Source code for SCons.Environment

# MIT License
#
# Copyright The SCons Foundation
#
# Permission is hereby granted, free of charge, to any person obtaining
# a copy of this software and associated documentation files (the
# "Software"), to deal in the Software without restriction, including
# without limitation the rights to use, copy, modify, merge, publish,
# distribute, sublicense, and/or sell copies of the Software, and to
# permit persons to whom the Software is furnished to do so, subject to
# the following conditions:
#
# The above copyright notice and this permission notice shall be included
# in all copies or substantial portions of the Software.
#
# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY
# KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE
# WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
# NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE
# LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
# OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
# WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.

"""Base class for construction Environments.

These are the primary objects used to communicate dependency and
construction information to the build engine.

Keyword arguments supplied when the construction Environment is created
are construction variables used to initialize the Environment.
"""

import copy
import os
import sys
import re
import shlex
from collections import UserDict, deque
from subprocess import PIPE, DEVNULL
from typing import Optional

import SCons.Action
import SCons.Builder
import SCons.Debug
from SCons.Debug import logInstanceCreation
import SCons.Defaults
from SCons.Errors import UserError, BuildError
import SCons.Memoize
import SCons.Node
import SCons.Node.Alias
import SCons.Node.FS
import SCons.Node.Python
import SCons.Platform
import SCons.SConf
import SCons.SConsign
import SCons.Subst
import SCons.Tool
import SCons.Warnings
from SCons.Util import (
    AppendPath,
    CLVar,
    LogicalLines,
    MethodWrapper,
    PrependPath,
    Split,
    WhereIs,
    flatten,
    is_Dict,
    is_List,
    is_Scalar,
    is_Sequence,
    is_String,
    is_Tuple,
    semi_deepcopy,
    semi_deepcopy_dict,
    to_String_for_subst,
    uniquer_hashables,
)

[docs]class _Null: pass
_null = _Null _warn_copy_deprecated = True _warn_source_signatures_deprecated = True _warn_target_signatures_deprecated = True CleanTargets = {} CalculatorArgs = {}
[docs]def alias_builder(env, target, source) -> None: pass
AliasBuilder = SCons.Builder.Builder( action=alias_builder, target_factory=SCons.Node.Alias.default_ans.Alias, source_factory=SCons.Node.FS.Entry, multi=True, is_explicit=None, name='AliasBuilder', )
[docs]def apply_tools(env, tools, toolpath) -> None: # Store the toolpath in the Environment. # This is expected to work even if no tools are given, so do this first. if toolpath is not None: env['toolpath'] = toolpath if not tools: return # Filter out null tools from the list. for tool in [_f for _f in tools if _f]: if is_List(tool) or is_Tuple(tool): # toolargs should be a dict of kw args toolname, toolargs, *rest = tool _ = env.Tool(toolname, **toolargs) else: _ = env.Tool(tool)
# These names are (or will be) controlled by SCons; users should never # set or override them. The warning can optionally be turned off, # but scons will still ignore the illegal variable names even if it's off. reserved_construction_var_names = [ 'CHANGED_SOURCES', 'CHANGED_TARGETS', 'SOURCE', 'SOURCES', 'TARGET', 'TARGETS', 'UNCHANGED_SOURCES', 'UNCHANGED_TARGETS', ] future_reserved_construction_var_names = [ #'HOST_OS', #'HOST_ARCH', #'HOST_CPU', ]
[docs]def copy_non_reserved_keywords(dict): result = semi_deepcopy(dict) for k in result.copy().keys(): if k in reserved_construction_var_names: msg = "Ignoring attempt to set reserved variable `$%s'" SCons.Warnings.warn(SCons.Warnings.ReservedVariableWarning, msg % k) del result[k] return result
[docs]def _set_reserved(env, key, value) -> None: msg = "Ignoring attempt to set reserved variable `$%s'" SCons.Warnings.warn(SCons.Warnings.ReservedVariableWarning, msg % key)
[docs]def _set_future_reserved(env, key, value) -> None: env._dict[key] = value msg = "`$%s' will be reserved in a future release and setting it will become ignored" SCons.Warnings.warn(SCons.Warnings.FutureReservedVariableWarning, msg % key)
[docs]def _set_BUILDERS(env, key, value): try: bd = env._dict[key] for k in bd.copy().keys(): del bd[k] except KeyError: bd = BuilderDict(bd, env) env._dict[key] = bd for k, v in value.items(): if not SCons.Builder.is_a_Builder(v): raise UserError('%s is not a Builder.' % repr(v)) bd.update(value)
[docs]def _del_SCANNERS(env, key) -> None: del env._dict[key] env.scanner_map_delete()
[docs]def _set_SCANNERS(env, key, value) -> None: env._dict[key] = value env.scanner_map_delete()
[docs]def _delete_duplicates(l, keep_last): """Delete duplicates from a sequence, keeping the first or last.""" seen=set() result=[] if keep_last: # reverse in & out, then keep first l.reverse() for i in l: try: if i not in seen: result.append(i) seen.add(i) except TypeError: # probably unhashable. Just keep it. result.append(i) if keep_last: result.reverse() return result
[docs]def _add_cppdefines( env_dict: dict, val, # add annotation? prepend: bool = False, unique: bool = False, delete_existing: bool = False, ) -> None: """Adds to ``CPPDEFINES``, using the rules for C preprocessor macros. This is split out from regular construction variable addition because these entries can express either a macro with a replacement value or one without. A macro with replacement value can be supplied as *val* in three ways: as a combined string ``"name=value"``; as a tuple ``(name, value)``, or as an entry in a dictionary ``{"name": value}``. A list argument with multiple macros can also be given. Additions can be unconditional (duplicates allowed) or uniquing (no dupes). Note if a replacement value is supplied, *unique* requires a full match to decide uniqueness - both the macro name and the replacement. The inner :func:`_is_in` is used to figure that out. Args: env_dict: the dictionary containing the ``CPPDEFINES`` to be modified. val: the value to add, can be string, sequence or dict prepend: whether to put *val* in front or back. unique: whether to add *val* if it already exists. delete_existing: if *unique* is true, add *val* after removing previous. .. versionadded:: 4.5.0 """ def _add_define(item, defines: deque, prepend: bool = False) -> None: """Convenience function to prepend/append a single value. Sole purpose is to shorten code in the outer function. """ if prepend: defines.appendleft(item) else: defines.append(item) def _is_in(item, defines: deque): """Returns match for *item* if found in *defines*. Accounts for type differences: tuple ("FOO", "BAR"), list ["FOO", "BAR"], string "FOO=BAR" and dict {"FOO": "BAR"} all differ as far as Python equality comparison is concerned, but are the same for purposes of creating the preprocessor macro. Also an unvalued string should match something like ``("FOO", None)``. Since the caller may wish to remove a matched entry, we need to return it - cannot remove *item* itself unless it happened to be an exact (type) match. Called from a place we know *defines* is always a deque, and *item* will not be a dict, so don't need to do much type checking. If this ends up used more generally, would need to adjust that. Note implied assumption that members of a list-valued define will not be dicts - we cannot actually guarantee this, since if the initial add is a list its contents are not converted. """ def _macro_conv(v) -> list: """Normalizes a macro to a list for comparisons.""" if is_Tuple(v): return list(v) elif is_String(v): rv = v.split("=") if len(rv) == 1: return [v, None] return rv return v if item in defines: # cheap check first return item item = _macro_conv(item) for define in defines: if item == _macro_conv(define): return define return False key = 'CPPDEFINES' try: defines = env_dict[key] except KeyError: # This is a new entry, just save it as is. Defer conversion to # preferred type until someone tries to amend the value. # processDefines has no problem with unconverted values if it # gets called without any later additions. if is_String(val): env_dict[key] = val.split() else: env_dict[key] = val return # Convert type of existing to deque (if necessary) to simplify processing # of additions - inserting at either end is cheap. Deferred conversion # is also useful in case CPPDEFINES was set initially without calling # through here (e.g. Environment kwarg, or direct assignment). if isinstance(defines, deque): # Already a deque? do nothing. Explicit check is so we don't get # picked up by the is_list case below. pass elif is_String(defines): env_dict[key] = deque(defines.split()) elif is_Tuple(defines): if len(defines) > 2: raise SCons.Errors.UserError( f"Invalid tuple in CPPDEFINES: {defines!r}, must be a two-tuple" ) env_dict[key] = deque([defines]) elif is_List(defines): # a little extra work in case the initial container has dict # item(s) inside it, so those can be matched by _is_in(). result = deque() for define in defines: if is_Dict(define): result.extend(define.items()) else: result.append(define) env_dict[key] = result elif is_Dict(defines): env_dict[key] = deque(defines.items()) else: env_dict[key] = deque(defines) defines = env_dict[key] # in case we reassigned due to conversion # now actually do the addition. if is_Dict(val): # Unpack the dict while applying to existing for item in val.items(): if unique: match = _is_in(item, defines) if match and delete_existing: defines.remove(match) _add_define(item, defines, prepend) elif not match: _add_define(item, defines, prepend) else: _add_define(item, defines, prepend) elif is_String(val): for v in val.split(): if unique: match = _is_in(v, defines) if match and delete_existing: defines.remove(match) _add_define(v, defines, prepend) elif not match: _add_define(v, defines, prepend) else: _add_define(v, defines, prepend) # A tuple appended to anything should yield -Dkey=value elif is_Tuple(val): if len(val) > 2: raise SCons.Errors.UserError( f"Invalid tuple added to CPPDEFINES: {val!r}, " "must be a two-tuple" ) if len(val) == 1: val = (val[0], None) # normalize if not is_Scalar(val[0]) or not is_Scalar(val[1]): raise SCons.Errors.UserError( f"Invalid tuple added to CPPDEFINES: {val!r}, " "values must be scalar" ) if unique: match = _is_in(val, defines) if match and delete_existing: defines.remove(match) _add_define(val, defines, prepend) elif not match: _add_define(val, defines, prepend) else: _add_define(val, defines, prepend) elif is_List(val): tmp = [] for item in val: if unique: match = _is_in(item, defines) if match and delete_existing: defines.remove(match) tmp.append(item) elif not match: tmp.append(item) else: tmp.append(item) if prepend: defines.extendleft(tmp) else: defines.extend(tmp)
# else: # are there any other cases? processDefines doesn't think so. # The following is partly based on code in a comment added by Peter # Shannon at the following page (there called the "transplant" class): # # ASPN : Python Cookbook : Dynamically added methods to a class # https://code.activestate.com/recipes/81732/ # # We had independently been using the idiom as BuilderWrapper, but # factoring out the common parts into this base class, and making # BuilderWrapper a subclass that overrides __call__() to enforce specific # Builder calling conventions, simplified some of our higher-layer code. # # Note: MethodWrapper moved to SCons.Util as it was needed there # and otherwise we had a circular import problem.
[docs]class BuilderWrapper(MethodWrapper): """ A MethodWrapper subclass that that associates an environment with a Builder. This mainly exists to wrap the __call__() function so that all calls to Builders can have their argument lists massaged in the same way (treat a lone argument as the source, treat two arguments as target then source, make sure both target and source are lists) without having to have cut-and-paste code to do it. As a bit of obsessive backwards compatibility, we also intercept attempts to get or set the "env" or "builder" attributes, which were the names we used before we put the common functionality into the MethodWrapper base class. We'll keep this around for a while in case people shipped Tool modules that reached into the wrapper (like the Tool/qt.py module does, or did). There shouldn't be a lot attribute fetching or setting on these, so a little extra work shouldn't hurt. """ def __call__(self, target=None, source=_null, *args, **kw): if source is _null: source = target target = None if target is not None and not is_List(target): target = [target] if source is not None and not is_List(source): source = [source] return super().__call__(target, source, *args, **kw) def __repr__(self) -> str: return '<BuilderWrapper %s>' % repr(self.name) def __str__(self) -> str: return self.__repr__() def __getattr__(self, name): if name == 'env': return self.object elif name == 'builder': return self.method else: raise AttributeError(name) def __setattr__(self, name, value) -> None: if name == 'env': self.object = value elif name == 'builder': self.method = value else: self.__dict__[name] = value
# This allows a Builder to be executed directly # through the Environment to which it's attached. # In practice, we shouldn't need this, because # builders actually get executed through a Node. # But we do have a unit test for this, and can't # yet rule out that it would be useful in the # future, so leave it for now. #def execute(self, **kw): # kw['env'] = self.env # self.builder.execute(**kw)
[docs]class BuilderDict(UserDict): """This is a dictionary-like class used by an Environment to hold the Builders. We need to do this because every time someone changes the Builders in the Environment's BUILDERS dictionary, we must update the Environment's attributes.""" def __init__(self, mapping, env) -> None: # Set self.env before calling the superclass initialization, # because it will end up calling our other methods, which will # need to point the values in this dictionary to self.env. self.env = env super().__init__(mapping) def __semi_deepcopy__(self): # These cannot be copied since they would both modify the same builder object, and indeed # just copying would modify the original builder raise TypeError( 'cannot semi_deepcopy a BuilderDict' ) def __setitem__(self, item, val) -> None: try: method = getattr(self.env, item).method except AttributeError: pass else: self.env.RemoveMethod(method) super().__setitem__(item, val) BuilderWrapper(self.env, val, item) def __delitem__(self, item) -> None: super().__delitem__(item) delattr(self.env, item)
[docs] def update(self, mapping) -> None: for i, v in mapping.items(): self.__setitem__(i, v)
_is_valid_var = re.compile(r'[_a-zA-Z]\w*$')
[docs]def is_valid_construction_var(varstr) -> bool: """Return True if *varstr* is a legitimate construction variable.""" return _is_valid_var.match(varstr)
[docs]class SubstitutionEnvironment: """Base class for different flavors of construction environments. This class contains a minimal set of methods that handle construction variable expansion and conversion of strings to Nodes, which may or may not be actually useful as a stand-alone class. Which methods ended up in this class is pretty arbitrary right now. They're basically the ones which we've empirically determined are common to the different construction environment subclasses, and most of the others that use or touch the underlying dictionary of construction variables. Eventually, this class should contain all the methods that we determine are necessary for a "minimal" interface to the build engine. A full "native Python" SCons environment has gotten pretty heavyweight with all of the methods and Tools and construction variables we've jammed in there, so it would be nice to have a lighter weight alternative for interfaces that don't need all of the bells and whistles. (At some point, we'll also probably rename this class "Base," since that more reflects what we want this class to become, but because we've released comments that tell people to subclass Environment.Base to create their own flavors of construction environment, we'll save that for a future refactoring when this class actually becomes useful.) """ def __init__(self, **kw) -> None: """Initialization of an underlying SubstitutionEnvironment class. """ if SCons.Debug.track_instances: logInstanceCreation(self, 'Environment.SubstitutionEnvironment') self.fs = SCons.Node.FS.get_default_fs() self.ans = SCons.Node.Alias.default_ans self.lookup_list = SCons.Node.arg2nodes_lookups self._dict = kw.copy() self._init_special() self.added_methods = [] #self._memo = {}
[docs] def _init_special(self) -> None: """Initial the dispatch tables for special handling of special construction variables.""" self._special_del = {} self._special_del['SCANNERS'] = _del_SCANNERS self._special_set = {} for key in reserved_construction_var_names: self._special_set[key] = _set_reserved for key in future_reserved_construction_var_names: self._special_set[key] = _set_future_reserved self._special_set['BUILDERS'] = _set_BUILDERS self._special_set['SCANNERS'] = _set_SCANNERS # Freeze the keys of self._special_set in a list for use by # methods that need to check. self._special_set_keys = list(self._special_set.keys())
def __eq__(self, other): return self._dict == other._dict def __delitem__(self, key) -> None: special = self._special_del.get(key) if special: special(self, key) else: del self._dict[key] def __getitem__(self, key): return self._dict[key] def __setitem__(self, key, value): # This is heavily used. This implementation is the best we have # according to the timings in bench/env.__setitem__.py. # # The "key in self._special_set_keys" test here seems to perform # pretty well for the number of keys we have. A hard-coded # list worked a little better in Python 2.5, but that has the # disadvantage of maybe getting out of sync if we ever add more # variable names. # So right now it seems like a good trade-off, but feel free to # revisit this with bench/env.__setitem__.py as needed (and # as newer versions of Python come out). if key in self._special_set_keys: self._special_set[key](self, key, value) else: # If we already have the entry, then it's obviously a valid # key and we don't need to check. If we do check, using a # global, pre-compiled regular expression directly is more # efficient than calling another function or a method. if key not in self._dict and not _is_valid_var.match(key): raise UserError("Illegal construction variable `%s'" % key) self._dict[key] = value
[docs] def get(self, key, default=None): """Emulates the get() method of dictionaries.""" return self._dict.get(key, default)
def __contains__(self, key) -> bool: return key in self._dict
[docs] def keys(self): """Emulates the keys() method of dictionaries.""" return self._dict.keys()
[docs] def values(self): """Emulates the values() method of dictionaries.""" return self._dict.values()
[docs] def items(self): """Emulates the items() method of dictionaries.""" return self._dict.items()
[docs] def setdefault(self, key, default=None): """Emulates the setdefault() method of dictionaries.""" return self._dict.setdefault(key, default)
[docs] def arg2nodes(self, args, node_factory=_null, lookup_list=_null, **kw): if node_factory is _null: node_factory = self.fs.File if lookup_list is _null: lookup_list = self.lookup_list if not args: return [] args = flatten(args) nodes = [] for v in args: if is_String(v): n = None for l in lookup_list: n = l(v) if n is not None: break if n is not None: if is_String(n): # n = self.subst(n, raw=1, **kw) kw['raw'] = 1 n = self.subst(n, **kw) if node_factory: n = node_factory(n) if is_List(n): nodes.extend(n) else: nodes.append(n) elif node_factory: # v = node_factory(self.subst(v, raw=1, **kw)) kw['raw'] = 1 v = node_factory(self.subst(v, **kw)) if is_List(v): nodes.extend(v) else: nodes.append(v) else: nodes.append(v) return nodes
[docs] def gvars(self): return self._dict
[docs] def lvars(self): return {}
[docs] def subst(self, string, raw: int=0, target=None, source=None, conv=None, executor=None, overrides: Optional[dict] = None): """Recursively interpolates construction variables from the Environment into the specified string, returning the expanded result. Construction variables are specified by a $ prefix in the string and begin with an initial underscore or alphabetic character followed by any number of underscores or alphanumeric characters. The construction variable names may be surrounded by curly braces to separate the name from trailing characters. """ gvars = self.gvars() lvars = self.lvars() lvars['__env__'] = self if executor: lvars.update(executor.get_lvars()) return SCons.Subst.scons_subst(string, self, raw, target, source, gvars, lvars, conv, overrides=overrides)
[docs] def subst_kw(self, kw, raw: int=0, target=None, source=None): nkw = {} for k, v in kw.items(): k = self.subst(k, raw, target, source) if is_String(v): v = self.subst(v, raw, target, source) nkw[k] = v return nkw
[docs] def subst_list(self, string, raw: int=0, target=None, source=None, conv=None, executor=None, overrides: Optional[dict] = None): """Calls through to SCons.Subst.scons_subst_list(). See the documentation for that function. """ gvars = self.gvars() lvars = self.lvars() lvars['__env__'] = self if executor: lvars.update(executor.get_lvars()) return SCons.Subst.scons_subst_list(string, self, raw, target, source, gvars, lvars, conv, overrides=overrides)
[docs] def subst_path(self, path, target=None, source=None): """Substitute a path list. Turns EntryProxies into Nodes, leaving Nodes (and other objects) as-is. """ if not is_List(path): path = [path] def s(obj): """This is the "string conversion" routine that we have our substitutions use to return Nodes, not strings. This relies on the fact that an EntryProxy object has a get() method that returns the underlying Node that it wraps, which is a bit of architectural dependence that we might need to break or modify in the future in response to additional requirements.""" try: get = obj.get except AttributeError: obj = to_String_for_subst(obj) else: obj = get() return obj r = [] for p in path: if is_String(p): p = self.subst(p, target=target, source=source, conv=s) if is_List(p): if len(p) == 1: p = p[0] else: # We have an object plus a string, or multiple # objects that we need to smush together. No choice # but to make them into a string. p = ''.join(map(to_String_for_subst, p)) else: p = s(p) r.append(p) return r
subst_target_source = subst
[docs] def backtick(self, command) -> str: """Emulate command substitution. Provides behavior conceptually like POSIX Shell notation for running a command in backquotes (backticks) by running ``command`` and returning the resulting output string. This is not really a public API any longer, it is provided for the use of :meth:`ParseFlags` (which supports it using a syntax of !command) and :meth:`ParseConfig`. Raises: OSError: if the external command returned non-zero exit status. """ # common arguments kw = { "stdin": DEVNULL, "stdout": PIPE, "stderr": PIPE, "universal_newlines": True, } # if the command is a list, assume it's been quoted # othewise force a shell if not is_List(command): kw["shell"] = True # run constructed command cp = SCons.Action.scons_subproc_run(self, command, **kw) if cp.stderr: sys.stderr.write(cp.stderr) if cp.returncode: raise OSError(f'{command!r} exited {cp.returncode}') return cp.stdout
[docs] def AddMethod(self, function, name=None) -> None: """ Adds the specified function as a method of this construction environment with the specified name. If the name is omitted, the default name is the name of the function itself. """ method = MethodWrapper(self, function, name) self.added_methods.append(method)
[docs] def RemoveMethod(self, function) -> None: """ Removes the specified function's MethodWrapper from the added_methods list, so we don't re-bind it when making a clone. """ self.added_methods = [dm for dm in self.added_methods if dm.method is not function]
[docs] def Override(self, overrides): """ Produce a modified environment whose variables are overridden by the overrides dictionaries. "overrides" is a dictionary that will override the variables of this environment. This function is much more efficient than Clone() or creating a new Environment because it doesn't copy the construction environment dictionary, it just wraps the underlying construction environment, and doesn't even create a wrapper object if there are no overrides. """ if not overrides: return self o = copy_non_reserved_keywords(overrides) if not o: return self overrides = {} merges = None for key, value in o.items(): if key == 'parse_flags': merges = value else: overrides[key] = SCons.Subst.scons_subst_once(value, self, key) env = OverrideEnvironment(self, overrides) if merges: env.MergeFlags(merges) return env
[docs] def ParseFlags(self, *flags) -> dict: """Return a dict of parsed flags. Parse ``flags`` and return a dict with the flags distributed into the appropriate construction variable names. The flags are treated as a typical set of command-line flags for a GNU-style toolchain, such as might have been generated by one of the {foo}-config scripts, and used to populate the entries based on knowledge embedded in this method - the choices are not expected to be portable to other toolchains. If one of the ``flags`` strings begins with a bang (exclamation mark), it is assumed to be a command and the rest of the string is executed; the result of that evaluation is then added to the dict. """ mapping = { 'ASFLAGS' : CLVar(''), 'CFLAGS' : CLVar(''), 'CCFLAGS' : CLVar(''), 'CXXFLAGS' : CLVar(''), 'CPPDEFINES' : [], 'CPPFLAGS' : CLVar(''), 'CPPPATH' : [], 'FRAMEWORKPATH' : CLVar(''), 'FRAMEWORKS' : CLVar(''), 'LIBPATH' : [], 'LIBS' : [], 'LINKFLAGS' : CLVar(''), 'RPATH' : [], } def do_parse(arg) -> None: # if arg is a sequence, recurse with each element if not arg: return if not is_String(arg): for t in arg: do_parse(t) return # if arg is a command, execute it if arg[0] == '!': arg = self.backtick(arg[1:]) # utility function to deal with -D option def append_define(name, mapping=mapping) -> None: t = name.split('=') if len(t) == 1: mapping['CPPDEFINES'].append(name) else: mapping['CPPDEFINES'].append([t[0], '='.join(t[1:])]) # Loop through the flags and add them to the appropriate option. # This tries to strike a balance between checking for all possible # flags and keeping the logic to a finite size, so it doesn't # check for some that don't occur often. It particular, if the # flag is not known to occur in a config script and there's a way # of passing the flag to the right place (by wrapping it in a -W # flag, for example) we don't check for it. Note that most # preprocessor options are not handled, since unhandled options # are placed in CCFLAGS, so unless the preprocessor is invoked # separately, these flags will still get to the preprocessor. # Other options not currently handled: # -iqoutedir (preprocessor search path) # -u symbol (linker undefined symbol) # -s (linker strip files) # -static* (linker static binding) # -shared* (linker dynamic binding) # -symbolic (linker global binding) # -R dir (deprecated linker rpath) # IBM compilers may also accept -qframeworkdir=foo params = shlex.split(arg) append_next_arg_to = None # for multi-word args for arg in params: if append_next_arg_to: if append_next_arg_to == 'CPPDEFINES': append_define(arg) elif append_next_arg_to == '-include': t = ('-include', self.fs.File(arg)) mapping['CCFLAGS'].append(t) elif append_next_arg_to == '-imacros': t = ('-imacros', self.fs.File(arg)) mapping['CCFLAGS'].append(t) elif append_next_arg_to == '-isysroot': t = ('-isysroot', arg) mapping['CCFLAGS'].append(t) mapping['LINKFLAGS'].append(t) elif append_next_arg_to == '-isystem': t = ('-isystem', arg) mapping['CCFLAGS'].append(t) elif append_next_arg_to == '-iquote': t = ('-iquote', arg) mapping['CCFLAGS'].append(t) elif append_next_arg_to == '-idirafter': t = ('-idirafter', arg) mapping['CCFLAGS'].append(t) elif append_next_arg_to == '-arch': t = ('-arch', arg) mapping['CCFLAGS'].append(t) mapping['LINKFLAGS'].append(t) elif append_next_arg_to == '--param': t = ('--param', arg) mapping['CCFLAGS'].append(t) else: mapping[append_next_arg_to].append(arg) append_next_arg_to = None elif not arg[0] in ['-', '+']: mapping['LIBS'].append(self.fs.File(arg)) elif arg == '-dylib_file': mapping['LINKFLAGS'].append(arg) append_next_arg_to = 'LINKFLAGS' elif arg[:2] == '-L': if arg[2:]: mapping['LIBPATH'].append(arg[2:]) else: append_next_arg_to = 'LIBPATH' elif arg[:2] == '-l': if arg[2:]: mapping['LIBS'].append(arg[2:]) else: append_next_arg_to = 'LIBS' elif arg[:2] == '-I': if arg[2:]: mapping['CPPPATH'].append(arg[2:]) else: append_next_arg_to = 'CPPPATH' elif arg[:4] == '-Wa,': mapping['ASFLAGS'].append(arg[4:]) mapping['CCFLAGS'].append(arg) elif arg[:4] == '-Wl,': if arg[:11] == '-Wl,-rpath=': mapping['RPATH'].append(arg[11:]) elif arg[:7] == '-Wl,-R,': mapping['RPATH'].append(arg[7:]) elif arg[:6] == '-Wl,-R': mapping['RPATH'].append(arg[6:]) else: mapping['LINKFLAGS'].append(arg) elif arg[:4] == '-Wp,': mapping['CPPFLAGS'].append(arg) elif arg[:2] == '-D': if arg[2:]: append_define(arg[2:]) else: append_next_arg_to = 'CPPDEFINES' elif arg == '-framework': append_next_arg_to = 'FRAMEWORKS' elif arg[:14] == '-frameworkdir=': mapping['FRAMEWORKPATH'].append(arg[14:]) elif arg[:2] == '-F': if arg[2:]: mapping['FRAMEWORKPATH'].append(arg[2:]) else: append_next_arg_to = 'FRAMEWORKPATH' elif arg in ( '-mno-cygwin', '-pthread', '-openmp', '-fmerge-all-constants', '-fopenmp', ) or arg.startswith('-fsanitize'): mapping['CCFLAGS'].append(arg) mapping['LINKFLAGS'].append(arg) elif arg == '-mwindows': mapping['LINKFLAGS'].append(arg) elif arg[:5] == '-std=': if '++' in arg[5:]: key = 'CXXFLAGS' else: key = 'CFLAGS' mapping[key].append(arg) elif arg[0] == '+': mapping['CCFLAGS'].append(arg) mapping['LINKFLAGS'].append(arg) elif arg in [ '-include', '-imacros', '-isysroot', '-isystem', '-iquote', '-idirafter', '-arch', '--param', ]: append_next_arg_to = arg else: mapping['CCFLAGS'].append(arg) for arg in flags: do_parse(arg) return mapping
[docs] def MergeFlags(self, args, unique: bool=True) -> None: """Merge flags into construction variables. Merges the flags from *args* into this construction environent. If *args* is not a dict, it is first converted to one with flags distributed into appropriate construction variables. See :meth:`ParseFlags`. As a side effect, if *unique* is true, a new object is created for each modified construction variable by the loop at the end. This is silently expected by the :meth:`Override` *parse_flags* functionality, which does not want to share the list (or whatever) with the environment being overridden. Args: args: flags to merge unique: merge flags rather than appending (default: True). When merging, path variables are retained from the front, other construction variables from the end. """ if not is_Dict(args): args = self.ParseFlags(args) if not unique: self.Append(**args) return for key, value in args.items(): if not value: continue value = Split(value) try: orig = self[key] except KeyError: orig = value else: if not orig: orig = value elif value: # Add orig and value. The logic here was lifted from # part of env.Append() (see there for a lot of comments # about the order in which things are tried) and is # used mainly to handle coercion of strings to CLVar to # "do the right thing" given (e.g.) an original CCFLAGS # string variable like '-pipe -Wall'. try: orig = orig + value except (KeyError, TypeError): # If CPPDEFINES is a deque, adding value (a list) # results in TypeError, so we handle that case here. # Just in case we got called from Override, make # sure we make a copy, because we don't go through # the cleanup loops at the end of the outer for loop, # which implicitly gives us a new object. if isinstance(orig, deque): self[key] = self[key].copy() self.AppendUnique(CPPDEFINES=value, delete_existing=True) continue try: add_to_orig = orig.append except AttributeError: value.insert(0, orig) orig = value else: add_to_orig(value) t = [] if key[-4:] == 'PATH': ### keep left-most occurence for v in orig: if v not in t: t.append(v) else: ### keep right-most occurence for v in orig[::-1]: if v not in t: t.insert(0, v) self[key] = t
[docs]def default_decide_source(dependency, target, prev_ni, repo_node=None): f = SCons.Defaults.DefaultEnvironment().decide_source return f(dependency, target, prev_ni, repo_node)
[docs]def default_decide_target(dependency, target, prev_ni, repo_node=None): f = SCons.Defaults.DefaultEnvironment().decide_target return f(dependency, target, prev_ni, repo_node)
[docs]def default_copy_from_cache(env, src, dst): return SCons.CacheDir.CacheDir.copy_from_cache(env, src, dst)
[docs]def default_copy_to_cache(env, src, dst): return SCons.CacheDir.CacheDir.copy_to_cache(env, src, dst)
[docs]class Base(SubstitutionEnvironment): """Base class for "real" construction Environments. These are the primary objects used to communicate dependency and construction information to the build engine. Keyword arguments supplied when the construction Environment is created are construction variables used to initialize the Environment. """ ####################################################################### # This is THE class for interacting with the SCons build engine, # and it contains a lot of stuff, so we're going to try to keep this # a little organized by grouping the methods. ####################################################################### ####################################################################### # Methods that make an Environment act like a dictionary. These have # the expected standard names for Python mapping objects. Note that # we don't actually make an Environment a subclass of UserDict for # performance reasons. Note also that we only supply methods for # dictionary functionality that we actually need and use. ####################################################################### def __init__( self, platform=None, tools=None, toolpath=None, variables=None, parse_flags=None, **kw ) -> None: """Initialization of a basic SCons construction environment. Sets up special construction variables like BUILDER, PLATFORM, etc., and searches for and applies available Tools. Note that we do *not* call the underlying base class (SubsitutionEnvironment) initialization, because we need to initialize things in a very specific order that doesn't work with the much simpler base class initialization. """ if SCons.Debug.track_instances: logInstanceCreation(self, 'Environment.Base') self._memo = {} self.fs = SCons.Node.FS.get_default_fs() self.ans = SCons.Node.Alias.default_ans self.lookup_list = SCons.Node.arg2nodes_lookups self._dict = semi_deepcopy(SCons.Defaults.ConstructionEnvironment) self._init_special() self.added_methods = [] # We don't use AddMethod, or define these as methods in this # class, because we *don't* want these functions to be bound # methods. They need to operate independently so that the # settings will work properly regardless of whether a given # target ends up being built with a Base environment or an # OverrideEnvironment or what have you. self.decide_target = default_decide_target self.decide_source = default_decide_source self.cache_timestamp_newer = False self._dict['BUILDERS'] = BuilderDict(self._dict['BUILDERS'], self) if platform is None: platform = self._dict.get('PLATFORM', None) if platform is None: platform = SCons.Platform.Platform() if is_String(platform): platform = SCons.Platform.Platform(platform) self._dict['PLATFORM'] = str(platform) platform(self) # these should be set by the platform, backstop just in case self._dict['HOST_OS'] = self._dict.get('HOST_OS', None) self._dict['HOST_ARCH'] = self._dict.get('HOST_ARCH', None) # these are not currently set by the platform, give them a default self._dict['TARGET_OS'] = self._dict.get('TARGET_OS', None) self._dict['TARGET_ARCH'] = self._dict.get('TARGET_ARCH', None) # Apply the passed-in and customizable variables to the # environment before calling the tools, because they may use # some of them during initialization. if 'options' in kw: # Backwards compatibility: they may stll be using the # old "options" keyword. variables = kw['options'] del kw['options'] self.Replace(**kw) keys = list(kw.keys()) if variables: keys = keys + list(variables.keys()) variables.Update(self) save = {} for k in keys: try: save[k] = self._dict[k] except KeyError: # No value may have been set if they tried to pass in a # reserved variable name like TARGETS. pass SCons.Tool.Initializers(self) if tools is None: tools = self._dict.get('TOOLS', None) if tools is None: tools = ['default'] apply_tools(self, tools, toolpath) # Now restore the passed-in and customized variables # to the environment, since the values the user set explicitly # should override any values set by the tools. for key, val in save.items(): self._dict[key] = val # Finally, apply any flags to be merged in if parse_flags: self.MergeFlags(parse_flags) ####################################################################### # Utility methods that are primarily for internal use by SCons. # These begin with lower-case letters. #######################################################################
[docs] def get_builder(self, name): """Fetch the builder with the specified name from the environment. """ try: return self._dict['BUILDERS'][name] except KeyError: return None
[docs] def validate_CacheDir_class(self, custom_class=None): """Validate the passed custom CacheDir class, or if no args are passed, validate the custom CacheDir class from the environment. """ if custom_class is None: custom_class = self.get("CACHEDIR_CLASS", SCons.CacheDir.CacheDir) if not issubclass(custom_class, SCons.CacheDir.CacheDir): raise UserError("Custom CACHEDIR_CLASS %s not derived from CacheDir" % str(custom_class)) return custom_class
[docs] def get_CacheDir(self): try: path = self._CacheDir_path except AttributeError: path = SCons.Defaults.DefaultEnvironment()._CacheDir_path cachedir_class = self.validate_CacheDir_class() try: if (path == self._last_CacheDir_path # this checks if the cachedir class type has changed from what the # instantiated cache dir type is. If the are exactly the same we # can just keep using the existing one, otherwise the user is requesting # something new, so we will re-instantiate below. and type(self._last_CacheDir) is cachedir_class): return self._last_CacheDir except AttributeError: pass cd = cachedir_class(path) self._last_CacheDir_path = path self._last_CacheDir = cd return cd
[docs] def get_factory(self, factory, default: str='File'): """Return a factory function for creating Nodes for this construction environment. """ name = default try: is_node = issubclass(factory, SCons.Node.FS.Base) except TypeError: # The specified factory isn't a Node itself--it's # most likely None, or possibly a callable. pass else: if is_node: # The specified factory is a Node (sub)class. Try to # return the FS method that corresponds to the Node's # name--that is, we return self.fs.Dir if they want a Dir, # self.fs.File for a File, etc. try: name = factory.__name__ except AttributeError: pass else: factory = None if not factory: # They passed us None, or we picked up a name from a specified # class, so return the FS method. (Note that we *don't* # use our own self.{Dir,File} methods because that would # cause env.subst() to be called twice on the file name, # interfering with files that have $$ in them.) factory = getattr(self.fs, name) return factory
[docs] @SCons.Memoize.CountMethodCall def _gsm(self): try: return self._memo['_gsm'] except KeyError: pass result = {} try: scanners = self._dict['SCANNERS'] except KeyError: pass else: # Reverse the scanner list so that, if multiple scanners # claim they can scan the same suffix, earlier scanners # in the list will overwrite later scanners, so that # the result looks like a "first match" to the user. if not is_List(scanners): scanners = [scanners] else: scanners = scanners[:] # copy so reverse() doesn't mod original scanners.reverse() for scanner in scanners: for k in scanner.get_skeys(self): if k and self['PLATFORM'] == 'win32': k = k.lower() result[k] = scanner self._memo['_gsm'] = result return result
[docs] def get_scanner(self, skey): """Find the appropriate scanner given a key (usually a file suffix). """ if skey and self['PLATFORM'] == 'win32': skey = skey.lower() return self._gsm().get(skey)
[docs] def scanner_map_delete(self, kw=None) -> None: """Delete the cached scanner map (if we need to). """ try: del self._memo['_gsm'] except KeyError: pass
[docs] def _update(self, other) -> None: """Private method to update an environment's consvar dict directly. Bypasses the normal checks that occur when users try to set items. """ self._dict.update(other)
[docs] def _update_onlynew(self, other) -> None: """Private method to add new items to an environment's consvar dict. Only adds items from `other` whose keys do not already appear in the existing dict; values from `other` are not used for replacement. Bypasses the normal checks that occur when users try to set items. """ for k, v in other.items(): if k not in self._dict: self._dict[k] = v
####################################################################### # Public methods for manipulating an Environment. These begin with # upper-case letters. The essential characteristic of methods in # this section is that they do *not* have corresponding same-named # global functions. For example, a stand-alone Append() function # makes no sense, because Append() is all about appending values to # an Environment's construction variables. #######################################################################
[docs] def Append(self, **kw) -> None: """Append values to construction variables in an Environment. The variable is created if it is not already present. """ kw = copy_non_reserved_keywords(kw) for key, val in kw.items(): if key == 'CPPDEFINES': _add_cppdefines(self._dict, val) continue try: orig = self._dict[key] except KeyError: # No existing var in the environment, so set to the new value. self._dict[key] = val continue try: # Check if the original looks like a dict: has .update? update_dict = orig.update except AttributeError: try: # Just try to add them together. This will work # in most cases, when the original and new values # are compatible types. self._dict[key] = orig + val except (KeyError, TypeError): try: # Check if the original is a list: has .append? add_to_orig = orig.append except AttributeError: # The original isn't a list, but the new # value is (by process of elimination), # so insert the original in the new value # (if there's one to insert) and replace # the variable with it. if orig: val.insert(0, orig) self._dict[key] = val else: # The original is a list, so append the new # value to it (if there's a value to append). if val: add_to_orig(val) continue # The original looks like a dictionary, so update it # based on what we think the value looks like. # We can't just try adding the value because # dictionaries don't have __add__() methods, and # things like UserList will incorrectly coerce the # original dict to a list (which we don't want). if is_List(val): for v in val: orig[v] = None else: try: update_dict(val) except (AttributeError, TypeError, ValueError): if is_Dict(val): for k, v in val.items(): orig[k] = v else: orig[val] = None self.scanner_map_delete(kw)
[docs] def _canonicalize(self, path): """Allow Dirs and strings beginning with # for top-relative. Note this uses the current env's fs (in self). """ if not is_String(path): # typically a Dir path = str(path) if path and path[0] == '#': path = str(self.fs.Dir(path)) return path
[docs] def AppendENVPath(self, name, newpath, envname: str='ENV', sep=os.pathsep, delete_existing: bool=False) -> None: """Append path elements to the path *name* in the *envname* dictionary for this environment. Will only add any particular path once, and will normpath and normcase all paths to help assure this. This can also handle the case where the env variable is a list instead of a string. If *delete_existing* is False, a *newpath* element already in the path will not be moved to the end (it will be left where it is). """ orig = '' if envname in self._dict and name in self._dict[envname]: orig = self._dict[envname][name] nv = AppendPath(orig, newpath, sep, delete_existing, canonicalize=self._canonicalize) if envname not in self._dict: self._dict[envname] = {} self._dict[envname][name] = nv
[docs] def AppendUnique(self, delete_existing: bool=False, **kw) -> None: """Append values to existing construction variables in an Environment, if they're not already there. If delete_existing is True, removes existing values first, so values move to end. """ kw = copy_non_reserved_keywords(kw) for key, val in kw.items(): if key == 'CPPDEFINES': _add_cppdefines(self._dict, val, unique=True, delete_existing=delete_existing) continue if is_List(val): val = _delete_duplicates(val, delete_existing) if key not in self._dict or self._dict[key] in ('', None): self._dict[key] = val elif is_Dict(self._dict[key]) and is_Dict(val): self._dict[key].update(val) elif is_List(val): dk = self._dict[key] if not is_List(dk): dk = [dk] if delete_existing: dk = [x for x in dk if x not in val] else: val = [x for x in val if x not in dk] self._dict[key] = dk + val else: dk = self._dict[key] if is_List(dk): # By elimination, val is not a list. Since dk is a # list, wrap val in a list first. if delete_existing: dk = list(filter(lambda x, val=val: x not in val, dk)) self._dict[key] = dk + [val] else: if val not in dk: self._dict[key] = dk + [val] else: if delete_existing: dk = [x for x in dk if x not in val] self._dict[key] = dk + val self.scanner_map_delete(kw)
[docs] def Clone(self, tools=[], toolpath=None, parse_flags = None, **kw): """Return a copy of a construction Environment. The copy is like a Python "deep copy"--that is, independent copies are made recursively of each objects--except that a reference is copied when an object is not deep-copyable (like a function). There are no references to any mutable objects in the original Environment. """ builders = self._dict.get('BUILDERS', {}) clone = copy.copy(self) # BUILDERS is not safe to do a simple copy clone._dict = semi_deepcopy_dict(self._dict, ['BUILDERS']) clone._dict['BUILDERS'] = BuilderDict(builders, clone) # Check the methods added via AddMethod() and re-bind them to # the cloned environment. Only do this if the attribute hasn't # been overwritten by the user explicitly and still points to # the added method. clone.added_methods = [] for mw in self.added_methods: if mw == getattr(self, mw.name): clone.added_methods.append(mw.clone(clone)) clone._memo = {} # Apply passed-in variables before the tools # so the tools can use the new variables kw = copy_non_reserved_keywords(kw) new = {} for key, value in kw.items(): new[key] = SCons.Subst.scons_subst_once(value, self, key) clone.Replace(**new) apply_tools(clone, tools, toolpath) # apply them again in case the tools overwrote them clone.Replace(**new) # Finally, apply any flags to be merged in if parse_flags: clone.MergeFlags(parse_flags) if SCons.Debug.track_instances: logInstanceCreation(self, 'Environment.EnvironmentClone') return clone
[docs] def _changed_build(self, dependency, target, prev_ni, repo_node=None) -> bool: if dependency.changed_state(target, prev_ni, repo_node): return True return self.decide_source(dependency, target, prev_ni, repo_node)
[docs] def _changed_content(self, dependency, target, prev_ni, repo_node=None) -> bool: return dependency.changed_content(target, prev_ni, repo_node)
[docs] def _changed_timestamp_then_content(self, dependency, target, prev_ni, repo_node=None) -> bool: return dependency.changed_timestamp_then_content(target, prev_ni, repo_node)
[docs] def _changed_timestamp_newer(self, dependency, target, prev_ni, repo_node=None) -> bool: return dependency.changed_timestamp_newer(target, prev_ni, repo_node)
[docs] def _changed_timestamp_match(self, dependency, target, prev_ni, repo_node=None) -> bool: return dependency.changed_timestamp_match(target, prev_ni, repo_node)
[docs] def Decider(self, function): self.cache_timestamp_newer = False if function in ('MD5', 'content'): # TODO: Handle if user requests MD5 and not content with deprecation notice function = self._changed_content elif function in ('MD5-timestamp', 'content-timestamp'): function = self._changed_timestamp_then_content elif function in ('timestamp-newer', 'make'): function = self._changed_timestamp_newer self.cache_timestamp_newer = True elif function == 'timestamp-match': function = self._changed_timestamp_match elif not callable(function): raise UserError("Unknown Decider value %s" % repr(function)) # We don't use AddMethod because we don't want to turn the # function, which only expects three arguments, into a bound # method, which would add self as an initial, fourth argument. self.decide_target = function self.decide_source = function
[docs] def Detect(self, progs): """Return the first available program from one or more possibilities. Args: progs (str or list): one or more command names to check for """ if not is_List(progs): progs = [progs] for prog in progs: path = self.WhereIs(prog) if path: return prog return None
[docs] def Dictionary(self, *args): r"""Return construction variables from an environment. Args: \*args (optional): variable names to look up Returns: If `args` omitted, the dictionary of all construction variables. If one arg, the corresponding value is returned. If more than one arg, a list of values is returned. Raises: KeyError: if any of `args` is not in the construction environment. """ if not args: return self._dict dlist = [self._dict[x] for x in args] if len(dlist) == 1: dlist = dlist[0] return dlist
[docs] def Dump(self, key=None, format: str='pretty'): """ Return construction variables serialized to a string. Args: key (optional): if None, format the whole dict of variables. Else format the value of `key` (Default value = None) format (str, optional): specify the format to serialize to. `"pretty"` generates a pretty-printed string, `"json"` a JSON-formatted string. (Default value = `"pretty"`) """ if key: cvars = self.Dictionary(key) else: cvars = self.Dictionary() fmt = format.lower() if fmt == 'pretty': import pprint pp = pprint.PrettyPrinter(indent=2) # TODO: pprint doesn't do a nice job on path-style values # if the paths contain spaces (i.e. Windows), because the # algorithm tries to break lines on spaces, while breaking # on the path-separator would be more "natural". Is there # a better way to format those? return pp.pformat(cvars) elif fmt == 'json': import json def non_serializable(obj): return str(type(obj).__qualname__) return json.dumps(cvars, indent=4, default=non_serializable) else: raise ValueError("Unsupported serialization format: %s." % fmt)
[docs] def FindIxes(self, paths, prefix, suffix): """Search a list of paths for something that matches the prefix and suffix. Args: paths: the list of paths or nodes. prefix: construction variable for the prefix. suffix: construction variable for the suffix. Returns: the matched path or None """ suffix = self.subst('$'+suffix) prefix = self.subst('$'+prefix) for path in paths: name = os.path.basename(str(path)) if name[:len(prefix)] == prefix and name[-len(suffix):] == suffix: return path
[docs] def ParseConfig(self, command, function=None, unique: bool=True): """Parse the result of running a command to update construction vars. Use ``function`` to parse the output of running ``command`` in order to modify the current environment. Args: command: a string or a list of strings representing a command and its arguments. function: called to process the result of ``command``, which will be passed as ``args``. If ``function`` is omitted or ``None``, :meth:`MergeFlags` is used. Takes 3 args ``(env, args, unique)`` unique: whether no duplicate values are allowed (default true) """ if function is None: def parse_conf(env, cmd, unique=unique): return env.MergeFlags(cmd, unique) function = parse_conf if is_List(command): command = ' '.join(command) command = self.subst(command) return function(self, self.backtick(command), unique)
[docs] def ParseDepends(self, filename, must_exist=None, only_one: bool=False): """ Parse a mkdep-style file for explicit dependencies. This is completely abusable, and should be unnecessary in the "normal" case of proper SCons configuration, but it may help make the transition from a Make hierarchy easier for some people to swallow. It can also be genuinely useful when using a tool that can write a .d file, but for which writing a scanner would be too complicated. """ filename = self.subst(filename) try: with open(filename) as fp: lines = LogicalLines(fp).readlines() except OSError: if must_exist: raise return lines = [l for l in lines if l[0] != '#'] tdlist = [] for line in lines: try: target, depends = line.split(':', 1) except (AttributeError, ValueError): # Throws AttributeError if line isn't a string. Can throw # ValueError if line doesn't split into two or more elements. pass else: tdlist.append((target.split(), depends.split())) if only_one: targets = [] for td in tdlist: targets.extend(td[0]) if len(targets) > 1: raise UserError( "More than one dependency target found in `%s': %s" % (filename, targets)) for target, depends in tdlist: self.Depends(target, depends)
[docs] def Platform(self, platform): platform = self.subst(platform) return SCons.Platform.Platform(platform)(self)
[docs] def Prepend(self, **kw) -> None: """Prepend values to construction variables in an Environment. The variable is created if it is not already present. """ kw = copy_non_reserved_keywords(kw) for key, val in kw.items(): if key == 'CPPDEFINES': _add_cppdefines(self._dict, val, prepend=True) continue try: orig = self._dict[key] except KeyError: # No existing var in the environment so set to the new value. self._dict[key] = val continue try: # Check if the original looks like a dict: has .update? update_dict = orig.update except AttributeError: try: # Just try to add them together. This will work # in most cases, when the original and new values # are compatible types. self._dict[key] = val + orig except (KeyError, TypeError): try: # Check if the added value is a list: has .append? add_to_val = val.append except AttributeError: # The added value isn't a list, but the # original is (by process of elimination), # so insert the the new value in the original # (if there's one to insert). if val: orig.insert(0, val) else: # The added value is a list, so append # the original to it (if there's a value # to append) and replace the original. if orig: add_to_val(orig) self._dict[key] = val continue # The original looks like a dictionary, so update it # based on what we think the value looks like. # We can't just try adding the value because # dictionaries don't have __add__() methods, and # things like UserList will incorrectly coerce the # original dict to a list (which we don't want). if is_List(val): for v in val: orig[v] = None else: try: update_dict(val) except (AttributeError, TypeError, ValueError): if is_Dict(val): for k, v in val.items(): orig[k] = v else: orig[val] = None self.scanner_map_delete(kw)
[docs] def PrependENVPath(self, name, newpath, envname: str='ENV', sep=os.pathsep, delete_existing: bool=True) -> None: """Prepend path elements to the path *name* in the *envname* dictionary for this environment. Will only add any particular path once, and will normpath and normcase all paths to help assure this. This can also handle the case where the env variable is a list instead of a string. If *delete_existing* is False, a *newpath* component already in the path will not be moved to the front (it will be left where it is). """ orig = '' if envname in self._dict and name in self._dict[envname]: orig = self._dict[envname][name] nv = PrependPath(orig, newpath, sep, delete_existing, canonicalize=self._canonicalize) if envname not in self._dict: self._dict[envname] = {} self._dict[envname][name] = nv
[docs] def PrependUnique(self, delete_existing: bool=False, **kw) -> None: """Prepend values to existing construction variables in an Environment, if they're not already there. If delete_existing is True, removes existing values first, so values move to front. """ kw = copy_non_reserved_keywords(kw) for key, val in kw.items(): if key == 'CPPDEFINES': _add_cppdefines(self._dict, val, unique=True, prepend=True, delete_existing=delete_existing) continue if is_List(val): val = _delete_duplicates(val, not delete_existing) if key not in self._dict or self._dict[key] in ('', None): self._dict[key] = val elif is_Dict(self._dict[key]) and is_Dict(val): self._dict[key].update(val) elif is_List(val): dk = self._dict[key] if not is_List(dk): dk = [dk] if delete_existing: dk = [x for x in dk if x not in val] else: val = [x for x in val if x not in dk] self._dict[key] = val + dk else: dk = self._dict[key] if is_List(dk): # By elimination, val is not a list. Since dk is a # list, wrap val in a list first. if delete_existing: dk = [x for x in dk if x not in val] self._dict[key] = [val] + dk else: if val not in dk: self._dict[key] = [val] + dk else: if delete_existing: dk = [x for x in dk if x not in val] self._dict[key] = val + dk self.scanner_map_delete(kw)
[docs] def Replace(self, **kw) -> None: """Replace existing construction variables in an Environment with new construction variables and/or values. """ try: kwbd = kw['BUILDERS'] except KeyError: pass else: kwbd = BuilderDict(kwbd,self) del kw['BUILDERS'] self.__setitem__('BUILDERS', kwbd) kw = copy_non_reserved_keywords(kw) self._update(semi_deepcopy(kw)) self.scanner_map_delete(kw)
[docs] def ReplaceIxes(self, path, old_prefix, old_suffix, new_prefix, new_suffix): """ Replace old_prefix with new_prefix and old_suffix with new_suffix. env - Environment used to interpolate variables. path - the path that will be modified. old_prefix - construction variable for the old prefix. old_suffix - construction variable for the old suffix. new_prefix - construction variable for the new prefix. new_suffix - construction variable for the new suffix. """ old_prefix = self.subst('$'+old_prefix) old_suffix = self.subst('$'+old_suffix) new_prefix = self.subst('$'+new_prefix) new_suffix = self.subst('$'+new_suffix) dir,name = os.path.split(str(path)) if name[:len(old_prefix)] == old_prefix: name = name[len(old_prefix):] if name[-len(old_suffix):] == old_suffix: name = name[:-len(old_suffix)] return os.path.join(dir, new_prefix+name+new_suffix)
[docs] def SetDefault(self, **kw) -> None: for k in list(kw.keys()): if k in self._dict: del kw[k] self.Replace(**kw)
[docs] def _find_toolpath_dir(self, tp): return self.fs.Dir(self.subst(tp)).srcnode().get_abspath()
[docs] def Tool(self, tool, toolpath=None, **kwargs) -> SCons.Tool.Tool: """Find and run tool module *tool*. .. versionchanged:: 4.2 returns the tool module rather than ``None``. """ if is_String(tool): tool = self.subst(tool) if toolpath is None: toolpath = self.get('toolpath', []) toolpath = list(map(self._find_toolpath_dir, toolpath)) tool = SCons.Tool.Tool(tool, toolpath, **kwargs) tool(self) return tool
[docs] def WhereIs(self, prog, path=None, pathext=None, reject=None): """Find prog in the path. """ if not prog: # nothing to search for, just give up return None if path is None: try: path = self['ENV']['PATH'] except KeyError: pass elif is_String(path): path = self.subst(path) if pathext is None: try: pathext = self['ENV']['PATHEXT'] except KeyError: pass elif is_String(pathext): pathext = self.subst(pathext) prog = CLVar(self.subst(prog)) # support "program --with-args" path = WhereIs(prog[0], path, pathext, reject) if path: return path return None
####################################################################### # Public methods for doing real "SCons stuff" (manipulating # dependencies, setting attributes on targets, etc.). These begin # with upper-case letters. The essential characteristic of methods # in this section is that they all *should* have corresponding # same-named global functions. #######################################################################
[docs] def Action(self, *args, **kw): def subst_string(a, self=self): if is_String(a): a = self.subst(a) return a nargs = list(map(subst_string, args)) nkw = self.subst_kw(kw) return SCons.Action.Action(*nargs, **nkw)
[docs] def AddPreAction(self, files, action): nodes = self.arg2nodes(files, self.fs.Entry) action = SCons.Action.Action(action) uniq = {} for executor in [n.get_executor() for n in nodes]: uniq[executor] = 1 for executor in uniq.keys(): executor.add_pre_action(action) return nodes
[docs] def AddPostAction(self, files, action): nodes = self.arg2nodes(files, self.fs.Entry) action = SCons.Action.Action(action) uniq = {} for executor in [n.get_executor() for n in nodes]: uniq[executor] = 1 for executor in uniq.keys(): executor.add_post_action(action) return nodes
[docs] def Alias(self, target, source=[], action=None, **kw): tlist = self.arg2nodes(target, self.ans.Alias) if not is_List(source): source = [source] source = [_f for _f in source if _f] if not action: if not source: # There are no source files and no action, so just # return a target list of classic Alias Nodes, without # any builder. The externally visible effect is that # this will make the wrapping Script.BuildTask class # say that there's "Nothing to be done" for this Alias, # instead of that it's "up to date." return tlist # No action, but there are sources. Re-call all the target # builders to add the sources to each target. result = [] for t in tlist: bld = t.get_builder(AliasBuilder) result.extend(bld(self, t, source)) return result nkw = self.subst_kw(kw) nkw.update({ 'action' : SCons.Action.Action(action), 'source_factory' : self.fs.Entry, 'multi' : 1, 'is_explicit' : None, }) bld = SCons.Builder.Builder(**nkw) # Apply the Builder separately to each target so that the Aliases # stay separate. If we did one "normal" Builder call with the # whole target list, then all of the target Aliases would be # associated under a single Executor. result = [] for t in tlist: # Calling the convert() method will cause a new Executor to be # created from scratch, so we have to explicitly initialize # it with the target's existing sources, plus our new ones, # so nothing gets lost. b = t.get_builder() if b is None or b is AliasBuilder: b = bld else: nkw['action'] = b.action + action b = SCons.Builder.Builder(**nkw) t.convert() result.extend(b(self, t, t.sources + source)) return result
[docs] def AlwaysBuild(self, *targets): tlist = [] for t in targets: tlist.extend(self.arg2nodes(t, self.fs.Entry)) for t in tlist: t.set_always_build() return tlist
[docs] def Builder(self, **kw): nkw = self.subst_kw(kw) return SCons.Builder.Builder(**nkw)
[docs] def CacheDir(self, path, custom_class=None) -> None: if path is not None: path = self.subst(path) self._CacheDir_path = path if custom_class: self['CACHEDIR_CLASS'] = self.validate_CacheDir_class(custom_class) if SCons.Action.execute_actions: # Only initialize the CacheDir if -n/-no_exec was NOT specified. # Now initialized the CacheDir and prevent a race condition which can # happen when there's no existing cache dir and you are building with # multiple threads, but initializing it before the task walk starts self.get_CacheDir()
[docs] def Clean(self, targets, files) -> None: global CleanTargets tlist = self.arg2nodes(targets, self.fs.Entry) flist = self.arg2nodes(files, self.fs.Entry) for t in tlist: try: CleanTargets[t].extend(flist) except KeyError: CleanTargets[t] = flist
[docs] def Configure(self, *args, **kw): nargs = [self] if args: nargs = nargs + self.subst_list(args)[0] nkw = self.subst_kw(kw) nkw['_depth'] = kw.get('_depth', 0) + 1 try: nkw['custom_tests'] = self.subst_kw(nkw['custom_tests']) except KeyError: pass return SCons.SConf.SConf(*nargs, **nkw)
[docs] def Command(self, target, source, action, **kw): """Builds the supplied target files from the supplied source files using the supplied action. Action may be any type that the Builder constructor will accept for an action.""" bkw = { 'action': action, 'target_factory': self.fs.Entry, 'source_factory': self.fs.Entry, } # source scanner try: bkw['source_scanner'] = kw['source_scanner'] except KeyError: pass else: del kw['source_scanner'] # target scanner try: bkw['target_scanner'] = kw['target_scanner'] except KeyError: pass else: del kw['target_scanner'] # source factory try: bkw['source_factory'] = kw['source_factory'] except KeyError: pass else: del kw['source_factory'] # target factory try: bkw['target_factory'] = kw['target_factory'] except KeyError: pass else: del kw['target_factory'] bld = SCons.Builder.Builder(**bkw) return bld(self, target, source, **kw)
[docs] def Depends(self, target, dependency): """Explicity specify that 'target's depend on 'dependency'.""" tlist = self.arg2nodes(target, self.fs.Entry) dlist = self.arg2nodes(dependency, self.fs.Entry) for t in tlist: t.add_dependency(dlist) return tlist
[docs] def Dir(self, name, *args, **kw): """ """ s = self.subst(name) if is_Sequence(s): result=[] for e in s: result.append(self.fs.Dir(e, *args, **kw)) return result return self.fs.Dir(s, *args, **kw)
[docs] def PyPackageDir(self, modulename): s = self.subst(modulename) if is_Sequence(s): result=[] for e in s: result.append(self.fs.PyPackageDir(e)) return result return self.fs.PyPackageDir(s)
[docs] def NoClean(self, *targets): """Tags a target so that it will not be cleaned by -c""" tlist = [] for t in targets: tlist.extend(self.arg2nodes(t, self.fs.Entry)) for t in tlist: t.set_noclean() return tlist
[docs] def NoCache(self, *targets): """Tags a target so that it will not be cached""" tlist = [] for t in targets: tlist.extend(self.arg2nodes(t, self.fs.Entry)) for t in tlist: t.set_nocache() return tlist
[docs] def Entry(self, name, *args, **kw): """ """ s = self.subst(name) if is_Sequence(s): result=[] for e in s: result.append(self.fs.Entry(e, *args, **kw)) return result return self.fs.Entry(s, *args, **kw)
[docs] def Environment(self, **kw): return SCons.Environment.Environment(**self.subst_kw(kw))
[docs] def Execute(self, action, *args, **kw): """Directly execute an action through an Environment """ action = self.Action(action, *args, **kw) result = action([], [], self) if isinstance(result, BuildError): errstr = result.errstr if result.filename: errstr = result.filename + ': ' + errstr sys.stderr.write("scons: *** %s\n" % errstr) return result.status else: return result
[docs] def File(self, name, *args, **kw): """ """ s = self.subst(name) if is_Sequence(s): result=[] for e in s: result.append(self.fs.File(e, *args, **kw)) return result return self.fs.File(s, *args, **kw)
[docs] def FindFile(self, file, dirs): file = self.subst(file) nodes = self.arg2nodes(dirs, self.fs.Dir) return SCons.Node.FS.find_file(file, tuple(nodes))
[docs] def Flatten(self, sequence): return flatten(sequence)
[docs] def GetBuildPath(self, files): result = list(map(str, self.arg2nodes(files, self.fs.Entry))) if is_List(files): return result else: return result[0]
[docs] def Glob(self, pattern, ondisk: bool=True, source: bool=False, strings: bool=False, exclude=None): return self.fs.Glob(self.subst(pattern), ondisk, source, strings, exclude)
[docs] def Ignore(self, target, dependency): """Ignore a dependency.""" tlist = self.arg2nodes(target, self.fs.Entry) dlist = self.arg2nodes(dependency, self.fs.Entry) for t in tlist: t.add_ignore(dlist) return tlist
[docs] def Literal(self, string): return SCons.Subst.Literal(string)
[docs] def Local(self, *targets): ret = [] for targ in targets: if isinstance(targ, SCons.Node.Node): targ.set_local() ret.append(targ) else: for t in self.arg2nodes(targ, self.fs.Entry): t.set_local() ret.append(t) return ret
[docs] def Precious(self, *targets): tlist = [] for t in targets: tlist.extend(self.arg2nodes(t, self.fs.Entry)) for t in tlist: t.set_precious() return tlist
[docs] def Pseudo(self, *targets): tlist = [] for t in targets: tlist.extend(self.arg2nodes(t, self.fs.Entry)) for t in tlist: t.set_pseudo() return tlist
[docs] def Repository(self, *dirs, **kw) -> None: dirs = self.arg2nodes(list(dirs), self.fs.Dir) self.fs.Repository(*dirs, **kw)
[docs] def Requires(self, target, prerequisite): """Specify that 'prerequisite' must be built before 'target', (but 'target' does not actually depend on 'prerequisite' and need not be rebuilt if it changes).""" tlist = self.arg2nodes(target, self.fs.Entry) plist = self.arg2nodes(prerequisite, self.fs.Entry) for t in tlist: t.add_prerequisite(plist) return tlist
[docs] def Scanner(self, *args, **kw): nargs = [] for arg in args: if is_String(arg): arg = self.subst(arg) nargs.append(arg) nkw = self.subst_kw(kw) return SCons.Scanner.ScannerBase(*nargs, **nkw)
[docs] def SConsignFile(self, name=SCons.SConsign.current_sconsign_filename(), dbm_module=None) -> None: if name is not None: name = self.subst(name) if not os.path.isabs(name): name = os.path.join(str(self.fs.SConstruct_dir), name) if name: name = os.path.normpath(name) sconsign_dir = os.path.dirname(name) if sconsign_dir and not os.path.exists(sconsign_dir): self.Execute(SCons.Defaults.Mkdir(sconsign_dir)) SCons.SConsign.File(name, dbm_module)
[docs] def SideEffect(self, side_effect, target): """Tell scons that side_effects are built as side effects of building targets.""" side_effects = self.arg2nodes(side_effect, self.fs.Entry) targets = self.arg2nodes(target, self.fs.Entry) added_side_effects = [] for side_effect in side_effects: if side_effect.multiple_side_effect_has_builder(): raise UserError("Multiple ways to build the same target were specified for: %s" % str(side_effect)) side_effect.add_source(targets) side_effect.side_effect = 1 self.Precious(side_effect) added = False for target in targets: if side_effect not in target.side_effects: target.side_effects.append(side_effect) added = True if added: added_side_effects.append(side_effect) return added_side_effects
[docs] def Split(self, arg): """This function converts a string or list into a list of strings or Nodes. This makes things easier for users by allowing files to be specified as a white-space separated list to be split. The input rules are: - A single string containing names separated by spaces. These will be split apart at the spaces. - A single Node instance - A list containing either strings or Node instances. Any strings in the list are not split at spaces. In all cases, the function returns a list of Nodes and strings.""" if is_List(arg): return list(map(self.subst, arg)) elif is_String(arg): return self.subst(arg).split() else: return [self.subst(arg)]
[docs] def Value(self, value, built_value=None, name=None): """Return a Value (Python expression) node. .. versionchanged:: 4.0 the *name* parameter was added. """ return SCons.Node.Python.ValueWithMemo(value, built_value, name)
[docs] def VariantDir(self, variant_dir, src_dir, duplicate: int=1) -> None: variant_dir = self.arg2nodes(variant_dir, self.fs.Dir)[0] src_dir = self.arg2nodes(src_dir, self.fs.Dir)[0] self.fs.VariantDir(variant_dir, src_dir, duplicate)
[docs] def FindSourceFiles(self, node: str='.') -> list: """Return a list of all source files.""" node = self.arg2nodes(node, self.fs.Entry)[0] sources = [] def build_source(ss) -> None: for s in ss: if isinstance(s, SCons.Node.FS.Dir): build_source(s.all_children()) elif s.has_builder(): build_source(s.sources) elif isinstance(s.disambiguate(), SCons.Node.FS.File): sources.append(s) build_source(node.all_children()) def final_source(node): while node != node.srcnode(): node = node.srcnode() return node sources = list(map(final_source, sources)) # remove duplicates return list(set(sources))
[docs] def FindInstalledFiles(self): """ returns the list of all targets of the Install and InstallAs Builder. """ from SCons.Tool import install if install._UNIQUE_INSTALLED_FILES is None: install._UNIQUE_INSTALLED_FILES = uniquer_hashables(install._INSTALLED_FILES) return install._UNIQUE_INSTALLED_FILES
[docs]class OverrideEnvironment(Base): """A proxy that overrides variables in a wrapped construction environment by returning values from an overrides dictionary in preference to values from the underlying subject environment. This is a lightweight (I hope) proxy that passes through most use of attributes to the underlying Environment.Base class, but has just enough additional methods defined to act like a real construction environment with overridden values. It can wrap either a Base construction environment, or another OverrideEnvironment, which can in turn nest arbitrary OverrideEnvironments... Note that we do *not* call the underlying base class (SubsitutionEnvironment) initialization, because we get most of those from proxying the attributes of the subject construction environment. But because we subclass SubstitutionEnvironment, this class also has inherited arg2nodes() and subst*() methods; those methods can't be proxied because they need *this* object's methods to fetch the values from the overrides dictionary. """ def __init__(self, subject, overrides=None) -> None: if SCons.Debug.track_instances: logInstanceCreation(self, 'Environment.OverrideEnvironment') self.__dict__['__subject'] = subject if overrides is None: self.__dict__['overrides'] = {} else: self.__dict__['overrides'] = overrides # Methods that make this class act like a proxy. def __getattr__(self, name): attr = getattr(self.__dict__['__subject'], name) # Here we check if attr is one of the Wrapper classes. For # example when a pseudo-builder is being called from an # OverrideEnvironment. # # These wrappers when they're constructed capture the # Environment they are being constructed with and so will not # have access to overrided values. So we rebuild them with the # OverrideEnvironment so they have access to overrided values. if isinstance(attr, MethodWrapper): return attr.clone(self) else: return attr def __setattr__(self, name, value) -> None: setattr(self.__dict__['__subject'], name, value) # Methods that make this class act like a dictionary. def __getitem__(self, key): try: return self.__dict__['overrides'][key] except KeyError: return self.__dict__['__subject'].__getitem__(key) def __setitem__(self, key, value): if not is_valid_construction_var(key): raise UserError("Illegal construction variable `%s'" % key) self.__dict__['overrides'][key] = value def __delitem__(self, key): try: del self.__dict__['overrides'][key] except KeyError: deleted = 0 else: deleted = 1 try: result = self.__dict__['__subject'].__delitem__(key) except KeyError: if not deleted: raise result = None return result
[docs] def get(self, key, default=None): """Emulates the get() method of dictionaries.""" try: return self.__dict__['overrides'][key] except KeyError: return self.__dict__['__subject'].get(key, default)
def __contains__(self, key) -> bool: if key in self.__dict__['overrides']: return True return key in self.__dict__['__subject']
[docs] def Dictionary(self, *args): d = self.__dict__['__subject'].Dictionary().copy() d.update(self.__dict__['overrides']) if not args: return d dlist = [d[x] for x in args] if len(dlist) == 1: dlist = dlist[0] return dlist
[docs] def items(self): """Emulates the items() method of dictionaries.""" return self.Dictionary().items()
[docs] def keys(self): """Emulates the keys() method of dictionaries.""" return self.Dictionary().keys()
[docs] def values(self): """Emulates the values() method of dictionaries.""" return self.Dictionary().values()
[docs] def setdefault(self, key, default=None): """Emulates the setdefault() method of dictionaries.""" try: return self.__getitem__(key) except KeyError: self.__dict__['overrides'][key] = default return default
# Overridden private construction environment methods.
[docs] def _update(self, other) -> None: self.__dict__['overrides'].update(other)
[docs] def _update_onlynew(self, other) -> None: """Update a dict with new keys. Unlike the .update method, if the key is already present, it is not replaced. """ for k, v in other.items(): if k not in self.__dict__['overrides']: self.__dict__['overrides'][k] = v
[docs] def gvars(self): return self.__dict__['__subject'].gvars()
[docs] def lvars(self): lvars = self.__dict__['__subject'].lvars() lvars.update(self.__dict__['overrides']) return lvars
# Overridden public construction environment methods.
[docs] def Replace(self, **kw) -> None: kw = copy_non_reserved_keywords(kw) self.__dict__['overrides'].update(semi_deepcopy(kw))
# The entry point that will be used by the external world # to refer to a construction environment. This allows the wrapper # interface to extend a construction environment for its own purposes # by subclassing SCons.Environment.Base and then assigning the # class to SCons.Environment.Environment. Environment = Base
[docs]def NoSubstitutionProxy(subject): """ An entry point for returning a proxy subclass instance that overrides the subst*() methods so they don't actually perform construction variable substitution. This is specifically intended to be the shim layer in between global function calls (which don't want construction variable substitution) and the DefaultEnvironment() (which would substitute variables if left to its own devices). We have to wrap this in a function that allows us to delay definition of the class until it's necessary, so that when it subclasses Environment it will pick up whatever Environment subclass the wrapper interface might have assigned to SCons.Environment.Environment. """ class _NoSubstitutionProxy(Environment): def __init__(self, subject) -> None: self.__dict__['__subject'] = subject def __getattr__(self, name): return getattr(self.__dict__['__subject'], name) def __setattr__(self, name, value): return setattr(self.__dict__['__subject'], name, value) def executor_to_lvars(self, kwdict) -> None: if 'executor' in kwdict: kwdict['lvars'] = kwdict['executor'].get_lvars() del kwdict['executor'] else: kwdict['lvars'] = {} def raw_to_mode(self, mapping) -> None: try: raw = mapping['raw'] except KeyError: pass else: del mapping['raw'] mapping['mode'] = raw def subst(self, string, *args, **kwargs): return string def subst_kw(self, kw, *args, **kwargs): return kw def subst_list(self, string, *args, **kwargs): nargs = (string, self,) + args nkw = kwargs.copy() nkw['gvars'] = {} self.executor_to_lvars(nkw) self.raw_to_mode(nkw) return SCons.Subst.scons_subst_list(*nargs, **nkw) def subst_target_source(self, string, *args, **kwargs): nargs = (string, self,) + args nkw = kwargs.copy() nkw['gvars'] = {} self.executor_to_lvars(nkw) self.raw_to_mode(nkw) return SCons.Subst.scons_subst(*nargs, **nkw) return _NoSubstitutionProxy(subject)
# Local Variables: # tab-width:4 # indent-tabs-mode:nil # End: # vim: set expandtab tabstop=4 shiftwidth=4: