X-Git-Url: https://git.korewanetadesu.com/?p=python-collate.git;a=blobdiff_plain;f=collate%2F_strings.py;h=d872ed4a1bca483944cf6b3a4a86b870562dd314;hp=fd18bb917008dcb56aa06246dc86d1a721fe601c;hb=7c67e10286c784b572703666a980e85b39b858ee;hpb=53e1676b8d68cccd2b0692654d3871e44e0ba6b6

diff --git a/collate/_strings.py b/collate/_strings.py
index fd18bb9..d872ed4 100644
--- a/collate/_strings.py
+++ b/collate/_strings.py
@@ -1,11 +1,24 @@
 import unicodedata
 
+def strip_punc(string):
+    return filter(lambda c: unicodedata.category(c)[0] not in "PS", string)
+
+def strip_ends(string):
+    while string and unicodedata.category(string[0])[0] in "ZPS":
+        string = string[1:]
+    while string and unicodedata.category(string[-1])[0] in "ZPS":
+        string = string[:-1]
+    return string
+
 def alnumsplit(string):
+    if not string:
+        return []
     string = unicode(string)
     strings = []
-    word = []
     numeric = None
-    for char in string:
+    start = 0
+    for i, char in enumerate(string):
+        category = unicodedata.category(char)
         if numeric is None:
             broke = False
             if char.isnumeric():
@@ -14,16 +27,18 @@ def alnumsplit(string):
                 numeric = False
         elif numeric and char.isalpha():
             broke = True
+            numeric = False
+        elif numeric and category in ["Zs", "Ps", "Pe"]:
+            broke = True
+            numeric = None
         elif not numeric and char.isnumeric():
             broke = True
+            numeric = True
         if broke:
-            if word:
-                strings.append(u"".join(word))
-                word = []
-            numeric = None
-        word.append(char)
-    if word:
-        strings.append(u"".join(word))
+            strings.append(strip_ends(string[start:i]))
+            start = i
+            broke = False
+    strings.append(strip_ends(string[start:i + 1]))
     return strings
 
 def wordlike(string):
@@ -44,6 +59,7 @@ def wordlike(string):
 def numeric(orig, invalid=float('inf')):
     if not orig:
         return (invalid, '')
+
     string = unicode(orig)
     for c in string:
         if c.isnumeric():
@@ -57,6 +73,9 @@ def numeric(orig, invalid=float('inf')):
             mult = -mult
         string = string[1:]
 
+    if not string[:1].isnumeric():
+        return (invalid, orig)
+
     # Early out if possible.
     try:
         return (float(string) * mult, orig)
@@ -70,7 +89,7 @@ def numeric(orig, invalid=float('inf')):
         total = 0
         for c in string:
             v = unicodedata.numeric(c)
-            if v >= 1:
+            if v >= 1 or v == 0:
                 total *= 10
             total += v
         return total