def feed_list(self,table):
""" Feeds a table (list of rows) which is converted
to CSV.
No more than len(columns) items are written for each
row. All rows are filled up with "" entries to have an
equal number of items. None entries are converted to empty
strings, all other objects are stringified.
"""
columns = self.columns
if columns:
rowlen = len(columns)
else:
# Calculate the max. number of columns in the table
rowlen = max(map(len,table))
# Prepare an empty table
t = [None] * len(table)
_quote = self._quote
# Fill in data
for i,row in Tools.irange(table):
row = _quote(row[:rowlen])
if len(row) < rowlen:
row[len(row):] = ['""'] * (rowlen - len(row))
t[i] = self.separator.join(row)
# Add final CRLF and add as CSV text
t.append('')
self.text = self.text + self.lineend.join(t)
class _modinit:
import math
l = Tools.frange(0, 1, MAXLOCALITY)
for i, factor in Tools.irange(l):
_weights[i] = int((math.exp(factor) - 1.0) * 8192)
if _debug:
print i, '. weight =', _weights[i]
def feed_list(self, table):
""" Feeds a table (list of rows) which is converted
to CSV.
No more than len(columns) items are written for each
row. All rows are filled up with "" entries to have an
equal number of items. None entries are converted to empty
strings, all other objects are stringified.
"""
columns = self.columns
if columns:
rowlen = len(columns)
else:
# Calculate the max. number of columns in the table
rowlen = max(map(len, table))
# Prepare an empty table
t = [None] * len(table)
_quote = self._quote
# Fill in data
for i, row in Tools.irange(table):
row = _quote(row[:rowlen])
if len(row) < rowlen:
row[len(row):] = ['""'] * (rowlen - len(row))
t[i] = self.separator.join(row)
# Add final CRLF and add as CSV text
t.append('')
self.text = self.text + self.lineend.join(t)
def feed_objects(self, objects, getattr=getattr):
""" Feeds a sequence of objects which is converted to CSV.
For each object the set column names are interpreted as
object attributes and used as basis for the CSV data.
None values are converted to empty strings, all other
attributes are added stringified.
"""
columns = self.columns
if not columns:
raise Error, 'no output columns set'
rowlen = len(columns)
# Create an emtpy table
rows = len(objects)
rowindices = Tools.trange(rows)
t = [None] * rows
for i in rowindices:
t[i] = [None] * rowlen
# Fill the table
icols = Tools.irange(columns)
for i in rowindices:
obj = objects[i]
for j, name in icols:
t[i][j] = str(getattr(obj, name))
# Quote and join lines
t = [self.separator.join(self._quote(x)) for x in t]
# Add final CRLF and store CSV text
t.append('')
self.text = self.text + self.lineend.join(t)
def cut(self, NOM=NOM, DENOM=DENOM):
""" Force a cut of the cache's contents.
This will make room for at least one new entry.
"""
if _debug:
print ' Cutting down cache size...'
cachesize = self.cachesize
# Cut the cache down to the entries in recent get history
newdata = {}
known_key = newdata.has_key
data = self.data
for id in self.get_history[-self.locality:]:
if known_key(id):
continue
try:
newdata[id] = data[id]
except KeyError:
pass
cachesize = len(newdata)
if _debug:
print ' Size after cut to recent history:', cachesize
# Check
if cachesize * NOM >= self.max_cachesize * DENOM:
# Calculate weights
d = {}
weights = _weights
d_get = d.get
for i, id in Tools.irange(self.get_history[-self.locality:]):
if not known_key(id):
continue
d[id] = d_get(id, 0) + weights[i]
# Delete all entries left from median
ranking = Tools.sortedby(d.items(), 1)
if _debug:
print ' Ranking:', ranking
for id, weight in ranking[:len(d) / 2]:
if _debug:
print ' Deleting', id, 'with weight =', weight
del newdata[id]
# Check
cachesize = len(newdata)
if cachesize * NOM >= self.max_cachesize * DENOM:
# Ok, so the smart way didn't work...
if _debug:
print ' Did not work, going the hard way...'
newdata.clear()
cachesize = 0
self.data = newdata
self.cachesize = cachesize
self.cuts = self.cuts + 1
def _unquote(self, line):
""" Unquote a CSV style quoted line of text.
Internal method. Do not use directly.
"""
for i, text in Tools.irange(line):
if text[:1] == '"' and text[-1:] == '"':
text = text[1:-1]
line[i] = text.replace('""', '"')
return line
def _quote(self, line, str=str):
""" CSV style quote the given line of text.
"""
nline = ['""'] * len(line)
for i, item in Tools.irange(line):
if item is not None:
text = str(item)
else:
text = ''
nline[i] = '"%s"' % text.replace('"', '""')
return nline
def rpcdecode(url,
prefix='',
decode=1,
splitat=TextTools.splitat,
charsplit=TextTools.charsplit,
len=len,
tuple=tuple,
urldecode=urldecode):
""" Decode a RPC encoded function/method call.
Returns a tuple (name,args,kws) where args is a tuple of
string arguments and kws is a dictionary containing the given
keyword parameters or None. All parameters are returned as
strings; it is up to the caller to decode them into
e.g. integers, etc.
If prefix is given and found it is removed from the name prior
to returning it. decode can be set to false to prevent the url
from being urldecoded prior to processing.
The decode function also supports the syntax 'method' instead
of 'method()' for calls without arguments.
"""
if decode:
url = urldecode(url)
# Decode the method: method[(arg0,arg1,...,kw0=val0,kw1=val1,...)]
name, rawargs = splitat(url, '(')
if rawargs:
# Cut out the pure argument part, ignoring any character after
# the final ')'
rawargs, rest = splitat(rawargs, ')', -1)
# Argument list: split at ','
args = charsplit(rawargs, ',')
if '=' in rawargs:
kws = {}
for i, arg in Tools.reverse(Tools.irange(args)):
if '=' in arg:
k, v = splitat(arg, '=')
kws[k] = v
del args[i]
else:
kws = None
args = tuple(args)
else:
args = ()
kws = None
if prefix:
if name[:len(prefix)] == prefix:
name = name[len(prefix):]
return name, args, kws
else:
return name, args, kws
def _unquote(self,line):
""" Unquote a CSV style quoted line of text.
Internal method. Do not use directly.
"""
for i,text in Tools.irange(line):
if text[:1] == '"' and text[-1:] == '"':
text = text[1:-1]
line[i] = text.replace('""','"')
return line
def dictionary(self):
""" Return the current data as dictionary of lists of strings,
with one entry for each column.
.columns must have been set using .set_columns() or by
processing a given CSV header.
"""
table = {}
lines = self.lines
keys = self.columns
if keys is None:
raise Error, 'no columns set'
rows = len(lines)
for k in keys:
table[k] = [None] * rows
for i, key in Tools.irange(keys):
column = table[key]
for j, row in Tools.irange(lines):
if len(row) > i:
column[j] = row[i]
return table
def filter_header(self, header, lower=TextTools.lower):
""" Filter the given header line.
The base class converts the column names to all lowercase
and removes any whitespace included in the header.
This method is only called in case the header was read
from the data provided to the object.
"""
l = [''] * len(header)
for i, column in Tools.irange(header):
l[i] = ''.join(lower(column).split())
return l
def dictionary(self):
""" Return the current data as dictionary of lists of strings,
with one entry for each column.
.columns must have been set using .set_columns() or by
processing a given CSV header.
"""
table = {}
lines = self.lines
keys = self.columns
if keys is None:
raise Error,'no columns set'
rows = len(lines)
for k in keys:
table[k] = [None] * rows
for i, key in Tools.irange(keys):
column = table[key]
for j, row in Tools.irange(lines):
if len(row) > i:
column[j] = row[i]
return table
def _quote(self, line,
str=str):
""" CSV style quote the given line of text.
"""
nline = ['""'] * len(line)
for i,item in Tools.irange(line):
if item is not None:
text = str(item)
else:
text = ''
nline[i] = '"%s"' % text.replace('"','""')
return nline
def rpcdecode(url,prefix='',decode=1,
splitat=TextTools.splitat,charsplit=TextTools.charsplit,
len=len,tuple=tuple,urldecode=urldecode):
""" Decode a RPC encoded function/method call.
Returns a tuple (name,args,kws) where args is a tuple of
string arguments and kws is a dictionary containing the given
keyword parameters or None. All parameters are returned as
strings; it is up to the caller to decode them into
e.g. integers, etc.
If prefix is given and found it is removed from the name prior
to returning it. decode can be set to false to prevent the url
from being urldecoded prior to processing.
The decode function also supports the syntax 'method' instead
of 'method()' for calls without arguments.
"""
if decode:
url = urldecode(url)
# Decode the method: method[(arg0,arg1,...,kw0=val0,kw1=val1,...)]
name,rawargs = splitat(url,'(')
if rawargs:
# Cut out the pure argument part, ignoring any character after
# the final ')'
rawargs,rest = splitat(rawargs,')',-1)
# Argument list: split at ','
args = charsplit(rawargs,',')
if '=' in rawargs:
kws = {}
for i,arg in Tools.reverse(Tools.irange(args)):
if '=' in arg:
k,v = splitat(arg,'=')
kws[k] = v
del args[i]
else:
kws = None
args = tuple(args)
else:
args = ()
kws = None
if prefix:
if name[:len(prefix)] == prefix:
name = name[len(prefix):]
return name,args,kws
else:
return name,args,kws
def list(self):
""" Return the current data as list of lists, each having
self.width string entries.
Missing entries are set to None.
"""
width = self.width
lines = self.lines
table = [None] * len(lines)
for i, row in Tools.irange(lines):
row = row[:]
if len(row) < width:
row[len(row):] = [None] * (width - len(row))
table[i] = row
return table
def _quote(self, line,
str=str):
""" CSV style quote the given line of text.
"""
nline = ['""'] * len(line)
for i,item in Tools.irange(line):
if item is None:
text = ''
elif isinstance(item, unicode):
text = item.encode(self.encoding)
else:
text = str(item)
nline[i] = '"%s"' % text.replace('"','""')
return nline
def list(self):
""" Return the current data as list of lists, each having
self.width string entries.
Missing entries are set to None.
"""
width = self.width
lines = self.lines
table = [None] * len(lines)
for i, row in Tools.irange(lines):
row = row[:]
if len(row) < width:
row[len(row):] = [None]*(width-len(row))
table[i] = row
return table
def filter_header(self, header,
lower=TextTools.lower):
""" Filter the given header line.
The base class converts the column names to all lowercase
and removes any whitespace included in the header.
This method is only called in case the header was read
from the data provided to the object.
"""
l = [''] * len(header)
for i,column in Tools.irange(header):
l[i] = ''.join(lower(column).split())
return l
def __str__(self):
lines = self.list()
desc = self.description()
width = 0
output = []
write = output.append
for col in desc:
write('%-*s|' % (col[1], col[0]))
write('\n')
for col in desc:
write('=' * col[1] + '+')
write('\n')
for line in lines:
for i, item in Tools.irange(line):
write('%-*s|' % (desc[i][1], item))
write('\n')
return ''.join(output)
def objects(self, constructor):
""" Builds a list of objects by calling the given constructor
with keywords defined by mapping column names to values for
each input line.
.columns must have been set using .set_columns() or by
processing a given CSV header.
"""
lines = self.lines
keys = self.columns
if keys is None:
raise Error, 'no columns set'
objs = [None] * len(lines)
for i, line in Tools.irange(lines):
kws = dict(Tools.tuples(keys, line))
objs[i] = apply(constructor, (), kws)
return objs
def __str__(self):
lines = self.list()
desc = self.description()
width = 0
output = []
write = output.append
for col in desc:
write('%-*s|' % (col[1],col[0]))
write('\n')
for col in desc:
write('=' * col[1] + '+')
write('\n')
for line in lines:
for i,item in Tools.irange(line):
write('%-*s|' % (desc[i][1],item))
write('\n')
return ''.join(output)
def objects(self,constructor):
""" Builds a list of objects by calling the given constructor
with keywords defined by mapping column names to values for
each input line.
.columns must have been set using .set_columns() or by
processing a given CSV header.
"""
lines = self.lines
keys = self.columns
if keys is None:
raise Error,'no columns set'
objs = [None] * len(lines)
for i,line in Tools.irange(lines):
kws = dict(Tools.tuples(keys, line))
objs[i] = apply(constructor,(),kws)
return objs
def feed_dict(self,table,rows=None):
""" Feeds a table (dict of lists) which is converted
to CSV.
Only the keys set as column names are used to form the CSV
data.
All lists in the dictionary must have equal length or at
least rows number of entries, if rows is given. None
entries are converted to empty strings, all other objects
are stringified.
"""
columns = self.columns
if not columns:
raise Error,'no output columns set'
rowlen = len(columns)
# Create an emtpy table
if not rows:
rows = 0
for column in columns:
nrows = len(table[column])
if nrows > rows:
rows = nrows
rowindices = Tools.trange(rows)
t = [None] * rows
for i in rowindices:
t[i] = [None] * rowlen
# Fill the table
for j,k in Tools.irange(columns):
for i in rowindices:
t[i][j] = table[k][i]
# Quote and join lines
t = [self.separator.join(self._quote(x)) for x in t]
# Add final CRLF and store CSV text
t.append('')
self.text = self.text + self.lineend.join(t)
def feed_dict(self,table,rows=None):
""" Feeds a table (dict of lists) which is converted
to CSV.
Only the keys set as column names are used to form the CSV
data.
All lists in the dictionary must have equal length or at
least rows number of entries, if rows is given. None
entries are converted to empty strings, all other objects
are stringified.
"""
columns = self.columns
if not columns:
raise Error,'no output columns set'
rowlen = len(columns)
# Create an emtpy table
if not rows:
rows = 0
for column in columns:
nrows = len(table[column])
if nrows > rows:
rows = nrows
rowindices = Tools.trange(rows)
t = [None] * rows
for i in rowindices:
t[i] = [None] * rowlen
# Fill the table
for j,k in Tools.irange(columns):
for i in rowindices:
t[i][j] = table[k][i]
# Quote and join lines
t = [self.separator.join(self._quote(x)) for x in t]
# Add final CRLF and store CSV text
t.append('')
self.text = self.text + self.lineend.join(t)
def description(self, header=1):
""" Return a list of tuples (column name, max length) found in the
data.
If header is true (default), the column names themselves
are included in the calculation.
"""
lines = self.lines
columns = self.columns
width = len(columns)
if header:
lengths = []
for column in columns:
lengths.append(len(column))
else:
lengths = [0] * width
for row in self.lines:
for i, o in Tools.irange(row[:width]):
if len(o) > lengths[i]:
lengths[i] = len(o)
return map(None, columns, lengths)
def feed_objects(self,objects,
getattr=getattr):
""" Feeds a sequence of objects which is converted to CSV.
For each object the set column names are interpreted as
object attributes and used as basis for the CSV data.
None values are converted to empty strings, all other
attributes are added stringified.
"""
columns = self.columns
if not columns:
raise Error,'no output columns set'
rowlen = len(columns)
# Create an emtpy table
rows = len(objects)
rowindices = Tools.trange(rows)
t = [None] * rows
for i in rowindices:
t[i] = [None] * rowlen
# Fill the table
icols = Tools.irange(columns)
for i in rowindices:
obj = objects[i]
for j,name in icols:
t[i][j] = str(getattr(obj, name))
# Quote and join lines
t = [self.separator.join(self._quote(x)) for x in t]
# Add final CRLF and store CSV text
t.append('')
self.text = self.text + self.lineend.join(t)
def description(self, header=1):
""" Return a list of tuples (column name, max length) found in the
data.
If header is true (default), the column names themselves
are included in the calculation.
"""
lines = self.lines
columns = self.columns
width = len(columns)
if header:
lengths = []
for column in columns:
lengths.append(len(column))
else:
lengths = [0] * width
for row in self.lines:
for i,o in Tools.irange(row[:width]):
if len(o) > lengths[i]:
lengths[i] = len(o)
return map(None,columns,lengths)
def cut(self,
NOM=NOM,DENOM=DENOM):
""" Force a cut of the cache's contents.
This will make room for at least one new entry.
"""
if _debug:
print ' Cutting down cache size...'
cachesize = self.cachesize
# Cut the cache down to the entries in recent get history
newdata = {}
known_key = newdata.has_key
data = self.data
for id in self.get_history[-self.locality:]:
if known_key(id):
continue
try:
newdata[id] = data[id]
except KeyError:
pass
cachesize = len(newdata)
if _debug:
print ' Size after cut to recent history:',cachesize
# Check
if cachesize * NOM >= self.max_cachesize * DENOM:
# Calculate weights
d = {}
weights = _weights
d_get = d.get
for i,id in Tools.irange(self.get_history[-self.locality:]):
if not known_key(id):
continue
d[id] = d_get(id,0) + weights[i]
# Delete all entries left from median
ranking = Tools.sortedby(d.items(),1)
if _debug:
print ' Ranking:',ranking
for id,weight in ranking[:len(d)/2]:
if _debug:
print ' Deleting',id,'with weight =',weight
del newdata[id]
# Check
cachesize = len(newdata)
if cachesize * NOM >= self.max_cachesize * DENOM:
# Ok, so the smart way didn't work...
if _debug:
print ' Did not work, going the hard way...'
newdata.clear()
cachesize = 0
self.data = newdata
self.cachesize = cachesize
self.cuts = self.cuts + 1
