Fixing bad deserialization following inclusion of a default for Punctuation. (#884)

* Fixing bad deserialization following inclusion of a default for
`Punctuation`.

* don't remove the type now...

* Adding slow test to run on all the tokenizers of the hub.

* `PartialEq` everywhere.

* Forcing `type` to exist on the `pre_tokenizers`.

bindings/node/native/Cargo.lock

@@ -1668,7 +1668,7 @@ checksum = "cda74da7e1a664f795bb1f8a87ec406fb89a02522cf6e50620d016add6dbbf5c"
 
 [[package]]
 name = "tokenizers"
-version = "0.11.0"
+version = "0.11.1"
 dependencies = [
  "aho-corasick",
  "cached-path",

bindings/python/tests/test_serialization.py

@@ -1,5 +1,10 @@
-from tokenizers import Tokenizer, models, normalizers
+from tokenizers import Tokenizer
+import os
+import unittest
 from .utils import data_dir, albert_base
+import json
+from huggingface_hub import HfApi, hf_hub_url, cached_download
+import tqdm
 
 
 class TestSerialization:
@@ -8,3 +13,70 @@ class TestSerialization:
         # This used to fail because of BufReader that would fail because the
         # file exceeds the buffer capacity
         tokenizer = Tokenizer.from_file(albert_base)
+
+
+def check(tokenizer_file) -> bool:
+    with open(tokenizer_file, "r") as f:
+        data = json.load(f)
+    if "pre_tokenizer" not in data:
+        return True
+    if "type" not in data["pre_tokenizer"]:
+        return False
+    if data["pre_tokenizer"]["type"] == "Sequence":
+        for pre_tok in data["pre_tokenizer"]["pretokenizers"]:
+            if "type" not in pre_tok:
+                return False
+    return True
+
+
+def slow(test_case):
+    """
+    Decorator marking a test as slow.
+
+    Slow tests are skipped by default. Set the RUN_SLOW environment variable to a truthy value to run them.
+
+    """
+    if os.getenv("RUN_SLOW") != "1":
+        return unittest.skip("use `RUN_SLOW=1` to run")(test_case)
+    else:
+        return test_case
+
+
+@slow
+class TestFullDeserialization(unittest.TestCase):
+    def test_full_deserialization_hub(self):
+        # Check we can read this file.
+        # This used to fail because of BufReader that would fail because the
+        # file exceeds the buffer capacity
+        api = HfApi()
+
+        not_loadable = []
+        invalid_pre_tokenizer = []
+
+        # models = api.list_models(filter="transformers")
+        # for model in tqdm.tqdm(models):
+        #     model_id = model.modelId
+        #     for model_file in model.siblings:
+        #         filename = model_file.rfilename
+        #         if filename == "tokenizer.json":
+        #             all_models.append((model_id, filename))
+
+        all_models = [("HueyNemud/das22-10-camembert_pretrained", "tokenizer.json")]
+        for (model_id, filename) in tqdm.tqdm(all_models):
+            tokenizer_file = cached_download(hf_hub_url(model_id, filename=filename))
+
+            is_ok = check(tokenizer_file)
+            if not is_ok:
+                print(f"{model_id} is affected by no type")
+                invalid_pre_tokenizer.append(model_id)
+            try:
+                Tokenizer.from_file(tokenizer_file)
+            except Exception as e:
+                print(f"{model_id} is not loadable: {e}")
+                not_loadable.append(model_id)
+            except:
+                print(f"{model_id} is not loadable: Rust error")
+                not_loadable.append(model_id)
+
+            self.assertEqual(invalid_pre_tokenizer, [])
+            self.assertEqual(not_loadable, [])

tokenizers/src/pre_tokenizers/bert.rs

@@ -5,7 +5,7 @@ fn is_bert_punc(x: char) -> bool {
     char::is_ascii_punctuation(&x) || x.is_punctuation()
 }
 
-#[derive(Copy, Clone, Debug)]
+#[derive(Copy, Clone, Debug, PartialEq)]
 pub struct BertPreTokenizer;
 impl_serde_unit_struct!(BertVisitor, BertPreTokenizer);
 

tokenizers/src/pre_tokenizers/byte_level.rs

@@ -1,7 +1,7 @@
 use std::collections::{HashMap, HashSet};
 
 use onig::Regex;
-use serde::{Deserialize, Serialize};
+use serde::{Deserialize, Deserializer, Serialize};
 
 use crate::tokenizer::{
     Decoder, Encoding, PostProcessor, PreTokenizedString, PreTokenizer, Result,
@@ -40,7 +40,7 @@ lazy_static! {
         bytes_char().into_iter().map(|(c, b)| (b, c)).collect();
 }
 
-#[derive(Deserialize, Serialize, Copy, Clone, Debug, PartialEq)]
+#[derive(Serialize, Copy, Clone, Debug, PartialEq)]
 /// Provides all the necessary steps to handle the BPE tokenization at the byte-level. Takes care
 /// of all the required processing steps to transform a UTF-8 string as needed before and after the
 /// BPE model does its job.
@@ -53,6 +53,30 @@ pub struct ByteLevel {
     /// Whether the post processing step should trim offsets to avoid including whitespaces.
     pub trim_offsets: bool,
 }
+
+impl<'de> Deserialize<'de> for ByteLevel {
+    fn deserialize<D>(deserializer: D) -> std::result::Result<Self, D::Error>
+    where
+        D: Deserializer<'de>,
+    {
+        #[derive(Deserialize)]
+        enum Type {
+            ByteLevel,
+        }
+
+        #[derive(Deserialize)]
+        pub struct ByteLevelHelper {
+            #[serde(rename = "type")]
+            _type: Type,
+            add_prefix_space: bool,
+            trim_offsets: bool,
+        }
+
+        let helper = ByteLevelHelper::deserialize(deserializer)?;
+        Ok(ByteLevel::new(helper.add_prefix_space, helper.trim_offsets))
+    }
+}
+
 impl Default for ByteLevel {
     fn default() -> Self {
         Self {

tokenizers/src/pre_tokenizers/delimiter.rs

@@ -1,14 +1,36 @@
-use serde::{Deserialize, Serialize};
+use serde::{Deserialize, Deserializer, Serialize};
 
 use crate::tokenizer::{PreTokenizedString, PreTokenizer, Result, SplitDelimiterBehavior};
 
-#[derive(Copy, Clone, Debug, Deserialize, Serialize)]
+#[derive(Copy, Clone, Debug, Serialize, PartialEq)]
 #[serde(tag = "type")]
 #[non_exhaustive]
 pub struct CharDelimiterSplit {
     pub delimiter: char,
 }
 
+impl<'de> Deserialize<'de> for CharDelimiterSplit {
+    fn deserialize<D>(deserializer: D) -> std::result::Result<Self, D::Error>
+    where
+        D: Deserializer<'de>,
+    {
+        #[derive(Deserialize)]
+        enum Type {
+            CharDelimiterSplit,
+        }
+
+        #[derive(Deserialize)]
+        pub struct CharDelimiterSplitHelper {
+            #[serde(rename = "type")]
+            _type: Type,
+            delimiter: char,
+        }
+
+        let helper = CharDelimiterSplitHelper::deserialize(deserializer)?;
+        Ok(CharDelimiterSplit::new(helper.delimiter))
+    }
+}
+
 impl CharDelimiterSplit {
     pub fn new(delimiter: char) -> Self {
         CharDelimiterSplit { delimiter }

tokenizers/src/pre_tokenizers/digits.rs

@@ -1,8 +1,8 @@
-use serde::{Deserialize, Serialize};
+use serde::{Deserialize, Deserializer, Serialize};
 
 use crate::tokenizer::{PreTokenizedString, PreTokenizer, Result, SplitDelimiterBehavior};
 
-#[derive(Serialize, Deserialize, Clone, Debug)]
+#[derive(Serialize, Clone, Debug, PartialEq)]
 /// Pre tokenizes the numbers into single tokens. If individual_digits is set
 /// to true, then all digits are splitted into individual tokens.
 #[serde(tag = "type")]
@@ -11,6 +11,28 @@ pub struct Digits {
     pub individual_digits: bool,
 }
 
+impl<'de> Deserialize<'de> for Digits {
+    fn deserialize<D>(deserializer: D) -> std::result::Result<Self, D::Error>
+    where
+        D: Deserializer<'de>,
+    {
+        #[derive(Deserialize)]
+        enum Type {
+            Digits,
+        }
+
+        #[derive(Deserialize)]
+        pub struct DigitsHelper {
+            #[serde(rename = "type")]
+            _type: Type,
+            individual_digits: bool,
+        }
+
+        let helper = DigitsHelper::deserialize(deserializer)?;
+        Ok(Digits::new(helper.individual_digits))
+    }
+}
+
 impl Digits {
     pub fn new(individual_digits: bool) -> Self {
         Self { individual_digits }

tokenizers/src/pre_tokenizers/metaspace.rs

@@ -1,11 +1,11 @@
-use serde::{Deserialize, Serialize};
+use serde::{Deserialize, Deserializer, Serialize};
 
 use crate::tokenizer::{Decoder, PreTokenizedString, PreTokenizer, Result, SplitDelimiterBehavior};
 
-#[derive(Debug, Clone, PartialEq, Serialize, Deserialize)]
+#[derive(Debug, Clone, PartialEq, Serialize)]
 /// Replaces all the whitespaces by the provided meta character and then
 /// splits on this character
-#[serde(tag = "type", from = "MetaspaceDeserializer")]
+#[serde(tag = "type")]
 pub struct Metaspace {
     replacement: char,
     pub add_prefix_space: bool,
@@ -13,17 +13,28 @@ pub struct Metaspace {
     str_rep: String,
 }
 
-#[doc(hidden)]
-#[derive(Deserialize)]
-#[serde(tag = "type")]
-pub struct MetaspaceDeserializer {
-    replacement: char,
-    add_prefix_space: bool,
-}
+impl<'de> Deserialize<'de> for Metaspace {
+    fn deserialize<D>(deserializer: D) -> std::result::Result<Self, D::Error>
+    where
+        D: Deserializer<'de>,
+    {
+        #[derive(Deserialize)]
+        enum Type {
+            Metaspace,
+        }
 
-impl From<MetaspaceDeserializer> for Metaspace {
-    fn from(v: MetaspaceDeserializer) -> Metaspace {
-        Metaspace::new(v.replacement, v.add_prefix_space)
+        #[derive(Deserialize)]
+        pub struct MetaspaceHelper {
+            #[serde(rename = "type")]
+            _type: Type,
+            replacement: char,
+            pub add_prefix_space: bool,
+            #[serde(skip, rename = "str_rep")]
+            _str_rep: String,
+        }
+
+        let helper = MetaspaceHelper::deserialize(deserializer)?;
+        Ok(Metaspace::new(helper.replacement, helper.add_prefix_space))
     }
 }
 
@@ -109,6 +120,12 @@ mod tests {
             serde_json::from_str::<Metaspace>(metaspace_s).unwrap(),
             metaspace
         );
+
+        let metaspace_parsed: Metaspace = serde_json::from_str(
+            r#"{"type":"Metaspace","replacement":"_","add_prefix_space":true}"#,
+        )
+        .unwrap();
+        assert_eq!(metaspace_parsed, metaspace);
     }
 
     #[test]

tokenizers/src/pre_tokenizers/mod.rs

@@ -23,7 +23,7 @@ use crate::pre_tokenizers::unicode_scripts::UnicodeScripts;
 use crate::pre_tokenizers::whitespace::{Whitespace, WhitespaceSplit};
 use crate::{PreTokenizedString, PreTokenizer};
 
-#[derive(Serialize, Deserialize, Clone, Debug)]
+#[derive(Deserialize, Serialize, Clone, Debug, PartialEq)]
 #[serde(untagged)]
 pub enum PreTokenizerWrapper {
     BertPreTokenizer(BertPreTokenizer),
@@ -68,3 +68,51 @@ impl_enum_from!(Metaspace, PreTokenizerWrapper, Metaspace);
 impl_enum_from!(WhitespaceSplit, PreTokenizerWrapper, WhitespaceSplit);
 impl_enum_from!(Digits, PreTokenizerWrapper, Digits);
 impl_enum_from!(UnicodeScripts, PreTokenizerWrapper, UnicodeScripts);
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[test]
+    fn test_deserialize() {
+        let pre_tokenizer: PreTokenizerWrapper = serde_json::from_str(r#"{"type":"Sequence","pretokenizers":[{"type":"WhitespaceSplit"},{"type":"Metaspace","replacement":"▁","str_rep":"▁","add_prefix_space":true}]}"#).unwrap();
+
+        assert_eq!(
+            pre_tokenizer,
+            PreTokenizerWrapper::Sequence(Sequence::new(vec![
+                PreTokenizerWrapper::WhitespaceSplit(WhitespaceSplit {}),
+                PreTokenizerWrapper::Metaspace(Metaspace::new('▁', true))
+            ]))
+        );
+
+        let pre_tokenizer: PreTokenizerWrapper = serde_json::from_str(
+            r#"{"type":"Metaspace","replacement":"▁","add_prefix_space":true}"#,
+        )
+        .unwrap();
+
+        assert_eq!(
+            pre_tokenizer,
+            PreTokenizerWrapper::Metaspace(Metaspace::new('▁', true))
+        );
+
+        let pre_tokenizer: PreTokenizerWrapper = serde_json::from_str(r#"{"type":"Sequence","pretokenizers":[{"type":"WhitespaceSplit"},{"type":"Metaspace","replacement":"▁","add_prefix_space":true}]}"#).unwrap();
+
+        assert_eq!(
+            pre_tokenizer,
+            PreTokenizerWrapper::Sequence(Sequence::new(vec![
+                PreTokenizerWrapper::WhitespaceSplit(WhitespaceSplit {}),
+                PreTokenizerWrapper::Metaspace(Metaspace::new('▁', true))
+            ]))
+        );
+    }
+
+    #[test]
+    fn test_deserialize_whitespace_split() {
+        let pre_tokenizer: PreTokenizerWrapper =
+            serde_json::from_str(r#"{"type":"WhitespaceSplit"}"#).unwrap();
+        assert_eq!(
+            pre_tokenizer,
+            PreTokenizerWrapper::WhitespaceSplit(WhitespaceSplit {})
+        );
+    }
+}

tokenizers/src/pre_tokenizers/punctuation.rs

@@ -1,4 +1,4 @@
-use serde::{Deserialize, Serialize};
+use serde::{Deserialize, Deserializer, Serialize};
 
 use crate::tokenizer::{PreTokenizedString, PreTokenizer, Result, SplitDelimiterBehavior};
 use unicode_categories::UnicodeCategories;
@@ -7,13 +7,35 @@ fn is_punc(x: char) -> bool {
     char::is_ascii_punctuation(&x) || x.is_punctuation()
 }
 
-#[derive(Serialize, Deserialize, Copy, Clone, Debug)]
+#[derive(Serialize, Copy, Clone, Debug, PartialEq)]
 #[serde(tag = "type")]
 pub struct Punctuation {
-    #[serde(default = "default_split")]
     behavior: SplitDelimiterBehavior,
 }
 
+impl<'de> Deserialize<'de> for Punctuation {
+    fn deserialize<D>(deserializer: D) -> std::result::Result<Self, D::Error>
+    where
+        D: Deserializer<'de>,
+    {
+        #[derive(Deserialize)]
+        enum Type {
+            Punctuation,
+        }
+
+        #[derive(Deserialize)]
+        pub struct PunctuationHelper {
+            #[serde(rename = "type")]
+            _type: Type,
+            #[serde(default = "default_split")]
+            behavior: SplitDelimiterBehavior,
+        }
+
+        let helper = PunctuationHelper::deserialize(deserializer)?;
+        Ok(Punctuation::new(helper.behavior))
+    }
+}
+
 fn default_split() -> SplitDelimiterBehavior {
     SplitDelimiterBehavior::Isolated
 }
@@ -65,6 +87,18 @@ mod tests {
 
     #[test]
     fn deserialization() {
-        let _punctuation: Punctuation = serde_json::from_str(r#"{"type": "punctuation"}"#).unwrap();
+        let punctuation: Punctuation = serde_json::from_str(r#"{"type": "Punctuation"}"#).unwrap();
+        assert_eq!(punctuation, Punctuation::default());
+        assert_eq!(
+            punctuation,
+            Punctuation::new(SplitDelimiterBehavior::Isolated)
+        );
+    }
+
+    #[test]
+    #[should_panic]
+    fn deserialization_erroneous() {
+        let _punctuation: Punctuation =
+            serde_json::from_str(r#"{"type": "WhitespaceSplit"}"#).unwrap();
     }
 }

tokenizers/src/pre_tokenizers/sequence.rs

@@ -1,13 +1,35 @@
 use crate::pre_tokenizers::PreTokenizerWrapper;
 use crate::tokenizer::{PreTokenizedString, PreTokenizer, Result};
-use serde::{Deserialize, Serialize};
+use serde::{Deserialize, Deserializer, Serialize};
 
-#[derive(Clone, Debug, Serialize, Deserialize)]
+#[derive(Clone, Debug, Serialize, PartialEq)]
 #[serde(tag = "type")]
 pub struct Sequence {
     pretokenizers: Vec<PreTokenizerWrapper>,
 }
 
+impl<'de> Deserialize<'de> for Sequence {
+    fn deserialize<D>(deserializer: D) -> std::result::Result<Self, D::Error>
+    where
+        D: Deserializer<'de>,
+    {
+        #[derive(Deserialize)]
+        enum Type {
+            Sequence,
+        }
+
+        #[derive(Deserialize)]
+        pub struct SequenceHelper {
+            #[serde(rename = "type")]
+            _type: Type,
+            pretokenizers: Vec<PreTokenizerWrapper>,
+        }
+
+        let helper = SequenceHelper::deserialize(deserializer)?;
+        Ok(Sequence::new(helper.pretokenizers))
+    }
+}
+
 impl Sequence {
     pub fn new(pretokenizers: Vec<PreTokenizerWrapper>) -> Self {
         Self { pretokenizers }

tokenizers/src/pre_tokenizers/split.rs

@@ -1,5 +1,5 @@
 use onig::Regex;
-use serde::{Deserialize, Serialize};
+use serde::{Deserialize, Deserializer, Serialize};
 
 use crate::tokenizer::{
     pattern::Invert, PreTokenizedString, PreTokenizer, Result, SplitDelimiterBehavior,
@@ -24,26 +24,8 @@ impl From<&str> for SplitPattern {
     }
 }
 
-/// We use this custom deserializer to provide the value for `regex` for `Split`
-#[doc(hidden)]
-#[derive(Deserialize)]
+#[derive(Debug, Serialize)]
 #[serde(tag = "type")]
-struct SplitDeserializer {
-    pattern: SplitPattern,
-    behavior: SplitDelimiterBehavior,
-    invert: bool,
-}
-
-impl std::convert::TryFrom<SplitDeserializer> for Split {
-    type Error = Box<dyn std::error::Error + Send + Sync>;
-
-    fn try_from(v: SplitDeserializer) -> Result<Self> {
-        Split::new(v.pattern, v.behavior, v.invert)
-    }
-}
-
-#[derive(Debug, Serialize, Deserialize)]
-#[serde(tag = "type", try_from = "SplitDeserializer")]
 pub struct Split {
     pattern: SplitPattern,
     #[serde(skip)]
@@ -52,6 +34,30 @@ pub struct Split {
     invert: bool,
 }
 
+impl<'de> Deserialize<'de> for Split {
+    fn deserialize<D>(deserializer: D) -> std::result::Result<Self, D::Error>
+    where
+        D: Deserializer<'de>,
+    {
+        #[derive(Deserialize)]
+        enum Type {
+            Split,
+        }
+
+        #[derive(Deserialize)]
+        pub struct SplitHelper {
+            #[serde(rename = "type")]
+            _type: Type,
+            pattern: SplitPattern,
+            behavior: SplitDelimiterBehavior,
+            invert: bool,
+        }
+
+        let helper = SplitHelper::deserialize(deserializer)?;
+        Split::new(helper.pattern, helper.behavior, helper.invert).map_err(serde::de::Error::custom)
+    }
+}
+
 impl Clone for Split {
     fn clone(&self) -> Self {
         Split::new(self.pattern.clone(), self.behavior, self.invert).unwrap()

tokenizers/src/pre_tokenizers/unicode_scripts/pre_tokenizer.rs

@@ -1,7 +1,7 @@
 use crate::pre_tokenizers::unicode_scripts::scripts::{get_script, Script};
 use crate::tokenizer::{normalizer::Range, PreTokenizedString, PreTokenizer, Result};
 
-#[derive(Clone, Debug)]
+#[derive(Clone, Debug, PartialEq)]
 pub struct UnicodeScripts;
 impl_serde_unit_struct!(UnicodeScriptsVisitor, UnicodeScripts);
 

tokenizers/src/pre_tokenizers/whitespace.rs

@@ -4,7 +4,7 @@ use crate::tokenizer::{
     pattern::Invert, PreTokenizedString, PreTokenizer, Result, SplitDelimiterBehavior,
 };
 
-#[derive(Clone, Debug)]
+#[derive(Clone, Debug, PartialEq)]
 pub struct Whitespace;
 impl_serde_unit_struct!(WhitespaceVisitor, Whitespace);
 
@@ -27,7 +27,7 @@ impl PreTokenizer for Whitespace {
     }
 }
 
-#[derive(Copy, Clone, Debug)]
+#[derive(Copy, Clone, Debug, PartialEq)]
 pub struct WhitespaceSplit;
 impl_serde_unit_struct!(WhitespaceSplitVisitor, WhitespaceSplit);
 

tokenizers/src/tokenizer/mod.rs

@@ -403,7 +403,8 @@ impl Tokenizer {
     }
     pub fn from_file<P: AsRef<Path>>(file: P) -> Result<Self> {
         let content = read_to_string(file)?;
-        Ok(serde_json::from_str(&content)?)
+        let tokenizer = serde_json::from_str(&content)?;
+        Ok(tokenizer)
     }
     #[cfg(feature = "http")]
     pub fn from_pretrained<S: AsRef<str>>(
@@ -1131,7 +1132,8 @@ where
     /// Instantiate a new Tokenizer from the given file
     pub fn from_file<P: AsRef<Path>>(file: P) -> Result<Self> {
         let content = read_to_string(file)?;
-        Ok(serde_json::from_str(&content)?)
+        let tokenizer = serde_json::from_str(&content)?;
+        Ok(tokenizer)
     }
 }