From e144d472f95e1b9810b345dd1eb4f518896af83f Mon Sep 17 00:00:00 2001
From: Alexander Mironov <sh1yo@tuta.io>
Date: Wed, 7 Jul 2021 17:23:12 +0400
Subject: [PATCH] Improve heuristic

---
 Cargo.toml   |  2 +-
 src/utils.rs | 12 +++++++++---
 2 files changed, 10 insertions(+), 4 deletions(-)

diff --git a/Cargo.toml b/Cargo.toml
index 7915b2c..2f9da62 100644
--- a/Cargo.toml
+++ b/Cargo.toml
@@ -1,6 +1,6 @@
 [package]
 name = "x8"
-version = "2.2.1"
+version = "2.3.0"
 authors = ["Alexander Mironov <sh1yo@tuta.io>"]
 edition = "2018"
 license = "GPL-3.0-or-later"
diff --git a/src/utils.rs b/src/utils.rs
index fa34644..b40f3de 100644
--- a/src/utils.rs
+++ b/src/utils.rs
@@ -68,10 +68,12 @@ pub fn compare(
     (code, diffs)
 }
 
-//get possible parameters from the page source code
+//get possible parameters from the page code
 pub fn heuristic(body: &str) -> Vec<String> {
     let mut found: Vec<String> = Vec::new();
 
+    let re_special_chars = Regex::new(r#"[\W]"#).unwrap();
+
     let re_name = Regex::new(r#"(?i)name=("|')?"#).unwrap();
     let re_inputs = Regex::new(r#"(?i)name=("|')?[\w-]+"#).unwrap();
     for cap in re_inputs.captures_iter(body) {
@@ -84,10 +86,14 @@ pub fn heuristic(body: &str) -> Vec<String> {
         found.push(re_var.replace_all(&cap[0], "").to_string());
     }
 
-    let re_quotes = Regex::new(r#"("|')"#).unwrap();
     let re_words_in_quotes = Regex::new(r#"("|')\w{3,20}('|")"#).unwrap();
     for cap in re_words_in_quotes.captures_iter(body) {
-        found.push(re_quotes.replace_all(&cap[0], "").to_string());
+        found.push(re_special_chars.replace_all(&cap[0], "").to_string());
+    }
+
+    let re_words_within_objects = Regex::new(r#"[\{,]\s*[[:alpha:]]\w{2,25}:"#).unwrap();
+    for cap in re_words_within_objects.captures_iter(body){
+        found.push(re_special_chars.replace_all(&cap[0], "").to_string());
     }
 
     found.sort();