X-Git-Url: https://git.korewanetadesu.com/?p=python-collate.git;a=blobdiff_plain;f=collate%2Fstrings.py;h=fd6f71d852fe4c86f9260eb9a5d83b073dbe7e8d;hp=60685c57a3415e763a63d05545340429113ec222;hb=29aa14bcc723a78fb2f172488c65c4893f6797a3;hpb=ee7317abef6e6c235c6344f3f173af134e52cf8c

diff --git a/collate/strings.py b/collate/strings.py
index 60685c5..fd6f71d 100644
--- a/collate/strings.py
+++ b/collate/strings.py
@@ -21,6 +21,25 @@ INFINITY = float('inf')
 KEEP_IN_NUMBERS = u"'.,"
 ALLOWED_IN_NUMBERS = KEEP_IN_NUMBERS + u"_"
 
+ROMAN = {
+    u"i": 1,
+    u"v": 5,
+    u"x": 10,
+    u"l": 50,
+    u"c": 100,
+    u"d": 500,
+    u"m": 1000,
+    u"\u2180": 1000,
+    u"\u2181": 5000,
+    u"\u2182": 10000,
+    u"\u2183": 100,
+    u"\u2184": 100,
+    u"\u2185": 6,
+    u"\u2186": 50,
+    u"\u2187": 50000,
+    u"\u2188": 100000,
+    }
+
 def stripends(word):
     """Strip punctuation and symbols from the ends of a string."""
     while word and unicodedata.category(word[0])[0] in "PS":
@@ -154,6 +173,10 @@ def numeric(orig, invalid=INFINITY):
     else:
         return invalid
 
+    for char in string:
+        if u"\u2160" <= char <= u"\u2188":
+            return deroman(string)
+
     mult = 1
     while string[:1] == u"-" or string[:1] == u"+":
         if string[:1] == u"-":
@@ -276,3 +299,20 @@ def normalize_number(string):
         string = string.replace(u".", u"")
 
     return string or "NaN"
+
+def deroman(string):
+    """Turn a Roman numeral into an integer."""
+    string = unicodedata.normalize('NFKD', unicode(string)).lower()
+    previous = 0
+    building = 0
+    for char in reversed(string):
+        try:
+            value = ROMAN[char]
+        except KeyError:
+            continue
+        if value < previous:
+            building -= value
+        else:
+            building += value
+        previous = value
+    return building