what the hell

src/general.rs

@@ -0,0 +1,309 @@
+//! # Generalized Levenshtein Distance
+//! This is a simple generalized Levenshtein Distance implementation.
+
+
+/// costs for different edit operations
+///
+/// each with the letters involved: so you can further tweak the cost in ur own [`CustomCost`]s impl  
+/// position? idk
+#[derive(Debug, Clone, PartialEq)]
+pub enum EditOperation<T>{
+    Insert { elem: T }, // Insert a character from the target string
+    Delete { elem: T }, // Delete a character from the source string
+    Substitute { from: T, to: T }, // Replace one character with another
+    Match { elem: T },  // Characters match, no operation
+}
+
+pub struct EditCost {
+    pub insert: i32,
+    pub delete: i32,
+    pub substitute: i32,
+}
+
+impl Default for EditCost {
+    fn default() -> Self {
+        EditCost {
+            insert: 1,
+            delete: 1,
+            substitute: 1,
+        }
+    }
+}
+
+impl EditCost {
+    pub fn new(insert: i32, delete: i32, substitute: i32) -> Self {
+        EditCost {
+            insert,
+            delete,
+            substitute,
+        }
+    }
+}
+/// custom cost for each [`EditOperation`]s operation
+pub trait CustomCost<T> {
+    fn cost(&self, operation: &EditOperation<T>) -> i32;
+}
+
+impl<T> CustomCost<T> for EditCost {
+    fn cost(&self, operation: &EditOperation<T>) -> i32 {
+        match operation {
+            EditOperation::Insert { .. } => self.insert,
+            EditOperation::Delete { .. } => self.delete,
+            EditOperation::Substitute { .. } => self.substitute,
+            EditOperation::Match { .. } => 0,
+        }
+    }
+}
+
+/// cost of operation and the operation itself
+#[derive(Debug, Clone)]
+pub struct CostOperation<T> {
+    pub cost: i32,
+    pub operation: Option<EditOperation<T>>,
+}
+// absolutely phenomenal naming
+use std::fmt::{Debug, Display};
+
+impl<T: Debug> Display for EditOperation<T> {
+    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
+        match self {
+            EditOperation::Insert { elem } => write!(f, "Insert {:#?}", elem),
+            EditOperation::Delete { elem } => write!(f, "Delete {:#?}", elem),
+            EditOperation::Substitute { from, to } => {
+                write!(f, "Substitute {:#?} with {:#?}", from, to)
+            }
+            EditOperation::Match { elem } => write!(f, "Match {:#?}", elem),
+        }
+    }
+}
+
+impl<T: Debug> Display for CostOperation<T> {
+    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
+        write!(f, "Cost: {}, Operation: {:#?}", self.cost, self.operation)
+    }
+}
+/// returns the distance + the dp table for who needs it
+///
+/// this the slow version yes
+pub fn generalized_levenshtein<T, F>(
+    seq1: &[T],
+    seq2: &[T],
+    cost_calc: &F,
+) -> (i32, Vec<Vec<CostOperation<T>>>)
+where
+    T: PartialEq + Clone + Debug,
+    F: CustomCost<T>,
+{
+    let len1 = seq1.len();
+    let len2 = seq2.len();
+
+    let mut dp = vec![
+        vec![
+            CostOperation {
+                cost: 0,
+                operation: None
+            };
+            len2 + 1
+        ];
+        len1 + 1
+    ];
+
+    // Fill the base cases
+    for i in 0..=len1 {
+        dp[i][0] = CostOperation {
+            cost: i as i32,
+            operation: if i > 0 {
+                Some(EditOperation::Delete {
+                    elem: seq1[i - 1].clone(),
+                })
+            } else {
+                None
+            },
+        };
+    }
+    for j in 0..=len2 {
+        dp[0][j] = CostOperation {
+            cost: j as i32,
+            operation: if j > 0 {
+                Some(EditOperation::Insert {
+                    elem: seq2[j - 1].clone(),
+                })
+            } else {
+                None
+            },
+        };
+    }
+
+    // Fill the DP table
+    for i in 1..=len1 {
+        for j in 1..=len2 {
+            let cost = seq1[i - 1] == seq2[j - 1];
+            let delete_op = EditOperation::Delete {
+                elem: seq1[i - 1].clone(),
+            };
+            let insert_op = EditOperation::Insert {
+                elem: seq2[j - 1].clone(),
+            };
+            let sub_op = if cost {
+                EditOperation::Match {
+                    elem: seq1[i - 1].clone(),
+                }
+            } else {
+                EditOperation::Substitute {
+                    from: seq1[i - 1].clone(),
+                    to: seq2[j - 1].clone(),
+                }
+            };
+
+            let delete = CostOperation {
+                cost: dp[i - 1][j].cost + cost_calc.cost(&delete_op),
+                operation: Some(delete_op),
+            };
+            let insert = CostOperation {
+                cost: dp[i][j - 1].cost + cost_calc.cost(&insert_op),
+                operation: Some(insert_op),
+            };
+            let substitute = CostOperation {
+                cost: dp[i - 1][j - 1].cost + cost_calc.cost(&sub_op),
+                operation: Some(sub_op),
+            };
+
+            dp[i][j] = [delete, insert, substitute]
+                .into_iter()
+                .min_by_key(|x| x.cost)
+                .unwrap();
+        }
+    }
+
+    (dp[len1][len2].cost, dp)
+}
+
+///extract the optimized edit sequence from the dp table
+///
+/// strs there just for the len() lmfao
+pub fn edit_sequence<T: Clone>(dp: &[Vec<CostOperation<T>>]) -> Vec<EditOperation<T>> {
+    let mut i = dp.len() - 1;
+    let mut j = dp[0].len() - 1;
+    let mut funny = vec![];
+
+    while i > 0 || j > 0 {
+        if let Some(operation) = &dp[i][j].operation {
+            match operation {
+                EditOperation::Insert { .. } => {
+                    funny.push(operation.clone());
+                    j -= 1;
+                }
+                EditOperation::Delete { .. } => {
+                    funny.push(operation.clone());
+                    i -= 1;
+                }
+                EditOperation::Substitute { .. } | EditOperation::Match { .. } => {
+                    funny.push(operation.clone());
+                    i -= 1;
+                    j -= 1;
+                }
+            }
+        }
+    }
+
+    funny.reverse();
+    funny
+}
+
+/// alignment  
+/// is ass dont use
+pub fn generate_alignment<T: Clone + Debug>(
+    operations: &[Vec<CostOperation<T>>],
+) -> (Vec<T>, Vec<T>) {
+    let mut aligned_seq1 = Vec::new();
+    let mut aligned_seq2 = Vec::new();
+
+    let mut i = operations.len() - 1;
+    let mut j = operations[0].len() - 1;
+
+    while i > 0 || j > 0 {
+        let op = &operations[i][j];
+        match op.operation {
+            Some(EditOperation::Insert { ref elem }) => {
+                aligned_seq1.push(None);
+                aligned_seq2.push(Some(elem.clone()));
+                j -= 1;
+            }
+            Some(EditOperation::Delete { ref elem }) => {
+                aligned_seq1.push(Some(elem.clone()));
+                aligned_seq2.push(None);
+                i -= 1;
+            }
+            Some(EditOperation::Substitute { ref from, ref to }) => {
+                aligned_seq1.push(Some(from.clone()));
+                aligned_seq2.push(Some(to.clone()));
+                i -= 1;
+                j -= 1;
+            }
+            Some(EditOperation::Match { ref elem }) => {
+                aligned_seq1.push(Some(elem.clone()));
+                aligned_seq2.push(Some(elem.clone()));
+                i -= 1;
+                j -= 1;
+            }
+            None => break,
+        }
+    }
+
+    aligned_seq1.reverse();
+    aligned_seq2.reverse();
+
+    (aligned_seq1.into_iter().filter_map(|x| x).collect(), aligned_seq2.into_iter().filter_map(|x| x).collect())
+}
+
+/// calculate distance, also get u the [`edit_sequence`]  
+/// use this one
+///
+/// what this does is call this ass [`levenshtein_with_operations`] and then the [`edit_sequence`] from the generated dp table
+pub fn levenstein_distance<T: PartialEq + Clone + Debug>(
+    seq1: &[T],
+    seq2: &[T],
+    cost: &impl CustomCost<T>,
+) -> (i32, Vec<EditOperation<T>>) {
+    let (c, d) = generalized_levenshtein(seq1, seq2, cost);
+    (c, edit_sequence(&d))
+}
+
+/// get source str from [`edit_sequence`]
+pub fn reconstruct_from<T: Clone>(vector: &Vec<EditOperation<T>>) -> Vec<T> {
+    let mut result = Vec::new();
+    for operation in vector {
+        match operation {
+            EditOperation::Insert { .. } => {}
+            EditOperation::Delete { elem } => {
+                result.push(elem.clone());
+            }
+            EditOperation::Substitute { from, .. } => {
+                result.push(from.clone());
+            }
+            EditOperation::Match { elem } => {
+                result.push(elem.clone());
+            }
+        }
+    }
+    result
+}
+/// get changed str from [`edit_sequence`]
+pub fn reconstruct_to<T: Clone>(vector: &Vec<EditOperation<T>>) -> Vec<T> {
+    let mut result = Vec::new();
+    for operation in vector {
+        match operation {
+            EditOperation::Insert { elem } => {
+                result.push(elem.clone());
+            }
+            EditOperation::Delete { .. } => {}
+            EditOperation::Substitute { to, .. } => {
+                result.push(to.clone());
+            }
+            EditOperation::Match { elem } => {
+                result.push(elem.clone());
+            }
+        }
+    }
+    result
+}

src/lib.rs

@@ -1,5 +1,7 @@
 //! # Levenshtein Distance
 //! This is a simple implementation of the Levenshtein distance algorithm with operations.
+//! 
+pub mod general;
 
 /// costs for different edit operations
 ///
@@ -90,8 +92,10 @@ pub fn levenshtein_with_operations(
     str2: &str,
     cost_calc: &impl CustomCost,
 ) -> (i32, Vec<Vec<CostOperation>>) {
-    let len1 = str1.len();
-    let len2 = str2.len();
+    let chars1: Vec<char> = str1.chars().collect();
+    let chars2: Vec<char> = str2.chars().collect();
+    let len1: usize = chars1.len();
+    let len2 = chars2.len();
     let mut dp = vec![
         vec![
             CostOperation {
@@ -102,11 +106,15 @@ pub fn levenshtein_with_operations(
         ];
         len1 + 1
     ];
-
-    let chars1: Vec<char> = str1.chars().collect();
-    let chars2: Vec<char> = str2.chars().collect();
+    // println!("len1: {}, len2: {}", str1.len(), str2.len());
+    // println!("len char1: {}, len char2: {}", chars1.len(), chars2.len());
     // Fill the base cases
     for i in 0..=len1 {
+        // if i > 3068 && i < 3075 {
+        //     println!("len1: {}, len2: {}", str1.len(), str2.len());
+        //     println!("len char1: {}, len char2: {}", chars1.len(), chars2.len());
+        //     println!("i: {}, char: {:#?}", i, chars1[i - 1]);
+        // }
         dp[i][0] = CostOperation {
             cost: i as i32,
             operation: if i > 0 {
@@ -178,9 +186,9 @@ pub fn levenshtein_with_operations(
 ///extract the optimized edit sequence from the dp table
 ///
 /// strs there just for the len() lmfao
-pub fn edit_sequence(dp: &[Vec<CostOperation>], str1: &str, str2: &str) -> Vec<EditOperation> {
-    let mut i = str1.len();
-    let mut j = str2.len();
+pub fn edit_sequence(dp: &[Vec<CostOperation>]) -> Vec<EditOperation> {
+    let mut i = dp.len() - 1;
+    let mut j = dp[0].len() - 1; // guaranteed
     let mut funny = vec![];
 
     while i > 0 || j > 0 {
@@ -214,15 +222,13 @@ pub fn edit_sequence(dp: &[Vec<CostOperation>], str1: &str, str2: &str) -> Vec<E
 /// alignment  
 /// is ass dont use
 pub fn generate_alignment(
-    str1: &str,
-    str2: &str,
     operations: &[Vec<CostOperation>],
 ) -> (String, String) {
     let mut aligned_str1 = String::new();
     let mut aligned_str2 = String::new();
 
-    let mut i = str1.len();
-    let mut j = str2.len();
+    let mut i = operations.len() - 1;
+    let mut j = operations[0].len() - 1;
 
     while i > 0 || j > 0 {
         let op = &operations[i][j];
@@ -266,7 +272,7 @@ pub fn generate_alignment(
 /// what this does is call this ass [`levenshtein_with_operations`] and then the [`edit_sequence`] from the generated dp table
 pub fn levenstein_distance(str1: &str, str2: &str, cost: &EditCost) -> (i32, Vec<EditOperation>) {
     let (c, d) = levenshtein_with_operations(str1, str2, cost);
-    (c, edit_sequence(&d, str1, str2))
+    (c, edit_sequence(&d))
 }
 
 /// get source str from [`edit_sequence`]
@@ -413,7 +419,7 @@ mod tests {
         let str1 = "kitten";
         let str2 = "sitting";
         let (distance, operations) = levenshtein_with_operations(str1, str2, &cost);
-        let operation = edit_sequence(&operations, str1, str2);
+        let operation = edit_sequence(&operations);
         assert_eq!(distance, 3);
         assert_eq!(
             operation,
@@ -428,7 +434,7 @@ mod tests {
             ]
         );
 
-        let (aligned_str1, aligned_str2) = generate_alignment(str1, str2, &operations);
+        let (aligned_str1, aligned_str2) = generate_alignment(&operations);
         assert_eq!(aligned_str1, "kitten-");
         assert_eq!(aligned_str2, "sitting");
     }
@@ -438,7 +444,7 @@ mod tests {
         let cost = EditCost::default();
         let (str1, str2) = ("flaw", "lawn");
         let (distance, operations) = levenshtein_with_operations(str1, str2, &cost);
-        let operation = edit_sequence(&operations, str1, str2);
+        let operation = edit_sequence(&operations);
         assert_eq!(distance, 2);
         assert_eq!(
             operation,
@@ -451,7 +457,7 @@ mod tests {
             ]
         );
 
-        let (aligned_str1, aligned_str2) = generate_alignment("flaw", "lawn", &operations);
+        let (aligned_str1, aligned_str2) = generate_alignment(&operations);
         assert_eq!(aligned_str1, "flaw-");
         assert_eq!(aligned_str2, "-lawn");
     }
@@ -461,7 +467,7 @@ mod tests {
         let cost = EditCost::default();
         let (str1, str2) = ("gumbo", "gambol");
         let (distance, operations) = levenshtein_with_operations(str1, str2, &cost);
-        let operation = edit_sequence(&operations, str1, str2);
+        let operation = edit_sequence(&operations);
         assert_eq!(distance, 2);
         assert_eq!(
             operation,
@@ -475,7 +481,7 @@ mod tests {
             ]
         );
 
-        let (aligned_str1, aligned_str2) = generate_alignment("gumbo", "gambol", &operations);
+        let (aligned_str1, aligned_str2) = generate_alignment(&operations);
         assert_eq!(aligned_str1, "gumbo-");
         assert_eq!(aligned_str2, "gambol");
     }
@@ -485,7 +491,7 @@ mod tests {
         let cost = EditCost::default();
         let (str1, str2) = ("test", "test");
         let (distance, operations) = levenshtein_with_operations(str1, str2, &cost);
-        let operation = edit_sequence(&operations, str1, str2);
+        let operation = edit_sequence(&operations);
         assert_eq!(distance, 0);
         assert_eq!(
             operation,
@@ -497,9 +503,9 @@ mod tests {
             ]
         );
 
-        let (aligned_str1, aligned_str2) = generate_alignment("test", "test", &operations);
+        let (aligned_str1, aligned_str2) = generate_alignment(&operations);
         assert_eq!(aligned_str1, "test");
         assert_eq!(aligned_str2, "test");
     }
 }
-pub mod wrapper_;
+pub mod wrapper_;

src/main.rs

@@ -20,6 +20,6 @@ fn main() {
 
     let (d2, dp) = levenshtein_with_operations(str1, str2, &EditCost::default());
     assert!(d2 == distance);
-    let (al1, al2) = generate_alignment(str1, str2, &dp);
+    let (al1, al2) = generate_alignment( &dp);
     println!("Alignment:\n{:#?}\n{:#?}", al1, al2);
 }