Updated test suite and expanded doctests

LucaCappelletti94 · Apr 11, 2024 · 45a097e · 45a097e
1 parent 57dac7b
commit 45a097e
Show file tree

Hide file tree

Showing 17 changed files with 238 additions and 141 deletions.
diff --git a/src/bi_webgraph.rs b/src/bi_webgraph.rs
@@ -146,7 +146,7 @@ impl WeightedBipartiteGraph for BiWebgraph {
     #[inline(always)]
     /// Returns the number of source nodes.
     ///
-    /// # Example
+    /// # Examples
     /// In this example, we create the trigram corpus associated
     /// to the ANIMALS dataset which we provide within this crate,
     /// and then we convert it to webgraph format. Secondarily,
@@ -171,7 +171,7 @@ impl WeightedBipartiteGraph for BiWebgraph {
     #[inline(always)]
     /// Returns the number of destination nodes.
     ///
-    /// # Example
+    /// # Examples
     /// In this example, we create the trigram corpus associated
     /// to the ANIMALS dataset which we provide within this crate,
     /// and then we convert it to webgraph format. Secondarily,
@@ -196,7 +196,7 @@ impl WeightedBipartiteGraph for BiWebgraph {
     #[inline(always)]
     /// Returns the number of edges.
     ///
-    /// # Example
+    /// # Examples
     /// In this example, we create the trigram corpus associated
     /// to the ANIMALS dataset which we provide within this crate,
     /// and then we convert it to webgraph format. Secondarily,
@@ -224,7 +224,7 @@ impl WeightedBipartiteGraph for BiWebgraph {
     /// # Arguments
     /// * `src_id`: A `usize` which is the source node identifier.
     ///
-    /// # Example
+    /// # Examples
     /// In this example, we create the trigram corpus associated
     /// to the ANIMALS dataset which we provide within this crate,
     /// and then we convert it to webgraph format. Secondarily,
@@ -257,7 +257,7 @@ impl WeightedBipartiteGraph for BiWebgraph {
     /// # Arguments
     /// * `dst_id`: A `usize` which is the destination node identifier.
     ///
-    /// # Example
+    /// # Examples
     /// In this example, we create the trigram corpus associated
     /// to the ANIMALS dataset which we provide within this crate,
     /// and then we convert it to webgraph format. Secondarily,
@@ -292,7 +292,7 @@ impl WeightedBipartiteGraph for BiWebgraph {
     /// # Arguments
     /// * `dst_id`: A `usize` which is the destination node identifier.
     ///
-    /// # Example
+    /// # Examples
     /// In this example, we create the trigram corpus associated
     /// to the ANIMALS dataset which we provide within this crate,
     /// and then we convert it to webgraph format. Secondarily,
@@ -330,7 +330,7 @@ impl WeightedBipartiteGraph for BiWebgraph {
     /// # Arguments
     /// * `src_id`: A `usize` which is the source node identifier.
     ///
-    /// # Example
+    /// # Examples
     /// In this example, we create the trigram corpus associated
     /// to the ANIMALS dataset which we provide within this crate,
     /// and then we convert it to webgraph format. Secondarily,
@@ -371,7 +371,7 @@ impl WeightedBipartiteGraph for BiWebgraph {
     /// # Arguments
     /// * `dst_id`: A `usize` which is the destination node identifier.
     ///
-    /// # Example
+    /// # Examples
     /// In this example, we create the trigram corpus associated
     /// to the ANIMALS dataset which we provide within this crate,
     /// and then we convert it to webgraph format. Secondarily,
@@ -407,7 +407,7 @@ impl WeightedBipartiteGraph for BiWebgraph {
     #[inline(always)]
     /// Returns the weights of the edges.
     ///
-    /// # Example
+    /// # Examples
     /// In this example, we create the trigram corpus associated
     /// to the ANIMALS dataset which we provide within this crate,
     /// and then we convert it to webgraph format. Secondarily,
@@ -442,7 +442,7 @@ impl WeightedBipartiteGraph for BiWebgraph {
     #[inline(always)]
     /// Returns the degrees of the nodes.
     ///
-    /// # Example
+    /// # Examples
     /// In this example, we create the trigram corpus associated
     /// to the ANIMALS dataset which we provide within this crate,
     /// and then we convert it to webgraph format. Secondarily,

diff --git a/src/corpus.rs b/src/corpus.rs
@@ -97,7 +97,7 @@ where
 
     /// Returns a reference to underlying graph.
     ///
-    /// # Example
+    /// # Examples
     ///
     /// ```rust
     /// use ngrammatic::prelude::*;
@@ -127,7 +127,7 @@ where
     #[inline(always)]
     /// Returns the number of keys in the corpus.
     ///
-    /// # Example
+    /// # Examples
     ///
     /// ```rust
     /// use ngrammatic::prelude::*;
@@ -145,7 +145,7 @@ where
     #[inline(always)]
     /// Returns the number of ngrams in the corpus.
     ///
-    /// # Example
+    /// # Examples
     ///
     /// ```rust
     /// use ngrammatic::prelude::*;
@@ -166,16 +166,16 @@ where
     /// # Arguments
     /// * `key_id` - The id of the key to get.
     ///
-    /// # Example
+    /// # Examples
     ///
     /// ```rust
     /// use ngrammatic::prelude::*;
     ///
     /// let animals: Corpus<_, TriGram<ASCIIChar>> = Corpus::from(ANIMALS);
     ///
-    /// assert_eq!(animals.key_from_id(0), "Aardvark");
-    /// assert_eq!(animals.key_from_id(1), "Abyssinian");
-    /// assert_eq!(animals.key_from_id(20), "Alligator");
+    /// assert_eq!(animals.key_from_id(0), &"Aardvark");
+    /// assert_eq!(animals.key_from_id(1), &"Abyssinian");
+    /// assert_eq!(animals.key_from_id(20), &"Alligator");
     /// ```
     pub fn key_from_id(
         &self,
@@ -190,7 +190,7 @@ where
     /// # Arguments
     /// * `ngram_id` - The id of the ngram to get.
     ///
-    /// # Example
+    /// # Examples
     ///
     /// ```rust
     /// use ngrammatic::prelude::*;
@@ -218,7 +218,7 @@ where
     /// # Arguments
     /// * `ngram` - The ngram to get the id from.
     ///
-    /// # Example
+    /// # Examples
     ///
     /// ```rust
     /// use ngrammatic::prelude::*;
@@ -245,7 +245,7 @@ where
     /// # Arguments
     /// * `key_id` - The id of the key to get the number of ngrams from.
     ///
-    /// # Example
+    /// # Examples
     ///
     /// ```rust
     /// use ngrammatic::prelude::*;
@@ -266,7 +266,7 @@ where
     /// # Arguments
     /// * `ngram_id` - The id of the ngram to get the number of keys from.
     ///
-    /// # Example
+    /// # Examples
     ///
     /// ```rust
     /// use ngrammatic::prelude::*;
@@ -287,7 +287,7 @@ where
     /// # Arguments
     /// * `ngram_id` - The id of the ngram to get the key ids from.
     ///
-    /// # Example
+    /// # Examples
     ///
     /// ```rust
     /// use ngrammatic::prelude::*;
@@ -308,7 +308,7 @@ where
     /// # Arguments
     /// * `key_id` - The id of the key to get the ngram ids from.
     ///
-    /// # Example
+    /// # Examples
     ///
     /// ```rust
     /// use ngrammatic::prelude::*;
@@ -329,7 +329,7 @@ where
     /// # Arguments
     /// * `key_id` - The id of the key to get the ngram co-occurrences from.
     ///
-    /// # Example
+    /// # Examples
     /// We check that all values are greater than 0.
     ///
     /// ```rust
@@ -355,7 +355,7 @@ where
     #[inline(always)]
     /// Returns all co-occurrences.
     ///
-    /// # Example
+    /// # Examples
     /// We check that all values are greater than 0.
     ///
     /// ```rust
@@ -376,7 +376,7 @@ where
     /// # Arguments
     /// * `key_id` - The id of the key to get the ngrams and their co-occurrences from.
     ///
-    /// # Example
+    /// # Examples
     ///
     /// ```rust
     /// use ngrammatic::prelude::*;
@@ -401,7 +401,7 @@ where
     /// # Arguments
     /// * `key_id` - The id of the key to get the ngrams and their co-occurrences from.
     ///
-    /// # Example
+    /// # Examples
     /// We check that all of the ngrams returned appear in the corpus and
     /// that all of the co-occurrences are greater than 0.
     ///
@@ -442,7 +442,7 @@ where
     /// # Arguments
     /// * `key_id` - The id of the key to get the ngrams from.
     ///
-    /// # Example
+    /// # Examples
     /// We check that all of the ngrams returned appear in the corpus.
     ///
     /// ```rust
@@ -480,7 +480,7 @@ where
     /// # Returns
     /// An iterator over the keys associated to the ngram.
     ///
-    /// # Example
+    /// # Examples
     /// We check that the keys returned by the keys_from_ngram_id method are the
     /// exactly same keys returned keys_from_ngram method.
     ///
@@ -515,7 +515,7 @@ where
     /// # Arguments
     /// * `ngram` - The ngram to get the number of keys from.
     ///
-    /// # Example
+    /// # Examples
     ///
     /// ```rust
     /// use ngrammatic::prelude::*;
@@ -568,7 +568,7 @@ where
     /// # Returns
     /// An iterator over the keys associated to the ngram.
     ///
-    /// # Example
+    /// # Examples
     ///
     /// ```rust
     /// use ngrammatic::prelude::*;
@@ -606,7 +606,7 @@ where
     /// # Implementative details
     /// This function is implemented using a Binary Heap.
     ///
-    /// # Example
+    /// # Examples
     ///
     /// ```rust
     /// use ngrammatic::prelude::*;

diff --git a/src/corpus_par_from.rs b/src/corpus_par_from.rs
@@ -19,7 +19,7 @@ where
     /// # Arguments
     /// * `keys` - The keys to create the corpus from.
     ///
-    /// # Example
+    /// # Examples
     /// In the following example, we create a corpus from the set of keys
     /// defined by the `ANIMALS` constant array. We provide several synonims
     /// for arrays, such as MonoGrams, BiGrams, TriGrams, and so on. This is
@@ -64,13 +64,13 @@ where
     ///
     /// let animals = vec!["cat", "dog", "bird", "fish", "lion"];
     ///
-    /// let bigram_corpus: Corpus<&[&str], BiGram<char>> = Corpus::par_from(&animals);
-    /// let trigram_corpus: Corpus<&[&str], TriGram<char>> = Corpus::par_from(&animals);
-    /// let tetragram_corpus: Corpus<&[&str], TetraGram<char>> = Corpus::par_from(&animals);
-    /// let pentagram_corpus: Corpus<&[&str], PentaGram<char>> = Corpus::par_from(&animals);
-    /// let hexagram_corpus: Corpus<&[&str], HexaGram<char>> = Corpus::par_from(&animals);
-    /// let heptagram_corpus: Corpus<&[&str], HeptaGram<char>> = Corpus::par_from(&animals);
-    /// let octagram_corpus: Corpus<&[&str], OctaGram<char>> = Corpus::par_from(&animals);
+    /// let bigram_corpus: Corpus<Vec<&str>, BiGram<char>> = Corpus::par_from(animals.clone());
+    /// let trigram_corpus: Corpus<Vec<&str>, TriGram<char>> = Corpus::par_from(animals.clone());
+    /// let tetragram_corpus: Corpus<Vec<&str>, TetraGram<char>> = Corpus::par_from(animals.clone());
+    /// let pentagram_corpus: Corpus<Vec<&str>, PentaGram<char>> = Corpus::par_from(animals.clone());
+    /// let hexagram_corpus: Corpus<Vec<&str>, HexaGram<char>> = Corpus::par_from(animals.clone());
+    /// let heptagram_corpus: Corpus<Vec<&str>, HeptaGram<char>> = Corpus::par_from(animals.clone());
+    /// let octagram_corpus: Corpus<Vec<&str>, OctaGram<char>> = Corpus::par_from(animals.clone());
     /// ```
     ///
     /// And references of arrays:
@@ -80,13 +80,13 @@ where
     ///
     /// let animals = ["cat", "dog", "bird", "fish", "lion"];
     ///
-    /// let bigram_corpus: Corpus<&[&str; 5], BiGram<char>> = Corpus::par_from(&animals);
-    /// let trigram_corpus: Corpus<&[&str; 5], TriGram<char>> = Corpus::par_from(&animals);
-    /// let tetragram_corpus: Corpus<&[&str; 5], TetraGram<char>> = Corpus::par_from(&animals);
-    /// let pentagram_corpus: Corpus<&[&str; 5], PentaGram<char>> = Corpus::par_from(&animals);
-    /// let hexagram_corpus: Corpus<&[&str; 5], HexaGram<char>> = Corpus::par_from(&animals);
-    /// let heptagram_corpus: Corpus<&[&str; 5], HeptaGram<char>> = Corpus::par_from(&animals);
-    /// let octagram_corpus: Corpus<&[&str; 5], OctaGram<char>> = Corpus::par_from(&animals);
+    /// let bigram_corpus: Corpus<[&str; 5], BiGram<char>> = Corpus::par_from(animals);
+    /// let trigram_corpus: Corpus<[&str; 5], TriGram<char>> = Corpus::par_from(animals);
+    /// let tetragram_corpus: Corpus<[&str; 5], TetraGram<char>> = Corpus::par_from(animals);
+    /// let pentagram_corpus: Corpus<[&str; 5], PentaGram<char>> = Corpus::par_from(animals);
+    /// let hexagram_corpus: Corpus<[&str; 5], HexaGram<char>> = Corpus::par_from(animals);
+    /// let heptagram_corpus: Corpus<[&str; 5], HeptaGram<char>> = Corpus::par_from(animals);
+    /// let octagram_corpus: Corpus<[&str; 5], OctaGram<char>> = Corpus::par_from(animals);
     /// ```
     ///
     /// In all of these examples, we have used char-based grams. We can also use u8-based grams:
@@ -96,13 +96,13 @@ where
     ///
     /// let animals = vec!["cat", "dog", "bird", "fish", "lion"];
     ///
-    /// let bigram_corpus: Corpus<&[&str], BiGram<u8>> = Corpus::par_from(&animals);
-    /// let trigram_corpus: Corpus<&[&str], TriGram<u8>> = Corpus::par_from(&animals);
-    /// let tetragram_corpus: Corpus<&[&str], TetraGram<u8>> = Corpus::par_from(&animals);
-    /// let pentagram_corpus: Corpus<&[&str], PentaGram<u8>> = Corpus::par_from(&animals);
-    /// let hexagram_corpus: Corpus<&[&str], HexaGram<u8>> = Corpus::par_from(&animals);
-    /// let heptagram_corpus: Corpus<&[&str], HeptaGram<u8>> = Corpus::par_from(&animals);
-    /// let octagram_corpus: Corpus<&[&str], OctaGram<u8>> = Corpus::par_from(&animals);
+    /// let bigram_corpus: Corpus<Vec<&str>, BiGram<u8>> = Corpus::par_from(animals.clone());
+    /// let trigram_corpus: Corpus<Vec<&str>, TriGram<u8>> = Corpus::par_from(animals.clone());
+    /// let tetragram_corpus: Corpus<Vec<&str>, TetraGram<u8>> = Corpus::par_from(animals.clone());
+    /// let pentagram_corpus: Corpus<Vec<&str>, PentaGram<u8>> = Corpus::par_from(animals.clone());
+    /// let hexagram_corpus: Corpus<Vec<&str>, HexaGram<u8>> = Corpus::par_from(animals.clone());
+    /// let heptagram_corpus: Corpus<Vec<&str>, HeptaGram<u8>> = Corpus::par_from(animals.clone());
+    /// let octagram_corpus: Corpus<Vec<&str>, OctaGram<u8>> = Corpus::par_from(animals.clone());
     /// ```
     ///
     /// It is also pretty easy to define normalizations for the keys. For instance, you can
@@ -113,18 +113,18 @@ where
     ///
     /// let animals = vec!["cat", "dog", "bIrd", "Fish", "Lion"];
     ///
-    /// let bigram_corpus: Corpus<&[&str], BiGram<char>, Lowercase<str>> = Corpus::par_from(&animals);
-    /// let trigram_corpus: Corpus<&[&str], TriGram<char>, Lowercase<str>> = Corpus::par_from(&animals);
-    /// let tetragram_corpus: Corpus<&[&str], TetraGram<char>, Lowercase<str>> =
-    ///     Corpus::par_from(&animals);
-    /// let pentagram_corpus: Corpus<&[&str], PentaGram<char>, Lowercase<str>> =
-    ///     Corpus::par_from(&animals);
-    /// let hexagram_corpus: Corpus<&[&str], HexaGram<char>, Lowercase<str>> =
-    ///     Corpus::par_from(&animals);
-    /// let heptagram_corpus: Corpus<&[&str], HeptaGram<char>, Lowercase<str>> =
-    ///     Corpus::par_from(&animals);
-    /// let octagram_corpus: Corpus<&[&str], OctaGram<char>, Lowercase<str>> =
-    ///     Corpus::par_from(&animals);
+    /// let bigram_corpus: Corpus<Vec<&str>, BiGram<char>, Lowercase<str>> = Corpus::par_from(animals.clone());
+    /// let trigram_corpus: Corpus<Vec<&str>, TriGram<char>, Lowercase<str>> = Corpus::par_from(animals.clone());
+    /// let tetragram_corpus: Corpus<Vec<&str>, TetraGram<char>, Lowercase<str>> =
+    ///     Corpus::par_from(animals.clone());
+    /// let pentagram_corpus: Corpus<Vec<&str>, PentaGram<char>, Lowercase<str>> =
+    ///     Corpus::par_from(animals.clone());
+    /// let hexagram_corpus: Corpus<Vec<&str>, HexaGram<char>, Lowercase<str>> =
+    ///     Corpus::par_from(animals.clone());
+    /// let heptagram_corpus: Corpus<Vec<&str>, HeptaGram<char>, Lowercase<str>> =
+    ///     Corpus::par_from(animals.clone());
+    /// let octagram_corpus: Corpus<Vec<&str>, OctaGram<char>, Lowercase<str>> =
+    ///     Corpus::par_from(animals.clone());
     /// ```
     pub fn par_from(keys: KS) -> Self {
         // We start by parsing the keys to extract the ngrams, the cooccurrences, the key offsets,

diff --git a/src/lib.rs b/src/lib.rs
@@ -42,4 +42,5 @@ pub mod prelude {
     pub use crate::bi_webgraph::*;
     pub use crate::ngram_search::*;
     pub use crate::tfidf::*;
+    pub use crate::search::*;
 }