Merge pull request #37 from snipsco/release/0.55.2

Release 0.55.2
snipsco · Apr 20, 2018 · 6a48f86 · 6a48f86
2 parents da54c23 + b08ccc2
commit 6a48f86
Show file tree

Hide file tree

Showing 15 changed files with 50 additions and 58 deletions.
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -1,8 +1,15 @@
 # Changelog
 All notable changes to this project will be documented in this file.
 
-## [0.55.1] - 2018-04-10
+## [0.55.2] - 2018-04-20
+### Changed
+- Make configurations and pipeline objects public
+- Bump snips-nlu-ontology to `0.54.3`
 
+### Fixed
+- Bug with prefix and suffix features
+
+## [0.55.1] - 2018-04-10
 ### Added
 - Add support for the `length` feature function in slot filling feature extractrion
 
@@ -23,5 +30,6 @@ All notable changes to this project will be documented in this file.
 - Rename python package to `snips_nlu_rust`
 
 
+[0.55.2]: https://github.com/snipsco/snips-nlu-rs/compare/0.55.1...0.55.2
 [0.55.1]: https://github.com/snipsco/snips-nlu-rs/compare/0.55.0...0.55.1
 [0.55.0]: https://github.com/snipsco/snips-nlu-rs/compare/0.54.0...0.55.0
diff --git a/snips-nlu-ffi/Cargo.toml b/snips-nlu-ffi/Cargo.toml
@@ -1,14 +1,14 @@
 [package]
 name = "snips-nlu-ffi"
-version = "0.55.1"
+version = "0.55.2"
 authors = [
     "Kevin Lefevre <[email protected]>",
     "Thibaut Lorrain <[email protected]>"
 ]
 
 [dependencies]
 snips-nlu-lib = { path = "../snips-nlu-lib" }
-snips-nlu-ontology-ffi-macros = { git = "https://github.com/snipsco/snips-nlu-ontology", tag = "0.54.2" }
+snips-nlu-ontology-ffi-macros = { git = "https://github.com/snipsco/snips-nlu-ontology", tag = "0.54.3" }
 failure = "0.1"
 lazy_static = "1.0"
 libc = "0.2"

diff --git a/snips-nlu-ffi/kotlin/build.gradle b/snips-nlu-ffi/kotlin/build.gradle
@@ -11,7 +11,7 @@ buildscript {
 
 apply plugin: 'kotlin'
 
-version = "0.55.1"
+version = "0.55.2"
 group = "ai.snips"
 
 repositories {

diff --git a/snips-nlu-ffi/python/snips-nlu-python-ffi/Cargo.toml b/snips-nlu-ffi/python/snips-nlu-python-ffi/Cargo.toml
@@ -9,4 +9,4 @@ crate-type = ["cdylib"]
 
 [dependencies]
 libc = "0.2"
-snips-nlu-ffi = { git = "https://github.com/snipsco/snips-nlu-rs", tag = "0.55.1" }
+snips-nlu-ffi = { git = "https://github.com/snipsco/snips-nlu-rs", tag = "0.55.2" }
diff --git a/snips-nlu-ffi/python/snips_nlu_rust/__version__ b/snips-nlu-ffi/python/snips_nlu_rust/__version__
@@ -1 +1 @@
-0.55.1
+0.55.2
diff --git a/snips-nlu-ffi/swift/SnipsNlu/Dependencies/build.sh b/snips-nlu-ffi/swift/SnipsNlu/Dependencies/build.sh
@@ -4,7 +4,7 @@
 
 set -e
 
-VERSION="0.55.1"
+VERSION="0.55.2"
 SYSTEM=$1
 
 if [ $SYSTEM != ios ] && [ $SYSTEM != macos ]; then

diff --git a/snips-nlu-lib/Cargo.toml b/snips-nlu-lib/Cargo.toml
@@ -1,6 +1,6 @@
 [package]
 name = "snips-nlu-lib"
-version = "0.55.1"
+version = "0.55.2"
 authors = [
     "Thibaut Lorrain <[email protected]>",
     "Kevin Lefevre <[email protected]>"
@@ -11,9 +11,9 @@ description = "Rust implementation of Snips NLU"
 [dependencies]
 snips-nlu-resources-packed = { path = "../snips-nlu-resources-packed" }
 crfsuite = { git = "https://github.com/snipsco/crfsuite-rs", rev = "b18d95c" }
-snips-nlu-ontology = { git = "https://github.com/snipsco/snips-nlu-ontology", tag = "0.54.2" }
-snips-nlu-ontology-parsers = { git = "https://github.com/snipsco/snips-nlu-ontology", tag = "0.54.2" }
-snips-nlu-utils = { git = "https://github.com/snipsco/snips-nlu-utils", tag = "0.6.0" }
+snips-nlu-ontology = { git = "https://github.com/snipsco/snips-nlu-ontology", tag = "0.54.3" }
+snips-nlu-ontology-parsers = { git = "https://github.com/snipsco/snips-nlu-ontology", tag = "0.54.3" }
+snips-nlu-utils = { git = "https://github.com/snipsco/snips-nlu-utils", tag = "0.6.1" }
 dinghy-test = "0.3"
 failure = "0.1"
 base64 = "0.9"

diff --git a/snips-nlu-lib/src/intent_classifier/log_reg_intent_classifier.rs b/snips-nlu-lib/src/intent_classifier/log_reg_intent_classifier.rs
@@ -101,6 +101,15 @@ impl IntentClassifier for LogRegIntentClassifier {
     }
 }
 
+impl LogRegIntentClassifier {
+    pub fn compute_features(&self, input: &str) -> Result<Array1<f32>> {
+        self.featurizer
+            .as_ref()
+            .map(|featurizer| featurizer.transform(input))
+            .unwrap_or_else(|| Ok(Array::from_iter(vec![])))
+    }
+}
+
 #[cfg(test)]
 mod tests {
     use super::*;

diff --git a/snips-nlu-lib/src/lib.rs b/snips-nlu-lib/src/lib.rs
@@ -38,9 +38,11 @@ mod utils;
 
 pub const MODEL_VERSION: &str = "0.14.0";
 
-pub use configurations::{FileBasedConfiguration, NluEngineConfiguration,
-                         NluEngineConfigurationConvertible, ZipBasedConfiguration};
+pub use configurations::*;
 pub use errors::*;
+pub use intent_classifier::{IntentClassifier, LogRegIntentClassifier};
+pub use intent_parser::{DeterministicIntentParser, IntentParser, ProbabilisticIntentParser};
 pub use nlu_engine::SnipsNluEngine;
+pub use slot_filler::{CRFSlotFiller, SlotFiller};
 pub use nlu_utils::token::{compute_all_ngrams, tokenize_light};
 pub use utils::file_path; // This is used by benches
diff --git a/snips-nlu-lib/src/slot_filler/crf_slot_filler.rs b/snips-nlu-lib/src/slot_filler/crf_slot_filler.rs
@@ -142,6 +142,16 @@ impl SlotFiller for CRFSlotFiller {
     }
 }
 
+impl CRFSlotFiller {
+    pub fn compute_features(&self, text: &str) -> Vec<Vec<(String, String)>> {
+        let tokens = tokenize(text, NluUtilsLanguage::from_language(self.language));
+        if tokens.is_empty() {
+            return vec![];
+        };
+        self.feature_processor.compute_features(&&*tokens)
+    }
+}
+
 // We need to use base64 encoding to ensure ascii encoding because of encoding issues in
 // python-crfsuite
 

diff --git a/snips-nlu-lib/src/slot_filler/feature_processor.rs b/snips-nlu-lib/src/slot_filler/feature_processor.rs
@@ -173,7 +173,7 @@ fn prefix_feature_function(
 ) -> Result<FeatureFunction> {
     let n = parse_as_u64(args, "prefix_size")? as usize;
     Ok(FeatureFunction::new(
-        &format!("prefix-{}", n),
+        &format!("prefix_{}", n),
         offsets,
         move |t, i| features::prefix(&t[i].value, n),
     ))
@@ -185,7 +185,7 @@ fn suffix_feature_function(
 ) -> Result<FeatureFunction> {
     let n = parse_as_u64(args, "suffix_size")? as usize;
     Ok(FeatureFunction::new(
-        &format!("suffix-{}", n),
+        &format!("suffix_{}", n),
         offsets,
         move |t, i| features::suffix(&t[i].value, n),
     ))

diff --git a/snips-nlu-lib/src/slot_filler/features.rs b/snips-nlu-lib/src/slot_filler/features.rs
@@ -1,9 +1,9 @@
 use itertools::Itertools;
 
 use super::crf_utils::{get_scheme_prefix, TaggingScheme};
-use super::features_utils::{get_shape, get_word_chunk, initial_string_from_tokens};
+use super::features_utils::{get_word_chunk, initial_string_from_tokens};
 use nlu_utils::range::ranges_overlap;
-use nlu_utils::string::normalize;
+use nlu_utils::string::{get_shape, normalize};
 use nlu_utils::token::{compute_all_ngrams, Token};
 use resources::gazetteer::Gazetteer;
 #[cfg(test)]

diff --git a/snips-nlu-lib/src/slot_filler/features_utils.rs b/snips-nlu-lib/src/slot_filler/features_utils.rs
@@ -24,30 +24,6 @@ pub fn get_word_chunk(
     }
 }
 
-pub fn get_shape(string: &str) -> String {
-    if string.chars().all(char::is_lowercase) {
-        "xxx".to_string()
-    } else if string.chars().all(char::is_uppercase) {
-        "XXX".to_string()
-    } else if is_title_case(string) {
-        "Xxx".to_string()
-    } else {
-        "xX".to_string()
-    }
-}
-
-fn is_title_case(string: &str) -> bool {
-    let mut first = true;
-    for c in string.chars() {
-        match (first, c.is_uppercase()) {
-            (true, true) => first = false,
-            (false, false) => continue,
-            _ => return false,
-        }
-    }
-    !first
-}
-
 pub fn initial_string_from_tokens(tokens: &[Token]) -> String {
     let mut current_index = 0;
     let mut chunks: Vec<String> = Vec::with_capacity(2 * tokens.len() - 1);
@@ -114,19 +90,6 @@ mod tests {
         assert_eq!(word_chunk, None);
     }
 
-    #[test]
-    fn get_shape_works() {
-        // Given
-        let inputs = vec!["héllo", "Héllo", "HÉLLO", "hélLo", "!!", "Éllo", "É"];
-
-        // When
-        let actual_shapes: Vec<String> = inputs.into_iter().map(|i| get_shape(i)).collect();
-
-        // Then
-        let expected_shapes = vec!["xxx", "Xxx", "XXX", "xX", "xX", "Xxx", "XXX"];
-        assert_eq!(actual_shapes, expected_shapes)
-    }
-
     #[test]
     fn initial_string_from_tokens_works() {
         // Given

diff --git a/snips-nlu-resources-packed/Cargo.toml b/snips-nlu-resources-packed/Cargo.toml
@@ -1,14 +1,14 @@
 [package]
 name = "snips-nlu-resources-packed"
-version = "0.55.1"
+version = "0.55.2"
 authors = ["Thibaut Lorrain <[email protected]>"]
 build = "build.rs"
 
 [dependencies]
 failure = "0.1"
 lazy_static = "1.0"
 phf = { version = "0.7.21", features = ["unicase"] }
-snips-nlu-ontology = { git = "https://github.com/snipsco/snips-nlu-ontology", tag = "0.54.2" }
+snips-nlu-ontology = { git = "https://github.com/snipsco/snips-nlu-ontology", tag = "0.54.3" }
 
 [build-dependencies]
 snips-nlu-resources = { path = "../snips-nlu-resources"}

diff --git a/snips-nlu-resources/Cargo.toml b/snips-nlu-resources/Cargo.toml
@@ -1,11 +1,11 @@
 [package]
 name = "snips-nlu-resources"
-version = "0.55.1"
+version = "0.55.2"
 authors = ["Thibaut Lorrain <[email protected]>"]
 
 [dependencies]
 csv = "0.15"
 failure = "0.1"
 itertools = { version = "0.7", default-features = false }
 lazy_static = "1.0"
-snips-nlu-utils = { git = "https://github.com/snipsco/snips-nlu-utils", tag = "0.6.0" }
+snips-nlu-utils = { git = "https://github.com/snipsco/snips-nlu-utils", tag = "0.6.1" }