Cargo feature to avoid regex dependency (#1102)

regex is a large crate, and is only used in a few specific spots. Users should
have the ability to avoid this transitive dependency if not using the features
in question.

libafl/Cargo.toml

@@ -12,8 +12,8 @@ edition = "2021"
 categories = ["development-tools::testing", "emulators", "embedded", "os", "no-std"]
 
 [features]
-default = ["std", "derive", "llmp_compression", "llmp_small_maps", "llmp_broker_timeouts", "rand_trait", "fork", "prelude", "gzip"]
-std = ["serde_json", "serde_json/std", "hostname", "nix", "serde/std", "bincode", "wait-timeout", "regex", "byteorder", "once_cell", "uuid", "tui_monitor", "ctor", "backtrace", "uds"] # print, env, launcher ... support
+default = ["std", "derive", "llmp_compression", "llmp_small_maps", "llmp_broker_timeouts", "rand_trait", "fork", "prelude", "gzip", "regex"]
+std = ["serde_json", "serde_json/std", "hostname", "nix", "serde/std", "bincode", "wait-timeout", "byteorder", "once_cell", "uuid", "tui_monitor", "ctor", "backtrace", "uds"] # print, env, launcher ... support
 derive = ["libafl_derive"] # provide derive(SerdeAny) macro.
 fork = [] # uses the fork() syscall to spawn children, instead of launching a new command, if supported by the OS (has no effect on Windows, no_std).
 rand_trait = ["rand_core"] # If set, libafl's rand implementations will implement `rand::Rng`
@@ -31,6 +31,7 @@ errors_backtrace = ["backtrace"] # Create backtraces at Error creation
 cmin = ["z3"] # for corpus minimisation
 corpus_btreemap = [] # Switches from HashMap to BTreeMap for CorpusId
 gzip = ["miniz_oxide"] # Enables gzip compression in certain parts of the lib
+regex = ["std", "dep:regex"] # enables the NaiveTokenizer and StacktraceObserver
 
 # features hiding dependencies licensed under GPL
 gpl = []

libafl/src/executors/forkserver.rs

@@ -35,14 +35,14 @@ use crate::{
     executors::{Executor, ExitKind, HasObservers},
     inputs::{HasTargetBytes, Input, UsesInput},
     mutators::Tokens,
-    observers::{
-        get_asan_runtime_flags_with_log_path, AsanBacktraceObserver, MapObserver, Observer,
-        ObserversTuple, UsesObservers,
-    },
+    observers::{MapObserver, Observer, ObserversTuple, UsesObservers},
     state::UsesState,
     Error,
 };
 
+#[cfg(feature = "regex")]
+use crate::observers::{get_asan_runtime_flags_with_log_path, AsanBacktraceObserver};
+
 const FORKSRV_FD: i32 = 198;
 #[allow(clippy::cast_possible_wrap)]
 const FS_OPT_ENABLED: i32 = 0x80000001_u32 as i32;
@@ -232,9 +232,11 @@ impl Forkserver {
             command.env("__AFL_DEFER_FORKSRV", "1");
         }
 
+        #[cfg(feature = "regex")]
+        command.env("ASAN_OPTIONS", get_asan_runtime_flags_with_log_path());
+
         match command
             .env("LD_BIND_NOW", "1")
-            .env("ASAN_OPTIONS", get_asan_runtime_flags_with_log_path())
             .envs(envs)
             .setlimit(memlimit)
             .setsid()
@@ -1129,6 +1131,7 @@ where
 
         if libc::WIFSIGNALED(self.forkserver.status()) {
             exit_kind = ExitKind::Crash;
+            #[cfg(feature = "regex")]
             if self.has_asan_observer.is_none() {
                 self.has_asan_observer = Some(
                     self.observers()
@@ -1136,6 +1139,7 @@ where
                         .is_some(),
                 );
             }
+            #[cfg(feature = "regex")]
             if self.has_asan_observer.unwrap() {
                 self.observers_mut()
                     .match_name_mut::<AsanBacktraceObserver>("AsanBacktraceObserver")

libafl/src/inputs/encoded.rs

@@ -15,7 +15,7 @@ use core::{
 
 use ahash::RandomState;
 use hashbrown::HashMap;
-#[cfg(feature = "std")]
+#[cfg(feature = "regex")]
 use regex::Regex;
 use serde::{Deserialize, Serialize};
 
@@ -108,7 +108,7 @@ impl Default for TokenInputEncoderDecoder {
 }
 
 /// A naive tokenizer struct
-#[cfg(feature = "std")]
+#[cfg(feature = "regex")]
 #[derive(Clone, Debug)]
 pub struct NaiveTokenizer {
     /// Ident regex
@@ -119,7 +119,7 @@ pub struct NaiveTokenizer {
     string_re: Regex,
 }
 
-#[cfg(feature = "std")]
+#[cfg(feature = "regex")]
 impl NaiveTokenizer {
     /// Creates a new [`NaiveTokenizer`]
     #[must_use]
@@ -132,7 +132,7 @@ impl NaiveTokenizer {
     }
 }
 
-#[cfg(feature = "std")]
+#[cfg(feature = "regex")]
 impl Default for NaiveTokenizer {
     fn default() -> Self {
         Self {
@@ -146,7 +146,7 @@ impl Default for NaiveTokenizer {
     }
 }
 
-#[cfg(feature = "std")]
+#[cfg(feature = "regex")]
 impl Tokenizer for NaiveTokenizer {
     fn tokenize(&self, bytes: &[u8]) -> Result<Vec<String>, Error> {
         let mut tokens = vec![];
@@ -259,7 +259,7 @@ impl EncodedInput {
     }
 }
 
-#[cfg(feature = "std")]
+#[cfg(feature = "regex")]
 #[cfg(test)]
 mod tests {
     use alloc::borrow::ToOwned;

libafl/src/observers/mod.rs

@@ -11,9 +11,9 @@ pub mod stdio;
 #[cfg(feature = "std")]
 pub use stdio::{StdErrObserver, StdOutObserver};
 
-#[cfg(feature = "std")]
+#[cfg(feature = "regex")]
 pub mod stacktrace;
-#[cfg(feature = "std")]
+#[cfg(feature = "regex")]
 pub use stacktrace::*;
 
 pub mod concolic;