🎨 supress pep 0515, fix #48

CHANGELOG.rst

@@ -1,6 +1,14 @@
 Change log
 ================================================================================
 
+0.5.7 - unreleased
+--------------------------------------------------------------------------------
+
+fixed
+^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+
+#. `#48 <https://github.com/pyexcel/pyexcel-io/issues/48>`_, turn off pep 0515
+
 0.5.6 - 11.01.2017
 --------------------------------------------------------------------------------
 

changelog.yml

@@ -1,6 +1,12 @@
 name: pyexcel-io
 organisation: pyexcel
 releases:
+- changes:
+  - action: fixed
+    details:
+    - '`#48`, turn off pep 0515'
+  date: unreleased
+  version: 0.5.7
 - changes:
   - action: fixed
     details:

pyexcel_io/readers/csvr.py

@@ -130,6 +130,7 @@ class CSVSheetReader(SheetReader):
         ignore_infinity=True,
         auto_detect_int=True,
         auto_detect_datetime=True,
+        pep_0515_off=True,
         **keywords
     ):
         SheetReader.__init__(self, sheet, **keywords)
@@ -139,6 +140,7 @@ class CSVSheetReader(SheetReader):
         self.__ignore_infinity = ignore_infinity
         self.__auto_detect_datetime = auto_detect_datetime
         self.__file_handle = None
+        self.__pep_0515_off = pep_0515_off
 
     def get_file_handle(self):
         """ return me unicde reader for csv """
@@ -159,9 +161,11 @@ class CSVSheetReader(SheetReader):
     def __convert_cell(self, csv_cell_text):
         ret = None
         if self.__auto_detect_int:
-            ret = service.detect_int_value(csv_cell_text)
+            ret = service.detect_int_value(csv_cell_text, self.__pep_0515_off)
         if ret is None and self.__auto_detect_float:
-            ret = service.detect_float_value(csv_cell_text)
+            ret = service.detect_float_value(
+                csv_cell_text, self.__pep_0515_off
+            )
             shall_we_ignore_the_conversion = (
                 (ret in [float("inf"), float("-inf")])
                 and self.__ignore_infinity

pyexcel_io/service.py

@@ -39,27 +39,35 @@ def detect_date_value(cell_text):
     return ret
 
 
-def detect_float_value(cell_text):
-    try:
-        should_we_skip_it = (
-            cell_text.startswith("0") and cell_text.startswith("0.") is False
-        )
-        if should_we_skip_it:
-            # do not convert if a number starts with 0
-            # e.g. 014325
+def detect_float_value(cell_text, pep_0515_off=True):
+    should_we_skip_it = (
+        cell_text.startswith("0") and cell_text.startswith("0.") is False
+    )
+    if should_we_skip_it:
+        # do not convert if a number starts with 0
+        # e.g. 014325
+        return None
+
+    if pep_0515_off:
+        pattern = "([0-9]+_)+[0-9]+.[0-9]*$"
+        if re.match(pattern, cell_text):
             return None
 
-        else:
-            return float(cell_text)
-
+    try:
+        return float(cell_text)
     except ValueError:
         return None
 
 
-def detect_int_value(cell_text):
+def detect_int_value(cell_text, pep_0515_off=True):
     if cell_text.startswith("0") and len(cell_text) > 1:
         return None
 
+    if pep_0515_off:
+        pattern = "([0-9]+_)+[0-9]+$"
+        if re.match(pattern, cell_text):
+            return None
+
     try:
         return int(cell_text)
 

tests/test_issues.py

@@ -170,5 +170,14 @@ def test_issue_43():
     p.get_book(url="https://github.com/pyexcel/pyexcel-xls/raw/master/tests/fixtures/file_with_an_empty_sheet.xls");  # flake8: noqa
 
 
+def test_pyexcel_issue_138():
+    array = [['123_122', '123_1.', '123_1.0']]
+    save_data('test.csv', array)
+    data = get_data('test.csv')
+    expected = [['123_122', '123_1.', '123_1.0']]
+    eq_(data['test.csv'], expected)
+    os.unlink('test.csv')
+
+
 def get_fixture(file_name):
     return os.path.join("tests", "fixtures", file_name)

tests/test_service.py

@@ -1,5 +1,7 @@
 from nose.tools import eq_, raises
 from pyexcel_io.service import date_value, time_value
+from pyexcel_io.service import detect_int_value
+from pyexcel_io.service import detect_float_value
 
 
 def test_date_util_parse():
@@ -50,3 +52,25 @@ def test_fake_date_time_20():
 def test_issue_1_error():
     result = time_value('PT1111')
     eq_(result, None)
+
+
+def test_detect_int_value():
+    result = detect_int_value('123')
+    eq_(result, 123)
+
+
+def test_detect_float_value():
+    result = detect_float_value('123.1')
+    eq_(result, 123.1)
+
+
+def test_suppression_of_pep_0515_int():
+    result = detect_int_value('123_123')
+    eq_(result, None)
+
+
+def test_suppression_of_pep_0515_float():
+    result = detect_float_value('123_123.')
+    eq_(result, None)
+    result = detect_float_value('123_123.1')
+    eq_(result, None)