Implement `Normalize` for `android::StringValue`

2 files changed, 32 insertions, 0 deletions

diff --git a/android/translations-converter/src/android/mod.rs b/android/translations-converter/src/android/mod.rs
index b43277c509..8bc9f8e41e 100644
--- a/android/translations-converter/src/android/mod.rs
+++ b/android/translations-converter/src/android/mod.rs
@@ -4,5 +4,6 @@ mod strings;
 
 pub use self::{
     plurals::{PluralQuantity, PluralResource, PluralResources},
+    string_value::StringValue,
     strings::{StringResource, StringResources},
 };
diff --git a/android/translations-converter/src/normalize.rs b/android/translations-converter/src/normalize.rs
index e9a2eb3757..395f0186cf 100644
--- a/android/translations-converter/src/normalize.rs
+++ b/android/translations-converter/src/normalize.rs
@@ -1,6 +1,37 @@
+use lazy_static::lazy_static;
+use regex::Regex;
+
 pub trait Normalize {
     /// Normalize the string value into a common format.
     ///
     /// Makes it possible to compare different representations of translation messages.
     fn normalize(&self) -> String;
 }
+
+mod android {
+    use super::*;
+    use crate::android::StringValue;
+
+    lazy_static! {
+        static ref LINE_BREAKS: Regex = Regex::new(r"\s*\n\s*").unwrap();
+        static ref APOSTROPHES: Regex = Regex::new(r"\\'").unwrap();
+        static ref DOUBLE_QUOTES: Regex = Regex::new(r#"\\""#).unwrap();
+        static ref PARAMETERS: Regex = Regex::new(r"%[0-9]*\$").unwrap();
+    }
+
+    impl Normalize for StringValue {
+        fn normalize(&self) -> String {
+            // Collapse line breaks present in the XML file
+            let value = LINE_BREAKS.replace_all(&*self, " ");
+            // Unescape apostrophes
+            let value = APOSTROPHES.replace_all(&value, "'");
+            // Unescape double quotes
+            let value = DOUBLE_QUOTES.replace_all(&value, r#"""#);
+            // Mark where parameters are positioned, removing the parameter index
+            let value = PARAMETERS.replace_all(&value, "%");
+
+            // Unescape XML characters
+            htmlize::unescape(value.as_bytes())
+        }
+    }
+}