From 3570cc365f03f99520f8150c9f56da2f714756a0 Mon Sep 17 00:00:00 2001
From: Ed Page <eopage@gmail.com>
Date: Fri, 19 Apr 2024 15:50:52 -0500
Subject: [PATCH] feat(filter): Allow regexes for substitutions

Cherry pick b5220cd90edbb654f958b74c86a8b94c59c3c90b (#281)
Cherry pick 38c4fb8887414b53d2bc462e6e754d26289a9a14 (#285)
---
 Cargo.lock                              |  1 +
 crates/snapbox/Cargo.toml               |  3 +
 crates/snapbox/src/filter/redactions.rs | 93 ++++++++++++++++++++++++-
 3 files changed, 96 insertions(+), 1 deletion(-)

diff --git a/Cargo.lock b/Cargo.lock
index b4554403..6f750544 100644
--- a/Cargo.lock
+++ b/Cargo.lock
@@ -808,6 +808,7 @@ dependencies = [
  "libc",
  "normalize-line-endings",
  "os_pipe",
+ "regex",
  "serde",
  "serde_json",
  "similar",
diff --git a/crates/snapbox/Cargo.toml b/crates/snapbox/Cargo.toml
index 0e4e8752..ea231fc1 100644
--- a/crates/snapbox/Cargo.toml
+++ b/crates/snapbox/Cargo.toml
@@ -42,6 +42,8 @@ path = ["dir"]
 cmd = ["dep:os_pipe", "dep:wait-timeout", "dep:libc", "dep:windows-sys"]
 ## Building of examples for snapshotting
 examples = ["dep:escargot"]
+## Regex text substitutions
+regex = ["dep:regex"]
 
 ## Snapshotting of json
 json = ["structured-data", "dep:serde_json", "dep:serde"]
@@ -92,6 +94,7 @@ document-features = { version = "0.2.6", optional = true }
 serde_json = { version = "1.0.85", optional = true}
 anstyle-svg = { version = "0.1.3", optional = true }
 serde = { version = "1.0.198", optional = true }
+regex = { version = "1.10.4", optional = true, default-features = false, features = ["std"] }
 
 [target.'cfg(windows)'.dependencies]
 windows-sys = { version = "0.52.0", features = ["Win32_Foundation"], optional = true }
diff --git a/crates/snapbox/src/filter/redactions.rs b/crates/snapbox/src/filter/redactions.rs
index c4a3e3c2..5dda69d5 100644
--- a/crates/snapbox/src/filter/redactions.rs
+++ b/crates/snapbox/src/filter/redactions.rs
@@ -32,6 +32,17 @@ impl Redactions {
     /// let mut subst = snapbox::Redactions::new();
     /// subst.insert("[EXE]", std::env::consts::EXE_SUFFIX);
     /// ```
+    ///
+    /// With the `regex` feature, you can define patterns using regexes.
+    /// You can choose to replace a subset of the regex by giving it the named capture group
+    /// `redacted`.
+    ///
+    /// ```rust
+    /// # #[cfg(feature = "regex")] {
+    /// let mut subst = snapbox::Redactions::new();
+    /// subst.insert("[OBJECT]", regex::Regex::new("(?<redacted>(world|moon))").unwrap());
+    /// # }
+    /// ```
     pub fn insert(
         &mut self,
         placeholder: &'static str,
@@ -108,10 +119,12 @@ pub struct RedactedValue {
     inner: Option<RedactedValueInner>,
 }
 
-#[derive(Clone, Debug, PartialOrd, Ord, PartialEq, Eq)]
+#[derive(Clone, Debug)]
 enum RedactedValueInner {
     Str(&'static str),
     String(String),
+    #[cfg(feature = "regex")]
+    Regex(regex::Regex),
 }
 
 impl RedactedValueInner {
@@ -119,6 +132,21 @@ impl RedactedValueInner {
         match self {
             Self::Str(s) => buffer.find(s).map(|offset| offset..(offset + s.len())),
             Self::String(s) => buffer.find(s).map(|offset| offset..(offset + s.len())),
+            #[cfg(feature = "regex")]
+            Self::Regex(r) => {
+                let captures = r.captures(buffer)?;
+                let m = captures.name("redacted").or_else(|| captures.get(0))?;
+                Some(m.range())
+            }
+        }
+    }
+
+    fn as_cmp(&self) -> &str {
+        match self {
+            Self::Str(s) => s,
+            Self::String(s) => s,
+            #[cfg(feature = "regex")]
+            Self::Regex(s) => s.as_str(),
         }
     }
 }
@@ -166,6 +194,42 @@ impl From<Cow<'static, str>> for RedactedValue {
     }
 }
 
+#[cfg(feature = "regex")]
+impl From<regex::Regex> for RedactedValue {
+    fn from(inner: regex::Regex) -> Self {
+        Self {
+            inner: Some(RedactedValueInner::Regex(inner)),
+        }
+    }
+}
+
+#[cfg(feature = "regex")]
+impl From<&'_ regex::Regex> for RedactedValue {
+    fn from(inner: &'_ regex::Regex) -> Self {
+        inner.clone().into()
+    }
+}
+
+impl PartialOrd for RedactedValueInner {
+    fn partial_cmp(&self, other: &Self) -> Option<std::cmp::Ordering> {
+        Some(self.as_cmp().cmp(other.as_cmp()))
+    }
+}
+
+impl Ord for RedactedValueInner {
+    fn cmp(&self, other: &Self) -> std::cmp::Ordering {
+        self.as_cmp().cmp(other.as_cmp())
+    }
+}
+
+impl PartialEq for RedactedValueInner {
+    fn eq(&self, other: &Self) -> bool {
+        self.as_cmp().eq(other.as_cmp())
+    }
+}
+
+impl Eq for RedactedValueInner {}
+
 /// Replacements is `(from, to)`
 fn replace_many<'a>(
     buffer: &mut String,
@@ -478,4 +542,31 @@ mod test {
         let actual = normalize(input, pattern, &sub);
         assert_eq!(actual, pattern);
     }
+
+    #[test]
+    #[cfg(feature = "regex")]
+    fn substitute_regex_unnamed() {
+        let input = "Hello world!";
+        let pattern = "Hello [OBJECT]!";
+        let mut sub = Redactions::new();
+        sub.insert("[OBJECT]", regex::Regex::new("world").unwrap())
+            .unwrap();
+        let actual = normalize(input, pattern, &sub);
+        assert_eq!(actual, pattern);
+    }
+
+    #[test]
+    #[cfg(feature = "regex")]
+    fn substitute_regex_named() {
+        let input = "Hello world!";
+        let pattern = "Hello [OBJECT]!";
+        let mut sub = Redactions::new();
+        sub.insert(
+            "[OBJECT]",
+            regex::Regex::new("(?<redacted>world)!").unwrap(),
+        )
+        .unwrap();
+        let actual = normalize(input, pattern, &sub);
+        assert_eq!(actual, pattern);
+    }
 }