From 6cc963ccd37f3ad40c46dcc0a5cae305dcfad87d Mon Sep 17 00:00:00 2001
From: Igor Matuszewski <xanewok@gmail.com>
Date: Sat, 9 Mar 2024 16:12:18 +0100
Subject: [PATCH] Skip contracts using exotic parser bugs or wrongly scraped

---
 crates/solidity/testing/sanctuary/src/main.rs |  3 +-
 .../solidity/testing/sanctuary/src/tests.rs   | 50 ++++++++++++++++---
 2 files changed, 45 insertions(+), 8 deletions(-)

diff --git a/crates/solidity/testing/sanctuary/src/main.rs b/crates/solidity/testing/sanctuary/src/main.rs
index 2834694e6f..362e8dc606 100644
--- a/crates/solidity/testing/sanctuary/src/main.rs
+++ b/crates/solidity/testing/sanctuary/src/main.rs
@@ -92,7 +92,8 @@ fn main() -> Result<()> {
     if failure_count > 0 {
         println!();
         println!(
-            "Found '{failure_count}' failures. Please check the logs above for more information."
+            "Found {failure_count} failure{suffix}. Please check the logs above for more information.",
+            suffix = if failure_count == 1 { "" } else { "s" },
         );
         println!();
 
diff --git a/crates/solidity/testing/sanctuary/src/tests.rs b/crates/solidity/testing/sanctuary/src/tests.rs
index a516ac5c9a..db866af619 100644
--- a/crates/solidity/testing/sanctuary/src/tests.rs
+++ b/crates/solidity/testing/sanctuary/src/tests.rs
@@ -1,4 +1,5 @@
 use std::cmp::min;
+use std::path::Path;
 
 use anyhow::Result;
 use infra_utils::paths::PathExtensions;
@@ -59,26 +60,40 @@ pub(crate) fn select_tests<'d>(
 }
 
 pub fn run_test(file: &SourceFile, events: &Events) -> Result<()> {
+    if !file.path.exists() {
+        // Index can be out of date:
+        events.test(TestOutcome::NotFound);
+        return Ok(());
+    }
+
+    // Ignore contracts that rely on obvious, exotic parser bugs fixed in later versions:
+    if uses_exotic_parser_bug(&file.path) {
+        events.test(TestOutcome::Incompatible);
+        return Ok(());
+    }
+
     let Some(version) = extract_compiler_version(&file.compiler) else {
         events.test(TestOutcome::Incompatible);
         return Ok(());
     };
 
-    if !file.path.exists() {
-        // Index can be out of date:
-        events.test(TestOutcome::NotFound);
+    let source = file.path.read_to_string()?;
+    // Heuristic: ignore wrongly scraped sanctuary files that
+    // contain HTML with a Cloudflare email obfuscation attribute:
+    if source.contains("data-cfemail=") {
+        events.test(TestOutcome::Incompatible);
         return Ok(());
     }
 
-    let source = file
-        .path
-        .read_to_string()?
+    let source = source
         // Some files are null terminated. Remove the null character:
         // https://github.com/tintinweb/smart-contract-sanctuary/issues/30
         .trim_end_matches('\0')
         // Remove unicode character inserted by sanctuary (illegal in Solidity):
         // https://github.com/tintinweb/smart-contract-sanctuary/issues/31
-        .replace("[email protected]", "[email-protected]");
+        .replace("[email protected]", "[email-protected]")
+        // Select contracts from Sanctuary were probably incorrectly web-scraped:
+        .replace("&#39;", "\"");
 
     let language = Language::new(version.clone())?;
     let output = language.parse(RuleKind::SourceUnit, &source);
@@ -123,3 +138,24 @@ fn extract_compiler_version(compiler: &str) -> Option<Version> {
 
     Some(version)
 }
+
+fn uses_exotic_parser_bug(file: &Path) -> bool {
+    static CONTRACTS_WITH_EXOTIC_PARSER_BUGS: &[&str] = &[
+        // 0.4.24: // Accepts malformed `* /` in multi-line comments:
+        // Fixed in 0.4.25: https://github.com/ethereum/solidity/pull/4937
+        "ethereum/contracts/mainnet/79/79bb6f4492d5cb13fad8ca0ecfbccd9e2c26ac42_Gateway.sol",
+        // 0.4.18: // Accepts unfinished multi-line comments at the end of the file:
+        // Fixed in 0.4.25: https://github.com/ethereum/solidity/pull/4937
+        "ethereum/contracts/mainnet/7d/7d81c361d6ac60634117dd81ab1b01b8dc795a9d_LILITHCOIN.sol",
+        // 0.8.15: Relies on invalidly accepting `indexed indexed` in the event declaration:
+        // Fixed in 0.8.18: https://github.com/ethereum/solidity/pull/13816
+        "ethereum/contracts/mainnet/da/d4559E5F507eD935F19208A5D50637898c192Ab3_Factory.sol",
+        // 0.4.24: Relies on silently ignoring invalid `\`-escapes in string literals:
+        // Fixed in 0.4.25: https://github.com/ethereum/solidity/pull/4937
+        "ethereum/contracts/mainnet/e0/e0b1bfe763b2cd43c652471bf126f5a577a81567_ExchangeProvider.sol",
+    ];
+
+    CONTRACTS_WITH_EXOTIC_PARSER_BUGS
+        .iter()
+        .any(|path| file.ends_with(path))
+}