Merge pull request #2 from AACTools/feature/ipaconvert

feat: X-SAMPA / Praat / SIL / Branner → IPA phonetic-alphabet modifiers

Author: willwade

Cargo.toml

@@ -15,6 +15,11 @@ quick-xml = "0.37"
 serde = { version = "1.0", features = ["derive"] }
 serde_json = "1.0"
 thiserror = "1.0"
+ipa-translate = { version = "0.2", optional = true }
+
+[features]
+default = ["phonetic-translation"]
+phonetic-translation = ["dep:ipa-translate"]
 
 [dev-dependencies]
 # Removed problematic benchmark dependencies for now

src/formatters/ssml/base.rs

@@ -600,6 +600,14 @@ impl SsmlFormatterBase {
                 }
                 attrs
             })),
+            "xsampa" | "praat" | "sil" | "branner" => {
+                let translated = translate_to_ipa(&key.to_lowercase(), value)?;
+                let mut attrs = vec![("alphabet".to_string(), "ipa".to_string())];
+                if !translated.is_empty() {
+                    attrs.push(("ph".to_string(), translated));
+                }
+                Some(("phoneme".to_string(), attrs))
+            }
             "sub" => {
                 if !value.is_empty() {
                     attributes.push(("alias".to_string(), value.to_string()));
@@ -735,6 +743,33 @@ impl SsmlFormatterBase {
     }
 }
 
+/// Translate a phonetic-alphabet value to IPA.
+///
+/// Recognized keys (lowercase): `xsampa`, `praat`, `sil`, `branner`. Returns
+/// `None` for any unrecognized key, or for every key when the
+/// `phonetic-translation` feature is disabled — callers treat `None` as
+/// "drop this modifier."
+///
+/// Conversion is infallible (garbage in, garbage out) per the upstream crate.
+#[cfg(feature = "phonetic-translation")]
+pub fn translate_to_ipa(key: &str, value: &str) -> Option<String> {
+    if value.is_empty() {
+        return Some(String::new());
+    }
+    match key {
+        "xsampa" => Some(ipa_translate::xsampa_to_ipa(value)),
+        "praat" => Some(ipa_translate::praat_to_ipa(value)),
+        "sil" => Some(ipa_translate::sil_to_ipa(value)),
+        "branner" => Some(ipa_translate::branner_to_ipa(value)),
+        _ => None,
+    }
+}
+
+#[cfg(not(feature = "phonetic-translation"))]
+pub fn translate_to_ipa(_key: &str, _value: &str) -> Option<String> {
+    None
+}
+
 pub fn format_attr_string_ordered(tag_name: &str, attributes: &TagAttrs) -> String {
     let fixed_order: Vec<&str> = match tag_name {
         "say-as" => vec!["interpret-as", "format"],
@@ -789,3 +824,64 @@ impl Formatter for SsmlFormatterBase {
         self.format_node_with_tags(node)
     }
 }
+
+#[cfg(test)]
+mod phonetic_alphabet_tests {
+    use super::*;
+    use crate::formatters::base::FormatterOptions;
+
+    fn fmt() -> SsmlFormatterBase {
+        SsmlFormatterBase::new(FormatterOptions::default())
+    }
+
+    #[cfg(feature = "phonetic-translation")]
+    #[test]
+    fn xsampa_value_becomes_ipa_phoneme_tag() {
+        let (tag, attrs) = fmt().attribute_to_tag("xsampa", "spitS").unwrap();
+        assert_eq!(tag, "phoneme");
+        assert_eq!(attrs_get(&attrs, "alphabet"), Some("ipa"));
+        assert_eq!(attrs_get(&attrs, "ph"), Some("spitʃ"));
+    }
+
+    #[cfg(feature = "phonetic-translation")]
+    #[test]
+    fn praat_sil_branner_all_emit_ipa_phoneme() {
+        let cases = [
+            ("praat", r"p\rta\:ft\^h", "pɹaːtʰ"),
+            ("sil", "si=l", "sɪl"),
+            ("branner", "br&ae):nE&r^", "bɹæːnɜ˞"),
+        ];
+        for (key, src, expected_ipa) in cases {
+            let (tag, attrs) = fmt().attribute_to_tag(key, src).unwrap();
+            assert_eq!(tag, "phoneme", "key {}", key);
+            assert_eq!(attrs_get(&attrs, "alphabet"), Some("ipa"), "key {}", key);
+            assert_eq!(
+                attrs_get(&attrs, "ph"),
+                Some(expected_ipa),
+                "key {}",
+                key
+            );
+        }
+    }
+
+    #[cfg(not(feature = "phonetic-translation"))]
+    #[test]
+    fn phonetic_keys_dropped_when_feature_disabled() {
+        for key in ["xsampa", "praat", "sil", "branner"] {
+            assert!(
+                fmt().attribute_to_tag(key, "anything").is_none(),
+                "{} should be dropped without phonetic-translation feature",
+                key
+            );
+        }
+    }
+
+    #[cfg(feature = "phonetic-translation")]
+    #[test]
+    fn empty_value_emits_phoneme_without_ph_attr() {
+        let (tag, attrs) = fmt().attribute_to_tag("xsampa", "").unwrap();
+        assert_eq!(tag, "phoneme");
+        assert_eq!(attrs_get(&attrs, "alphabet"), Some("ipa"));
+        assert_eq!(attrs_get(&attrs, "ph"), None);
+    }
+}

src/formatters/ssml/google_assistant.rs

@@ -26,7 +26,7 @@ impl GoogleAssistantSsmlFormatter {
             }
             "excited" | "disappointed" => None,
             "voice" | "lang" => None,
-            "ipa" => None,
+            "ipa" | "xsampa" | "praat" | "sil" | "branner" => None,
             "style" => {
                 if !value.is_empty() {
                     attributes.push(("name".to_string(), value.to_string()));
@@ -40,19 +40,24 @@ impl GoogleAssistantSsmlFormatter {
     fn format_google_text_modifier(&self, node: &AstNode) -> Result<String> {
         let mut tags: Vec<TagInfo> = Vec::new();
         let mut last_say_as: Option<TagInfo> = None;
-        let mut has_ipa = false;
-        let mut non_ipa_count = 0;
+        let mut phonetic_key: Option<String> = None;
+        let mut non_phonetic_count = 0;
 
         for key in &node.attribute_keys {
             let value = match node.attributes.get(key) {
                 Some(v) => v,
                 None => continue,
             };
 
-            if key.to_lowercase() == "ipa" {
-                has_ipa = true;
+            if matches!(
+                key.to_lowercase().as_str(),
+                "ipa" | "xsampa" | "praat" | "sil" | "branner"
+            ) {
+                if phonetic_key.is_none() {
+                    phonetic_key = Some(key.clone());
+                }
             } else {
-                non_ipa_count += 1;
+                non_phonetic_count += 1;
             }
 
             if let Some(tag_info) = self.google_attribute_to_tag(key, value) {
@@ -71,15 +76,13 @@ impl GoogleAssistantSsmlFormatter {
             }
         }
 
-        if has_ipa && non_ipa_count == 0 {
+        if phonetic_key.is_some() && non_phonetic_count == 0 {
             return Ok(node.text.clone());
         }
 
-        if has_ipa {
-            if let Some(ipa_tag) = self
-                .base
-                .attribute_to_tag("ipa", node.attributes.get("ipa").unwrap_or(&String::new()))
-            {
+        if let Some(key) = phonetic_key {
+            let value = node.attributes.get(&key).cloned().unwrap_or_default();
+            if let Some(ipa_tag) = self.base.attribute_to_tag(&key, &value) {
                 tags.push(ipa_tag);
             }
         }

tests/integration_test.rs

@@ -31,6 +31,21 @@ fn test_all_test_cases() {
             .and_then(|n| n.to_str())
             .unwrap_or("unknown");
 
+        // Fixtures that exercise the phonetic-translation feature; only
+        // run them when that feature is on.
+        #[cfg(not(feature = "phonetic-translation"))]
+        if matches!(
+            test_name,
+            "xsampa-standard"
+                | "xsampa-stress"
+                | "praat-standard"
+                | "sil-standard"
+                | "branner-standard"
+                | "combo-xsampa-emphasis"
+        ) {
+            continue;
+        }
+
         // Read the .smd input file
         let smd_file = test_dir.join(format!("{}.smd", test_name));
         if !smd_file.exists() {