From 447c57c942f51881e85b87a1e07bab887f11fafa Mon Sep 17 00:00:00 2001
From: Simon Laux <mobile.info@simonlaux.de>
Date: Wed, 8 Jan 2025 13:22:35 +0100
Subject: [PATCH] move parenthesis counter into dedicated file and change to
 more descriptive name also add some tests

---
 src/parser/link_url/mod.rs                 |  1 +
 src/parser/link_url/parenthesis_counter.rs | 81 ++++++++++++++++++++++
 src/parser/link_url/parse_link.rs          | 65 +----------------
 3 files changed, 85 insertions(+), 62 deletions(-)
 create mode 100644 src/parser/link_url/parenthesis_counter.rs

diff --git a/src/parser/link_url/mod.rs b/src/parser/link_url/mod.rs
index 9473ba2..fee9c8e 100644
--- a/src/parser/link_url/mod.rs
+++ b/src/parser/link_url/mod.rs
@@ -1,4 +1,5 @@
 mod ip;
+mod parenthesis_counter;
 mod parse_link;
 
 use nom::{
diff --git a/src/parser/link_url/parenthesis_counter.rs b/src/parser/link_url/parenthesis_counter.rs
new file mode 100644
index 0000000..dd0dbf6
--- /dev/null
+++ b/src/parser/link_url/parenthesis_counter.rs
@@ -0,0 +1,81 @@
+use nom::Slice;
+
+macro_rules! adjust_balance {
+    ($a: expr, $b: expr, $c: expr, $d: expr) => {
+        // for opening ones
+        {
+            $a = $a.saturating_add(1);
+            if $d.slice($c..).find($b).is_none() {
+                return Some($c);
+            }
+        }
+    };
+    ($a: expr, $b: expr) => {
+        // for closing ones
+        {
+            if $a == 0 {
+                return Some($b);
+            } else {
+                $a = $a.saturating_sub(1);
+            }
+        }
+    };
+}
+
+/// finds unbalanced closing parenthesesis and returns distance to it.
+/// unbalanced means it was closed but not opened before in the given string
+pub(super) fn count_chars_in_complete_parenthesis(input: &str) -> Option<usize> {
+    let mut parenthes = 0usize; // ()
+    let mut curly_bracket = 0usize; // {}
+    let mut bracket = 0usize; // []
+    let mut angle = 0usize; // <>
+
+    for (i, ch) in input.chars().enumerate() {
+        match ch {
+            '(' => {
+                adjust_balance!(parenthes, ')', i, input);
+            }
+            '{' => {
+                adjust_balance!(curly_bracket, '}', i, input);
+            }
+            '[' => {
+                adjust_balance!(bracket, ']', i, input);
+            }
+            '<' => {
+                adjust_balance!(angle, '>', i, input);
+            }
+            ')' => {
+                adjust_balance!(parenthes, i);
+            }
+            ']' => {
+                adjust_balance!(bracket, i);
+            }
+            '}' => {
+                adjust_balance!(curly_bracket, i);
+            }
+            '>' => {
+                adjust_balance!(angle, i);
+            }
+            _ => continue,
+        }
+    }
+    None
+}
+
+#[test]
+fn test_count_parenthesis() {
+    assert_eq!(count_chars_in_complete_parenthesis("{}"), None);
+    assert_eq!(count_chars_in_complete_parenthesis("{} test"), None);
+    assert_eq!(count_chars_in_complete_parenthesis("(test) test"), None);
+    assert_eq!(count_chars_in_complete_parenthesis("(test)) test"), Some(6));
+}
+
+#[test]
+fn test_count_different_types_invalid() {
+    assert_eq!(count_chars_in_complete_parenthesis("(({(})))"), None);
+}
+
+#[test]
+fn test_count_different_types_invalid2() {
+    assert_eq!(count_chars_in_complete_parenthesis("}(({(})))"), Some(0));
+}
diff --git a/src/parser/link_url/parse_link.rs b/src/parser/link_url/parse_link.rs
index 74776fb..0913a1c 100644
--- a/src/parser/link_url/parse_link.rs
+++ b/src/parser/link_url/parse_link.rs
@@ -22,6 +22,8 @@ use crate::parser::{
     },
 };
 
+use super::parenthesis_counter::count_chars_in_complete_parenthesis;
+
 /// determines which generic schemes (without '://') get linkifyed
 fn is_allowed_generic_scheme(scheme: &str) -> bool {
     matches!(
@@ -272,67 +274,6 @@ fn ifragment(input: &str) -> IResult<&str, &str, CustomError<&str>> {
     recognize(tuple((char('#'), take_while_ifragment)))(input)
 }
 
-macro_rules! link_correct {
-    ($a: expr, $b: expr, $c: expr, $d: expr) => {
-        // for opening ones
-        {
-            $a = $a.saturating_add(1);
-            if $d.slice($c..).find($b).is_none() {
-                return Some($c);
-            }
-        }
-    };
-    ($a: expr, $b: expr) => {
-        // for closing ones
-        {
-            if $a == 0 {
-                return Some($b);
-            } else {
-                $a = $a.saturating_sub(1);
-            }
-        }
-    };
-}
-
-// TODO: better name for this function
-fn get_correct_link(link: &str) -> Option<usize> {
-    let mut parenthes = 0usize; // ()
-    let mut curly_bracket = 0usize; // {}
-    let mut bracket = 0usize; // []
-    let mut angle = 0usize; // <>
-
-    for (i, ch) in link.chars().enumerate() {
-        match ch {
-            '(' => {
-                link_correct!(parenthes, ')', i, link);
-            }
-            '{' => {
-                link_correct!(curly_bracket, '}', i, link);
-            }
-            '[' => {
-                link_correct!(bracket, ']', i, link);
-            }
-            '<' => {
-                link_correct!(angle, '>', i, link);
-            }
-            ')' => {
-                link_correct!(parenthes, i);
-            }
-            ']' => {
-                link_correct!(bracket, i);
-            }
-            '}' => {
-                link_correct!(curly_bracket, i);
-            }
-            '>' => {
-                link_correct!(angle, i);
-            }
-            _ => continue,
-        }
-    }
-    None
-}
-
 fn parse_ipath_abempty(input: &str) -> IResult<&str, &str, CustomError<&str>> {
     recognize(many0(tuple((char('/'), opt(take_while_ipchar1)))))(input)
 }
@@ -406,7 +347,7 @@ fn parse_iri(input: &str) -> IResult<&str, LinkDestination, CustomError<&str>> {
                 host = input_.slice(scheme.len().saturating_add(3)..input_.len().saturating_sub(1));
             }
         }
-        len = get_correct_link(link).unwrap_or(len);
+        len = count_chars_in_complete_parenthesis(link).unwrap_or(len);
         let link = input_.slice(0..len);
         let input = input_.slice(len..);