Merge pull request #4 from dylanwal/bug_fix

update to scientific notation handling

Author: dylanwal

dev_files/simple.py

@@ -0,0 +1,9 @@
+import logging
+
+from unit_parse import parser, logger
+
+logger.setLevel(logging.DEBUG)
+
+result = parser("3.6E+00004 mg/L")
+print(result)
+

setup.cfg

@@ -1,6 +1,6 @@
 [metadata]
 name = unit_parse
-version = 0.1.6
+version = 0.1.7
 description = Parse units from strings. From mess to order!
 long_description = file: README.md
 long_description_content_type = text/markdown

src/unit_parse/pre_processing_substitution.py

@@ -192,8 +192,8 @@ def sub_sci_notation(text_in: str) -> str:
         text_in = text_in.replace(exp, exp_new)
 
     # remove leading zero in powers ( 5*10**-05 --> 5*10**-5)
-    text_in = re.sub('(?<=[*]{2}[-])0(?=[0-9])', "", text_in)  # neg. power
-    text_in = re.sub('(?<=[*]{2})0(?=[0-9])', "", text_in)  # pos. power
+    text_in = re.sub('(?<=[*]{2}[-])0+(?=[0-9])', "", text_in)  # neg. power
+    text_in = re.sub('(?<=[*]{2})0+(?=[0-9])', "", text_in)  # pos. power
 
     return text_in.strip()
 

tests/test_pre_processing_substitution.py

@@ -72,6 +72,7 @@ def test_remove_words(input_, output_):
     ['5.3e1 g/mol', '5.3*10**1 g/mol'],
     ['5.3E1 g/mol', '5.3*10**1 g/mol'],
     ['5.3 109 g/mol', '5.3*10**9 g/mol'],
+    ["3.6E+00004 mg/L", "3.6*10**4 mg/L"],
 
     # negative control (no changes made)
     ['5.3*10**1 g/mol', '5.3*10**1 g/mol'],