[libcasr] Add JS stacktrace parsing (#158)

ispras · Oct 21, 2023 · f654bcd · f654bcd
1 parent 0ac3deb
commit f654bcd
Show file tree

Hide file tree

Showing 9 changed files with 468 additions and 8 deletions.
diff --git a/libcasr/fuzz/fuzz_targets/parse_stacktrace.rs b/libcasr/fuzz/fuzz_targets/parse_stacktrace.rs
@@ -9,6 +9,7 @@ use libcasr::{
     init_ignored_frames,
     java::JavaStacktrace,
     python::PythonStacktrace,
+    js::JsStacktrace,
     stacktrace::{CrashLineExt, Filter, ParseStacktrace, Stacktrace},
 };
 
@@ -17,8 +18,8 @@ fuzz_target!(|data: &[u8]| {
         return;
     }
     let s = String::from_utf8_lossy(&data[1..]);
-    init_ignored_frames!("cpp", "rust", "python", "go", "java");
-    match data[0] % 5 {
+    init_ignored_frames!("cpp", "rust", "python", "go", "java", "js");
+    match data[0] % 6 {
         0 => {
             // Asan
             if let Ok(raw) = AsanStacktrace::extract_stacktrace(&s) {
@@ -51,6 +52,14 @@ fuzz_target!(|data: &[u8]| {
                 }
             }
         }
+        4 => {
+            // JS
+            if let Ok(raw) = JsStacktrace::extract_stacktrace(&s) {
+                if let Ok(st) = JsStacktrace::parse_stacktrace(&raw) {
+                    let _ = st.crash_line();
+                }
+            }
+        }
         _ => {
             // Gdb
             if let Ok(raw) = GdbStacktrace::extract_stacktrace(&s) {

diff --git a/libcasr/fuzz/init_corpus/gdb b/libcasr/fuzz/init_corpus/gdb
@@ -1,4 +1,4 @@
-     #0  0x00000000005e3099 in xlnt::detail::compound_document::read_directory (this=0x7fffffffcee0) at /xlnt/source/detail/cryptography/compound_document.cpp:975
+     #0  0x00000000005e3099 in xlnt::detail::compound_document::read_directory (this=0x7fffffffcee0) at /xlnt/source/detail/cryptography/compound_document.cpp:975
      #1  0x00000000005e2956 in xlnt::detail::compound_document::compound_document (this=0x7fffffffcee0, in=...) at /xlnt/source/detail/cryptography/compound_document.cpp:517
      #2  0x000000000048a45c in (anonymous namespace)::decrypt_xlsx (bytes=std::vector of length 18655, capacity 32768 = {...}, password=u\"VelvetSweatshop\") at /xlnt/source/detail/cryptography/xlsx_crypto_consumer.cpp:320
      #3  0x000000000048a2d9 in xlnt::detail::decrypt_xlsx (data=std::vector of length 18655, capacity 32768 = {...}, password=) at /xlnt/source/detail/cryptography/xlsx_crypto_consumer.cpp:339

diff --git a/libcasr/fuzz/init_corpus/js b/libcasr/fuzz/init_corpus/js
@@ -0,0 +1,13 @@
+Error: 1
+    at eval (eval at <anonymous> (eval at g (/fuzz/FuzzTarget_jazzer.js:7:7)), <anonymous>:4:23)
+    at eval (eval at g (/fuzz/FuzzTarget_jazzer.js:7:7), <anonymous>:8:13)
+    at g (/fuzz/FuzzTarget_jazzer.js:14:13)
+    at fuzz (/fuzz/FuzzTarget_jazzer.js:29:9)
+    at module.exports.fuzz (/fuzz/FuzzTarget_jazzer.js:51:5)
+    at result (/fuzz/node_modules/@jazzer.js/core/core.ts:335:15)
+Error: 2
+    at eval (eval at g (/fuzz/FuzzTarget_jazzer.js:7:7), <anonymous>:9:54)
+    at g (/fuzz/FuzzTarget_jazzer.js:14:13)
+    at fuzz (/fuzz/FuzzTarget_jazzer.js:29:9)
+    at module.exports.fuzz (/fuzz/FuzzTarget_jazzer.js:51:5)
+    at result (/fuzz/node_modules/@jazzer.js/core/core.ts:335:15)
diff --git a/libcasr/src/constants.rs b/libcasr/src/constants.rs
@@ -31,6 +31,12 @@ pub const STACK_FRAME_FUNCTION_IGNORE_REGEXES_JAVA: &[&str] = &[
     r"^javax\.",
 ];
 
+/// Regular expressions for JS functions to be ignored.
+pub const STACK_FRAME_FUNCTION_IGNORE_REGEXES_JS: &[&str] = &[
+    // TODO
+    r"^<anonymous>$",
+];
+
 /// Regular expressions for python functions to be ignored.
 pub const STACK_FRAME_FUNCTION_IGNORE_REGEXES_PYTHON: &[&str] = &[
     // TODO
@@ -228,6 +234,18 @@ pub const STACK_FRAME_FILEPATH_IGNORE_REGEXES_JAVA: &[&str] = &[
     r"^[^.]$",
 ];
 
+/// Regular expressions for paths to JS files that should be ignored.
+pub const STACK_FRAME_FILEPATH_IGNORE_REGEXES_JS: &[&str] = &[
+    // TODO
+    // Anonymous functions
+    r"^<anonymous>$",
+    // Native locations (within V8’s libraries)
+    r"^native$",
+    // JS internal modules
+    r"^(|node:)internal/?",
+    r"^(|node:)events/?",
+];
+
 /// Regular expressions for paths to python files that should be ignored.
 pub const STACK_FRAME_FILEPATH_IGNORE_REGEXES_PYTHON: &[&str] = &[
     // TODO

diff --git a/libcasr/src/java.rs b/libcasr/src/java.rs
@@ -20,7 +20,7 @@ impl ParseStacktrace for JavaStacktrace {
             conv_counter: usize,
         }
         // Get java stack trace.
-        let re = Regex::new(r"(?m)^(?:Caused by:|Exception in thread|== Java Exception:)(?:.|\n)*?((?:\n(?:\s|\t)+at .*\(.*\))+)(?:\n(?:\s|\t)+\.\.\. (\d+) more)?").unwrap();
+        let re = Regex::new(r"(?m)^(?:Caused by:|Exception in thread|== Java Exception:)(?:.|\n)*?((?:\n\s+at .*\(.*\))+)(?:\n\s+\.\.\. (\d+) more)?").unwrap();
         let mut blocks = Vec::new();
         for cap in re.captures_iter(stream) {
             let body: Vec<&'_ str> = cap
@@ -66,7 +66,7 @@ impl ParseStacktrace for JavaStacktrace {
     }
 
     fn parse_stacktrace_entry(entry: &str) -> Result<StacktraceEntry> {
-        let re = Regex::new(r"(?:\s|\t)*at (.*)\((.*)\)").unwrap();
+        let re = Regex::new(r"\s*at (.*)\((.*)\)").unwrap();
 
         let Some(cap) = re.captures(entry) else {
             return Err(Error::Casr(format!(