!2580 BasicTokenizer do not case fold on preserved words

Merge pull request !2580 from qianlong21st/fix_basic_tokenizer

mindspore/ccsrc/dataset/text/kernels/basic_tokenizer_op.cc

@@ -15,11 +15,16 @@
  */
 #include "dataset/text/kernels/basic_tokenizer_op.h"
 #include <memory>
+#include <queue>
 #include <string>
 #include <string_view>
 #include <utility>
 #include <vector>
 
+#include "unicode/errorcode.h"
+#include "unicode/normalizer2.h"
+#include "unicode/utypes.h"
+
 namespace mindspore {
 namespace dataset {
 const bool BasicTokenizerOp::kDefLowerCase = false;
@@ -40,8 +45,8 @@ const char BasicTokenizerOp::kCommonPattern[] =
   "|[\\x{2B820}-\\x{2CEAF}]"
   "|[\\x{F900}-\\x{FAFF}]"
   "|[\\x{2F800}-\\x{2FA1F}]";
-const char BasicTokenizerOp::kUnusedPattern[] = "\\[CLS\\]|\\[SEP\\]|\\[UNK\\]|\\[PAD\\]|\\[MASK\\]|";
-
+const char BasicTokenizerOp::kUnusedPattern[] = "\\[CLS\\]|\\[SEP\\]|\\[UNK\\]|\\[PAD\\]|\\[MASK\\]|\\[unused\\d+\\]|";
+const std::unordered_set<std::string> BasicTokenizerOp::kUnusedWords{"[CLS]", "[SEP]", "[UNK]", "[PAD]", "[MASK]"};
 BasicTokenizerOp::BasicTokenizerOp(bool lower_case, bool keep_whitespace, NormalizeForm normalization_form,
                                    bool preserve_unused_token)
     : lower_case_(lower_case),
@@ -67,6 +72,69 @@ BasicTokenizerOp::BasicTokenizerOp(bool lower_case, bool keep_whitespace, Normal
   regex_tokenizer_ = std::make_unique<RegexTokenizerOp>(delim_pattern, keep_delim_pattern);
 }
 
+Status BasicTokenizerOp::CaseFoldWithoutUnusedWords(const std::string_view &text,
+                                                    const std::unordered_set<std::string> &unused_words,
+                                                    std::string *outupt) {
+  icu::ErrorCode error;
+  const icu::Normalizer2 *nfkc_case_fold = icu::Normalizer2::getNFKCCasefoldInstance(error);
+  CHECK_FAIL_RETURN_UNEXPECTED(error.isSuccess(), "getNFKCCasefoldInstance failed.");
+  outupt->clear();
+
+  // 1. get start and end offsets of not case fold strs
+  std::queue<std::pair<int, int>> offsets;  // offsets of not used words
+  int start = -1;
+  int len = 0;
+  for (int i = 0; i < text.length(); i++) {
+    if (text[i] == '[') {
+      start = i;
+      ++len;
+    } else if (text[i] == ']' && start >= 0) {
+      ++len;
+      std::string word(text.substr(start, len));
+      if (unused_words.find(word) != unused_words.end()) {
+        offsets.push(std::make_pair(start, start + len - 1));
+      }
+      start = -1;
+      len = 0;
+    } else if (start >= 0) {
+      ++len;
+    }
+  }
+
+  // 2. Do not apply case fold on `unused_words`
+  start = 0;
+  for (int i = 0; i < text.length();) {
+    std::string_view process_text;
+    std::string preserve_token;
+    if (offsets.empty()) {
+      i = text.length();
+      process_text = text.substr(start, i - start);
+    } else {
+      preserve_token = text.substr(offsets.front().first, offsets.front().second - offsets.front().first + 1);
+      process_text = text.substr(start, offsets.front().first - start);
+      i = offsets.front().second + 1;
+      offsets.pop();
+    }
+    std::string temp;
+    icu::StringByteSink<std::string> sink(&temp);
+    nfkc_case_fold->normalizeUTF8(0, icu::StringPiece(process_text.data(), process_text.size()), sink, nullptr, error);
+    *outupt += temp + preserve_token;
+  }
+  return Status::OK();
+}
+
+Status BasicTokenizerOp::CaseFoldWithoutUnusedWords(const std::shared_ptr<Tensor> &input,
+                                                    std::shared_ptr<Tensor> *output) {
+  IO_CHECK(input, output);
+  std::vector<std::string> strs(input->Size());
+  int i = 0;
+  for (auto iter = input->begin<std::string_view>(); iter != input->end<std::string_view>(); iter++) {
+    RETURN_IF_NOT_OK(CaseFoldWithoutUnusedWords(*iter, kUnusedWords, &strs[i++]));
+  }
+  *output = std::make_shared<Tensor>(std::move(strs), input->shape());
+  return Status::OK();
+}
+
 Status BasicTokenizerOp::Compute(const std::shared_ptr<Tensor> &input, std::shared_ptr<Tensor> *output) {
   IO_CHECK(input, output);
   if (input->Rank() != 0 || input->type() != DataType::DE_STRING) {
@@ -75,8 +143,13 @@ Status BasicTokenizerOp::Compute(const std::shared_ptr<Tensor> &input, std::shar
   std::shared_ptr<Tensor> cur_input;
   std::shared_ptr<Tensor> processed_tensor;
   if (lower_case_) {
-    // to lower case
-    RETURN_IF_NOT_OK(case_fold_->Compute(input, &processed_tensor));
+    if (!preserve_unused_token_) {
+      // to lower case
+      RETURN_IF_NOT_OK(case_fold_->Compute(input, &processed_tensor));
+    } else {
+      // to lower case except words in kUnusedWords
+      RETURN_IF_NOT_OK(CaseFoldWithoutUnusedWords(input, &processed_tensor));
+    }
     cur_input = processed_tensor;
     // strip accent characters
     RETURN_IF_NOT_OK(nfd_normalize_->Compute(cur_input, &processed_tensor));

mindspore/ccsrc/dataset/text/kernels/basic_tokenizer_op.h

@@ -17,6 +17,7 @@
 #define DATASET_TEXT_KERNELS_BASIC_TOKENIZER_OP_H_
 #include <memory>
 #include <string>
+#include <unordered_set>
 
 #include "dataset/core/tensor.h"
 #include "dataset/kernels/tensor_op.h"
@@ -45,9 +46,15 @@ class BasicTokenizerOp : public TensorOp {
 
   Status Compute(const std::shared_ptr<Tensor> &input, std::shared_ptr<Tensor> *output) override;
 
+ protected:
+  Status CaseFoldWithoutUnusedWords(const std::string_view &text, const std::unordered_set<std::string> &unused_words,
+                                    std::string *outupt);
+  Status CaseFoldWithoutUnusedWords(const std::shared_ptr<Tensor> &input, std::shared_ptr<Tensor> *output);
+
  private:
   static const char kCommonPattern[];
   static const char kUnusedPattern[];
+  static const std::unordered_set<std::string> kUnusedWords;
   bool lower_case_;
   bool keep_whitespace_;
   NormalizeForm normalization_form_;

tests/ut/data/dataset/testTokenizerData/bert_tokenizer.txt

@@ -10,5 +10,6 @@ unused [SEP]
 unused [UNK]
 unused [PAD]
 unused [MASK]
-12+/-28=40/-16
-Hello World!
\ No newline at end of file
+[unused1]
+[unused10]
+12+/-28=40/-16
\ No newline at end of file

tests/ut/python/dataset/test_bert_tokenizer.py

@@ -27,7 +27,7 @@ vocab_bert = [
     "繁", "體", "字", "嘿", "哈", "大", "笑", "嘻",
     "i", "am", "mak", "make", "small", "mistake", "##s", "during", "work", "##ing", "hour",
     "😀", "😃", "😄", "😁", "+", "/", "-", "=", "12", "28", "40", "16", " ", "I",
-    "[CLS]", "[SEP]", "[UNK]", "[PAD]", "[MASK]"
+    "[CLS]", "[SEP]", "[UNK]", "[PAD]", "[MASK]", "[unused1]", "[unused10]"
 ]
 pad = '<pad>'
 test_paras = [
@@ -69,22 +69,40 @@ test_paras = [
     # test preserved tokens
     dict(
         first=8,
-        last=12,
+        last=14,
         expect_str=[
             ['[UNK]', '[CLS]'],
             ['[UNK]', '[SEP]'],
             ['[UNK]', '[UNK]'],
             ['[UNK]', '[PAD]'],
             ['[UNK]', '[MASK]'],
+            ['[unused1]'],
+            ['[unused10]']
         ],
         lower_case=False,
         vocab_list=vocab_bert,
         preserve_unused_token=True,
     ),
+    dict(
+        first=8,
+        last=14,
+        expect_str=[
+            ['[UNK]', '[CLS]'],
+            ['[UNK]', '[SEP]'],
+            ['[UNK]', '[UNK]'],
+            ['[UNK]', '[PAD]'],
+            ['[UNK]', '[MASK]'],
+            ['[unused1]'],
+            ['[unused10]']
+        ],
+        lower_case=True,
+        vocab_list=vocab_bert,
+        preserve_unused_token=True,
+    ),
     # test special symbol
     dict(
-        first=13,
-        last=13,
+        first=15,
+        last=15,
         expect_str=[['12', '+', '/', '-', '28', '=', '40', '/', '-', '16']],
         preserve_unused_token=True,
         vocab_list=vocab_bert