Add LogDog stream name normalization function.

client/libs/logdog/streamname.py

Index: client/libs/logdog/streamname.py
diff --git a/client/libs/logdog/streamname.py b/client/libs/logdog/streamname.py
index 3b92fa72980a69d9ec3180c1e257c1ffe895d20b..0243bc214a818f9faf9c046a6d301a91073c6bd4 100644
--- a/client/libs/logdog/streamname.py
+++ b/client/libs/logdog/streamname.py
@@ -13,6 +13,7 @@ _MAX_STREAM_NAME_LENGTH = 4096
 _MAX_TAG_KEY_LENGTH = 64
 _MAX_TAG_VALUE_LENGTH = 4096
 
+
 def validate_stream_name(v, maxlen=None):
   """Verifies that a given stream name is valid.
 
@@ -42,3 +43,66 @@ def validate_tag(key, value):
   """
   validate_stream_name(key, maxlen=_MAX_TAG_KEY_LENGTH)
   validate_stream_name(value, maxlen=_MAX_TAG_VALUE_LENGTH)
+
+
+def normalize(v, prefix=None):
+  """Given a string, "v", mutate it into a valid stream name.
+
+  This operates by replacing invalid stream naem characters with underscores (_)
+  when encountered.
+
+  A special case is when "v" begins with an invalid character. In this case, we
+  will replace it with the "prefix", if one is supplied.
+
+  See _STREAM_NAME_RE for a description of a valid stream name.
+
+  Raises:
+    ValueError: If normalization could not be successfully performed.
+  """
+  if len(v) == 0:
+    if not prefix:
+      raise ValueError('Cannot normalize empty name with no prefix.')
+    v = prefix
+  else:
+    out = []
+    for i, ch in enumerate(v):
+      if i == 0 and not _is_valid_stream_char(ch, first=True):
+        # The first letter is special, and must be alphanumeric.
+        # If we have a prefix, prepend that to the resulting string.
+        if prefix is None:
+          raise ValueError('Name has invalid beginning, and no prefix was '
+                           'provided.')
+        out.append(prefix)

[martiniss, 2016/08/11 20:24:43]

Doing iteration over every character in a string like this, and appending to a
new list isn't the most efficient. This isn't on any super critical path though
AFAIK, so it should be fine.

[dnj, 2016/08/11 20:42:29]

If you know a better way to do this, LMK.

+
+      if not _is_valid_stream_char(ch):
+        ch = '_'
+      out.append(ch)
+    v = ''.join(out)
+
+  # Validate the resulting string.
+  validate_stream_name(v)
+  return v
+
+
+def _is_valid_stream_char(ch, first=False):
+  """Returns (bool): True if a character is alphanumeric.
+
+  The first character must be alphanumeric, matching [a-zA-Z0-9].
+  Additional characters must either be alphanumeric or one of: (: _ - .).
+
+  Args:
+    ch (str): the character to evaluate.
+    first (bool): if true, apply special first-character constraints.
+  """
+  # Alphanumeric check.
+  ch_ord = ord(ch)
+  if ((ch_ord >= ord('0') and ch_ord <= ord('9')) or

[martiniss, 2016/08/11 20:24:43]

Why not use https://docs.python.org/2/library/string.html#string-constants ?

[dnj, 2016/08/11 20:42:29]

This seems faster, but might as well.

+      (ch_ord >= ord('a') and ch_ord <= ord('z')) or
+      (ch_ord >= ord('A') and ch_ord <= ord('Z'))):
+    return True
+  if first:
+    # The first character must be alphanumeric.
+    return False
+
+  # Check additional middle-name characters:
+  return ch in ':_-./'