• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
1:mod:`ast` --- Abstract Syntax Trees
2====================================
3
4.. module:: ast
5   :synopsis: Abstract Syntax Tree classes and manipulation.
6
7.. sectionauthor:: Martin v. Löwis <martin@v.loewis.de>
8.. sectionauthor:: Georg Brandl <georg@python.org>
9
10**Source code:** :source:`Lib/ast.py`
11
12--------------
13
14The :mod:`ast` module helps Python applications to process trees of the Python
15abstract syntax grammar.  The abstract syntax itself might change with each
16Python release; this module helps to find out programmatically what the current
17grammar looks like.
18
19An abstract syntax tree can be generated by passing :data:`ast.PyCF_ONLY_AST` as
20a flag to the :func:`compile` built-in function, or using the :func:`parse`
21helper provided in this module.  The result will be a tree of objects whose
22classes all inherit from :class:`ast.AST`.  An abstract syntax tree can be
23compiled into a Python code object using the built-in :func:`compile` function.
24
25
26Node classes
27------------
28
29.. class:: AST
30
31   This is the base of all AST node classes.  The actual node classes are
32   derived from the :file:`Parser/Python.asdl` file, which is reproduced
33   :ref:`below <abstract-grammar>`.  They are defined in the :mod:`_ast` C
34   module and re-exported in :mod:`ast`.
35
36   There is one class defined for each left-hand side symbol in the abstract
37   grammar (for example, :class:`ast.stmt` or :class:`ast.expr`).  In addition,
38   there is one class defined for each constructor on the right-hand side; these
39   classes inherit from the classes for the left-hand side trees.  For example,
40   :class:`ast.BinOp` inherits from :class:`ast.expr`.  For production rules
41   with alternatives (aka "sums"), the left-hand side class is abstract: only
42   instances of specific constructor nodes are ever created.
43
44   .. index:: single: ? (question mark); in AST grammar
45   .. index:: single: * (asterisk); in AST grammar
46
47   .. attribute:: _fields
48
49      Each concrete class has an attribute :attr:`_fields` which gives the names
50      of all child nodes.
51
52      Each instance of a concrete class has one attribute for each child node,
53      of the type as defined in the grammar.  For example, :class:`ast.BinOp`
54      instances have an attribute :attr:`left` of type :class:`ast.expr`.
55
56      If these attributes are marked as optional in the grammar (using a
57      question mark), the value might be ``None``.  If the attributes can have
58      zero-or-more values (marked with an asterisk), the values are represented
59      as Python lists.  All possible attributes must be present and have valid
60      values when compiling an AST with :func:`compile`.
61
62   .. attribute:: lineno
63                  col_offset
64                  end_lineno
65                  end_col_offset
66
67      Instances of :class:`ast.expr` and :class:`ast.stmt` subclasses have
68      :attr:`lineno`, :attr:`col_offset`, :attr:`lineno`, and :attr:`col_offset`
69      attributes.  The :attr:`lineno` and :attr:`end_lineno` are the first and
70      last line numbers of source text span (1-indexed so the first line is line 1)
71      and the :attr:`col_offset` and :attr:`end_col_offset` are the corresponding
72      UTF-8 byte offsets of the first and last tokens that generated the node.
73      The UTF-8 offset is recorded because the parser uses UTF-8 internally.
74
75      Note that the end positions are not required by the compiler and are
76      therefore optional. The end offset is *after* the last symbol, for example
77      one can get the source segment of a one-line expression node using
78      ``source_line[node.col_offset : node.end_col_offset]``.
79
80   The constructor of a class :class:`ast.T` parses its arguments as follows:
81
82   * If there are positional arguments, there must be as many as there are items
83     in :attr:`T._fields`; they will be assigned as attributes of these names.
84   * If there are keyword arguments, they will set the attributes of the same
85     names to the given values.
86
87   For example, to create and populate an :class:`ast.UnaryOp` node, you could
88   use ::
89
90      node = ast.UnaryOp()
91      node.op = ast.USub()
92      node.operand = ast.Constant()
93      node.operand.value = 5
94      node.operand.lineno = 0
95      node.operand.col_offset = 0
96      node.lineno = 0
97      node.col_offset = 0
98
99   or the more compact ::
100
101      node = ast.UnaryOp(ast.USub(), ast.Constant(5, lineno=0, col_offset=0),
102                         lineno=0, col_offset=0)
103
104.. deprecated:: 3.8
105
106   Class :class:`ast.Constant` is now used for all constants. Old classes
107   :class:`ast.Num`, :class:`ast.Str`, :class:`ast.Bytes`,
108   :class:`ast.NameConstant` and :class:`ast.Ellipsis` are still available,
109   but they will be removed in future Python releases.
110
111
112.. _abstract-grammar:
113
114Abstract Grammar
115----------------
116
117The abstract grammar is currently defined as follows:
118
119.. literalinclude:: ../../Parser/Python.asdl
120   :language: none
121
122
123:mod:`ast` Helpers
124------------------
125
126Apart from the node classes, the :mod:`ast` module defines these utility functions
127and classes for traversing abstract syntax trees:
128
129.. function:: parse(source, filename='<unknown>', mode='exec', *, type_comments=False, feature_version=None)
130
131   Parse the source into an AST node.  Equivalent to ``compile(source,
132   filename, mode, ast.PyCF_ONLY_AST)``.
133
134   If ``type_comments=True`` is given, the parser is modified to check
135   and return type comments as specified by :pep:`484` and :pep:`526`.
136   This is equivalent to adding :data:`ast.PyCF_TYPE_COMMENTS` to the
137   flags passed to :func:`compile()`.  This will report syntax errors
138   for misplaced type comments.  Without this flag, type comments will
139   be ignored, and the ``type_comment`` field on selected AST nodes
140   will always be ``None``.  In addition, the locations of ``# type:
141   ignore`` comments will be returned as the ``type_ignores``
142   attribute of :class:`Module` (otherwise it is always an empty list).
143
144   In addition, if ``mode`` is ``'func_type'``, the input syntax is
145   modified to correspond to :pep:`484` "signature type comments",
146   e.g. ``(str, int) -> List[str]``.
147
148   Also, setting ``feature_version`` to a tuple ``(major, minor)``
149   will attempt to parse using that Python version's grammar.
150   Currently ``major`` must equal to ``3``.  For example, setting
151   ``feature_version=(3, 4)`` will allow the use of ``async`` and
152   ``await`` as variable names.  The lowest supported version is
153   ``(3, 4)``; the highest is ``sys.version_info[0:2]``.
154
155   .. warning::
156      It is possible to crash the Python interpreter with a
157      sufficiently large/complex string due to stack depth limitations
158      in Python's AST compiler.
159
160   .. versionchanged:: 3.8
161      Added ``type_comments``, ``mode='func_type'`` and ``feature_version``.
162
163
164.. function:: literal_eval(node_or_string)
165
166   Safely evaluate an expression node or a string containing a Python literal or
167   container display.  The string or node provided may only consist of the
168   following Python literal structures: strings, bytes, numbers, tuples, lists,
169   dicts, sets, booleans, and ``None``.
170
171   This can be used for safely evaluating strings containing Python values from
172   untrusted sources without the need to parse the values oneself.  It is not
173   capable of evaluating arbitrarily complex expressions, for example involving
174   operators or indexing.
175
176   .. warning::
177      It is possible to crash the Python interpreter with a
178      sufficiently large/complex string due to stack depth limitations
179      in Python's AST compiler.
180
181   .. versionchanged:: 3.2
182      Now allows bytes and set literals.
183
184
185.. function:: get_docstring(node, clean=True)
186
187   Return the docstring of the given *node* (which must be a
188   :class:`FunctionDef`, :class:`AsyncFunctionDef`, :class:`ClassDef`,
189   or :class:`Module` node), or ``None`` if it has no docstring.
190   If *clean* is true, clean up the docstring's indentation with
191   :func:`inspect.cleandoc`.
192
193   .. versionchanged:: 3.5
194      :class:`AsyncFunctionDef` is now supported.
195
196
197.. function:: get_source_segment(source, node, *, padded=False)
198
199   Get source code segment of the *source* that generated *node*.
200   If some location information (:attr:`lineno`, :attr:`end_lineno`,
201   :attr:`col_offset`, or :attr:`end_col_offset`) is missing, return ``None``.
202
203   If *padded* is ``True``, the first line of a multi-line statement will
204   be padded with spaces to match its original position.
205
206   .. versionadded:: 3.8
207
208
209.. function:: fix_missing_locations(node)
210
211   When you compile a node tree with :func:`compile`, the compiler expects
212   :attr:`lineno` and :attr:`col_offset` attributes for every node that supports
213   them.  This is rather tedious to fill in for generated nodes, so this helper
214   adds these attributes recursively where not already set, by setting them to
215   the values of the parent node.  It works recursively starting at *node*.
216
217
218.. function:: increment_lineno(node, n=1)
219
220   Increment the line number and end line number of each node in the tree
221   starting at *node* by *n*. This is useful to "move code" to a different
222   location in a file.
223
224
225.. function:: copy_location(new_node, old_node)
226
227   Copy source location (:attr:`lineno`, :attr:`col_offset`, :attr:`end_lineno`,
228   and :attr:`end_col_offset`) from *old_node* to *new_node* if possible,
229   and return *new_node*.
230
231
232.. function:: iter_fields(node)
233
234   Yield a tuple of ``(fieldname, value)`` for each field in ``node._fields``
235   that is present on *node*.
236
237
238.. function:: iter_child_nodes(node)
239
240   Yield all direct child nodes of *node*, that is, all fields that are nodes
241   and all items of fields that are lists of nodes.
242
243
244.. function:: walk(node)
245
246   Recursively yield all descendant nodes in the tree starting at *node*
247   (including *node* itself), in no specified order.  This is useful if you only
248   want to modify nodes in place and don't care about the context.
249
250
251.. class:: NodeVisitor()
252
253   A node visitor base class that walks the abstract syntax tree and calls a
254   visitor function for every node found.  This function may return a value
255   which is forwarded by the :meth:`visit` method.
256
257   This class is meant to be subclassed, with the subclass adding visitor
258   methods.
259
260   .. method:: visit(node)
261
262      Visit a node.  The default implementation calls the method called
263      :samp:`self.visit_{classname}` where *classname* is the name of the node
264      class, or :meth:`generic_visit` if that method doesn't exist.
265
266   .. method:: generic_visit(node)
267
268      This visitor calls :meth:`visit` on all children of the node.
269
270      Note that child nodes of nodes that have a custom visitor method won't be
271      visited unless the visitor calls :meth:`generic_visit` or visits them
272      itself.
273
274   Don't use the :class:`NodeVisitor` if you want to apply changes to nodes
275   during traversal.  For this a special visitor exists
276   (:class:`NodeTransformer`) that allows modifications.
277
278   .. deprecated:: 3.8
279
280      Methods :meth:`visit_Num`, :meth:`visit_Str`, :meth:`visit_Bytes`,
281      :meth:`visit_NameConstant` and :meth:`visit_Ellipsis` are deprecated
282      now and will not be called in future Python versions.  Add the
283      :meth:`visit_Constant` method to handle all constant nodes.
284
285
286.. class:: NodeTransformer()
287
288   A :class:`NodeVisitor` subclass that walks the abstract syntax tree and
289   allows modification of nodes.
290
291   The :class:`NodeTransformer` will walk the AST and use the return value of
292   the visitor methods to replace or remove the old node.  If the return value
293   of the visitor method is ``None``, the node will be removed from its
294   location, otherwise it is replaced with the return value.  The return value
295   may be the original node in which case no replacement takes place.
296
297   Here is an example transformer that rewrites all occurrences of name lookups
298   (``foo``) to ``data['foo']``::
299
300      class RewriteName(NodeTransformer):
301
302          def visit_Name(self, node):
303              return copy_location(Subscript(
304                  value=Name(id='data', ctx=Load()),
305                  slice=Index(value=Constant(value=node.id)),
306                  ctx=node.ctx
307              ), node)
308
309   Keep in mind that if the node you're operating on has child nodes you must
310   either transform the child nodes yourself or call the :meth:`generic_visit`
311   method for the node first.
312
313   For nodes that were part of a collection of statements (that applies to all
314   statement nodes), the visitor may also return a list of nodes rather than
315   just a single node.
316
317   Usually you use the transformer like this::
318
319      node = YourTransformer().visit(node)
320
321
322.. function:: dump(node, annotate_fields=True, include_attributes=False)
323
324   Return a formatted dump of the tree in *node*.  This is mainly useful for
325   debugging purposes.  If *annotate_fields* is true (by default),
326   the returned string will show the names and the values for fields.
327   If *annotate_fields* is false, the result string will be more compact by
328   omitting unambiguous field names.  Attributes such as line
329   numbers and column offsets are not dumped by default.  If this is wanted,
330   *include_attributes* can be set to true.
331
332.. seealso::
333
334    `Green Tree Snakes <https://greentreesnakes.readthedocs.io/>`_, an external documentation resource, has good
335    details on working with Python ASTs.
336