Refactor Android string resource normalization

3 files changed, 38 insertions, 10 deletions

diff --git a/android/translations-converter/Cargo.toml b/android/translations-converter/Cargo.toml
index fab9c4484d..2b32cd9303 100644
--- a/android/translations-converter/Cargo.toml
+++ b/android/translations-converter/Cargo.toml
@@ -7,6 +7,7 @@ edition = "2018"
 # See more keys and their definitions at https://doc.rust-lang.org/cargo/reference/manifest.html
 
 [dependencies]
+lazy_static = "1"
 regex = "1"
 serde = { version = "1", features = ["derive"] }
 serde-xml-rs = "0.4"
diff --git a/android/translations-converter/src/android.rs b/android/translations-converter/src/android.rs
index 9a7fd1c99a..2481d27a4e 100644
--- a/android/translations-converter/src/android.rs
+++ b/android/translations-converter/src/android.rs
@@ -1,9 +1,16 @@
+use lazy_static::lazy_static;
+use regex::Regex;
 use serde::{Deserialize, Serialize};
 use std::{
     fmt::{self, Display, Formatter},
     ops::{Deref, DerefMut},
 };
 
+lazy_static! {
+    static ref LINE_BREAKS: Regex = Regex::new(r"\s*\n\s*").unwrap();
+    static ref APOSTROPHES: Regex = Regex::new(r"\\'").unwrap();
+}
+
 /// Contents of an Android string resources file.
 ///
 /// This type can be created directly deserializing the `strings.xml` file.
@@ -31,6 +38,15 @@ impl StringResources {
             entries: Vec::new(),
         }
     }
+
+    /// Normalize the strings into a common format.
+    ///
+    /// Allows the string values to be compared to the gettext messages.
+    pub fn normalize(&mut self) {
+        for entry in &mut self.entries {
+            entry.normalize();
+        }
+    }
 }
 
 impl Deref for StringResources {
@@ -68,6 +84,18 @@ impl StringResource {
 
         StringResource { name, value }
     }
+
+    /// Normalize the string value into a common format.
+    ///
+    /// Makes it possible to compare the Android strings with the gettext messages.
+    pub fn normalize(&mut self) {
+        // Collapse line breaks present in the XML file
+        let value = LINE_BREAKS.replace_all(&self.value, " ");
+        // Unescape apostrophes
+        let value = APOSTROPHES.replace_all(&value, "'");
+
+        self.value = value.into_owned();
+    }
 }
 
 // Unfortunately, direct serialization to XML isn't working correctly.
diff --git a/android/translations-converter/src/main.rs b/android/translations-converter/src/main.rs
index 09296f87fe..24bf06a574 100644
--- a/android/translations-converter/src/main.rs
+++ b/android/translations-converter/src/main.rs
@@ -3,8 +3,12 @@
 //! The procedure for converting the translations is relatively simple. The base Android string
 //! resources file is first loaded, and then each gettext translation file is loaded and compared to
 //! the Android base strings. For every translation string that matches exactly the Android base
-//! string value, the translated string is used in the new Android strings file for the respective
-//! locale.
+//! string value (after a normalization pass described below), the translated string is used in the
+//! new Android strings file for the respective locale.
+//!
+//! To make the comparison work on most strings, the Android and gettext messages are normalized
+//! first. This means that new lines in the XML files are removed and collapsed into a single space
+//! and apostrophes are unescaped.
 //!
 //! Note that this conversion procedure is very raw and likely very brittle, so while it works for
 //! most cases, it is important to keep in mind that this is just a helper tool and manual steps are
@@ -13,7 +17,6 @@
 mod android;
 mod gettext;
 
-use regex::Regex;
 use std::{
     collections::HashMap,
     fs::{self, File},
@@ -24,21 +27,17 @@ fn main() {
     let resources_dir = Path::new("../src/main/res");
     let strings_file = File::open(resources_dir.join("values/strings.xml"))
         .expect("Failed to open string resources file");
-    let string_resources: android::StringResources =
+    let mut string_resources: android::StringResources =
         serde_xml_rs::from_reader(strings_file).expect("Failed to read string resources file");
 
-    let line_breaks = Regex::new(r"\s*\n\s*").unwrap();
-    let apostrophes = Regex::new(r"\\'").unwrap();
+    string_resources.normalize();
 
     let known_strings: HashMap<_, _> = string_resources
         .into_iter()
         .map(|string| {
             let android_id = string.name;
-            let without_line_breaks = line_breaks.replace_all(&string.value, " ");
-            let without_escaped_apostrophes = apostrophes.replace_all(&without_line_breaks, "'");
-            let string_value = without_escaped_apostrophes.into_owned();
 
-            (string_value, android_id)
+            (string.value, android_id)
         })
         .collect();