diff options
Diffstat (limited to 'mesonbuild/rewriter.py')
| -rw-r--r-- | mesonbuild/rewriter.py | 366 |
1 files changed, 228 insertions, 138 deletions
diff --git a/mesonbuild/rewriter.py b/mesonbuild/rewriter.py index 981dfd326..9023033b9 100644 --- a/mesonbuild/rewriter.py +++ b/mesonbuild/rewriter.py @@ -11,13 +11,15 @@ from __future__ import annotations from .ast import IntrospectionInterpreter, BUILD_TARGET_FUNCTIONS, AstConditionLevel, AstIDGenerator, AstIndentationGenerator, AstPrinter from .ast.interpreter import IntrospectionBuildTarget, IntrospectionDependency, _symbol -from .interpreterbase import TV_func -from mesonbuild.mesonlib import MesonException, setup_vsenv +from .interpreterbase import UnknownValue, TV_func +from .interpreterbase.helpers import flatten +from mesonbuild.mesonlib import MesonException, setup_vsenv, relpath from . import mlog, environment from functools import wraps -from .mparser import Token, ArrayNode, ArgumentNode, AssignmentNode, BaseNode, StringNode, BooleanNode, ElementaryNode, IdNode, FunctionNode, PlusAssignmentNode -import json, os, re, sys +from .mparser import Token, ArrayNode, ArgumentNode, ArithmeticNode, AssignmentNode, BaseNode, StringNode, BooleanNode, ElementaryNode, IdNode, FunctionNode, PlusAssignmentNode +import json, os, re, sys, codecs import typing as T +from pathlib import Path if T.TYPE_CHECKING: import argparse @@ -633,6 +635,166 @@ class Rewriter: return ass return None + def affects_no_other_targets(self, candidate: BaseNode) -> bool: + affected = self.interpreter.dataflow_dag.reachable({candidate}, False) + affected_targets = [x for x in affected if isinstance(x, FunctionNode) and x.func_name.value in BUILD_TARGET_FUNCTIONS] + return len(affected_targets) == 1 + + def get_relto(self, target_node: BaseNode, node: BaseNode) -> Path: + cwd = Path(os.getcwd()) + all_paths = self.interpreter.dataflow_dag.find_all_paths(node, target_node) + # len(all_paths) == 0 would imply that data does not flow from node to + # target_node. This would imply that adding sources to node would not + # add the source to the target. + assert all_paths + if len(all_paths) > 1: + return None + return (cwd / next(x for x in all_paths[0] if isinstance(x, FunctionNode)).filename).parent + + def add_src_or_extra(self, op: str, target: IntrospectionBuildTarget, newfiles: T.List[str], to_sort_nodes: T.List[T.Union[FunctionNode, ArrayNode]]) -> None: + assert op in {'src_add', 'extra_files_add'} + + if op == 'src_add': + old: T.Set[T.Union[BaseNode, UnknownValue]] = set(target.source_nodes) + elif op == 'extra_files_add': + if target.extra_files is None: + old = set() + else: + old = {target.extra_files} + tgt_function: FunctionNode = target.node + + cwd = Path(os.getcwd()) + target_dir_abs = cwd / os.path.dirname(target.node.filename) + source_root_abs = cwd / self.interpreter.source_root + + candidates1 = self.interpreter.dataflow_dag.reachable(old, True) + # A node is a member of the set `candidates1` exactly if data from this node + # flow into one of the `dest` nodes. We assume that this implies that if we + # add `foo.c` to this node, then 'foo.c' will be added to one of these + # nodes. This assumption is not always true: + # ar = ['a.c', 'b.c'] + # srcs = ar[1] + # executable('name', srcs) + # Data flows from `ar` to `srcs`, but if we add 'foo.c': + # ar = ['a.c', 'b.c', 'foo.c'] + # srcs = ar[1] + # executable('name', srcs) + # this does not add 'foo.c' to `srcs`. This is a known bug/limitation of + # the meson rewriter that could be fixed by replacing `reachable` with a + # more advanced analysis. But this is a lot of work and I think e.g. + # `srcs = ar[1]` is rare in real-world projects, so I will just leave + # this for now. + + candidates2 = {x for x in candidates1 if isinstance(x, (FunctionNode, ArrayNode))} + + # If we have this meson.build file: + # shared = ['shared.c'] + # executable('foo', shared + ['foo.c']) + # executable('bar', shared + ['bar.c']) + # and we are tasked with adding 'new.c' to 'foo', we should do e.g this: + # shared = ['shared.c'] + # executable('foo', shared + ['foo.c', 'new.c']) + # executable('bar', shared + ['bar.c']) + # but never this: + # shared = ['shared.c', 'new.c'] + # executable('foo', shared + ['foo.c']) + # executable('bar', shared + ['bar.c']) + # We do this by removing the `['shared.c']`-node from `candidates2`. + candidates2 = {x for x in candidates2 if self.affects_no_other_targets(x)} + + def path_contains_unknowns(candidate: BaseNode) -> bool: + all_paths = self.interpreter.dataflow_dag.find_all_paths(candidate, target.node) + for path in all_paths: + for el in path: + if isinstance(el, UnknownValue): + return True + return False + + candidates2 = {x for x in candidates2 if not path_contains_unknowns(x)} + + candidates2 = {x for x in candidates2 if self.get_relto(target.node, x) is not None} + + chosen: T.Union[FunctionNode, ArrayNode] = None + new_kwarg_flag = False + if len(candidates2) > 0: + # So that files(['a', 'b']) gets modified to files(['a', 'b', 'c']) instead of files(['a', 'b'], 'c') + if len({x for x in candidates2 if isinstance(x, ArrayNode)}) > 0: + candidates2 = {x for x in candidates2 if isinstance(x, ArrayNode)} + + # We choose one more or less arbitrary candidate + chosen = min(candidates2, key=lambda x: (x.lineno, x.colno)) + elif op == 'src_add': + chosen = target.node + elif op == 'extra_files_add': + chosen = ArrayNode(_symbol('['), ArgumentNode(Token('', tgt_function.filename, 0, 0, 0, None, '[]')), _symbol(']')) + + # this is fundamentally error prone + self.interpreter.dataflow_dag.add_edge(chosen, target.node) + + extra_files_idnode = IdNode(Token('string', tgt_function.filename, 0, 0, 0, None, 'extra_files')) + if tgt_function not in self.modified_nodes: + self.modified_nodes += [tgt_function] + new_extra_files_node: BaseNode + if target.node.args.get_kwarg_or_default('extra_files', None) is None: + # Target has no extra_files kwarg, create one + new_kwarg_flag = True + new_extra_files_node = chosen + else: + new_kwarg_flag = True + old_extra_files = target.node.args.get_kwarg_or_default('extra_files', None) + target.node.args.kwargs = {k: v for k, v in target.node.args.kwargs.items() if not (isinstance(k, IdNode) and k.value == 'extra_files')} + new_extra_files_node = ArithmeticNode('add', old_extra_files, _symbol('+'), chosen) + + tgt_function.args.kwargs[extra_files_idnode] = new_extra_files_node + + newfiles_relto = self.get_relto(target.node, chosen) + old_src_list: T.List[T.Any] = flatten([self.interpreter.node_to_runtime_value(sn) for sn in old]) + + if op == 'src_add': + name = 'Source' + elif op == 'extra_files_add': + name = 'Extra file' + # Generate the new String nodes + to_append = [] + added = [] + + old_src_list = [(target_dir_abs / x).resolve() if isinstance(x, str) else x.to_abs_path(source_root_abs) for x in old_src_list if not isinstance(x, UnknownValue)] + for _newf in sorted(set(newfiles)): + newf = Path(_newf) + if os.path.isabs(newf): + newf = Path(newf) + else: + newf = source_root_abs / newf + if newf in old_src_list: + mlog.log(' -- ', name, mlog.green(str(newf)), 'is already defined for the target --> skipping') + continue + + mlog.log(' -- Adding ', name.lower(), mlog.green(str(newf)), 'at', + mlog.yellow(f'{chosen.filename}:{chosen.lineno}')) + added.append(newf) + mocktarget = self.interpreter.funcvals[target.node] + assert isinstance(mocktarget, IntrospectionBuildTarget) + # print("adding ", str(newf), 'to', mocktarget.name) todo: should we write something to stderr? + + path = relpath(newf, newfiles_relto) + path = codecs.encode(path, 'unicode_escape').decode() # Because the StringNode constructor does the inverse + token = Token('string', chosen.filename, 0, 0, 0, None, path) + to_append += [StringNode(token)] + + assert isinstance(chosen, (FunctionNode, ArrayNode)) + arg_node = chosen.args + # Append to the AST at the right place + arg_node.arguments += to_append + + # Mark the node as modified + if chosen not in to_sort_nodes: + to_sort_nodes += [chosen] + # If the extra_files array is newly created, i.e. if new_kwarg_flag is + # True, don't mark it as its parent function node already is, otherwise + # this would cause double modification. + if chosen not in self.modified_nodes and not new_kwarg_flag: + self.modified_nodes += [chosen] + # Utility function to get a list of the sources from a node def arg_list_from_node(self, n: BaseNode) -> T.List[BaseNode]: args = [] @@ -646,18 +808,26 @@ class Rewriter: args = n.arguments return args - def rm_src_or_extra(self, op: str, target: IntrospectionBuildTarget, to_be_removed: T.List[str], to_sort_nodes: T.List[ArgumentNode]) -> None: + def rm_src_or_extra(self, op: str, target: IntrospectionBuildTarget, to_be_removed: T.List[str], to_sort_nodes: T.List[T.Union[FunctionNode, ArrayNode]]) -> None: + assert op in {'src_rm', 'extra_files_rm'} + cwd = Path(os.getcwd()) + source_root_abs = cwd / self.interpreter.source_root + # Helper to find the exact string node and its parent - def find_node(src: str) -> T.Tuple[BaseNode, StringNode]: + def find_node(src: str) -> T.Tuple[T.Optional[BaseNode], T.Optional[StringNode]]: if op == 'src_rm': - nodes = target.source_nodes + nodes = self.interpreter.dataflow_dag.reachable(set(target.source_nodes), True).union({target.node}) elif op == 'extra_files_rm': - nodes = target.extra_files + nodes = self.interpreter.dataflow_dag.reachable({target.extra_files}, True) for i in nodes: - for j in self.arg_list_from_node(i): - if isinstance(j, StringNode): - if j.value == src: - return i, j + if isinstance(i, UnknownValue): + continue + relto = self.get_relto(target.node, i) + if relto is not None: + for j in self.arg_list_from_node(i): + if isinstance(j, StringNode): + if os.path.normpath(relto / j.value) == os.path.normpath(source_root_abs / src): + return i, j return None, None if op == 'src_rm': @@ -671,22 +841,22 @@ class Rewriter: if root is None: mlog.warning(' -- Unable to find', name, mlog.green(i), 'in the target') continue + if not self.affects_no_other_targets(string_node): + mlog.warning(' -- Removing the', name, mlog.green(i), 'is too compilicated') + continue - arg_node = None - if isinstance(root, (FunctionNode, ArrayNode)): - arg_node = root.args - elif isinstance(root, ArgumentNode): - arg_node = root - assert arg_node is not None + if not isinstance(root, (FunctionNode, ArrayNode)): + raise NotImplementedError # I'm lazy # Remove the found string node from the argument list + arg_node = root.args mlog.log(' -- Removing', name, mlog.green(i), 'from', mlog.yellow(f'{string_node.filename}:{string_node.lineno}')) arg_node.arguments.remove(string_node) # Mark the node as modified - if arg_node not in to_sort_nodes and not isinstance(root, FunctionNode): - to_sort_nodes += [arg_node] + if root not in to_sort_nodes: + to_sort_nodes += [root] if root not in self.modified_nodes: self.modified_nodes += [root] @@ -711,106 +881,10 @@ class Rewriter: if target is not None: cmd['sources'] = [rel_source(x) for x in cmd['sources']] - to_sort_nodes = [] - - if cmd['operation'] == 'src_add': - node = None - if target.source_nodes: - node = target.source_nodes[0] - else: - node = target.node - assert node is not None - - # Generate the current source list - src_list = [] - for src_node in target.source_nodes: - for j in self.arg_list_from_node(src_node): - if isinstance(j, StringNode): - src_list += [j.value] - - # Generate the new String nodes - to_append = [] - for i in sorted(set(cmd['sources'])): - if i in src_list: - mlog.log(' -- Source', mlog.green(i), 'is already defined for the target --> skipping') - continue - mlog.log(' -- Adding source', mlog.green(i), 'at', - mlog.yellow(f'{node.filename}:{node.lineno}')) - token = Token('string', node.filename, 0, 0, 0, None, i) - to_append += [StringNode(token)] - - # Append to the AST at the right place - arg_node = None - if isinstance(node, (FunctionNode, ArrayNode)): - arg_node = node.args - elif isinstance(node, ArgumentNode): - arg_node = node - assert arg_node is not None - arg_node.arguments += to_append - - # Mark the node as modified - if arg_node not in to_sort_nodes and not isinstance(node, FunctionNode): - to_sort_nodes += [arg_node] - if node not in self.modified_nodes: - self.modified_nodes += [node] - - elif cmd['operation'] == 'extra_files_add': - tgt_function: FunctionNode = target.node - mark_array = True - try: - node = target.extra_files[0] - except IndexError: - # Specifying `extra_files` with a list that flattens to empty gives an empty - # target.extra_files list, account for that. - try: - extra_files_key = next(k for k in tgt_function.args.kwargs.keys() if isinstance(k, IdNode) and k.value == 'extra_files') - node = tgt_function.args.kwargs[extra_files_key] - except StopIteration: - # Target has no extra_files kwarg, create one - node = ArrayNode(_symbol('['), ArgumentNode(Token('', tgt_function.filename, 0, 0, 0, None, '[]')), _symbol(']')) - tgt_function.args.kwargs[IdNode(Token('string', tgt_function.filename, 0, 0, 0, None, 'extra_files'))] = node - mark_array = False - if tgt_function not in self.modified_nodes: - self.modified_nodes += [tgt_function] - target.extra_files = [node] - if isinstance(node, IdNode): - cv = self.interpreter.get_cur_value(node.value) - assert isinstance(cv, BaseNode) - node = cv - target.extra_files = [node] - if not isinstance(node, ArrayNode): - mlog.error('Target', mlog.bold(cmd['target']), 'extra_files argument must be a list', *self.on_error()) - return self.handle_error() - - # Generate the current extra files list - extra_files_list = [] - for i in target.extra_files: - for j in self.arg_list_from_node(i): - if isinstance(j, StringNode): - extra_files_list += [j.value] - - # Generate the new String nodes - to_append = [] - for i in sorted(set(cmd['sources'])): - if i in extra_files_list: - mlog.log(' -- Extra file', mlog.green(i), 'is already defined for the target --> skipping') - continue - mlog.log(' -- Adding extra file', mlog.green(i), 'at', - mlog.yellow(f'{node.filename}:{node.lineno}')) - token = Token('string', node.filename, 0, 0, 0, None, i) - to_append += [StringNode(token)] + to_sort_nodes: T.List[T.Union[FunctionNode, ArrayNode]] = [] - # Append to the AST at the right place - arg_node = node.args - arg_node.arguments += to_append - - # Mark the node as modified - if arg_node not in to_sort_nodes: - to_sort_nodes += [arg_node] - # If the extra_files array is newly created, don't mark it as its parent function node already is, - # otherwise this would cause double modification. - if mark_array and node not in self.modified_nodes: - self.modified_nodes += [node] + if cmd['operation'] in {'src_add', 'extra_files_add'}: + self.add_src_or_extra(cmd['operation'], target, cmd['sources'], to_sort_nodes) elif cmd['operation'] in {'src_rm', 'extra_files_rm'}: self.rm_src_or_extra(cmd['operation'], target, cmd['sources'], to_sort_nodes) @@ -823,7 +897,7 @@ class Rewriter: id_base = re.sub(r'[- ]', '_', cmd['target']) target_id = id_base + '_exe' if cmd['target_type'] == 'executable' else '_lib' source_id = id_base + '_sources' - filename = os.path.join(cmd['subdir'], environment.build_filename) + filename = os.path.join(os.getcwd(), self.interpreter.source_root, cmd['subdir'], environment.build_filename) # Build src list src_arg_node = ArgumentNode(Token('string', filename, 0, 0, 0, None, '')) @@ -857,16 +931,16 @@ class Rewriter: elif cmd['operation'] == 'info': # T.List all sources in the target - src_list = [] - for i in target.source_nodes: - for j in self.arg_list_from_node(i): - if isinstance(j, StringNode): - src_list += [j.value] - extra_files_list = [] - for i in target.extra_files: - for j in self.arg_list_from_node(i): - if isinstance(j, StringNode): - extra_files_list += [j.value] + + cwd = Path(os.getcwd()) + source_root_abs = cwd / self.interpreter.source_root + + src_list = self.interpreter.nodes_to_pretty_filelist(source_root_abs, target.subdir, target.source_nodes) + extra_files_list = self.interpreter.nodes_to_pretty_filelist(source_root_abs, target.subdir, [target.extra_files] if target.extra_files else []) + + src_list = ['unknown' if isinstance(x, UnknownValue) else relpath(x, source_root_abs) for x in src_list] + extra_files_list = ['unknown' if isinstance(x, UnknownValue) else relpath(x, source_root_abs) for x in extra_files_list] + test_data = { 'name': target.name, 'sources': src_list, @@ -885,10 +959,16 @@ class Rewriter: def path_sorter(key: str) -> T.List[T.Tuple[bool, T.List[T.Union[int, str]]]]: return [(key.count('/') <= idx, alphanum_key(x)) for idx, x in enumerate(key.split('/'))] - unknown = [x for x in i.arguments if not isinstance(x, StringNode)] - sources = [x for x in i.arguments if isinstance(x, StringNode)] + if isinstance(i, FunctionNode) and i.func_name.value in BUILD_TARGET_FUNCTIONS: + src_args = i.args.arguments[1:] + target_name = [i.args.arguments[0]] + else: + src_args = i.args.arguments + target_name = [] + unknown: T.List[BaseNode] = [x for x in src_args if not isinstance(x, StringNode)] + sources: T.List[StringNode] = [x for x in src_args if isinstance(x, StringNode)] sources = sorted(sources, key=lambda x: path_sorter(x.value)) - i.arguments = unknown + T.cast(T.List[BaseNode], sources) + i.args.arguments = target_name + unknown + T.cast(T.List[BaseNode], sources) def process(self, cmd: T.Dict[str, T.Any]) -> None: if 'type' not in cmd: @@ -931,7 +1011,7 @@ class Rewriter: for i in str_list: if i['file'] in files: continue - fpath = os.path.realpath(os.path.join(self.sourcedir, T.cast(str, i['file']))) + fpath = os.path.realpath(T.cast(str, i['file'])) fdata = '' # Create an empty file if it does not exist if not os.path.exists(fpath): @@ -1077,12 +1157,22 @@ def run(options: argparse.Namespace) -> int: if not isinstance(commands, list): raise TypeError('Command is not a list') - for i in commands: - if not isinstance(i, object): + for i, cmd in enumerate(commands): + if not isinstance(cmd, object): raise TypeError('Command is not an object') - rewriter.process(i) + rewriter.process(cmd) + rewriter.apply_changes() + + if i == len(commands) - 1: # Improves the performance, is not necessary for correctness. + break + + rewriter.modified_nodes = [] + rewriter.to_remove_nodes = [] + rewriter.to_add_nodes = [] + # The AST changed, so we need to update every information that was derived from the AST + rewriter.interpreter = IntrospectionInterpreter(rewriter.sourcedir, '', rewriter.interpreter.backend, visitors = [AstIDGenerator(), AstIndentationGenerator(), AstConditionLevel()]) + rewriter.analyze_meson() - rewriter.apply_changes() rewriter.print_info() return 0 except Exception as e: |