Roman numeral parsing. More test cases. (Fixes issue #3)

diff --git a/NEWS.txt b/NEWS.txt
index 8bb6e70..823498d 100644 (file)
--- a/NEWS.txt
+++ b/NEWS.txt
@@ -1,5 +1,6 @@
 2010 - 0.2 - Dedicated to ２ (U+FF12)
  - 直感～ before 直感２～.
+ - Detect Unicode Roman numeral codepoints and sort them numerically.
 
 2010.02.22 - 0.1 - Dedicated to 勘 (U+52D8)
  - Initial release.

diff --git a/collate/strings.py b/collate/strings.py
index 60685c5..fd6f71d 100644 (file)
--- 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

diff --git a/tests/en/kanjinum.list.txt b/tests/en/kanjinum.list.txt
new file mode 100644 (file)
index 0000000..be3f783
--- /dev/null
+++ b/tests/en/kanjinum.list.txt
@@ -0,0 +1,4 @@
+白玉楼 ～ 幽冥の住人は割と少ない
+直感～時として恋は～ (live version)
+直感２～逃した魚は大きいぞ！～
+真っ赤なバラとジントニック

diff --git a/tests/en/romannumeral.list.txt b/tests/en/romannumeral.list.txt
new file mode 100644 (file)
index 0000000..00db739
--- /dev/null
+++ b/tests/en/romannumeral.list.txt
@@ -0,0 +1,10 @@
+ⅰ
+ⅱ
+ⅲ
+ⅳ
+ⅴ
+ⅵ
+ⅴⅰⅰ
+ⅹⅹ
+ⅿⅰⅹ
+ⅿⅹ