Move encoding/decoding functions to hexora_io

Author: rushter

Cargo.lock

@@ -30,7 +30,6 @@ path = "src/main.rs"
 [dependencies]
 ruff_python_ast = { git = "https://github.com/astral-sh/ruff", rev="3f63ea4b504d50eaf3c479de68e4017d2434a160" }
 ruff_python_parser = { git = "https://github.com/astral-sh/ruff", rev="3f63ea4b504d50eaf3c479de68e4017d2434a160" }
-ruff_source_file = { git = "https://github.com/astral-sh/ruff/", rev="3f63ea4b504d50eaf3c479de68e4017d2434a160"}
 ruff_python_stdlib = { git = "https://github.com/astral-sh/ruff/", rev="3f63ea4b504d50eaf3c479de68e4017d2434a160" }
 ruff_text_size = { git = "https://github.com/astral-sh/ruff/", rev="3f63ea4b504d50eaf3c479de68e4017d2434a160" }
 
@@ -49,8 +48,6 @@ strum_macros = "0.27.0"
 serde_json = "1.0.113"
 pythonize = "0.26.0"
 sha2 = "0.11.0-rc.2"
-encoding_rs = "0.8.35"
-base64 = "0.22.1"
 hexora_io = { path = "../hexora_io" }
 
 [dev-dependencies]

crates/hexora/Cargo.toml

@@ -1,8 +1,9 @@
 use crate::indexer::model::Transformation;
 use crate::indexer::node_transformer::NodeTransformer;
 use crate::indexer::taint::TaintState;
-use base64::{Engine as _, engine::general_purpose};
-use hexora_io::encoding::{decode_bytes, unescape_to_bytes};
+use hexora_io::encoding::{
+    base64_decode, bytes_to_escaped, decode_bytes, hex_to_escaped, unescape_to_bytes,
+};
 use ruff_python_ast::{
     self as ast, AtomicNodeIndex, ExprContext, HasNodeIndex, NodeIndex, Operator,
     StringLiteralFlags,
@@ -204,9 +205,7 @@ impl<'a> NodeTransformer<'a> {
 
         let arg_str = self.extract_string(&call.arguments.args[0])?;
         if name == "a2b_base64" {
-            let bytes = general_purpose::STANDARD
-                .decode(arg_str.trim().as_bytes())
-                .ok()?;
+            let bytes = base64_decode(&arg_str, false)?;
             return Some(self.make_string_expr(
                 call.range,
                 bytes_to_escaped(&bytes),
@@ -260,12 +259,7 @@ impl<'a> NodeTransformer<'a> {
         }
 
         let arg_str = self.extract_string(&call.arguments.args[0])?;
-        let bytes_arg = arg_str.trim().as_bytes();
-        let bytes = if name == "urlsafe_b64decode" {
-            general_purpose::URL_SAFE.decode(bytes_arg).ok()?
-        } else {
-            general_purpose::STANDARD.decode(bytes_arg).ok()?
-        };
+        let bytes = base64_decode(&arg_str, name == "urlsafe_b64decode")?;
 
         let escaped = bytes_to_escaped(&bytes);
         Some(self.make_string_expr(call.range, escaped, Some(Transformation::Base64)))
@@ -378,33 +372,3 @@ impl<'a> NodeTransformer<'a> {
         }
     }
 }
-
-#[inline]
-pub(crate) fn bytes_to_escaped(bytes: &[u8]) -> String {
-    bytes.iter().map(|b| format!("\\x{:02x}", b)).collect()
-}
-
-#[inline]
-pub(crate) fn hex_to_escaped(input: &str) -> Option<String> {
-    let filtered: String = input.chars().filter(|c| !c.is_ascii_whitespace()).collect();
-    if filtered.is_empty() || !filtered.len().is_multiple_of(2) {
-        return None;
-    }
-    filtered
-        .as_bytes()
-        .chunks(2)
-        .map(|chunk| {
-            let h = chunk[0] as char;
-            let l = chunk[1] as char;
-            if h.is_ascii_hexdigit() && l.is_ascii_hexdigit() {
-                Some(format!(
-                    "\\x{}{}",
-                    h.to_ascii_lowercase(),
-                    l.to_ascii_lowercase()
-                ))
-            } else {
-                None
-            }
-        })
-        .collect()
-}

crates/hexora/src/indexer/strings/transformer.rs

@@ -11,6 +11,7 @@ tar = "0.4"
 flate2 = "1.0"
 encoding_rs = "0.8.35"
 once_cell = "1.21.3"
+base64 = "0.22.1"
 ruff_source_file = { git = "https://github.com/astral-sh/ruff/", rev="3f63ea4b504d50eaf3c479de68e4017d2434a160"}
 ruff_text_size = { git = "https://github.com/astral-sh/ruff/", rev="3f63ea4b504d50eaf3c479de68e4017d2434a160" }
 

crates/hexora_io/Cargo.toml

@@ -1,3 +1,4 @@
+use base64::{Engine as _, engine::general_purpose};
 use encoding_rs;
 use once_cell::sync::Lazy;
 use std::collections::HashMap;
@@ -298,6 +299,45 @@ pub fn unescape_to_bytes(input: &str) -> Option<Vec<u8>> {
     Some(bytes)
 }
 
+#[inline]
+pub fn bytes_to_escaped(bytes: &[u8]) -> String {
+    bytes.iter().map(|b| format!("\\x{:02x}", b)).collect()
+}
+
+#[inline]
+pub fn hex_to_escaped(input: &str) -> Option<String> {
+    let filtered: String = input.chars().filter(|c| !c.is_ascii_whitespace()).collect();
+    if filtered.is_empty() || filtered.len() % 2 != 0 {
+        return None;
+    }
+    filtered
+        .as_bytes()
+        .chunks(2)
+        .map(|chunk| {
+            let h = chunk[0] as char;
+            let l = chunk[1] as char;
+            if h.is_ascii_hexdigit() && l.is_ascii_hexdigit() {
+                Some(format!(
+                    "\\x{}{}",
+                    h.to_ascii_lowercase(),
+                    l.to_ascii_lowercase()
+                ))
+            } else {
+                None
+            }
+        })
+        .collect()
+}
+
+pub fn base64_decode(input: &str, url_safe: bool) -> Option<Vec<u8>> {
+    let input = input.trim();
+    if url_safe {
+        general_purpose::URL_SAFE.decode(input).ok()
+    } else {
+        general_purpose::STANDARD.decode(input).ok()
+    }
+}
+
 #[cfg(test)]
 mod tests {
     use super::*;