我有一个分析 Perforce 服务器输出的 python 脚本。有了这个，我有两种不同的方式来输出给定时间正在运行的进程。
1.匹配进程缩进与进程缩进开始（P120将有119个缩进）
+P1
+P2
-P1
+P3
-P3
-P2
2.匹配进程缩进与正在运行的进程数
+P1
+P2
+P3
-P2
-P3
-P1
对于这两个函数，输入是我创建的流程对象列表。
类型 1 的代码：（慢）
def visual_output_match(processes):
output_string = ""
process_number = 1
indent_number = 0
running_processes = []
# Go through each process
for process in processes:
# Assign a process number to the new process that is about to start
process.set_process_number(process_number)
# This will append the process onto a list of processes that are currently running
running_processes.append(process)
# Get a list of processes that have ended before the new process started
ending_processes = check_ended_processes(running_processes, process)
# Delete the processes that have ended from the running_processes list
running_processes = remove_running_processes(running_processes, ending_processes)
# Print all the processes that finished before the new process began
for p in ending_processes:
# Tabs over the correct amount depending on how many processes are running
indent_number = p.get_process_number() - 1
output_string += ('\t' * indent_number)
output_string += ("-P" + str(p.get_process_number()) + " " + p.short_srting_summary_end() + '\n')
# Tabs over the correct amount depending on how many processes are running
indent_number = process.get_process_number() - 1
output_string += ('\t' * indent_number)
output_string += ("+P" + str(process.get_process_number()) + " " + process.short_srting_summary_start() + '\n')
process_number += 1
return output_string`
类型 2 的代码：（快速）
def visual_output_not_match(processes):
output_string = ""
number_tabs = 0
process_number = 1
running_processes = []
# Go through each process
for process in processes:
# Assign a process number to the new process that is about to start
process.set_process_number(process_number)
# This will append the process onto a list of processes that are currently running
running_processes.append(process)
# Get a list of processes that have ended before the new process started
ending_processes = check_ended_processes(running_processes, process)
# Delete the processes that have ended from the running_processes list
running_processes = remove_running_processes(running_processes, ending_processes)
# Print all the processes that finished before the new process began
for p in ending_processes:
number_tabs -= 1
# Tabs over the correct amount depending on how many processes are running
output_string += ('\t' * number_tabs)
output_string += ("-P" + str(p.get_process_number()) + " " + p.short_srting_summary_end() + '\n')
# Tabs over the correct amount depending on how many processes are running
output_string += ("\t" * number_tabs)
output_string += ("+P" + str(process.get_process_number()) + " " + process.short_srting_summary_start() + '\n')
number_tabs += 1
process_number += 1
return output_string
对于确切的一些进程，第一种类型将花费 11 分钟以上，而另一种只需要大约 1 秒。现在我意识到，对于类型 1，如果我有 11,000 个进程，那么我将在某个时候在给定的行上有 11,000 个选项卡，而对于类型 2，这是不正确的。虽然这是唯一让我的脚本变慢的事情吗？其他人是否看到任何其他严重错误。如果您需要查看我在此脚本中调用的其他一些方法，请告诉我。
现在我确实在这两个函数上运行了一个 cProfiler，这就是我得到的：
1：（慢）
3381490 function calls in 681.195 seconds
Ordered by: internal time
ncalls tottime percall cumtime percall filename:lineno(function)
1 679.175 679.175 681.182 681.182 visual_output.py:107(visual_output_match)
13706 0.568 0.000 0.769 0.000 visual_output.py:147(check_ended_processes)
13706 0.404 0.000 0.657 0.000 visual_output.py:162(remove_running_processes)
13706 0.258 0.000 0.264 0.000 server_info.py:398(short_srting_summary_start)
13702 0.257 0.000 0.264 0.000 server_info.py:402(short_srting_summary_end)
787125 0.206 0.000 0.206 0.000 server_info.py:67(__eq__)
800837 0.118 0.000 0.118 0.000 server_info.py:267(get_datetime_end)
800837 0.080 0.000 0.080 0.000 server_info.py:264(get_datetime_start)
800828 0.038 0.000 0.038 0.000 {len}
54816 0.028 0.000 0.028 0.000 server_info.py:308(get_process_number)
13706 0.018 0.000 0.018 0.000 server_info.py:406(set_process_number)
1 0.013 0.013 681.195 681.195 <string>:1(<module>)
27408 0.013 0.000 0.013 0.000 {method 'time' of 'datetime.datetime' objects}
27408 0.011 0.000 0.011 0.000 {method 'append' of 'list' objects}
13702 0.010 0.000 0.010 0.000 {method 'pop' of 'list' objects}
1 0.000 0.000 0.000 0.000 {method 'disable' of '_lsprof.Profiler' objects}
2：（快速）
3354269 function calls (3354219 primitive calls) in 1.981 seconds
Ordered by: internal time
ncalls tottime percall cumtime percall filename:lineno(function)
1 0.688 0.688 1.981 1.981 visual_output.py:67(visual_output_not_match)
13706 0.407 0.000 0.563 0.000 visual_output.py:147(check_ended_processes)
13706 0.375 0.000 0.603 0.000 visual_output.py:162(remove_running_processes)
787125 0.189 0.000 0.189 0.000 server_info.py:67(__eq__)
800837 0.079 0.000 0.079 0.000 server_info.py:267(get_datetime_end)
800837 0.075 0.000 0.075 0.000 server_info.py:264(get_datetime_start)
13706 0.059 0.000 0.061 0.000 server_info.py:398(short_srting_summary_start)
13702 0.053 0.000 0.055 0.000 server_info.py:402(short_srting_summary_end)
800850 0.035 0.000 0.035 0.000 {len}
13706 0.005 0.000 0.005 0.000 server_info.py:406(set_process_number)
27408 0.005 0.000 0.005 0.000 server_info.py:308(get_process_number)
13702 0.004 0.000 0.004 0.000 {method 'pop' of 'list' objects}
27434 0.003 0.000 0.003 0.000 {method 'append' of 'list' objects}
27408 0.003 0.000 0.003 0.000 {method 'time' of 'datetime.datetime' objects}
1 0.000 0.000 1.981 1.981 <string>:1(<module>)
2 0.000 0.000 0.000 0.000 {method 'send' of '_socket.socket' objects}
24/2 0.000 0.000 0.000 0.000 brine.py:202(_dump)
12/2 0.000 0.000 0.000 0.000 brine.py:179(_dump_tuple)
10 0.000 0.000 0.000 0.000 brine.py:106(_dump_int)
10/2 0.000 0.000 0.000 0.000 brine.py:360(dumpable)
2 0.000 0.000 0.000 0.000 protocol.py:220(_send)
2 0.000 0.000 0.000 0.000 protocol.py:227(_send_request)
2 0.000 0.000 0.000 0.000 channel.py:56(send)
2 0.000 0.000 0.000 0.000 <string>:531(write)
2 0.000 0.000 0.000 0.000 stream.py:173(write)
2 0.000 0.000 0.000 0.000 brine.py:332(dump)
14/8 0.000 0.000 0.000 0.000 brine.py:369(<genexpr>)
5 0.000 0.000 0.000 0.000 {method 'pack' of 'Struct' objects}
2 0.000 0.000 0.000 0.000 protocol.py:438(_async_request)
2 0.000 0.000 0.000 0.000 brine.py:150(_dump_str)
6/2 0.000 0.000 0.000 0.000 {all}
2 0.000 0.000 0.000 0.000 protocol.py:241(_box)
24 0.000 0.000 0.000 0.000 {method 'get' of 'dict' objects}
2 0.000 0.000 0.000 0.000 {method 'join' of 'str' objects}
2 0.000 0.000 0.000 0.000 brine.py:173(_dump_long)
2 0.000 0.000 0.000 0.000 {method 'acquire' of 'thread.lock' objects}
2 0.000 0.000 0.000 0.000 {next}
4 0.000 0.000 0.000 0.000 compat.py:17(BYTES_LITERAL)
2 0.000 0.000 0.000 0.000 {method 'release' of 'thread.lock' objects}
1 0.000 0.000 0.000 0.000 {method 'disable' of '_lsprof.Profiler' objects}
我试图理解这一点，看看它在哪里变慢了。我真正看到的唯一区别是该函数的第一次调用具有不同的 totime。
有没有人可以使用更好的代码分析器？
让我知道您是否还需要其他任何东西。