def main(self):
"""
IN: module_linos
OT: (A, B)
A: self.module_calls (updated)
B: prj_modules
"""
module_indexing = ModuleIndexing(
self.module_helper, self.ast_tree, self.ast_indents
)
prj_modules = module_indexing.find_prj_modules()
module_linos = module_indexing.indexing_module_linos()
assign_analyser = AssignAnalyser(
self.module_helper, self.ast_tree, self.ast_indents
)
self.line_parser = LineParser(
self.module_helper.get_top_module(),
assign_analyser.top_assigns
)
# ------------------------------------------------
for module, linos in module_linos.items():
var_reachables, parent_module = assign_analyser \
.indexing_assign_reachables(
module, module_linos
)
self.line_parser.reset(var_reachables, parent_module)
self.analyse_module(module, linos)
# ------------------------------------------------
return self.module_calls, prj_modules
def find_global_vars(self):
"""
哪些是全局变量:
runtime 层级的 Import, ImportFrom
runtime 层级的 Assign
行内的 `global xxx`
IN: module_linos: provided by src.module_analyser.ModuleIndexing
#indexing_module_linos
self.module_helper
self.ast_tree
self.ast_indents
OT: dict. {var: module}
"""
top_linos = tuple(
lino for lino, indent in self.ast_indents.items()
if indent == 0
)
lk.logt('[D3743]', self.top_module, top_linos)
# ------------------------------------------------
# runtime 层级的 Import, ImportFrom & runtime 层级的 Assign
line_parser = LineParser(self.top_module)
for lino in top_linos:
ast_line = self.ast_tree[lino]
lk.logt('[TEMPRINT]_20190811_214127', lino, ast_line)
line_parser.main(ast_line)
# line_parser 会自动帮我们处理 ast_line 涉及的 Import, ImportFrom,
# Assign 等的变量与 module 的对照关系.
# ------------------------------------------------
# 行内的 `global xxx`
for lino in self.ast_indents:
if lino in top_linos:
continue
pass # TODO
return line_parser.get_vars()
class ModuleAnalyser:
line_parser = None
def __init__(self, module_helper: ModuleHelper, ast_tree, ast_indents):
self.module_helper = module_helper
self.ast_tree = ast_tree
self.ast_indents = ast_indents
self.module_calls = {} # format: {module: [call, ...], ...}
def main(self):
"""
IN: module_linos
OT: (A, B)
A: self.module_calls (updated)
B: prj_modules
"""
module_indexing = ModuleIndexing(
self.module_helper, self.ast_tree, self.ast_indents
)
prj_modules = module_indexing.find_prj_modules()
module_linos = module_indexing.indexing_module_linos()
assign_analyser = AssignAnalyser(
self.module_helper, self.ast_tree, self.ast_indents
)
self.line_parser = LineParser(
self.module_helper.get_top_module(),
assign_analyser.top_assigns
)
# ------------------------------------------------
for module, linos in module_linos.items():
var_reachables, parent_module = assign_analyser \
.indexing_assign_reachables(
module, module_linos
)
self.line_parser.reset(var_reachables, parent_module)
self.analyse_module(module, linos)
# ------------------------------------------------
return self.module_calls, prj_modules
def analyse_module(self, module, linos):
"""
发现该 module 下的与其他 module 之间的调用关系.
"""
lk.logd('analyse_module', module, style='■')
# lk.logt('[TEMPRINT]_20190811_214927',
# self.line_parser.get_global_vars())
related_calls = []
for lino in linos:
ast_line = self.ast_tree[lino]
module_called = self.analyse_line(ast_line)
# lk.logt('[D3233]', module_called)
for m in module_called:
if m not in related_calls:
related_calls.append(m)
lk.logt('[I3259]', related_calls)
self.module_calls.update({module: tuple(related_calls)})
def analyse_line(self, ast_line):
return self.line_parser.main(ast_line)
class ModuleAnalyser:
line_parser = None
def __init__(self, module_helper: ModuleHelper, ast_tree, ast_indents):
self.module_helper = module_helper
self.ast_tree = ast_tree
self.ast_indents = ast_indents
self.module_calls = {} # format: {module: [call, ...], ...}
def main(self):
"""
IN: module_linos
OT: (A, B)
A: self.module_calls (updated)
B: prj_modules
"""
module_indexing = ModuleIndexing(
self.module_helper, self.ast_tree, self.ast_indents
)
assign_analyser = AssignAnalyser(
self.module_helper, self.ast_tree, self.ast_indents
)
prj_modules = module_indexing.find_prj_modules()
module_linos = module_indexing.indexing_module_linos()
self.line_parser = LineParser(assign_analyser.top_assigns)
# ------------------------------------------------
for module, linos in module_linos.items():
var_reachables = assign_analyser.indexing_assign_reachables(
module, module_linos
)
self.analyse_module(module, linos, var_reachables)
# ------------------------------------------------
return self.module_calls, prj_modules
def analyse_module(self, module, linos, var_reachables):
"""
发现该 module 下的与其他 module 之间的调用关系.
"""
lk.logd('analyse_module', module, style='■')
related_calls = []
self.line_parser.reset(var_reachables)
for lino in linos:
ast_line = self.ast_tree[lino]
modules = self.analyse_line(ast_line)
for m in modules:
if m not in related_calls:
related_calls.append(m)
self.module_calls.update({module: tuple(related_calls)})
def analyse_line(self, ast_line):
return self.line_parser.main(ast_line)
def indexing_assign_reachables(
self, target_module, module_linos
):
"""
IN: target_module: str. 'src.app.Init.main'
module_linos
OT: (<dict var_reachables>, <str parent_module>)
"""
if self.module_helper.is_runtime_module(target_module):
# 相当于返回 self.find_global_vars() 的结果.
return self.top_assigns.copy(), ''
if target_module not in module_linos:
lk.logt('[E2459]', target_module, module_linos)
raise Exception
lk.logt('[I0114]', target_module)
# ------------------------------------------------
"""
workflow:
1. 以 target_module 的 linos[0] 为起点, 向前找到第一个 indent 为 0 的
lino
2. 以 target_module 的 linos[-1] 为起点, 向后找到第一个 indent 为 0 的
lino
3. 在此区间内, 将所有 ast_defs 进行解析, 并认定为 var_reachables
"""
# ------------------------------------------------ lino reachables
target_linos = module_linos[target_module]
curr_module_lino = target_linos[0]
start_offset, end_offset = target_linos[0], target_linos[-1] + 1
# the start lino reachable
indent = self.ast_indents[start_offset]
# lk.logt('[TEMPRINT]20190811182309', target_module, start_offset,
# indent)
if indent == 0:
parent_module = ''
else:
while True:
parent_module = self.module_helper.get_parent_module(
target_module
)
# lk.logt('[TEMPRINT]20190811182549', target_module,
# parent_module)
parent_linos = module_linos[parent_module]
start_offset, end_offset = parent_linos[0], parent_linos[-1] + 1
parent_indent = self.ast_indents[start_offset]
if parent_indent == 0:
break
else:
continue
# -> parent_module = 'src.app.Init'
# the end lino reachable
while end_offset < self.max_lino:
if end_offset in self.ast_indents:
indent = self.ast_indents[end_offset]
if indent == 0:
break
end_offset += 1
# get lino reachalbes
lino_reachables = [
lino
for lino in range(start_offset, end_offset)
if lino in self.ast_indents and lino != curr_module_lino
]
"""
这里为什么要判断 `lino != curr_module_lino`?
因为 target_module 不能指任自身, 所以应去除.
例如 target_module = 'src.app.module', 在源码中, 不能因此自动产生 `module:
src.app.module` 的对应关系. 所以不能加入到 assigns 中.
"""
# parse vars
line_parser = LineParser(self.top_module)
ast_defs = ("<class '_ast.FunctionDef'>", "<class '_ast.ClassDef'>")
for lino in lino_reachables:
ast_line = self.eval_ast_line(lino)
if ast_line[0] in ast_defs:
line_parser.main(ast_line)
return line_parser.get_vars(), parent_module