileError>"
+- "name": "Generator::impl_template"
+  "params":
+  - "name": "ctx"
+    "type": "&Context<'a>"
+  - "name": "buf"
+    "type": "&mut Buffer"
+  "return_type": "Result"
+  "signature": "fn impl_template(&mut self, ctx: &Context<'a>, buf: &mut Buffer) -> Result<(), CompileError>"
+"language": "rust"
+"project": "askama"
+"target_name": "fuzz_filters"
+"target_path": "/src/askama/fuzz/fuzz_targets/fuzz_filters.rs"
diff --git a/benchmark-sets/rust-small/bincode.yaml b/benchmark-sets/rust-small/bincode.yaml
@@ -0,0 +1,29 @@
+"functions":
+- "name": "DeriveStruct::generate_decode"
+  "params":
+  - "name": "generator"
+    "type": "&mut Generator"
+  "return_type": "Result"
+  "signature": "pub fn generate_decode(self, generator: &mut Generator) -> Result<()>"
+- "name": "AllTypes::encode"
+  "params":
+  - "name": "encoder"
+    "type": "& mut __E"
+  "return_type": "core :: result :: Result"
+  "signature": "fn encode < __E : :: bincode :: enc :: Encoder >(& self, encoder : & mut __E) ->core :: result :: Result < (), :: bincode:: error :: EncodeError >"
+- "name": "DeriveEnum::generate_borrow_decode"
+  "params":
+  - "name": "generator"
+    "type": "&mut Generator"
+  "return_type": "Result"
+  "signature": "pub fn generate_borrow_decode(self, generator: &mut Generator) -> Result<()>"
+- "name": "AllTypes::borrow_decode"
+  "params":
+  - "name": "decoder"
+    "type": "& mut __D"
+  "return_type": "core :: result :: Result"
+  "signature": "fn borrow_decode < __D : :: bincode :: de :: BorrowDecoder < '__de > >(decoder : & mut __D) ->core :: result :: Result < Self, :: bincode ::error :: DecodeError >"
+"language": "rust"
+"project": "bincode"
+"target_name": "roundtrip"
+"target_path": "/src/bincode/fuzz/fuzz_targets/roundtrip.rs"
diff --git a/benchmark-sets/rust-small/httparse.yaml b/benchmark-sets/rust-small/httparse.yaml
@@ -0,0 +1,23 @@
+"functions":
+- "name": "parse_reason"
+  "params":
+  - "name": "bytes"
+    "type": "&mut Bytes<'a>"
+  "return_type": "Result"
+  "signature": "fn parse_reason<'a>(bytes: &mut Bytes<'a>) -> Result<&'a str>"
+- "name": "enable_simd"
+  "params":
+  - "name": "version"
+    "type": "Version"
+  "return_type": "void"
+  "signature": "fn enable_simd(version: Version)"
+- "name": "match_uri_vectored"
+  "params":
+  - "name": "bytes"
+    "type": "&mut Bytes"
+  "return_type": "void"
+  "signature": "fn match_uri_vectored(bytes: &mut Bytes)"
+"language": "rust"
+"project": "httparse"
+"target_name": "parse_chunk_size"
+"target_path": "Did-not-find-sourcefile"
diff --git a/benchmark-sets/rust-small/itoa.yaml b/benchmark-sets/rust-small/itoa.yaml
@@ -0,0 +1,25 @@
+"functions":
+- "name": "udivmod_1e19"
+  "params":
+  - "name": "n"
+    "type": "u128"
+  "return_type": "(u128, u64)"
+  "signature": "pub fn udivmod_1e19(n: u128) -> (u128, u64)"
+- "name": "u128_mulhi"
+  "params":
+  - "name": "x"
+    "type": "u128"
+  - "name": "y"
+    "type": "u128"
+  "return_type": "u128"
+  "signature": "fn u128_mulhi(x: u128, y: u128) -> u128"
+- "name": "Buffer::format"
+  "params":
+  - "name": "i"
+    "type": "I"
+  "return_type": "&str"
+  "signature": "pub fn format<I: Integer>(&mut self, i: I) -> &str"
+"language": "rust"
+"project": "itoa"
+"target_name": "fuzz_itoa"
+"target_path": "/src/itoa/fuzz/fuzz_targets/fuzz_itoa.rs"
diff --git a/benchmark-sets/rust-small/ryu.yaml b/benchmark-sets/rust-small/ryu.yaml
@@ -0,0 +1,35 @@
+"functions":
+- "name": "s2f"
+  "params":
+  - "name": "buffer"
+    "type": "&[u8]"
+  "return_type": "Result"
+  "signature": "pub fn s2f(buffer: &[u8]) -> Result<f32, Error>"
+- "name": "s2d"
+  "params":
+  - "name": "buffer"
+    "type": "&[u8]"
+  "return_type": "Result"
+  "signature": "pub fn s2d(buffer: &[u8]) -> Result<f64, Error>"
+- "name": "Buffer::format"
+  "params":
+  - "name": "f"
+    "type": "F"
+  "return_type": "&str"
+  "signature": "pub fn format<F: Float>(&mut self, f: F) -> &str"
+- "name": "compute_pow5"
+  "params":
+  - "name": "i"
+    "type": "u32"
+  "return_type": "(u64, u64)"
+  "signature": "pub unsafe fn compute_pow5(i: u32) -> (u64, u64)"
+- "name": "compute_inv_pow5"
+  "params":
+  - "name": "i"
+    "type": "u32"
+  "return_type": "(u64, u64)"
+  "signature": "pub unsafe fn compute_inv_pow5(i: u32) -> (u64, u64)"
+"language": "rust"
+"project": "ryu"
+"target_name": "fuzz_ryu"
+"target_path": "/src/ryu/fuzz/fuzz_targets/fuzz_ryu.rs"
diff --git a/benchmark-sets/rust-small/utf8parse.yaml b/benchmark-sets/rust-small/utf8parse.yaml
@@ -0,0 +1,45 @@
+"functions":
+- "name": "expand_state_table"
+  "params":
+  - "name": "cx"
+    "type": "&'cx mut ExtCtxt"
+  - "name": "sp"
+    "type": "Span"
+  - "name": "args"
+    "type": "&[TokenTree]"
+  "return_type": "Box"
+  "signature": "fn expand_state_table<'cx>(cx: &'cx mut ExtCtxt,sp: Span,args: &[TokenTree])-> Box<MacResult + 'cx>"
+- "name": "parse_raw_definitions"
+  "params":
+  - "name": "definitions"
+    "type": "Vec<TableDefinitionExprs>"
+  - "name": "cx"
+    "type": "&mut ExtCtxt"
+  "return_type": "Result"
+  "signature": "fn parse_raw_definitions(definitions: Vec<TableDefinitionExprs>,cx: &mut ExtCtxt) -> Result<Vec<TableDefinition>, ()>"
+- "name": "input_mapping_from_arm"
+  "params":
+  - "name": "arm"
+    "type": "Arm"
+  - "name": "cx"
+    "type": "&mut ExtCtxt"
+  "return_type": "Result"
+  "signature": "fn input_mapping_from_arm(arm: Arm, cx: &mut ExtCtxt) -> Result<InputMapping, ()>"
+- "name": "Transition::from_expr"
+  "params":
+  - "name": "expr"
+    "type": "&Expr"
+  - "name": "cx"
+    "type": "&mut ExtCtxt"
+  "return_type": "Result"
+  "signature": "fn from_expr(expr: &Expr, cx: &mut ExtCtxt) -> Result<Transition, ()>"
+- "name": "build_state_tables"
+  "params":
+  - "name": "defs"
+    "type": "T"
+  "return_type": "[[u8; 256]; 16]"
+  "signature": "fn build_state_tables<T>(defs: T) -> [[u8; 256]; 16]where T: AsRef<[TableDefinition]>"
+"language": "rust"
+"project": "utf8parse"
+"target_name": "parse"
+"target_path": "/src/vte/utf8parse/fuzz/fuzz_targets/parse.rs"
diff --git a/data_prep/introspector.py b/data_prep/introspector.py
@@ -57,7 +57,7 @@
 INTROSPECTOR_ORACLE_FAR_REACH = ''
 INTROSPECTOR_ORACLE_KEYWORD = ''
 INTROSPECTOR_ORACLE_EASY_PARAMS = ''
-INTROSPECTOR_ORACLE_ALL_JVM_PUBLIC_CANDIDATES = ''
+INTROSPECTOR_ORACLE_ALL_PUBLIC_CANDIDATES = ''
 INTROSPECTOR_ORACLE_OPTIMAL = ''
 INTROSPECTOR_ORACLE_ALL_TESTS = ''
 INTROSPECTOR_FUNCTION_SOURCE = ''
@@ -90,6 +90,7 @@ def get_oracle_dict() -> Dict[str, Any]:
       'jvm-public-candidates': query_introspector_jvm_all_public_candidates,
       'optimal-targets': query_introspector_for_optimal_targets,
       'test-migration': query_introspector_for_tests,
+      'all-public-candidates': query_introspector_all_public_candidates,
   }
   return oracle_dict
 
@@ -102,7 +103,7 @@ def set_introspector_endpoints(endpoint):
       INTROSPECTOR_ORACLE_KEYWORD, INTROSPECTOR_ADDR_TYPE, \
       INTROSPECTOR_ALL_HEADER_FILES, INTROSPECTOR_ALL_FUNC_TYPES, \
       INTROSPECTOR_SAMPLE_XREFS, INTROSPECTOR_ORACLE_EASY_PARAMS, \
-      INTROSPECTOR_ORACLE_ALL_JVM_PUBLIC_CANDIDATES, \
+      INTROSPECTOR_ORACLE_ALL_PUBLIC_CANDIDATES, \
       INTROSPECTOR_ALL_JVM_SOURCE_PATH, INTROSPECTOR_ORACLE_OPTIMAL, \
       INTROSPECTOR_HEADERS_FOR_FUNC, \
       INTROSPECTOR_FUNCTION_WITH_MATCHING_RETURN_TYPE, \
@@ -119,7 +120,7 @@ def set_introspector_endpoints(endpoint):
       f'{INTROSPECTOR_ENDPOINT}/far-reach-low-cov-fuzz-keyword')
   INTROSPECTOR_ORACLE_EASY_PARAMS = (
       f'{INTROSPECTOR_ENDPOINT}/easy-params-far-reach')
-  INTROSPECTOR_ORACLE_ALL_JVM_PUBLIC_CANDIDATES = (
+  INTROSPECTOR_ORACLE_ALL_PUBLIC_CANDIDATES = (
       f'{INTROSPECTOR_ENDPOINT}/all-public-candidates')
   INTROSPECTOR_ORACLE_OPTIMAL = f'{INTROSPECTOR_ENDPOINT}/optimal-targets'
   INTROSPECTOR_FUNCTION_SOURCE = f'{INTROSPECTOR_ENDPOINT}/function-source-code'
@@ -277,8 +278,17 @@ def query_introspector_jvm_all_public_candidates(project: str) -> list[dict]:
   """Queries Fuzz Introspector for all public accessible function or
   constructor candidates.
   """
-  return query_introspector_oracle(
-      project, INTROSPECTOR_ORACLE_ALL_JVM_PUBLIC_CANDIDATES)
+  return query_introspector_oracle(project,
+                                   INTROSPECTOR_ORACLE_ALL_PUBLIC_CANDIDATES)
+
+
+def query_introspector_all_public_candidates(project: str) -> list[dict]:
+  """Queries Fuzz Introspector for all public accessible function or
+  constructor candidates.
+  """
+  #TODO May combine this with query_introspector_jvm_all_public_candidates
+  return query_introspector_oracle(project,
+                                   INTROSPECTOR_ORACLE_ALL_PUBLIC_CANDIDATES)
 
 
 def query_introspector_for_targets(project, target_oracle) -> list[Dict]:
@@ -876,9 +886,9 @@ def populate_benchmarks_using_introspector(project: str, language: str,
         if not any(src_path.endswith(src_file) for src_path in src_path_list):
           logger.error('error: %s %s', filename, interesting.keys())
           continue
-    elif interesting and filename not in [
-        os.path.basename(i) for i in interesting.keys()
-    ]:
+
+    elif (language not in ['rust'] and interesting and
+          filename not in [os.path.basename(i) for i in interesting.keys()]):
       # TODO: Bazel messes up paths to include "/proc/self/cwd/..."
       logger.error('error: %s %s', filename, interesting.keys())
       continue

diff --git a/data_prep/project_src.py b/data_prep/project_src.py
@@ -96,6 +96,8 @@ def _get_harness(src_file: str, out: str, language: str) -> tuple[str, str]:
     return '', ''
   if language.lower() == 'python' and 'atheris.Fuzz()' not in content:
     return '', ''
+  if language.lower() == 'rust' and 'fuzz_target!' not in content:
+    return '', ''
 
   short_path = src_file[len(out):]
   return short_path, content
@@ -233,9 +235,9 @@ def _copy_project_src_from_local(project: str, out: str, language: str):
     # Sometimes the previous container need longer time to delete
     # If the next docker run is invoked before the previous container
     # completely removed, it will resulti n Conflict error.
-    # Sleep for 60 seconds and retry.
-    logger.warning('Failed to run OSS-Fuzz on %s, retry in 60 sec', project)
-    time.sleep(60)
+    # Sleep for 180 seconds and retry.
+    logger.warning('Failed to run OSS-Fuzz on %s, retry in 180 sec', project)
+    time.sleep(180)
     result = sp.run(run_container,
                     capture_output=True,
                     stdin=sp.DEVNULL,
@@ -307,6 +309,12 @@ def _identify_fuzz_targets(out: str, interesting_filenames: list[str],
           interesting_filepaths.append(path)
         if path.endswith('.py'):
           potential_harnesses.append(path)
+      elif language == 'rust':
+        # For Rust
+        if path.endswith(tuple(interesting_filenames)):
+          interesting_filepaths.append(path)
+        if path.endswith('.rs'):
+          potential_harnesses.append(path)
       else:
         # For C/C++
         short_path = path[len(out):]

diff --git a/experiment/benchmark.py b/experiment/benchmark.py
@@ -204,6 +204,13 @@ def __init__(self,
       # zipp-zipp.difference.
       self.id = self.id.replace('._', '.')
 
+    if self.language == 'rust':
+      # For rust projects, double colon (::) is sometime used to identify
+      # crate, impl or trait name of a function. This could affect the
+      # benchmark_id and cause OSS-Fuzz build failed.
+      # Special handling of benchmark_id is needed to avoid this situation.
+      self.id = self.id.replace('::', '-')
+
   def __str__(self):
     return (f'Benchmark<id={self.id}, project={self.project}, '
             f'language={self.language}, '