原始内容是一段可直接运行的 Python 脚本，这里补上用途说明、输入参数和典型使用方式。
这段脚本做什么
它主要做三件事：
- 递归收集
CMakeLists.txt - 解析
target_link_libraries(...)依赖关系 - 检测循环依赖，并导出 Graphviz
dot文件
如果你的工程模块很多、依赖关系已经开始难以靠肉眼判断，这类脚本很适合先把整体图拉出来。
使用方式
python3 cmake_analyzer.py <搜索路径或通配符> [选项]
常用场景：
python3 cmake_analyzer.py . -o cmake_deps.dot
python3 cmake_analyzer.py . --highlight 10
python3 cmake_analyzer.py . --top 30 -o pruned.dot
python3 cmake_analyzer.py . -I ./src --exclude ./third_party
支持的几个关键参数
-I, --include：只分析这些目录--exclude：排除某些目录-o, --output：指定导出的dot文件名--highlight：高亮入度大于等于 N 的节点--prune-in/--prune-out：按入度或出度裁剪图--top：只保留前 N 个关键节点--debug：输出调试信息
脚本
#!/usr/bin/env python3
import os
import re
import glob
import argparse
from collections import defaultdict, Counter
def normalize_dirs(dirs):
"""规范化目录路径，保证是绝对路径并以 / 结尾"""
if not dirs:
return None
result = []
for d in dirs:
absd = os.path.abspath(d)
if not absd.endswith(os.sep):
absd += os.sep
result.append(absd)
return result
def expand_cmakelists(paths, include_dirs=None, exclude_dirs=None, debug=False):
"""递归展开所有 CMakeLists.txt"""
files = []
for p in paths:
matches = glob.glob(p, recursive=True)
if not matches and os.path.isdir(p):
matches = glob.glob(os.path.join(p, "**/CMakeLists.txt"), recursive=True)
for m in matches:
absm = os.path.abspath(m)
if include_dirs and not any(absm.startswith(d) for d in include_dirs):
continue
if exclude_dirs and any(absm.startswith(d) for d in exclude_dirs):
if debug:
print(f"[DEBUG] 排除: {absm}")
continue
files.append(absm)
if debug:
print(f"[DEBUG] 找到 {len(files)} 个 CMakeLists.txt")
return files
def parse_cmake(files, debug=False):
"""解析 target_link_libraries 依赖"""
graph = defaultdict(list)
pattern = re.compile(r"target_link_libraries\s*\((\w+)\s+([^)]+)\)", re.IGNORECASE)
for f in files:
if debug:
print(f"[DEBUG] 解析 {f}")
with open(f, "r", encoding="utf-8", errors="ignore") as fh:
for line in fh:
m = pattern.search(line)
if m:
target = m.group(1)
deps = re.split(r"[\s;]+", m.group(2).strip())
deps = [d for d in deps if d and d.upper() not in ("PUBLIC", "PRIVATE", "INTERFACE")]
graph[target].extend(deps)
if debug:
print(f" {target} -> {deps}")
return graph
def detect_cycles(graph, max_cycles=20):
"""DFS 检测循环依赖"""
visited, stack, cycles = set(), [], []
def dfs(node):
if node in stack:
idx = stack.index(node)
cycles.append(stack[idx:] + [node])
return
if node in visited:
return
visited.add(node)
stack.append(node)
for dep in graph.get(node, []):
dfs(dep)
stack.pop()
for n in graph:
dfs(n)
if len(cycles) >= max_cycles:
break
return cycles
def prune_graph(graph, counts_in, counts_out, top_n=None, prune_in=None, prune_out=None, debug=False):
"""裁剪子图，只保留最关键部分"""
keep = set()
if top_n:
keep.update([n for n, _ in counts_in.most_common(top_n)])
if prune_in:
keep.update([n for n, c in counts_in.items() if c >= prune_in])
if prune_out:
keep.update([n for n, c in counts_out.items() if c >= prune_out])
if not keep:
return graph
new_graph = {}
for n, deps in graph.items():
if n in keep:
new_graph[n] = [d for d in deps if d in keep]
if debug:
print(f"[DEBUG] 裁剪: 原始 {len(graph)} 节点 -> 剩余 {len(new_graph)} 节点")
return new_graph
def export_to_dot(graph, counts_in, highlight=None, out_file="cmake_deps.dot"):
"""导出依赖关系为 Graphviz dot 文件"""
with open(out_file, "w", encoding="utf-8") as f:
f.write("digraph cmake_deps {\n")
f.write(" node [shape=box, style=filled, fillcolor=lightgray];\n")
for n, deps in graph.items():
color = "red" if highlight and counts_in[n] >= highlight else "lightgray"
f.write(f" \"{n}\" [fillcolor={color}];\n")
for d in deps:
f.write(f" \"{n}\" -> \"{d}\";\n")
f.write("}\n")
print(f"[OK] 依赖图已导出: {out_file}")
def main():
ap = argparse.ArgumentParser(description="CMake target 依赖循环检测工具")
ap.add_argument("sources", nargs="+", help="CMakeLists 搜索路径或通配符")
ap.add_argument("-I", "--include", action="append", help="只分析这些目录下的 CMakeLists")
ap.add_argument("--exclude", action="append", help="屏蔽某些目录")
ap.add_argument("-o", "--output", default="cmake_deps.dot", help="输出 dot 文件名")
ap.add_argument("--highlight", type=int, help="高亮依赖数 >= N 的节点")
ap.add_argument("--prune-in", type=int, help="保留入度 >= N 的节点")
ap.add_argument("--prune-out", type=int, help="保留出度 >= N 的节点")
ap.add_argument("--top", type=int, help="保留前 N 个入度最大的节点")
ap.add_argument("--debug", action="store_true", help="调试模式")
args = ap.parse_args()
include_dirs = normalize_dirs(args.include)
exclude_dirs = normalize_dirs(args.exclude)
files = expand_cmakelists(args.sources, include_dirs, exclude_dirs, args.debug)
graph = parse_cmake(files, args.debug)
counts_in, counts_out = Counter(), Counter()
for n, deps in graph.items():
counts_out[n] += len(deps)
for d in deps:
counts_in[d] += 1
cycles = detect_cycles(graph)
if cycles:
print("🔴 循环依赖检测到:")
for c in cycles:
print(" -> ".join(c))
else:
print("✅ 没有发现循环依赖")
graph = prune_graph(
graph,
counts_in,
counts_out,
top_n=args.top,
prune_in=args.prune_in,
prune_out=args.prune_out,
debug=args.debug,
)
export_to_dot(graph, counts_in, args.highlight, args.output)
if __name__ == "__main__":
main()
整理后的使用建议
- 先跑全量图，再做裁剪，不然容易一开始就漏掉关键依赖
- 如果输出太大，优先用
--top、--prune-in、--prune-out - 真正复杂的项目里，
target_link_libraries可能跨多行、含变量展开，这类脚本更适合做快速体检，而不是替代完整 CMake 语义解析
