diff options
author | David Robillard <d@drobilla.net> | 2019-04-21 22:53:32 +0200 |
---|---|---|
committer | David Robillard <d@drobilla.net> | 2019-04-21 22:55:04 +0200 |
commit | 4be819c27f627d0fad0f63cb5817bfca8460541a (patch) | |
tree | 46e2d6880484bc5621a7b148775a2aa47b838d9b /waflib/Task.py | |
parent | 9b7bfdd92d9a12b0d7db59f0ec0bb790fb827406 (diff) | |
download | lv2-4be819c27f627d0fad0f63cb5817bfca8460541a.tar.xz |
Switch to using a submodule for autowaf
Diffstat (limited to 'waflib/Task.py')
m--------- | waflib | 0 | ||||
-rw-r--r-- | waflib/Task.py | 1406 |
2 files changed, 0 insertions, 1406 deletions
diff --git a/waflib b/waflib new file mode 160000 +Subproject 2314e236ca6e7d94a26c3c17091da0f25f5867f diff --git a/waflib/Task.py b/waflib/Task.py deleted file mode 100644 index cb49a73..0000000 --- a/waflib/Task.py +++ /dev/null @@ -1,1406 +0,0 @@ -#!/usr/bin/env python -# encoding: utf-8 -# Thomas Nagy, 2005-2018 (ita) - -""" -Tasks represent atomic operations such as processes. -""" - -import os, re, sys, tempfile, traceback -from waflib import Utils, Logs, Errors - -# task states -NOT_RUN = 0 -"""The task was not executed yet""" - -MISSING = 1 -"""The task has been executed but the files have not been created""" - -CRASHED = 2 -"""The task execution returned a non-zero exit status""" - -EXCEPTION = 3 -"""An exception occurred in the task execution""" - -CANCELED = 4 -"""A dependency for the task is missing so it was cancelled""" - -SKIPPED = 8 -"""The task did not have to be executed""" - -SUCCESS = 9 -"""The task was successfully executed""" - -ASK_LATER = -1 -"""The task is not ready to be executed""" - -SKIP_ME = -2 -"""The task does not need to be executed""" - -RUN_ME = -3 -"""The task must be executed""" - -CANCEL_ME = -4 -"""The task cannot be executed because of a dependency problem""" - -COMPILE_TEMPLATE_SHELL = ''' -def f(tsk): - env = tsk.env - gen = tsk.generator - bld = gen.bld - cwdx = tsk.get_cwd() - p = env.get_flat - def to_list(xx): - if isinstance(xx, str): return [xx] - return xx - tsk.last_cmd = cmd = \'\'\' %s \'\'\' % s - return tsk.exec_command(cmd, cwd=cwdx, env=env.env or None) -''' - -COMPILE_TEMPLATE_NOSHELL = ''' -def f(tsk): - env = tsk.env - gen = tsk.generator - bld = gen.bld - cwdx = tsk.get_cwd() - def to_list(xx): - if isinstance(xx, str): return [xx] - return xx - def merge(lst1, lst2): - if lst1 and lst2: - return lst1[:-1] + [lst1[-1] + lst2[0]] + lst2[1:] - return lst1 + lst2 - lst = [] - %s - if '' in lst: - lst = [x for x in lst if x] - tsk.last_cmd = lst - return tsk.exec_command(lst, cwd=cwdx, env=env.env or None) -''' - -COMPILE_TEMPLATE_SIG_VARS = ''' -def f(tsk): - sig = tsk.generator.bld.hash_env_vars(tsk.env, tsk.vars) - tsk.m.update(sig) - env = tsk.env - gen = tsk.generator - bld = gen.bld - cwdx = tsk.get_cwd() - p = env.get_flat - buf = [] - %s - tsk.m.update(repr(buf).encode()) -''' - -classes = {} -""" -The metaclass :py:class:`waflib.Task.store_task_type` stores all class tasks -created by user scripts or Waf tools to this dict. It maps class names to class objects. -""" - -class store_task_type(type): - """ - Metaclass: store the task classes into the dict pointed by the - class attribute 'register' which defaults to :py:const:`waflib.Task.classes`, - - The attribute 'run_str' is compiled into a method 'run' bound to the task class. - """ - def __init__(cls, name, bases, dict): - super(store_task_type, cls).__init__(name, bases, dict) - name = cls.__name__ - - if name != 'evil' and name != 'Task': - if getattr(cls, 'run_str', None): - # if a string is provided, convert it to a method - (f, dvars) = compile_fun(cls.run_str, cls.shell) - cls.hcode = Utils.h_cmd(cls.run_str) - cls.orig_run_str = cls.run_str - # change the name of run_str or it is impossible to subclass with a function - cls.run_str = None - cls.run = f - # process variables - cls.vars = list(set(cls.vars + dvars)) - cls.vars.sort() - if cls.vars: - fun = compile_sig_vars(cls.vars) - if fun: - cls.sig_vars = fun - elif getattr(cls, 'run', None) and not 'hcode' in cls.__dict__: - # getattr(cls, 'hcode') would look in the upper classes - cls.hcode = Utils.h_cmd(cls.run) - - # be creative - getattr(cls, 'register', classes)[name] = cls - -evil = store_task_type('evil', (object,), {}) -"Base class provided to avoid writing a metaclass, so the code can run in python 2.6 and 3.x unmodified" - -class Task(evil): - """ - Task objects represents actions to perform such as commands to execute by calling the `run` method. - - Detecting when to execute a task occurs in the method :py:meth:`waflib.Task.Task.runnable_status`. - - Detecting which tasks to execute is performed through a hash value returned by - :py:meth:`waflib.Task.Task.signature`. The task signature is persistent from build to build. - """ - vars = [] - """ConfigSet variables that should trigger a rebuild (class attribute used for :py:meth:`waflib.Task.Task.sig_vars`)""" - - always_run = False - """Specify whether task instances must always be executed or not (class attribute)""" - - shell = False - """Execute the command with the shell (class attribute)""" - - color = 'GREEN' - """Color for the console display, see :py:const:`waflib.Logs.colors_lst`""" - - ext_in = [] - """File extensions that objects of this task class may use""" - - ext_out = [] - """File extensions that objects of this task class may create""" - - before = [] - """The instances of this class are executed before the instances of classes whose names are in this list""" - - after = [] - """The instances of this class are executed after the instances of classes whose names are in this list""" - - hcode = Utils.SIG_NIL - """String representing an additional hash for the class representation""" - - keep_last_cmd = False - """Whether to keep the last command executed on the instance after execution. - This may be useful for certain extensions but it can a lot of memory. - """ - - weight = 0 - """Optional weight to tune the priority for task instances. - The higher, the earlier. The weight only applies to single task objects.""" - - tree_weight = 0 - """Optional weight to tune the priority of task instances and whole subtrees. - The higher, the earlier.""" - - prio_order = 0 - """Priority order set by the scheduler on instances during the build phase. - You most likely do not need to set it. - """ - - __slots__ = ('hasrun', 'generator', 'env', 'inputs', 'outputs', 'dep_nodes', 'run_after') - - def __init__(self, *k, **kw): - self.hasrun = NOT_RUN - try: - self.generator = kw['generator'] - except KeyError: - self.generator = self - - self.env = kw['env'] - """:py:class:`waflib.ConfigSet.ConfigSet` object (make sure to provide one)""" - - self.inputs = [] - """List of input nodes, which represent the files used by the task instance""" - - self.outputs = [] - """List of output nodes, which represent the files created by the task instance""" - - self.dep_nodes = [] - """List of additional nodes to depend on""" - - self.run_after = set() - """Set of tasks that must be executed before this one""" - - def __lt__(self, other): - return self.priority() > other.priority() - def __le__(self, other): - return self.priority() >= other.priority() - def __gt__(self, other): - return self.priority() < other.priority() - def __ge__(self, other): - return self.priority() <= other.priority() - - def get_cwd(self): - """ - :return: current working directory - :rtype: :py:class:`waflib.Node.Node` - """ - bld = self.generator.bld - ret = getattr(self, 'cwd', None) or getattr(bld, 'cwd', bld.bldnode) - if isinstance(ret, str): - if os.path.isabs(ret): - ret = bld.root.make_node(ret) - else: - ret = self.generator.path.make_node(ret) - return ret - - def quote_flag(self, x): - """ - Surround a process argument by quotes so that a list of arguments can be written to a file - - :param x: flag - :type x: string - :return: quoted flag - :rtype: string - """ - old = x - if '\\' in x: - x = x.replace('\\', '\\\\') - if '"' in x: - x = x.replace('"', '\\"') - if old != x or ' ' in x or '\t' in x or "'" in x: - x = '"%s"' % x - return x - - def priority(self): - """ - Priority of execution; the higher, the earlier - - :return: the priority value - :rtype: a tuple of numeric values - """ - return (self.weight + self.prio_order, - getattr(self.generator, 'tg_idx_count', 0)) - - def split_argfile(self, cmd): - """ - Splits a list of process commands into the executable part and its list of arguments - - :return: a tuple containing the executable first and then the rest of arguments - :rtype: tuple - """ - return ([cmd[0]], [self.quote_flag(x) for x in cmd[1:]]) - - def exec_command(self, cmd, **kw): - """ - Wrapper for :py:meth:`waflib.Context.Context.exec_command`. - This version set the current working directory (``build.variant_dir``), - applies PATH settings (if self.env.PATH is provided), and can run long - commands through a temporary ``@argfile``. - - :param cmd: process command to execute - :type cmd: list of string (best) or string (process will use a shell) - :return: the return code - :rtype: int - - Optional parameters: - - #. cwd: current working directory (Node or string) - #. stdout: set to None to prevent waf from capturing the process standard output - #. stderr: set to None to prevent waf from capturing the process standard error - #. timeout: timeout value (Python 3) - """ - if not 'cwd' in kw: - kw['cwd'] = self.get_cwd() - - if hasattr(self, 'timeout'): - kw['timeout'] = self.timeout - - if self.env.PATH: - env = kw['env'] = dict(kw.get('env') or self.env.env or os.environ) - env['PATH'] = self.env.PATH if isinstance(self.env.PATH, str) else os.pathsep.join(self.env.PATH) - - if hasattr(self, 'stdout'): - kw['stdout'] = self.stdout - if hasattr(self, 'stderr'): - kw['stderr'] = self.stderr - - if not isinstance(cmd, str): - if Utils.is_win32: - # win32 compares the resulting length http://support.microsoft.com/kb/830473 - too_long = sum([len(arg) for arg in cmd]) + len(cmd) > 8192 - else: - # non-win32 counts the amount of arguments (200k) - too_long = len(cmd) > 200000 - - if too_long and getattr(self, 'allow_argsfile', True): - # Shunt arguments to a temporary file if the command is too long. - cmd, args = self.split_argfile(cmd) - try: - (fd, tmp) = tempfile.mkstemp() - os.write(fd, '\r\n'.join(args).encode()) - os.close(fd) - if Logs.verbose: - Logs.debug('argfile: @%r -> %r', tmp, args) - return self.generator.bld.exec_command(cmd + ['@' + tmp], **kw) - finally: - try: - os.remove(tmp) - except OSError: - # anti-virus and indexers can keep files open -_- - pass - return self.generator.bld.exec_command(cmd, **kw) - - def process(self): - """ - Runs the task and handles errors - - :return: 0 or None if everything is fine - :rtype: integer - """ - # remove the task signature immediately before it is executed - # so that the task will be executed again in case of failure - try: - del self.generator.bld.task_sigs[self.uid()] - except KeyError: - pass - - try: - ret = self.run() - except Exception: - self.err_msg = traceback.format_exc() - self.hasrun = EXCEPTION - else: - if ret: - self.err_code = ret - self.hasrun = CRASHED - else: - try: - self.post_run() - except Errors.WafError: - pass - except Exception: - self.err_msg = traceback.format_exc() - self.hasrun = EXCEPTION - else: - self.hasrun = SUCCESS - - if self.hasrun != SUCCESS and self.scan: - # rescan dependencies on next run - try: - del self.generator.bld.imp_sigs[self.uid()] - except KeyError: - pass - - def log_display(self, bld): - "Writes the execution status on the context logger" - if self.generator.bld.progress_bar == 3: - return - - s = self.display() - if s: - if bld.logger: - logger = bld.logger - else: - logger = Logs - - if self.generator.bld.progress_bar == 1: - c1 = Logs.colors.cursor_off - c2 = Logs.colors.cursor_on - logger.info(s, extra={'stream': sys.stderr, 'terminator':'', 'c1': c1, 'c2' : c2}) - else: - logger.info(s, extra={'terminator':'', 'c1': '', 'c2' : ''}) - - def display(self): - """ - Returns an execution status for the console, the progress bar, or the IDE output. - - :rtype: string - """ - col1 = Logs.colors(self.color) - col2 = Logs.colors.NORMAL - master = self.generator.bld.producer - - def cur(): - # the current task position, computed as late as possible - return master.processed - master.ready.qsize() - - if self.generator.bld.progress_bar == 1: - return self.generator.bld.progress_line(cur(), master.total, col1, col2) - - if self.generator.bld.progress_bar == 2: - ela = str(self.generator.bld.timer) - try: - ins = ','.join([n.name for n in self.inputs]) - except AttributeError: - ins = '' - try: - outs = ','.join([n.name for n in self.outputs]) - except AttributeError: - outs = '' - return '|Total %s|Current %s|Inputs %s|Outputs %s|Time %s|\n' % (master.total, cur(), ins, outs, ela) - - s = str(self) - if not s: - return None - - total = master.total - n = len(str(total)) - fs = '[%%%dd/%%%dd] %%s%%s%%s%%s\n' % (n, n) - kw = self.keyword() - if kw: - kw += ' ' - return fs % (cur(), total, kw, col1, s, col2) - - def hash_constraints(self): - """ - Identifies a task type for all the constraints relevant for the scheduler: precedence, file production - - :return: a hash value - :rtype: string - """ - return (tuple(self.before), tuple(self.after), tuple(self.ext_in), tuple(self.ext_out), self.__class__.__name__, self.hcode) - - def format_error(self): - """ - Returns an error message to display the build failure reasons - - :rtype: string - """ - if Logs.verbose: - msg = ': %r\n%r' % (self, getattr(self, 'last_cmd', '')) - else: - msg = ' (run with -v to display more information)' - name = getattr(self.generator, 'name', '') - if getattr(self, "err_msg", None): - return self.err_msg - elif not self.hasrun: - return 'task in %r was not executed for some reason: %r' % (name, self) - elif self.hasrun == CRASHED: - try: - return ' -> task in %r failed with exit status %r%s' % (name, self.err_code, msg) - except AttributeError: - return ' -> task in %r failed%s' % (name, msg) - elif self.hasrun == MISSING: - return ' -> missing files in %r%s' % (name, msg) - elif self.hasrun == CANCELED: - return ' -> %r canceled because of missing dependencies' % name - else: - return 'invalid status for task in %r: %r' % (name, self.hasrun) - - def colon(self, var1, var2): - """ - Enable scriptlet expressions of the form ${FOO_ST:FOO} - If the first variable (FOO_ST) is empty, then an empty list is returned - - The results will be slightly different if FOO_ST is a list, for example:: - - env.FOO = ['p1', 'p2'] - env.FOO_ST = '-I%s' - # ${FOO_ST:FOO} returns - ['-Ip1', '-Ip2'] - - env.FOO_ST = ['-a', '-b'] - # ${FOO_ST:FOO} returns - ['-a', '-b', 'p1', '-a', '-b', 'p2'] - """ - tmp = self.env[var1] - if not tmp: - return [] - - if isinstance(var2, str): - it = self.env[var2] - else: - it = var2 - if isinstance(tmp, str): - return [tmp % x for x in it] - else: - lst = [] - for y in it: - lst.extend(tmp) - lst.append(y) - return lst - - def __str__(self): - "string to display to the user" - name = self.__class__.__name__ - if self.outputs: - if name.endswith(('lib', 'program')) or not self.inputs: - node = self.outputs[0] - return node.path_from(node.ctx.launch_node()) - if not (self.inputs or self.outputs): - return self.__class__.__name__ - if len(self.inputs) == 1: - node = self.inputs[0] - return node.path_from(node.ctx.launch_node()) - - src_str = ' '.join([a.path_from(a.ctx.launch_node()) for a in self.inputs]) - tgt_str = ' '.join([a.path_from(a.ctx.launch_node()) for a in self.outputs]) - if self.outputs: - sep = ' -> ' - else: - sep = '' - return '%s: %s%s%s' % (self.__class__.__name__, src_str, sep, tgt_str) - - def keyword(self): - "Display keyword used to prettify the console outputs" - name = self.__class__.__name__ - if name.endswith(('lib', 'program')): - return 'Linking' - if len(self.inputs) == 1 and len(self.outputs) == 1: - return 'Compiling' - if not self.inputs: - if self.outputs: - return 'Creating' - else: - return 'Running' - return 'Processing' - - def __repr__(self): - "for debugging purposes" - try: - ins = ",".join([x.name for x in self.inputs]) - outs = ",".join([x.name for x in self.outputs]) - except AttributeError: - ins = ",".join([str(x) for x in self.inputs]) - outs = ",".join([str(x) for x in self.outputs]) - return "".join(['\n\t{task %r: ' % id(self), self.__class__.__name__, " ", ins, " -> ", outs, '}']) - - def uid(self): - """ - Returns an identifier used to determine if tasks are up-to-date. Since the - identifier will be stored between executions, it must be: - - - unique for a task: no two tasks return the same value (for a given build context) - - the same for a given task instance - - By default, the node paths, the class name, and the function are used - as inputs to compute a hash. - - The pointer to the object (python built-in 'id') will change between build executions, - and must be avoided in such hashes. - - :return: hash value - :rtype: string - """ - try: - return self.uid_ - except AttributeError: - m = Utils.md5(self.__class__.__name__) - up = m.update - for x in self.inputs + self.outputs: - up(x.abspath()) - self.uid_ = m.digest() - return self.uid_ - - def set_inputs(self, inp): - """ - Appends the nodes to the *inputs* list - - :param inp: input nodes - :type inp: node or list of nodes - """ - if isinstance(inp, list): - self.inputs += inp - else: - self.inputs.append(inp) - - def set_outputs(self, out): - """ - Appends the nodes to the *outputs* list - - :param out: output nodes - :type out: node or list of nodes - """ - if isinstance(out, list): - self.outputs += out - else: - self.outputs.append(out) - - def set_run_after(self, task): - """ - Run this task only after the given *task*. - - Calling this method from :py:meth:`waflib.Task.Task.runnable_status` may cause - build deadlocks; see :py:meth:`waflib.Tools.fc.fc.runnable_status` for details. - - :param task: task - :type task: :py:class:`waflib.Task.Task` - """ - assert isinstance(task, Task) - self.run_after.add(task) - - def signature(self): - """ - Task signatures are stored between build executions, they are use to track the changes - made to the input nodes (not to the outputs!). The signature hashes data from various sources: - - * explicit dependencies: files listed in the inputs (list of node objects) :py:meth:`waflib.Task.Task.sig_explicit_deps` - * implicit dependencies: list of nodes returned by scanner methods (when present) :py:meth:`waflib.Task.Task.sig_implicit_deps` - * hashed data: variables/values read from task.vars/task.env :py:meth:`waflib.Task.Task.sig_vars` - - If the signature is expected to give a different result, clear the cache kept in ``self.cache_sig``:: - - from waflib import Task - class cls(Task.Task): - def signature(self): - sig = super(Task.Task, self).signature() - delattr(self, 'cache_sig') - return super(Task.Task, self).signature() - - :return: the signature value - :rtype: string or bytes - """ - try: - return self.cache_sig - except AttributeError: - pass - - self.m = Utils.md5(self.hcode) - - # explicit deps - self.sig_explicit_deps() - - # env vars - self.sig_vars() - - # implicit deps / scanner results - if self.scan: - try: - self.sig_implicit_deps() - except Errors.TaskRescan: - return self.signature() - - ret = self.cache_sig = self.m.digest() - return ret - - def runnable_status(self): - """ - Returns the Task status - - :return: a task state in :py:const:`waflib.Task.RUN_ME`, - :py:const:`waflib.Task.SKIP_ME`, :py:const:`waflib.Task.CANCEL_ME` or :py:const:`waflib.Task.ASK_LATER`. - :rtype: int - """ - bld = self.generator.bld - if bld.is_install < 0: - return SKIP_ME - - for t in self.run_after: - if not t.hasrun: - return ASK_LATER - elif t.hasrun < SKIPPED: - # a dependency has an error - return CANCEL_ME - - # first compute the signature - try: - new_sig = self.signature() - except Errors.TaskNotReady: - return ASK_LATER - - # compare the signature to a signature computed previously - key = self.uid() - try: - prev_sig = bld.task_sigs[key] - except KeyError: - Logs.debug('task: task %r must run: it was never run before or the task code changed', self) - return RUN_ME - - if new_sig != prev_sig: - Logs.debug('task: task %r must run: the task signature changed', self) - return RUN_ME - - # compare the signatures of the outputs - for node in self.outputs: - sig = bld.node_sigs.get(node) - if not sig: - Logs.debug('task: task %r must run: an output node has no signature', self) - return RUN_ME - if sig != key: - Logs.debug('task: task %r must run: an output node was produced by another task', self) - return RUN_ME - if not node.exists(): - Logs.debug('task: task %r must run: an output node does not exist', self) - return RUN_ME - - return (self.always_run and RUN_ME) or SKIP_ME - - def post_run(self): - """ - Called after successful execution to record that the task has run by - updating the entry in :py:attr:`waflib.Build.BuildContext.task_sigs`. - """ - bld = self.generator.bld - for node in self.outputs: - if not node.exists(): - self.hasrun = MISSING - self.err_msg = '-> missing file: %r' % node.abspath() - raise Errors.WafError(self.err_msg) - bld.node_sigs[node] = self.uid() # make sure this task produced the files in question - bld.task_sigs[self.uid()] = self.signature() - if not self.keep_last_cmd: - try: - del self.last_cmd - except AttributeError: - pass - - def sig_explicit_deps(self): - """ - Used by :py:meth:`waflib.Task.Task.signature`; it hashes :py:attr:`waflib.Task.Task.inputs` - and :py:attr:`waflib.Task.Task.dep_nodes` signatures. - """ - bld = self.generator.bld - upd = self.m.update - - # the inputs - for x in self.inputs + self.dep_nodes: - upd(x.get_bld_sig()) - - # manual dependencies, they can slow down the builds - if bld.deps_man: - additional_deps = bld.deps_man - for x in self.inputs + self.outputs: - try: - d = additional_deps[x] - except KeyError: - continue - - for v in d: - try: - v = v.get_bld_sig() - except AttributeError: - if hasattr(v, '__call__'): - v = v() # dependency is a function, call it - upd(v) - - def sig_deep_inputs(self): - """ - Enable rebuilds on input files task signatures. Not used by default. - - Example: hashes of output programs can be unchanged after being re-linked, - despite the libraries being different. This method can thus prevent stale unit test - results (waf_unit_test.py). - - Hashing input file timestamps is another possibility for the implementation. - This may cause unnecessary rebuilds when input tasks are frequently executed. - Here is an implementation example:: - - lst = [] - for node in self.inputs + self.dep_nodes: - st = os.stat(node.abspath()) - lst.append(st.st_mtime) - lst.append(st.st_size) - self.m.update(Utils.h_list(lst)) - - The downside of the implementation is that it absolutely requires all build directory - files to be declared within the current build. - """ - bld = self.generator.bld - lst = [bld.task_sigs[bld.node_sigs[node]] for node in (self.inputs + self.dep_nodes) if node.is_bld()] - self.m.update(Utils.h_list(lst)) - - def sig_vars(self): - """ - Used by :py:meth:`waflib.Task.Task.signature`; it hashes :py:attr:`waflib.Task.Task.env` variables/values - When overriding this method, and if scriptlet expressions are used, make sure to follow - the code in :py:meth:`waflib.Task.Task.compile_sig_vars` to enable dependencies on scriptlet results. - - This method may be replaced on subclasses by the metaclass to force dependencies on scriptlet code. - """ - sig = self.generator.bld.hash_env_vars(self.env, self.vars) - self.m.update(sig) - - scan = None - """ - This method, when provided, returns a tuple containing: - - * a list of nodes corresponding to real files - * a list of names for files not found in path_lst - - For example:: - - from waflib.Task import Task - class mytask(Task): - def scan(self, node): - return ([], []) - - The first and second lists in the tuple are stored in :py:attr:`waflib.Build.BuildContext.node_deps` and - :py:attr:`waflib.Build.BuildContext.raw_deps` respectively. - """ - - def sig_implicit_deps(self): - """ - Used by :py:meth:`waflib.Task.Task.signature`; it hashes node signatures - obtained by scanning for dependencies (:py:meth:`waflib.Task.Task.scan`). - - The exception :py:class:`waflib.Errors.TaskRescan` is thrown - when a file has changed. In this case, the method :py:meth:`waflib.Task.Task.signature` is called - once again, and return here to call :py:meth:`waflib.Task.Task.scan` and searching for dependencies. - """ - bld = self.generator.bld - - # get the task signatures from previous runs - key = self.uid() - prev = bld.imp_sigs.get(key, []) - - # for issue #379 - if prev: - try: - if prev == self.compute_sig_implicit_deps(): - return prev - except Errors.TaskNotReady: - raise - except EnvironmentError: - # when a file was renamed, remove the stale nodes (headers in folders without source files) - # this will break the order calculation for headers created during the build in the source directory (should be uncommon) - # the behaviour will differ when top != out - for x in bld.node_deps.get(self.uid(), []): - if not x.is_bld() and not x.exists(): - try: - del x.parent.children[x.name] - except KeyError: - pass - del bld.imp_sigs[key] - raise Errors.TaskRescan('rescan') - - # no previous run or the signature of the dependencies has changed, rescan the dependencies - (bld.node_deps[key], bld.raw_deps[key]) = self.scan() - if Logs.verbose: - Logs.debug('deps: scanner for %s: %r; unresolved: %r', self, bld.node_deps[key], bld.raw_deps[key]) - - # recompute the signature and return it - try: - bld.imp_sigs[key] = self.compute_sig_implicit_deps() - except EnvironmentError: - for k in bld.node_deps.get(self.uid(), []): - if not k.exists(): - Logs.warn('Dependency %r for %r is missing: check the task declaration and the build order!', k, self) - raise - - def compute_sig_implicit_deps(self): - """ - Used by :py:meth:`waflib.Task.Task.sig_implicit_deps` for computing the actual hash of the - :py:class:`waflib.Node.Node` returned by the scanner. - - :return: a hash value for the implicit dependencies - :rtype: string or bytes - """ - upd = self.m.update - self.are_implicit_nodes_ready() - - # scanner returns a node that does not have a signature - # just *ignore* the error and let them figure out from the compiler output - # waf -k behaviour - for k in self.generator.bld.node_deps.get(self.uid(), []): - upd(k.get_bld_sig()) - return self.m.digest() - - def are_implicit_nodes_ready(self): - """ - For each node returned by the scanner, see if there is a task that creates it, - and infer the build order - - This has a low performance impact on null builds (1.86s->1.66s) thanks to caching (28s->1.86s) - """ - bld = self.generator.bld - try: - cache = bld.dct_implicit_nodes - except AttributeError: - bld.dct_implicit_nodes = cache = {} - - # one cache per build group - try: - dct = cache[bld.current_group] - except KeyError: - dct = cache[bld.current_group] = {} - for tsk in bld.cur_tasks: - for x in tsk.outputs: - dct[x] = tsk - - modified = False - for x in bld.node_deps.get(self.uid(), []): - if x in dct: - self.run_after.add(dct[x]) - modified = True - - if modified: - for tsk in self.run_after: - if not tsk.hasrun: - #print "task is not ready..." - raise Errors.TaskNotReady('not ready') -if sys.hexversion > 0x3000000: - def uid(self): - try: - return self.uid_ - except AttributeError: - m = Utils.md5(self.__class__.__name__.encode('latin-1', 'xmlcharrefreplace')) - up = m.update - for x in self.inputs + self.outputs: - up(x.abspath().encode('latin-1', 'xmlcharrefreplace')) - self.uid_ = m.digest() - return self.uid_ - uid.__doc__ = Task.uid.__doc__ - Task.uid = uid - -def is_before(t1, t2): - """ - Returns a non-zero value if task t1 is to be executed before task t2:: - - t1.ext_out = '.h' - t2.ext_in = '.h' - t2.after = ['t1'] - t1.before = ['t2'] - waflib.Task.is_before(t1, t2) # True - - :param t1: Task object - :type t1: :py:class:`waflib.Task.Task` - :param t2: Task object - :type t2: :py:class:`waflib.Task.Task` - """ - to_list = Utils.to_list - for k in to_list(t2.ext_in): - if k in to_list(t1.ext_out): - return 1 - - if t1.__class__.__name__ in to_list(t2.after): - return 1 - - if t2.__class__.__name__ in to_list(t1.before): - return 1 - - return 0 - -def set_file_constraints(tasks): - """ - Updates the ``run_after`` attribute of all tasks based on the task inputs and outputs - - :param tasks: tasks - :type tasks: list of :py:class:`waflib.Task.Task` - """ - ins = Utils.defaultdict(set) - outs = Utils.defaultdict(set) - for x in tasks: - for a in x.inputs: - ins[a].add(x) - for a in x.dep_nodes: - ins[a].add(x) - for a in x.outputs: - outs[a].add(x) - - links = set(ins.keys()).intersection(outs.keys()) - for k in links: - for a in ins[k]: - a.run_after.update(outs[k]) - - -class TaskGroup(object): - """ - Wrap nxm task order constraints into a single object - to prevent the creation of large list/set objects - - This is an optimization - """ - def __init__(self, prev, next): - self.prev = prev - self.next = next - self.done = False - - def get_hasrun(self): - for k in self.prev: - if not k.hasrun: - return NOT_RUN - return SUCCESS - - hasrun = property(get_hasrun, None) - -def set_precedence_constraints(tasks): - """ - Updates the ``run_after`` attribute of all tasks based on the after/before/ext_out/ext_in attributes - - :param tasks: tasks - :type tasks: list of :py:class:`waflib.Task.Task` - """ - cstr_groups = Utils.defaultdict(list) - for x in tasks: - h = x.hash_constraints() - cstr_groups[h].append(x) - - keys = list(cstr_groups.keys()) - maxi = len(keys) - - # this list should be short - for i in range(maxi): - t1 = cstr_groups[keys[i]][0] - for j in range(i + 1, maxi): - t2 = cstr_groups[keys[j]][0] - - # add the constraints based on the comparisons - if is_before(t1, t2): - a = i - b = j - elif is_before(t2, t1): - a = j - b = i - else: - continue - - a = cstr_groups[keys[a]] - b = cstr_groups[keys[b]] - - if len(a) < 2 or len(b) < 2: - for x in b: - x.run_after.update(a) - else: - group = TaskGroup(set(a), set(b)) - for x in b: - x.run_after.add(group) - -def funex(c): - """ - Compiles a scriptlet expression into a Python function - - :param c: function to compile - :type c: string - :return: the function 'f' declared in the input string - :rtype: function - """ - dc = {} - exec(c, dc) - return dc['f'] - -re_cond = re.compile(r'(?P<var>\w+)|(?P<or>\|)|(?P<and>&)') -re_novar = re.compile(r'^(SRC|TGT)\W+.*?$') -reg_act = re.compile(r'(?P<backslash>\\)|(?P<dollar>\$\$)|(?P<subst>\$\{(?P<var>\w+)(?P<code>.*?)\})', re.M) -def compile_fun_shell(line): - """ - Creates a compiled function to execute a process through a sub-shell - """ - extr = [] - def repl(match): - g = match.group - if g('dollar'): - return "$" - elif g('backslash'): - return '\\\\' - elif g('subst'): - extr.append((g('var'), g('code'))) - return "%s" - return None - line = reg_act.sub(repl, line) or line - dvars = [] - def add_dvar(x): - if x not in dvars: - dvars.append(x) - - def replc(m): - # performs substitutions and populates dvars - if m.group('and'): - return ' and ' - elif m.group('or'): - return ' or ' - else: - x = m.group('var') - add_dvar(x) - return 'env[%r]' % x - - parm = [] - app = parm.append - for (var, meth) in extr: - if var == 'SRC': - if meth: - app('tsk.inputs%s' % meth) - else: - app('" ".join([a.path_from(cwdx) for a in tsk.inputs])') - elif var == 'TGT': - if meth: - app('tsk.outputs%s' % meth) - else: - app('" ".join([a.path_from(cwdx) for a in tsk.outputs])') - elif meth: - if meth.startswith(':'): - add_dvar(var) - m = meth[1:] - if m == 'SRC': - m = '[a.path_from(cwdx) for a in tsk.inputs]' - elif m == 'TGT': - m = '[a.path_from(cwdx) for a in tsk.outputs]' - elif re_novar.match(m): - m = '[tsk.inputs%s]' % m[3:] - elif re_novar.match(m): - m = '[tsk.outputs%s]' % m[3:] - else: - add_dvar(m) - if m[:3] not in ('tsk', 'gen', 'bld'): - m = '%r' % m - app('" ".join(tsk.colon(%r, %s))' % (var, m)) - elif meth.startswith('?'): - # In A?B|C output env.A if one of env.B or env.C is non-empty - expr = re_cond.sub(replc, meth[1:]) - app('p(%r) if (%s) else ""' % (var, expr)) - else: - call = '%s%s' % (var, meth) - add_dvar(call) - app(call) - else: - add_dvar(var) - app("p('%s')" % var) - if parm: - parm = "%% (%s) " % (',\n\t\t'.join(parm)) - else: - parm = '' - - c = COMPILE_TEMPLATE_SHELL % (line, parm) - Logs.debug('action: %s', c.strip().splitlines()) - return (funex(c), dvars) - -reg_act_noshell = re.compile(r"(?P<space>\s+)|(?P<subst>\$\{(?P<var>\w+)(?P<code>.*?)\})|(?P<text>([^$ \t\n\r\f\v]|\$\$)+)", re.M) -def compile_fun_noshell(line): - """ - Creates a compiled function to execute a process without a sub-shell - """ - buf = [] - dvars = [] - merge = False - app = buf.append - - def add_dvar(x): - if x not in dvars: - dvars.append(x) - - def replc(m): - # performs substitutions and populates dvars - if m.group('and'): - return ' and ' - elif m.group('or'): - return ' or ' - else: - x = m.group('var') - add_dvar(x) - return 'env[%r]' % x - - for m in reg_act_noshell.finditer(line): - if m.group('space'): - merge = False - continue - elif m.group('text'): - app('[%r]' % m.group('text').replace('$$', '$')) - elif m.group('subst'): - var = m.group('var') - code = m.group('code') - if var == 'SRC': - if code: - app('[tsk.inputs%s]' % code) - else: - app('[a.path_from(cwdx) for a in tsk.inputs]') - elif var == 'TGT': - if code: - app('[tsk.outputs%s]' % code) - else: - app('[a.path_from(cwdx) for a in tsk.outputs]') - elif code: - if code.startswith(':'): - # a composed variable ${FOO:OUT} - add_dvar(var) - m = code[1:] - if m == 'SRC': - m = '[a.path_from(cwdx) for a in tsk.inputs]' - elif m == 'TGT': - m = '[a.path_from(cwdx) for a in tsk.outputs]' - elif re_novar.match(m): - m = '[tsk.inputs%s]' % m[3:] - elif re_novar.match(m): - m = '[tsk.outputs%s]' % m[3:] - else: - add_dvar(m) - if m[:3] not in ('tsk', 'gen', 'bld'): - m = '%r' % m - app('tsk.colon(%r, %s)' % (var, m)) - elif code.startswith('?'): - # In A?B|C output env.A if one of env.B or env.C is non-empty - expr = re_cond.sub(replc, code[1:]) - app('to_list(env[%r] if (%s) else [])' % (var, expr)) - else: - # plain code such as ${tsk.inputs[0].abspath()} - call = '%s%s' % (var, code) - add_dvar(call) - app('to_list(%s)' % call) - else: - # a plain variable such as # a plain variable like ${AR} - app('to_list(env[%r])' % var) - add_dvar(var) - if merge: - tmp = 'merge(%s, %s)' % (buf[-2], buf[-1]) - del buf[-1] - buf[-1] = tmp - merge = True # next turn - - buf = ['lst.extend(%s)' % x for x in buf] - fun = COMPILE_TEMPLATE_NOSHELL % "\n\t".join(buf) - Logs.debug('action: %s', fun.strip().splitlines()) - return (funex(fun), dvars) - -def compile_fun(line, shell=False): - """ - Parses a string expression such as '${CC} ${SRC} -o ${TGT}' and returns a pair containing: - - * The function created (compiled) for use as :py:meth:`waflib.Task.Task.run` - * The list of variables that must cause rebuilds when *env* data is modified - - for example:: - - from waflib.Task import compile_fun - compile_fun('cxx', '${CXX} -o ${TGT[0]} ${SRC} -I ${SRC[0].parent.bldpath()}') - - def build(bld): - bld(source='wscript', rule='echo "foo\\${SRC[0].name}\\bar"') - - The env variables (CXX, ..) on the task must not hold dicts so as to preserve a consistent order. - The reserved keywords ``TGT`` and ``SRC`` represent the task input and output nodes - - """ - if isinstance(line, str): - if line.find('<') > 0 or line.find('>') > 0 or line.find('&&') > 0: - shell = True - else: - dvars_lst = [] - funs_lst = [] - for x in line: - if isinstance(x, str): - fun, dvars = compile_fun(x, shell) - dvars_lst += dvars - funs_lst.append(fun) - else: - # assume a function to let through - funs_lst.append(x) - def composed_fun(task): - for x in funs_lst: - ret = x(task) - if ret: - return ret - return None - return composed_fun, dvars_lst - if shell: - return compile_fun_shell(line) - else: - return compile_fun_noshell(line) - -def compile_sig_vars(vars): - """ - This method produces a sig_vars method suitable for subclasses that provide - scriptlet code in their run_str code. - If no such method can be created, this method returns None. - - The purpose of the sig_vars method returned is to ensures - that rebuilds occur whenever the contents of the expression changes. - This is the case B below:: - - import time - # case A: regular variables - tg = bld(rule='echo ${FOO}') - tg.env.FOO = '%s' % time.time() - # case B - bld(rule='echo ${gen.foo}', foo='%s' % time.time()) - - :param vars: env variables such as CXXFLAGS or gen.foo - :type vars: list of string - :return: A sig_vars method relevant for dependencies if adequate, else None - :rtype: A function, or None in most cases - """ - buf = [] - for x in sorted(vars): - if x[:3] in ('tsk', 'gen', 'bld'): - buf.append('buf.append(%s)' % x) - if buf: - return funex(COMPILE_TEMPLATE_SIG_VARS % '\n\t'.join(buf)) - return None - -def task_factory(name, func=None, vars=None, color='GREEN', ext_in=[], ext_out=[], before=[], after=[], shell=False, scan=None): - """ - Returns a new task subclass with the function ``run`` compiled from the line given. - - :param func: method run - :type func: string or function - :param vars: list of variables to hash - :type vars: list of string - :param color: color to use - :type color: string - :param shell: when *func* is a string, enable/disable the use of the shell - :type shell: bool - :param scan: method scan - :type scan: function - :rtype: :py:class:`waflib.Task.Task` - """ - - params = { - 'vars': vars or [], # function arguments are static, and this one may be modified by the class - 'color': color, - 'name': name, - 'shell': shell, - 'scan': scan, - } - - if isinstance(func, str) or isinstance(func, tuple): - params['run_str'] = func - else: - params['run'] = func - - cls = type(Task)(name, (Task,), params) - classes[name] = cls - - if ext_in: - cls.ext_in = Utils.to_list(ext_in) - if ext_out: - cls.ext_out = Utils.to_list(ext_out) - if before: - cls.before = Utils.to_list(before) - if after: - cls.after = Utils.to_list(after) - - return cls - -def deep_inputs(cls): - """ - Task class decorator to enable rebuilds on input files task signatures - """ - def sig_explicit_deps(self): - Task.sig_explicit_deps(self) - Task.sig_deep_inputs(self) - cls.sig_explicit_deps = sig_explicit_deps - return cls - -TaskBase = Task -"Provided for compatibility reasons, TaskBase should not be used" - -class TaskSemaphore(object): - """ - Task semaphores provide a simple and efficient way of throttling the amount of - a particular task to run concurrently. The throttling value is capped - by the amount of maximum jobs, so for example, a `TaskSemaphore(10)` - has no effect in a `-j2` build. - - Task semaphores are typically specified on the task class level:: - - class compile(waflib.Task.Task): - semaphore = waflib.Task.TaskSemaphore(2) - run_str = 'touch ${TGT}' - - Task semaphores are meant to be used by the build scheduler in the main - thread, so there are no guarantees of thread safety. - """ - def __init__(self, num): - """ - :param num: maximum value of concurrent tasks - :type num: int - """ - self.num = num - self.locking = set() - self.waiting = set() - - def is_locked(self): - """Returns True if this semaphore cannot be acquired by more tasks""" - return len(self.locking) >= self.num - - def acquire(self, tsk): - """ - Mark the semaphore as used by the given task (not re-entrant). - - :param tsk: task object - :type tsk: :py:class:`waflib.Task.Task` - :raises: :py:class:`IndexError` in case the resource is already acquired - """ - if self.is_locked(): - raise IndexError('Cannot lock more %r' % self.locking) - self.locking.add(tsk) - - def release(self, tsk): - """ - Mark the semaphore as unused by the given task. - - :param tsk: task object - :type tsk: :py:class:`waflib.Task.Task` - :raises: :py:class:`KeyError` in case the resource is not acquired by the task - """ - self.locking.remove(tsk) - |