quickner - npm Package Compare versions

Comparing version 0.0.1a12 to 0.0.1a13

.gitignore

 /target
-/quickner-cli/target
+/quickner-core/target
 
@@ -4,0 +4,0 @@ # Byte-compiled / optimized / DLL files

Cargo.lock

@@ -548,3 +548,3 @@ # This file is automatically @generated by Cargo.
 name = "quickner"
-version = "0.0.1-alpha.12"
+version = "0.0.1-alpha.13"
 dependencies = [
@@ -558,3 +558,3 @@ "pyo3",
 name = "quickner-core"
-version = "0.0.1-alpha.12"
+version = "0.0.1-alpha.13"
 dependencies = [
@@ -561,0 +561,0 @@ "clap",

Cargo.toml

 [package]
 name = "quickner"
-version = "0.0.1-alpha.12"
+version = "0.0.1-alpha.13"
 edition = "2021"
@@ -5,0 +5,0 @@ authors = ["Omar MHAIMDAT"]

local_dependencies/quickner-core/Cargo.toml

 [package]
 name = "quickner-core"
-version = "0.0.1-alpha.12"
+version = "0.0.1-alpha.13"
 edition = "2021"
@@ -5,0 +5,0 @@ authors = ["Omar MHAIMDAT"]

local_dependencies/quickner-core/src/document.rs

@@ -22,6 +22,2 @@ use serde::{Deserialize, Serialize};
     }
-
-    fn ne(&self, other: &Self) -> bool {
-        !self.eq(other)
-    }
 }
@@ -28,0 +24,0 @@

local_dependencies/quickner-core/src/entity.rs

@@ -0,1 +1,3 @@
+use std::hash::Hash;
+
 use serde::{Deserialize, Serialize};
@@ -7,3 +9,3 @@
 /// annotate the text.
-#[derive(Eq, Hash, Serialize, Deserialize, Clone, Debug)]
+#[derive(Eq, Serialize, Deserialize, Clone, Debug)]
 pub struct Entity {
@@ -19,1 +21,8 @@ pub name: String,
 }
+
+impl Hash for Entity {
+    fn hash<H: std::hash::Hasher>(&self, state: &mut H) {
+        self.name.hash(state);
+        self.label.hash(state);
+    }
+}

local_dependencies/quickner-core/src/quickner.rs

@@ -27,3 +27,3 @@ use crate::{
     pub config: Config,
-    pub config_file: String,
+    pub config_file: Option<String>,
     pub documents: Vec<Document>,
@@ -40,3 +40,3 @@ pub entities: Vec<Entity>,
             config: Config::default(),
-            config_file: "./config.toml".to_string(),
+            config_file: Some("./config.toml".to_string()),
             documents: Vec::new(),
@@ -74,3 +74,3 @@ entities: Vec::new(),
         // let mut annotations = Vec::new();
-        let annotations = entities.iter().map(|entity| {
+        let annotations = entities.iter().filter_map(|entity| {
             let target_len = entity.name.len();
@@ -95,18 +95,12 @@ for (start, _) in text.match_indices(entity.name.as_str()) {
                 {
-                    return (start, start + target_len, entity.label.to_string());
+                    return Some((start, start + target_len, entity.label.to_string()));
                 }
             }
-            (0, 0, String::new())
+            None
         });
-        let annotations: Vec<(usize, usize, String)> = annotations
-            .filter(|(_, _, label)| !label.is_empty())
-            .collect();
         // Unique annotations
-        let mut annotations = annotations
-            .into_iter()
-            .collect::<HashSet<(usize, usize, String)>>()
-            .into_iter()
-            .collect::<Vec<(usize, usize, String)>>();
+        let mut annotations = annotations.collect::<Vec<(usize, usize, String)>>();
+        annotations.sort_by(|a, b| a.0.cmp(&b.0));
+        annotations.dedup();
         // Sort annotations by start index
-        annotations.sort_by(|a, b| a.0.cmp(&b.0));
         if !annotations.is_empty() {
@@ -196,6 +190,7 @@ Some(annotations)
         let config = Config::from_file(config_file.as_str());
-        let mut quick = Quickner::default();
-        quick.config = config;
-        quick.config_file = config_file;
-        quick
+        Quickner {
+            config,
+            config_file: Some(config_file),
+            ..Default::default()
+        }
     }
@@ -283,3 +278,3 @@
         info!("----------------------------------------");
-        if self.entities.len() == 0 {
+        if self.entities.is_empty() {
             let entities: HashSet<Entity> = self.entities(
@@ -292,3 +287,3 @@ config.entities.input.path.as_str(),
         }
-        if self.documents.len() == 0 {
+        if self.documents.is_empty() {
             let texts: HashSet<Text> = self.texts(
@@ -489,3 +484,3 @@ config.texts.input.path.as_str(),
             config: Config::default(),
-            config_file: String::from(""),
+            config_file: None,
             documents,
@@ -547,3 +542,3 @@ entities,
             config: Config::default(),
-            config_file: String::from(""),
+            config_file: None,
             documents,
@@ -585,3 +580,3 @@ entities,
 impl Quickner {
-    fn build_label_index(&mut self) {
+    pub fn build_label_index(&mut self) {
         let mut index: HashMap<String, Vec<String>> = HashMap::new();
@@ -597,3 +592,3 @@ for document in &self.documents {
 
-    fn build_entity_index(&mut self) {
+    pub fn build_entity_index(&mut self) {
         let mut index: HashMap<String, Vec<String>> = HashMap::new();
@@ -660,5 +655,5 @@ for document in &self.documents {
 
-    fn document_hash(documents: &Vec<Document>) -> HashMap<String, Document> {
+    pub fn document_hash(documents: &[Document]) -> HashMap<String, Document> {
         documents
-            .into_iter()
+            .iter()
             .map(|document| (document.id.clone(), document.clone()))
@@ -665,0 +660,0 @@ .collect::<HashMap<String, Document>>()

Makefile

@@ -0,1 +1,2 @@
+all: build test
 .PHONY: build
@@ -6,2 +7,2 @@ build:
 test:
-	python test.py
+	python tests/test.py

PKG-INFO

 Metadata-Version: 2.1
 Name: quickner
-Version: 0.0.1a12
+Version: 0.0.1a13
 Classifier: Programming Language :: Rust
@@ -5,0 +5,0 @@ Classifier: Programming Language :: Python :: Implementation :: CPython

src/pydocument.rs

@@ -6,3 +6,3 @@ use crate::{
 use pyo3::prelude::*;
-use quickner::{hash_string, Document, Entity};
+use quickner::{hash_string, Document};
 use serde::{Deserialize, Serialize};
@@ -78,6 +78,3 @@
         let mut annotation = Document::from_string(self.text.clone());
-        let entities = entities
-            .into_iter()
-            .map(|entity| Entity::from(entity))
-            .collect();
+        let entities = entities.into_iter().collect();
         annotation.annotate(entities, case_sensitive);
@@ -84,0 +81,0 @@ self.label.extend(

src/pyquickner.rs

@@ -37,6 +37,8 @@ use std::collections::HashMap;
 
+pub type SpacyFormat = Vec<(String, HashMap<String, Vec<(usize, usize, String)>>)>;
+
 #[pyclass(name = "SpacyGenerator")]
 pub struct PySpacyGenerator {
     #[pyo3(get)]
-    pub entities: Vec<Vec<(String, HashMap<String, Vec<(usize, usize, String)>>)>>,
+    pub entities: Vec<SpacyFormat>,
 }
@@ -48,3 +50,3 @@
     #[pyo3(signature = (entities))]
-    fn new(entities: Vec<Vec<(String, HashMap<String, Vec<(usize, usize, String)>>)>>) -> Self {
+    fn new(entities: Vec<SpacyFormat>) -> Self {
         PySpacyGenerator { entities }
@@ -57,5 +59,3 @@ }
 
-    fn __next__(
-        mut slf: PyRefMut<'_, Self>,
-    ) -> Option<Vec<(String, HashMap<String, Vec<(usize, usize, String)>>)>> {
+    fn __next__(mut slf: PyRefMut<'_, Self>) -> Option<SpacyFormat> {
         if slf.entities.is_empty() {
@@ -94,3 +94,3 @@ PyGeneratorExit::new_err("No more entities");
             Some(documents) => {
-                quickner.documents = documents.into_iter().map(|x| Document::from(x)).collect();
+                quickner.documents = documents.into_iter().collect();
             }
@@ -101,3 +101,3 @@ None => quickner.documents = Vec::new(),
             Some(entities) => {
-                quickner.entities = entities.into_iter().map(|x| Entity::from(x)).collect();
+                quickner.entities = entities.into_iter().collect();
             }
@@ -116,3 +116,7 @@ None => quickner.entities = Vec::new(),
     pub fn documents(&mut self, documents: Vec<PyDocument>) {
-        self.documents = documents.into_iter().map(|x| x.into()).collect();
+        self.documents = documents.clone();
+        self.quickner.documents = documents.into_iter().collect();
+        self.quickner.documents_hash = Quickner::document_hash(&self.quickner.documents);
+        self.quickner.build_label_index();
+        self.quickner.build_entity_index();
     }
@@ -122,3 +126,4 @@
     pub fn entities(&mut self, entities: Vec<PyEntity>) {
-        self.entities = entities.into_iter().map(|x| x.into()).collect();
+        self.entities = entities.clone();
+        self.quickner.entities = entities.into_iter().collect();
     }
@@ -130,5 +135,4 @@
         }
-        match self.documents {
-            ref mut documents => documents.push(document.clone()),
-        }
+        let documents = &mut self.documents;
+        documents.push(document.clone());
         let document = Document::from(document);
@@ -149,5 +153,4 @@ self.quickner.add_document(document);
         }
-        match self.entities {
-            ref mut entities => entities.push(entity.clone()),
-        }
+        let entities = &mut self.entities;
+        entities.push(entity.clone());
         let entity = Entity {
@@ -202,3 +205,3 @@ name: entity.name,
             .into_iter()
-            .map(|document| PyDocument::from(document))
+            .map(PyDocument::from)
             .collect::<Vec<PyDocument>>();
@@ -210,3 +213,3 @@ self.entities = self
             .into_iter()
-            .map(|entity| PyEntity::from(entity))
+            .map(PyEntity::from)
             .collect::<Vec<PyEntity>>();
@@ -308,5 +311,4 @@ Ok(())
     pub fn find_documents_by_label(&self, label: &str) -> Vec<PyDocument> {
-        let documents_index = match &self.quickner {
-            quickner => quickner.documents_label_index.to_owned(),
-        };
+        let quickner = &self.quickner;
+        let documents_index = quickner.documents_label_index.to_owned();
         let documents_ids = match documents_index.get(label) {
@@ -316,13 +318,12 @@ Some(documents_ids) => documents_ids,
         };
-        let documents = match &self.quickner {
-            quickner => {
-                let documents = documents_ids
-                    .into_iter()
-                    .map(|id| {
-                        let document = quickner.documents_hash.get(id).unwrap();
-                        PyDocument::from(document.to_owned())
-                    })
-                    .collect();
-                documents
-            }
+        let quickner = &self.quickner;
+        let documents = {
+            let documents = documents_ids
+                .iter()
+                .map(|id| {
+                    let document = quickner.documents_hash.get(id).unwrap();
+                    PyDocument::from(document.to_owned())
+                })
+                .collect();
+            documents
         };
@@ -334,5 +335,4 @@ documents
     pub fn find_documents_by_entity(&self, name: &str) -> Vec<PyDocument> {
-        let documents_entities_index = match &self.quickner {
-            quickner => quickner.documents_entities_index.to_owned(),
-        };
+        let quickner = &self.quickner;
+        let documents_entities_index = quickner.documents_entities_index.to_owned();
         let binding = name.to_lowercase();
@@ -344,13 +344,12 @@ let name = binding.as_str();
         };
-        let documents = match &self.quickner {
-            quickner => {
-                let documents = documents_ids
-                    .into_iter()
-                    .map(|id| {
-                        let document = quickner.documents_hash.get(id).unwrap();
-                        PyDocument::from(document.to_owned())
-                    })
-                    .collect();
-                documents
-            }
+        let quickner = &self.quickner;
+        let documents = {
+            let documents = documents_ids
+                .iter()
+                .map(|id| {
+                    let document = quickner.documents_hash.get(id).unwrap();
+                    PyDocument::from(document.to_owned())
+                })
+                .collect();
+            documents
         };
@@ -460,3 +459,3 @@ documents
             },
-            config_path: quickner.config_file,
+            config_path: quickner.config_file.unwrap_or("".to_string()),
             documents: quickner
@@ -463,0 +462,0 @@ .documents

Alert delta unavailable

Currently unable to show alert delta for PyPI packages.

Worsened metrics

Total package byte prevSize

20733702

0