MOD-14010 support Homogenues array floating point forcing(deserializa… (#17)

* MOD-13577 support Homogenues array floating point forcing(deserialization path only)

* add fallback option

* export FPHAConfig

* docs

* fmt

* comments

* lower fuzz time

* bring back old code

* change to lossy push

* Binary encoder/decoder

* remove fallback from FPHAConfig

* more fuzz tests

* tag to enum

* update fuzz parameters in ci

* fmt

* remove print in fuzz

* wrap with zstd compression

* misc

* move to cbor based implementation

* .

* fmt

* CR

* move to fork of ciborium

* fix CR

* fmt

* misc: JsonValue in fuzz tests to use serde

* fmt

* revert to manual string in fuzz, fix push_with_fp case

* cr

* remove dead code

* .

* CR

* move to arbitrary-json crate in fuzz tests

Author: AvivDavid23

.github/actions/fuzz_tests/action.yml

@@ -8,7 +8,7 @@ inputs:
   fuzz_time:
     description: 'Maximum time in seconds to run fuzzing'
     required: false
-    default: '180'
+    default: '300'
   cargo_fuzz_version:
     description: 'Version of cargo-fuzz to install'
     required: false
@@ -26,5 +26,5 @@ runs:
     - name: Run Fuzz Tests
       shell: bash
       working-directory: fuzz
-      run: cargo fuzz run ${{ inputs.fuzz_target }} --release -- -max_total_time=${{ inputs.fuzz_time }}
+      run: cargo fuzz run ${{ inputs.fuzz_target }} --release -- -max_total_time=${{ inputs.fuzz_time }} -ignore_ooms=1 -rss_limit_mb=0
 

.github/workflows/ci.yml

@@ -24,7 +24,8 @@ jobs:
       matrix:
         fuzz_target:
           - fuzz_json_de
-          # Add more fuzz targets here as needed
+          - fuzz_cbor_decode
+          - fuzz_cbor_roundtrip
     steps:
       - name: Checkout repository
         uses: actions/checkout@v4

Cargo.toml

@@ -30,10 +30,15 @@ serde = { workspace = true }
 serde_json = { workspace = true }
 ctor = { version = "0.1.16", optional = true }
 paste = "1.0.15"
-half = "2.0.0"
+ciborium = { git = "https://github.com/AvivDavid23/ciborium", branch = "main" }
+ciborium-ll = { git = "https://github.com/AvivDavid23/ciborium", branch = "main" }
+bytemuck = "1"
+half = { version = "2.0.0", features = ["bytemuck"] }
+thiserror = "2.0.18"
+zstd = "0.13"
 
 [dev-dependencies]
 mockalloc = "0.1.2"
 ctor = "0.1.16"
 rand = "0.8.4"
-
+zstd = "0.13"

fuzz/Cargo.toml

@@ -7,11 +7,16 @@ edition = "2021"
 [package.metadata]
 cargo-fuzz = true
 
+[lib]
+name = "ijson_fuzz"
+path = "src/lib.rs"
+
 [dependencies]
 libfuzzer-sys = "0.4"
 arbitrary = { version = "1.3", features = ["derive"] }
-serde = { workspace = true }
+serde = { workspace = true, features = ["derive"] }
 serde_json = { workspace = true }
+arbitrary-json = "=0.1.1"
 
 [dependencies.ijson]
 path = ".."
@@ -22,3 +27,17 @@ path = "fuzz_targets/fuzz_json_de.rs"
 test = false
 doc = false
 bench = false
+
+[[bin]]
+name = "fuzz_cbor_decode"
+path = "fuzz_targets/fuzz_cbor_decode.rs"
+test = false
+doc = false
+bench = false
+
+[[bin]]
+name = "fuzz_cbor_roundtrip"
+path = "fuzz_targets/fuzz_cbor_roundtrip.rs"
+test = false
+doc = false
+bench = false

fuzz/fuzz_targets/fuzz_cbor_decode.rs

@@ -0,0 +1,8 @@
+#![no_main]
+
+use ijson::cbor::decode;
+use libfuzzer_sys::fuzz_target;
+
+fuzz_target!(|data: &[u8]| {
+    let _ = decode(data);
+});

fuzz/fuzz_targets/fuzz_cbor_roundtrip.rs

@@ -0,0 +1,22 @@
+#![no_main]
+
+use arbitrary_json::ArbitraryValue;
+use ijson::{cbor, IValue};
+use libfuzzer_sys::fuzz_target;
+use serde::Deserialize;
+
+fuzz_target!(|value: ArbitraryValue| {
+    let json_string = value.to_string();
+    let mut deserializer = serde_json::Deserializer::from_str(&json_string);
+    let Ok(original) = IValue::deserialize(&mut deserializer) else {
+        return;
+    };
+
+    let encoded = cbor::encode(&original);
+    let decoded = cbor::decode(&encoded).expect("encode->decode round-trip must not fail");
+
+    assert_eq!(
+        original, decoded,
+        "round-trip mismatch for input: {json_string}"
+    );
+});

fuzz/fuzz_targets/fuzz_json_de.rs

@@ -1,54 +1,12 @@
 #![no_main]
 
-use arbitrary::Arbitrary;
+use arbitrary_json::ArbitraryValue;
 use ijson::IValue;
 use libfuzzer_sys::fuzz_target;
 use serde::Deserialize;
-use std::collections::HashMap;
 
-#[derive(Arbitrary, Debug)]
-enum JsonValue {
-    Null,
-    Bool(bool),
-    Number(f64),
-    String(String),
-    Array(Vec<JsonValue>),
-    Object(HashMap<String, JsonValue>),
-}
-
-impl JsonValue {
-    fn to_json_string(&self) -> String {
-        match self {
-            JsonValue::Null => "null".to_string(),
-            JsonValue::Bool(b) => b.to_string(),
-            JsonValue::Number(n) => {
-                if n.is_finite() {
-                    n.to_string()
-                } else {
-                    "0".to_string()
-                }
-            }
-            JsonValue::String(s) => format!("\"{}\"", s),
-            JsonValue::Array(arr) => {
-                let items: Vec<String> = arr.iter().map(|v| v.to_json_string()).collect();
-                format!("[{}]", items.join(","))
-            }
-            JsonValue::Object(obj) => {
-                let items: Vec<String> = obj
-                    .iter()
-                    .map(|(k, v)| {
-                        let key = k.clone();
-                        format!("\"{}\":{}", key, v.to_json_string())
-                    })
-                    .collect();
-                format!("{{{}}}", items.join(","))
-            }
-        }
-    }
-}
-
-fuzz_target!(|value: JsonValue| {
-    let json_string = value.to_json_string();
+fuzz_target!(|value: ArbitraryValue| {
+    let json_string = value.to_string();
     let mut deserializer = serde_json::Deserializer::from_str(&json_string);
     let _ = IValue::deserialize(&mut deserializer);
 });

fuzz/src/lib.rs

@@ -0,0 +1 @@
+