From 44fb5f91161af5a0ca798d237271e15b3eb13fe3 Mon Sep 17 00:00:00 2001
From: Chirag Jain <jain.chirag925@gmail.com>
Date: Fri, 5 Mar 2021 20:46:24 +0530
Subject: [PATCH 1/2] fix(v2/number): cast to int directly if no decimal point
 is found to avoid wrong values because of precision loss

---
 .../numeral/number/standard_number_detector.py   |  9 +++++++--
 ner_v2/detectors/numeral/utils.py                |  1 +
 .../numeral/number/en/number_ner_tests.yaml      | 16 +++++++++++++++-
 .../numeral/number/en/test_number_detection.py   |  3 +++
 4 files changed, 26 insertions(+), 3 deletions(-)

diff --git a/ner_v2/detectors/numeral/number/standard_number_detector.py b/ner_v2/detectors/numeral/number/standard_number_detector.py
index 0c454f64e..0b971a834 100644
--- a/ner_v2/detectors/numeral/number/standard_number_detector.py
+++ b/ner_v2/detectors/numeral/number/standard_number_detector.py
@@ -3,6 +3,7 @@
 
 import collections
 import os
+import decimal
 
 import pandas as pd
 from six.moves import zip
@@ -302,8 +303,12 @@ def _detect_number_from_digit(self, number_list=None, original_list=None):
                 scale = 1
 
             if number:
-                number = float(number) * scale
-                number = int(number) if number.is_integer() else number
+                if '.' not in number:
+                    number = int(number) * scale
+                else:
+                    number = float(number) * scale
+                    # FIXME: this conversion from float -> int is lossy, consider using Decimal class
+                    number = int(number) if number.is_integer() else number
                 unit = None
                 if self.unit_type:
                     unit, original_text = self._get_unit_from_text(original_text, processed_text)
diff --git a/ner_v2/detectors/numeral/utils.py b/ner_v2/detectors/numeral/utils.py
index 573605b05..7a748d314 100644
--- a/ner_v2/detectors/numeral/utils.py
+++ b/ner_v2/detectors/numeral/utils.py
@@ -26,6 +26,7 @@ def get_number_from_number_word(text, number_word_dict):
         [In]  >> _get_number_from_numerals('two hundred three four hundred three',  number_word_dict)
         [Out] >> (['103', '403'], ['one hundred three', 'four hundred three'])
     """
+    # FIXME: conversion from float -> int is lossy, consider using Decimal class
     detected_number_list = []
     detected_original_text_list = []
 
diff --git a/ner_v2/tests/numeral/number/en/number_ner_tests.yaml b/ner_v2/tests/numeral/number/en/number_ner_tests.yaml
index 24cfa727c..89ba0d73d 100644
--- a/ner_v2/tests/numeral/number/en/number_ner_tests.yaml
+++ b/ner_v2/tests/numeral/number/en/number_ner_tests.yaml
@@ -175,4 +175,18 @@ tests:
       - original_text: "50000 rupees"
         output_id: 3
         unit: rupees
-        value: "50000"
\ No newline at end of file
+        value: "50000"
+  - id: en_14
+    message: "here are some super large numbers 12345678901234567890 12345678 987654321012345678901 98765432101234567890123 "
+    unit_type: null
+    min_digit: 20
+    max_digit: 22
+    outputs:
+      - original_text: "12345678901234567890"
+        output_id: 1
+        unit: null
+        value: "12345678901234567890"
+      - original_text: "987654321012345678901"
+        output_id: 2
+        unit: null
+        value: "987654321012345678901"
\ No newline at end of file
diff --git a/ner_v2/tests/numeral/number/en/test_number_detection.py b/ner_v2/tests/numeral/number/en/test_number_detection.py
index bcd834583..79508ae2c 100644
--- a/ner_v2/tests/numeral/number/en/test_number_detection.py
+++ b/ner_v2/tests/numeral/number/en/test_number_detection.py
@@ -194,6 +194,9 @@ def run_test(self):
             message = testcase["message"]
             unit_type = testcase.get("unit_type", None)
             number_detector_object = NumberDetector(entity_name="number", language=language, unit_type=unit_type)
+            number_detector_object.set_min_max_digits(
+                min_digit=testcase.get('min_digit', number_detector_object.min_digit),
+                max_digit=testcase.get('max_digit', number_detector_object.max_digit))
             number_dicts, spans = number_detector_object.detect_entity(message)
 
             expected_number_dicts, expected_spans = parse_expected_outputs(testcase["outputs"])

From c12bf84a545763c3e6ae1e5d0a7e659602f0f10f Mon Sep 17 00:00:00 2001
From: Chirag Jain <jain.chirag925@gmail.com>
Date: Fri, 5 Mar 2021 20:51:21 +0530
Subject: [PATCH 2/2] style: remove unused decimal import

---
 ner_v2/detectors/numeral/number/standard_number_detector.py | 1 -
 1 file changed, 1 deletion(-)

diff --git a/ner_v2/detectors/numeral/number/standard_number_detector.py b/ner_v2/detectors/numeral/number/standard_number_detector.py
index 0b971a834..2df38a2f9 100644
--- a/ner_v2/detectors/numeral/number/standard_number_detector.py
+++ b/ner_v2/detectors/numeral/number/standard_number_detector.py
@@ -3,7 +3,6 @@
 
 import collections
 import os
-import decimal
 
 import pandas as pd
 from six.moves import zip