handeling brackets and + symbol in phone number

hellohaptik · Nov 9, 2022 · f8eb9ad · f8eb9ad
1 parent 414600d
commit f8eb9ad
Show file tree

Hide file tree

Showing 2 changed files with 17 additions and 9 deletions.
diff --git a/ner_v2/detectors/numeral/number/zh-TW/number_detection.py b/ner_v2/detectors/numeral/number/zh-TW/number_detection.py
@@ -42,7 +42,8 @@ def __init__(self, entity_name='number', unit_type=None):
 
         self.special_chars_mapping = {
             ',': '、',
-            '.': '點'
+            '.': '點',
+            '+': '加'
         }
 
     def _get_base_map_choices(self, base_map):
@@ -134,13 +135,13 @@ def get_number(self, original_text):
             return self.get_number_digit_by_digit(original_text)
         return self.get_number_with_digit_scaling(original_text)
 
-    def extract_digits_only(self, text, with_scale=False):
+    def extract_digits_only(self, text, rgx_pattern=None, with_special_chars=False):
         text = text or ''
-        rgx_pattern = r'[-,.+\s{}]+'
-        if not with_scale:
-            rgx_pattern = re.compile(rgx_pattern.format(self.base_numbers_map_choices))
-        else:
-            rgx_pattern = re.compile(rgx_pattern.format(self.base_numbers_map_choices_full))
+        rgx_pattern = rgx_pattern or r'[-,.+\s{}]+'
+        digit_choices = self.base_numbers_map_choices
+        if with_special_chars:
+            digit_choices += '|'.join(self.special_chars_mapping.values())
+        rgx_pattern = re.compile(rgx_pattern.format(digit_choices))
         return rgx_pattern.findall(text)
 
     def get_number_digit_by_digit(self, text=''):

diff --git a/ner_v2/detectors/pattern/phone_number/phone_number_detection.py b/ner_v2/detectors/pattern/phone_number/phone_number_detection.py
@@ -183,9 +183,16 @@ def _text_list_for_detection(self, text=None):
         return : list[string]
         """
         text = text or ''
-        matches = self.language_number_detector.extract_digits_only(text)
+        phone_number_format_regex = r'[-(),.+\s{}]+'
+        matches = self.language_number_detector.extract_digits_only(text, phone_number_format_regex, True)
         return matches
 
+    def _sanitize_text(self, text=None):
+        text = text or ''
+        sanitized_text = self.language_number_detector.replace_special_chars(text)
+        sanitized_text = self.language_number_detector.get_number_digit_by_digit(sanitized_text)
+        return sanitized_text
+
     def detect_entity(self, text, **kwargs):
         """
         This is to detect phone numbers from text by mapping chinese digits to numeric values
@@ -194,7 +201,7 @@ def detect_entity(self, text, **kwargs):
         self.phone, self.original_phone_text = [], []
         for _text in number_matches:
             original_text = " " + _text.lower().strip() + " "
-            sanitized_text = self.language_number_detector.get_number_digit_by_digit(original_text)
+            sanitized_text = self._sanitize_text(original_text)
             for match in phonenumbers.PhoneNumberMatcher(sanitized_text, self.country_code, leniency=0):
                 if match.number.country_code == phonenumbers.country_code_for_region(self.country_code):
                     self.phone.append(self.check_for_country_code(str(match.number.national_number)))