improved error handling

src/dsiParser.py

@@ -34,7 +34,7 @@ class dsiParser:
         warningMessages = []
         # Catch any double (triple...) \ before they annoy us
         while r'\\' in dsiString:
-            warningMessages.append(_warn(f"Double backslash found in string, treating as one backslash: \"{dsiString}\"", RuntimeWarning))
+            warningMessages.append(_warn(f"Double backslash found in string, treating as one backslash: >{dsiString}<", RuntimeWarning))
             dsiString = dsiString.replace(r'\\', '\\')
 
         if dsiString == "":
@@ -90,7 +90,7 @@ class dsiParser:
         if items[0] == '': #first item of List should be empty, remove it
             items.pop(0) 
         else:
-            warningMessages.append(_warn(f"string should start with \\, string given was \"{dsiString}\"", RuntimeWarning))
+            warningMessages.append(_warn(f"string should start with \\, string given was >{dsiString}<", RuntimeWarning))
         nodes = []
 
         (prefix, unit, exponent) = ('','','')
@@ -116,7 +116,7 @@ class dsiParser:
                     item = ''
             elif re.match(r'tothe\{.*\}', item):
                 exponent = item.split('{')[1].split('}')[0]
-                warningMessages.append(_warn(f"The exponent \"{exponent}\" is not a number!", RuntimeWarning))
+                warningMessages.append(_warn(f"The exponent >{exponent}< is not a number!", RuntimeWarning))
                 try:
                     item = items.pop(0)
                 except IndexError:
@@ -127,14 +127,20 @@ class dsiParser:
                     item = items.pop(0)
                 except IndexError:
                     item = ''
-                warningMessages.append(_warn(f"The identifier \"{unit}\" does not match any D-SI units!", RuntimeWarning))
+                closestMatches= _getClosestStr(unit)
+                if len (closestMatches) > 0:
+                    closestMatchesStr= ', \\'.join(closestMatches)
+                    closestMatchesStr= '\\' + closestMatchesStr
+                    warningMessages.append(_warn(fr"The identifier >{unit}< does not match any D-SI units! Did you mean one of these >{closestMatchesStr}< ?", RuntimeWarning))
+                else:
+                    warningMessages.append(_warn(fr"The identifier >{unit}< does not match any D-SI units!", RuntimeWarning))
             elif unit == '':
                 itemStr = ""
                 if prefix != "":
                     itemStr = itemStr + "\\" + prefix
                 if exponent != "":
                     itemStr = itemStr + r"\tothe{" + exponent + r"}"
-                warningMessages.append(_warn(f"This D-SI unit seems to be missing the base unit! \"{itemStr}\"", RuntimeWarning))
+                warningMessages.append(_warn(f"This D-SI unit seems to be missing the base unit! >{itemStr}<", RuntimeWarning))
 
             nodes.append(_node(prefix, unit, exponent))
             if (len(items) == 0) and (item == ''): break
@@ -258,7 +264,7 @@ def _warn(message: str, warningClass):
     warnings.warn(message, warningClass)
     return message
 
-def _getClosestStrAndType(unkownStr):
+def _getClosestStr(unkownStr):
     """returns the closest string and type of the given string
 
     Args:
@@ -268,19 +274,9 @@ def _getClosestStrAndType(unkownStr):
         str: closest string
         str: type of closest string
     """
-    closestStr = difflib.get_close_matches(unkownStr, _dsiPrefixesLatex.keys(), n=1)
-    if len(closestStr) == 0:
-        closestStr = difflib.get_close_matches(unkownStr, _dsiUnitsLatex.keys(), n=1)
-        if len(closestStr) == 0:
-            closestStr = difflib.get_close_matches(unkownStr, _dsiKeyWords.keys(), n=1)
-            if len(closestStr) == 0:
-                return (unkownStr, 'unknown')
-            else:
-                return (closestStr[0], 'keyWord')
-        else:
-            return (closestStr[0], 'unit')
-    else:
-        return (closestStr[0], 'prefix')
+    possibleDsiKeys = _dsiPrefixesLatex.keys() | _dsiUnitsLatex.keys() | _dsiKeyWords.keys()
+    closestStr = difflib.get_close_matches(unkownStr, possibleDsiKeys, n=3,cutoff=0.66)
+    return closestStr
 
 # mapping D-SI prefixes to latex
 _dsiPrefixesLatex = {

tests/test_main.py

 import pytest
-from dsiParser import dsiParser, _node, _getClosestStrAndType
+from dsiParser import dsiParser, _node, _getClosestStr
 
 p = dsiParser()
 
@@ -20,30 +20,30 @@ def test_baseCase():
 
 def test_robustness():
     # Unknown unit
-    with pytest.warns(RuntimeWarning, match='The identifier \"foo\" does not match any D-SI units!'):
+    with pytest.warns(RuntimeWarning, match='The identifier >foo< does not match any D-SI units!'):
         tree = p.parse(r'\foo')
         assert tree.toLatex() == r'$${\color{red}\mathrm{foo}}$$'  
         assert not tree.valid
         assert len(tree.warnings) == 1
-        assert tree.warnings == ['The identifier \"foo\" does not match any D-SI units!']
+        assert tree.warnings == ['The identifier >foo< does not match any D-SI units!']
     
     # Unknown string in the middle of input
-    with pytest.warns(RuntimeWarning, match='The identifier \"mini\" does not match any D-SI units!'):
+    with pytest.warns(RuntimeWarning, match=r'The identifier >mini< does not match any D-SI units! Did you mean one of these >\\milli< ?'):
         tree = p.parse(r'\kilo\metre\per\mini\second')
         print(tree.toLatex())
         assert tree.toLatex() == r'$$\frac{\mathrm{k}\mathrm{m}}{{\color{red}\mathrm{mini}}\,\mathrm{s}}$$'  
         assert not tree.valid
         assert len(tree.warnings) == 1
-        assert tree.warnings == ['The identifier \"mini\" does not match any D-SI units!']
+        assert tree.warnings == ['The identifier >mini< does not match any D-SI units! Did you mean one of these >\\milli< ?']
     
     # Base unit missing
-    with pytest.warns(RuntimeWarning, match=r'This D-SI unit seems to be missing the base unit! \"\\milli\\tothe\{2\}\"'):
+    with pytest.warns(RuntimeWarning, match=r'This D-SI unit seems to be missing the base unit! >\\milli\\tothe\{2\}<'):
         tree = p.parse(r'\milli\tothe{2}')
         print(tree.toLatex())
         assert tree.toLatex() == r'$${\color{red}\mathrm{m}{\color{red}\mathrm{}}^{2}}$$'  
         assert not tree.valid
         assert len(tree.warnings) == 1
-        assert tree.warnings == ['This D-SI unit seems to be missing the base unit! \"\\milli\\tothe{2}\"']
+        assert tree.warnings == ['This D-SI unit seems to be missing the base unit! >\\milli\\tothe{2}<']
 
 
 def test_power():
@@ -58,10 +58,10 @@ def test_power():
     assert p.parse(r'\metre\tothe{1337}').toLatex() == r'$$\mathrm{m}^{1337}$$'
 
     # Non-numerical power
-    with pytest.warns(RuntimeWarning, match='The exponent \"foo\" is not a number!'):
+    with pytest.warns(RuntimeWarning, match='The exponent >foo< is not a number!'):
         abcPowerTree = p.parse(r'\metre\tothe{foo}')
         assert abcPowerTree.toLatex() == r'$$\mathrm{m}^{{\color{red}\mathrm{foo}}}$$'
-        assert abcPowerTree.warnings == ['The exponent "foo" is not a number!']
+        assert abcPowerTree.warnings == ['The exponent >foo< is not a number!']
 
 def test_prefix():
     # D-SI prefix
@@ -104,9 +104,9 @@ def test_wrappers():
     assert parserFull.parse(r'\metre').toLatex() == r'@\mathrm{Prefix}\mathrm{m}\mathrm{Suffix}@'
 
 def test_getClosestMatch():
-    closestMatch = _getClosestStrAndType('\kiilo')
-    assert closestMatch == ('kilo','prefix')
-    closestMatch = _getClosestStrAndType('\mettre')
-    assert closestMatch == ('metre','unit')
-    closestMatch = _getClosestStrAndType(r'\ttothe')
-    assert closestMatch == ('tothe', 'keyWord')
+    closestMatch = _getClosestStr('\kiilo')
+    assert closestMatch == (['kilo'])
+    closestMatch = _getClosestStr('\mettre')
+    assert closestMatch == (['metre'])
+    closestMatch = _getClosestStr(r'\ttothe')
+    assert closestMatch == (['tothe'])