From 2f9da6773d71a213f90c022a162d76e687759d99 Mon Sep 17 00:00:00 2001
From: Navneet Aman <contact@navneetaman.com>
Date: Fri, 4 Oct 2024 10:54:05 +0530
Subject: [PATCH] Expose the substitute method of pcre as Regex::substitute and
 Regex::substitute_all.

---
 src/bytes.rs | 76 ++++++++++++++++++++++++++++++++++++++++++++++++++++
 src/ffi.rs   | 48 +++++++++++++++++++++++++++++++++
 2 files changed, 124 insertions(+)

diff --git a/src/bytes.rs b/src/bytes.rs
index 2129119..d5daab4 100644
--- a/src/bytes.rs
+++ b/src/bytes.rs
@@ -588,6 +588,44 @@ impl Regex {
     ) -> CaptureMatches<'r, 's> {
         CaptureMatches { re: self, subject, last_end: 0, last_match: None }
     }
+
+    /// Replaces the first match in `subject` with the `replacement`,
+    /// and puts the replaced string in `output`.
+    /// ```rust
+    /// # fn example() -> Result<(), ::pcre2::Error> {
+    /// use std::str;
+    ///
+    /// use pcre2::bytes::Regex;
+    ///
+    /// let re = Regex::new(r"mike")?;
+    /// let text = b"Hi mike, wait you are not mike.";
+    /// let mut output = Vec::new();
+    /// re.substitute(text, b"john", &mut output).unwrap();
+    /// assert_eq!(&output, b"Hi john, wait you are not mike.");
+    /// # Ok(()) }; example().unwrap()
+    /// ```
+    pub fn substitute(&self, subject: &[u8], replacement: &[u8], output: &mut Vec<u8>) -> Result<usize, Error>{
+        self.code.substitute(subject, replacement, output, 0)
+    }
+    /// Replaces all the matches in `subject` with the `replacement`,
+    /// and puts the replaced string in `output`.
+    /// ```rust
+    /// # fn example() -> Result<(), ::pcre2::Error> {
+    /// use std::str;
+    ///
+    /// use pcre2::bytes::Regex;
+    ///
+    /// let re = Regex::new(r"mike")?;
+    /// let text = b"Hi mike, wait you are not mike.";
+    /// let mut output = Vec::new();
+    /// re.substitute_all(text, b"john", &mut output).unwrap();
+    /// assert_eq!(&output, b"Hi john, wait you are not john.");
+    /// # Ok(()) }; example().unwrap()
+    /// ```
+    pub fn substitute_all(&self, subject: &[u8], replacement: &[u8], output: &mut Vec<u8>) -> Result<usize, Error>{
+        self.code.substitute(subject, replacement, output, pcre2_sys::PCRE2_SUBSTITUTE_GLOBAL)
+    }
+
 }
 
 /// Advanced or  "lower level" search methods.
@@ -1370,4 +1408,42 @@ mod tests {
         let matched = re.find(hay.as_bytes()).unwrap().unwrap();
         assert_eq!(matched.as_bytes(), "😀👍🏼🎉".as_bytes());
     }
+    #[test]
+    fn test_substitute() {
+        let hay = "0123456789😀👍🏼🎉abcdefghijklmnopqrst😀👍🏼🎉auvwxyzABCKLMNOPQRSTUVWXYZ";
+        let pattern = r"(*UTF)
+            (?x)    (?#: Allow comments and whitespace.)
+
+            [^\N{U+0000}-\N{U+007F}]    (?#: Non-ascii code points.)
+            +                           (?#: One or more times.)
+            ";
+        let re = RegexBuilder::new()
+            .extended(true)
+            .utf(true)
+            .jit_if_available(true)
+            .build(pattern)
+            .unwrap();
+        let mut output = Vec::new();
+        re.substitute(hay.as_bytes(), b"42", &mut output).unwrap();
+        assert_eq!(&output, "012345678942abcdefghijklmnopqrst😀👍🏼🎉auvwxyzABCKLMNOPQRSTUVWXYZ".as_bytes());
+    }
+    #[test]
+    fn test_substitute_all() {
+        let hay = "0123456789😀👍🏼🎉abcdefghijklmnopqrst😀👍🏼🎉auvwxyzABCKLMNOPQRSTUVWXYZ";
+        let pattern = r"(*UTF)
+            (?x)    (?#: Allow comments and whitespace.)
+
+            [^\N{U+0000}-\N{U+007F}]    (?#: Non-ascii code points.)
+            +                           (?#: One or more times.)
+            ";
+        let re = RegexBuilder::new()
+            .extended(true)
+            .utf(true)
+            .jit_if_available(true)
+            .build(pattern)
+            .unwrap();
+        let mut output = Vec::new();
+        re.substitute_all(hay.as_bytes(), b"42", &mut output).unwrap();
+        assert_eq!(&output, "012345678942abcdefghijklmnopqrst42auvwxyzABCKLMNOPQRSTUVWXYZ".as_bytes());
+    }
 }
diff --git a/src/ffi.rs b/src/ffi.rs
index aaabf74..ed7fb3c 100644
--- a/src/ffi.rs
+++ b/src/ffi.rs
@@ -258,6 +258,54 @@ impl Code {
             Ok(1 + count as usize)
         }
     }
+
+    /// Substitute the replacement pattern in subject and put the output in
+    /// output vec. Output vec is will be cleared before use.
+    pub(crate) fn substitute(&self, subject: &[u8], replacement: &[u8], output: &mut Vec<u8>, options: u32) -> Result<usize, Error>{
+        output.clear();
+        let mut output_length = output.capacity();
+        let mut rc = unsafe {
+            pcre2_substitute_8(
+                self.as_ptr(),
+                subject.as_ptr(),
+                subject.len(),
+                0, //startoffset
+                PCRE2_SUBSTITUTE_OVERFLOW_LENGTH | options,
+                ptr::null_mut(), //match_data
+                ptr::null_mut(), //match context
+                replacement.as_ptr(),
+                replacement.len(),
+                output.as_mut_ptr(),
+                &mut output_length,
+            )
+        };
+        if rc == PCRE2_ERROR_NOMEMORY {
+            output.reserve_exact(output_length + 1);
+            rc = unsafe {
+                pcre2_substitute_8(
+                    self.as_ptr(),
+                    subject.as_ptr(),
+                    subject.len(),
+                    0, //startoffset
+                    options,
+                    ptr::null_mut(), //match_data
+                    ptr::null_mut(), //match context
+                    replacement.as_ptr(),
+                    replacement.len(),
+                    output.as_mut_ptr(),
+                    &mut output_length,
+                )
+            }
+        }
+        if rc < 0 {
+            Err(Error::info(rc))
+        } else {
+            // Safety: pcre2_substitute_8 method above would set this
+            // field correctly.
+            unsafe { output.set_len(output_length) };
+            Ok(rc as usize)
+        }
+    }
 }
 
 /// A low level representation of PCRE2's compilation context.