1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
|
#
# TreeFragments - parsing of strings to trees
#
import re
from cStringIO import StringIO
from Scanning import PyrexScanner, StringSourceDescriptor
from Symtab import BuiltinScope, ModuleScope
import Symtab
import PyrexTypes
from Visitor import VisitorTransform
from Nodes import Node, StatListNode
from ExprNodes import NameNode
import Parsing
import Main
import UtilNodes
"""
Support for parsing strings into code trees.
"""
class StringParseContext(Main.Context):
def __init__(self, include_directories, name):
Main.Context.__init__(self, include_directories, {})
self.module_name = name
def find_module(self, module_name, relative_to = None, pos = None, need_pxd = 1):
if module_name != self.module_name:
raise AssertionError("Not yet supporting any cimports/includes from string code snippets")
return ModuleScope(module_name, parent_module = None, context = self)
def parse_from_strings(name, code, pxds={}, level=None, initial_pos=None):
"""
Utility method to parse a (unicode) string of code. This is mostly
used for internal Cython compiler purposes (creating code snippets
that transforms should emit, as well as unit testing).
code - a unicode string containing Cython (module-level) code
name - a descriptive name for the code source (to use in error messages etc.)
"""
# Since source files carry an encoding, it makes sense in this context
# to use a unicode string so that code fragments don't have to bother
# with encoding. This means that test code passed in should not have an
# encoding header.
assert isinstance(code, unicode), "unicode code snippets only please"
encoding = "UTF-8"
module_name = name
if initial_pos is None:
initial_pos = (name, 1, 0)
code_source = StringSourceDescriptor(name, code)
context = StringParseContext([], name)
scope = context.find_module(module_name, pos = initial_pos, need_pxd = 0)
buf = StringIO(code.encode(encoding))
scanner = PyrexScanner(buf, code_source, source_encoding = encoding,
scope = scope, context = context, initial_pos = initial_pos)
if level is None:
tree = Parsing.p_module(scanner, 0, module_name)
else:
tree = Parsing.p_code(scanner, level=level)
return tree
class TreeCopier(VisitorTransform):
def visit_Node(self, node):
if node is None:
return node
else:
c = node.clone_node()
self.visitchildren(c)
return c
class ApplyPositionAndCopy(TreeCopier):
def __init__(self, pos):
super(ApplyPositionAndCopy, self).__init__()
self.pos = pos
def visit_Node(self, node):
copy = super(ApplyPositionAndCopy, self).visit_Node(node)
copy.pos = self.pos
return copy
class TemplateTransform(VisitorTransform):
"""
Makes a copy of a template tree while doing substitutions.
A dictionary "substitutions" should be passed in when calling
the transform; mapping names to replacement nodes. Then replacement
happens like this:
- If an ExprStatNode contains a single NameNode, whose name is
a key in the substitutions dictionary, the ExprStatNode is
replaced with a copy of the tree given in the dictionary.
It is the responsibility of the caller that the replacement
node is a valid statement.
- If a single NameNode is otherwise encountered, it is replaced
if its name is listed in the substitutions dictionary in the
same way. It is the responsibility of the caller to make sure
that the replacement nodes is a valid expression.
Also a list "temps" should be passed. Any names listed will
be transformed into anonymous, temporary names.
Currently supported for tempnames is:
NameNode
(various function and class definition nodes etc. should be added to this)
Each replacement node gets the position of the substituted node
recursively applied to every member node.
"""
def __call__(self, node, substitutions, temps, pos):
self.substitutions = substitutions
self.pos = pos
tempmap = {}
temphandles = []
for temp in temps:
handle = UtilNodes.TempHandle(PyrexTypes.py_object_type)
tempmap[temp] = handle
temphandles.append(handle)
self.tempmap = tempmap
result = super(TemplateTransform, self).__call__(node)
if temps:
result = UtilNodes.TempsBlockNode(self.get_pos(node),
temps=temphandles,
body=result)
return result
def get_pos(self, node):
if self.pos:
return self.pos
else:
return node.pos
def visit_Node(self, node):
if node is None:
return None
else:
c = node.clone_node()
if self.pos is not None:
c.pos = self.pos
self.visitchildren(c)
return c
def try_substitution(self, node, key):
sub = self.substitutions.get(key)
if sub is not None:
pos = self.pos
if pos is None: pos = node.pos
return ApplyPositionAndCopy(pos)(sub)
else:
return self.visit_Node(node) # make copy as usual
def visit_NameNode(self, node):
temphandle = self.tempmap.get(node.name)
if temphandle:
# Replace name with temporary
return temphandle.ref(self.get_pos(node))
else:
return self.try_substitution(node, node.name)
def visit_ExprStatNode(self, node):
# If an expression-as-statement consists of only a replaceable
# NameNode, we replace the entire statement, not only the NameNode
if isinstance(node.expr, NameNode):
return self.try_substitution(node, node.expr.name)
else:
return self.visit_Node(node)
def copy_code_tree(node):
return TreeCopier()(node)
INDENT_RE = re.compile(ur"^ *")
def strip_common_indent(lines):
"Strips empty lines and common indentation from the list of strings given in lines"
# TODO: Facilitate textwrap.indent instead
lines = [x for x in lines if x.strip() != u""]
minindent = min([len(INDENT_RE.match(x).group(0)) for x in lines])
lines = [x[minindent:] for x in lines]
return lines
class TreeFragment(object):
def __init__(self, code, name="(tree fragment)", pxds={}, temps=[], pipeline=[], level=None, initial_pos=None):
if isinstance(code, unicode):
def fmt(x): return u"\n".join(strip_common_indent(x.split(u"\n")))
fmt_code = fmt(code)
fmt_pxds = {}
for key, value in pxds.iteritems():
fmt_pxds[key] = fmt(value)
mod = t = parse_from_strings(name, fmt_code, fmt_pxds, level=level, initial_pos=initial_pos)
if level is None:
t = t.body # Make sure a StatListNode is at the top
if not isinstance(t, StatListNode):
t = StatListNode(pos=mod.pos, stats=[t])
for transform in pipeline:
t = transform(t)
self.root = t
elif isinstance(code, Node):
if pxds != {}: raise NotImplementedError()
self.root = code
else:
raise ValueError("Unrecognized code format (accepts unicode and Node)")
self.temps = temps
def copy(self):
return copy_code_tree(self.root)
def substitute(self, nodes={}, temps=[], pos = None):
return TemplateTransform()(self.root,
substitutions = nodes,
temps = self.temps + temps, pos = pos)
|
