From 23b9e6de62035c101631ca350226e3548392b95f Mon Sep 17 00:00:00 2001
From: mjtnix <mjtnix@5914aa95-5b3a-0410-a3b5-7b719e7fe9b2>
Date: Fri, 12 Oct 2007 07:35:58 +0000
Subject: mql key quoting and unquoting routines

git-svn-id: http://freebase-python.googlecode.com/svn/trunk@5 5914aa95-5b3a-0410-a3b5-7b719e7fe9b2
---
 freebase/api/__init__.py |   4 +-
 freebase/api/mqlkey.py   | 102 +++++++++++++++++++++++++++++++++++++++++++++++
 2 files changed, 105 insertions(+), 1 deletion(-)
 create mode 100755 freebase/api/mqlkey.py

(limited to 'freebase/api')

diff --git a/freebase/api/__init__.py b/freebase/api/__init__.py
index 959259f..c3adebc 100755
--- a/freebase/api/__init__.py
+++ b/freebase/api/__init__.py
@@ -1,2 +1,4 @@
 
-from session import HTTPMetawebSession, MetawebError
+from session import HTTPMetawebSession, MetawebError, attrdict
+
+from mqlkey import quotekey, unquotekey
diff --git a/freebase/api/mqlkey.py b/freebase/api/mqlkey.py
new file mode 100755
index 0000000..0458693
--- /dev/null
+++ b/freebase/api/mqlkey.py
@@ -0,0 +1,102 @@
+
+import string
+import re
+
+def quotekey(ustr):
+    """
+    quote a unicode string to turn it into a valid namespace key
+    
+    """
+    valid_always = string.ascii_letters + string.digits
+    valid_interior_only = valid_always + '_-'
+
+    if isinstance(ustr, str):
+        s = unicode(ustr,'utf-8')        
+    elif isinstance(ustr, unicode):
+        s = ustr
+    else:
+        raise ValueError, 'quotekey() expects utf-8 string or unicode'
+
+    output = []
+    if s[0] in valid_always:
+        output.append(s[0])
+    else:
+        output.append('$%04X' % ord(s[0]))
+
+    for c in s[1:-1]:
+        if c in valid_interior_only:
+            output.append(c)
+        else:
+            output.append('$%04X' % ord(c))
+
+    if len(s) > 1:
+        if s[-1] in valid_always:
+            output.append(s[-1])
+        else:
+            output.append('$%04X' % ord(s[-1]))
+
+    return str(''.join(output))
+
+
+def unquotekey(key, encoding=None):
+    """
+    unquote a namespace key and turn it into a unicode string
+    """
+
+    valid_always = string.ascii_letters + string.digits
+
+    output = []
+    i = 0
+    while i < len(key):
+        if key[i] in valid_always:
+            output.append(key[i])
+            i += 1
+        elif key[i] in '_-' and i != 0 and i != len(key):
+            output.append(key[i])
+            i += 1
+        elif key[i] == '$' and i+4 < len(key):
+            # may raise ValueError if there are invalid characters
+            output.append(unichr(int(key[i+1:i+5],16)))
+            i += 5
+        else:
+            raise ValueError, "unquote key saw invalid character '%s' at position %d" % (key[i], i)
+
+    ustr = u''.join(output)
+    
+    if encoding is None:
+        return ustr
+
+    return ustr.encode(encoding)
+
+
+
+def id_to_urlid(id):
+    """
+    convert a mql id to an id suitable for embedding in a url path.
+    """
+
+    # XXX shouldn't be in metaweb.api!
+    from mw.formats.http import urlencode_pathseg
+
+    segs = id.split('/')
+
+    assert isinstance(id, str) and id != '', 'bad id "%s"' % id
+
+    if id[0] == '~':
+        assert len(segs) == 1
+        # assume valid, should check
+        return id
+
+    if id[0] == '#':
+        assert len(segs) == 1
+        # assume valid, should check
+        return '%23' + id[1:]
+
+    if id[0] != '/':
+        raise ValueError, 'unknown id format %s' % id
+
+    # ok, we have a slash-path
+    # requote components as keys and rejoin.
+    # urlids do not have leading slashes!!!
+    return '/'.join(urlencode_pathseg(unquotekey(seg)) for seg in segs[1:])
+
-- 
cgit v1.3.1