From 077684c68119a61736f4d859e44de19eab7d8d87 Mon Sep 17 00:00:00 2001
From: trivernis <trivernis@protonmail.com>
Date: Mon, 3 Aug 2020 11:02:05 +0200
Subject: [PATCH] Add string sequence matching

---
 Cargo.toml         |  2 +-
 src/lib.rs         | 31 ++++++++++++++++++--
 src/tapemachine.rs | 70 ++++++++++++++++++++++++++++++++++++++++++++++
 3 files changed, 99 insertions(+), 4 deletions(-)

diff --git a/Cargo.toml b/Cargo.toml
index 7213771..00ab8c2 100644
--- a/Cargo.toml
+++ b/Cargo.toml
@@ -1,6 +1,6 @@
 [package]
 name = "charred"
-version = "0.2.2"
+version = "0.3.0"
 authors = ["trivernis <trivernis@protonmail.com>"]
 edition = "2018"
 license-file = "LICENSE"
diff --git a/src/lib.rs b/src/lib.rs
index f25262f..b0d0a3c 100644
--- a/src/lib.rs
+++ b/src/lib.rs
@@ -1,9 +1,12 @@
+#![feature(test)]
+extern crate test;
 pub mod tapemachine;
 
 #[cfg(test)]
 mod tests {
-    use crate::tapemachine::{CharTapeMachine, TapeError};
     use crate::tapemachine::TapeResult;
+    use crate::tapemachine::{CharTapeMachine, TapeError};
+    use test::Bencher;
 
     const TEST_STRING: &str = "TEST STRING 1234 \\l \\n";
 
@@ -58,11 +61,13 @@ mod tests {
         ctm.seek_one().unwrap();
         ctm.assert_char(&'E', None)?;
         ctm.seek_one().unwrap();
-        ctm.assert_sequence(&['S', 'T', ' '], None)?;
+        ctm.assert_str_sequence("ST ", None)?;
         ctm.seek_one().unwrap();
         ctm.assert_any_sequence(&[&['C'], &['A'], &['A', 'B'], &['S', 'T', 'R']], None)?;
 
-        if let Ok(_) = ctm.assert_any_sequence(&[&['C'], &['A'], &['A', 'B'], &['S', 'T', 'R']], None) {
+        if let Ok(_) =
+            ctm.assert_any_sequence(&[&['C'], &['A'], &['A', 'B'], &['S', 'T', 'R']], None)
+        {
             Err(TapeError::new(0))
         } else {
             Ok(())
@@ -77,4 +82,24 @@ mod tests {
 
         Ok(())
     }
+
+    #[bench]
+    fn bench_assert_seek(b: &mut Bencher) {
+        let mut ctm = CharTapeMachine::new(TEST_STRING.chars().collect());
+        b.iter(|| {
+            ctm.check_char(&'T');
+            ctm.seek_one().unwrap();
+            ctm.check_char(&'E');
+            ctm.seek_one().unwrap();
+            ctm.check_char(&'F');
+            ctm.seek_one().unwrap();
+            ctm.check_any(&['A', 'B', 'C', 'D', 'E', '2']);
+            ctm.seek_one().unwrap();
+            ctm.seek_whitespace();
+            ctm.check_sequence(&['S', 'T', 'R', 'I', 'N', 'T']);
+            ctm.check_sequence(&['S', 'T', 'R', 'I', 'N', 'G']);
+            ctm.check_eof();
+            ctm.rewind(0);
+        })
+    }
 }
diff --git a/src/tapemachine.rs b/src/tapemachine.rs
index f9c15c5..2b5160f 100644
--- a/src/tapemachine.rs
+++ b/src/tapemachine.rs
@@ -185,6 +185,37 @@ impl CharTapeMachine {
         }
     }
 
+    // checks if the next characters mach a string sequence
+    pub fn check_str_sequence(&mut self, sequence: &str) -> bool {
+        let start_index = self.index;
+
+        if self.check_escaped() {
+            self.rewind(start_index);
+
+            false
+        } else {
+            let matches = sequence.chars().all(|sq_character| {
+                if self.current_char != sq_character {
+                    self.rewind(start_index);
+                    return false;
+                }
+                if self.next_char() == None {
+                    self.rewind(start_index);
+                    return false;
+                }
+                true
+            });
+            if !matches {
+                false
+            } else {
+                if self.index > 0 {
+                    self.rewind(self.index - 1);
+                }
+                true
+            }
+        }
+    }
+
     /// checks if the next characters match any given sequence
     #[inline]
     pub fn check_any_sequence(&mut self, sequences: &[&[char]]) -> bool {
@@ -197,6 +228,18 @@ impl CharTapeMachine {
         false
     }
 
+    /// checks if the next characters match any given sequence of strings
+    #[inline]
+    pub fn check_any_str_sequence(&mut self, sequences: &[&str]) -> bool {
+        for str_seq in sequences {
+            if self.check_str_sequence(str_seq) {
+                return true;
+            }
+        }
+
+        false
+    }
+
     /// returns an error on the current position and optionally rewinds
     /// if a rewind index is given
     #[inline]
@@ -243,6 +286,20 @@ impl CharTapeMachine {
         }
     }
 
+    /// returns an error if the next chars don't match a special sequence
+    #[inline]
+    pub fn assert_str_sequence(
+        &mut self,
+        sequence: &str,
+        rewind_index: Option<usize>,
+    ) -> TapeResult<()> {
+        if self.check_str_sequence(sequence) {
+            Ok(())
+        } else {
+            Err(self.assert_error(rewind_index))
+        }
+    }
+
     /// returns an error if the next chars don't match any given sequence
     pub fn assert_any_sequence(
         &mut self,
@@ -256,6 +313,19 @@ impl CharTapeMachine {
         }
     }
 
+    /// returns an error if the next chars don't match any given sequence
+    pub fn assert_any_str_sequence(
+        &mut self,
+        sequences: &[&str],
+        rewind_index: Option<usize>,
+    ) -> TapeResult<()> {
+        if self.check_any_str_sequence(sequences) {
+            Ok(())
+        } else {
+            Err(self.assert_error(rewind_index))
+        }
+    }
+
     /// returns the string until any given character is matched is matched.
     /// rewinds with error if it encounters a character form the error group
     #[inline]