上传新文件

1 files changed, 120 insertions, 0 deletions

diff --git a/connection_dataset_new.py b/connection_dataset_new.py
new file mode 100644
index 0000000..031f334
--- /dev/null
+++ b/connection_dataset_new.py
@@ -0,0 +1,120 @@
+import os

+import numpy as np

+def getHSFirstTime(path):

+    timestamp=None

+    with open(path,'r') as file:

+        lines=file.readlines()

+    for line in lines:

+        if('rly_data' in line and 'in' in line):

+            timestamp=line.split('\t')[0]

+            # print(timestamp)

+            break

+    return timestamp

+

+def getClientFirstTime(path):

+    timestamp=None

+    with open(path,'r') as file:

+        lines=file.readlines()

+    for line in lines:

+        if('rly_data' in line and 'out' in line):

+            timestamp=line.split('\t')[1]

+            # print(timestamp)

+            break

+    return timestamp      

+

+def getClientFirstTimeDic(path):

+    firstTimeList={}

+    filelist=os.listdir(path)

+    for filename in filelist:

+        trace_name=filename.split('-')[0]

+        if(len([x for x in filelist if trace_name in x])==1):

+            timestamp=getClientFirstTime(os.path.join(path,filename))

+            if(not timestamp is None):

+                firstTimeList[timestamp]=filename

+    return firstTimeList

+

+def getHSFirstTimeDic(path):

+    firstTimeList={}

+    for filename in os.listdir(path):

+        timestamp=getHSFirstTime(os.path.join(path,filename))

+        if(not timestamp is None):

+            firstTimeList[timestamp]=filename

+    return firstTimeList

+

+def getCellList(path,t):

+    celllist=[]

+    timelist=[]

+    with open(path,'r') as file:

+        lines=file.readlines()

+    for line in lines:

+        if('\trelay' not in line):

+            continue

+        if('in' in line):

+            cell=-1

+        elif('out' in line):

+            cell=1

+        if('client'==t):

+            timestamp=line.split('\t')[1]

+            timelist.append(float(timestamp))

+        elif('hs'==t):

+            timestamp=line.split('\t')[0]

+            timelist.append(float(timestamp))

+        celllist.append(cell)

+    return celllist,timelist

+

+def process(client_log_path,hs_log_path,save_path):

+    log_file=open('{}/process_relay.log'.format(save_path),'w+')

+    total_dic={}

+    website_list=os.listdir(client_log_path)

+    for website in website_list:

+        webclient_path=os.path.join(client_log_path,website)

+        webhs_path=os.path.join(hs_log_path,website)

+        if(os.path.exists(webclient_path) and os.path.exists(webhs_path)):

+            website_dic={'client':{'cell':[],'time':[]},'hs':{'cell':[],'time':[]}}

+            client_time_dic=getClientFirstTimeDic(webclient_path)

+            hs_time_dic=getHSFirstTimeDic(webhs_path)

+            for client_key in client_time_dic.keys():

+                tmp_key=0.0

+                is_multiple=False

+                # tmp_filename=''

+                for hs_key in hs_time_dic.keys():

+                    if(abs((float(client_key))-(float(hs_key)))<2):

+                        if(tmp_key==0.0):

+                            tmp_key=hs_key

+                            # tmp_filename=hs_time_dic[tmp_key]

+                        else:

+                            print('multiple similar first key which client file {} with key is {} and hs file {} with key is {},and hs tmp file is {} with key is {}'.format(client_time_dic[client_key],client_key,hs_time_dic[hs_key],hs_key,hs_time_dic[tmp_key],tmp_key))

+                            is_multiple=True

+                if(not tmp_key == 0.0 and not is_multiple):

+                    conn_client_filename=client_time_dic[client_key]

+                    conn_hs_filename=hs_time_dic[tmp_key]

+                    log_file.write('connection relationship : client is {} , hs is {}\n'.format(conn_client_filename,conn_hs_filename))

+                    conn_client_cell_list,conn_client_time_list=getCellList(os.path.join(webclient_path,conn_client_filename),'client')

+                    conn_hs_cell_list,conn_hs_time_list=getCellList(os.path.join(webhs_path,conn_hs_filename),'hs')

+                    website_dic['client']['cell'].append(conn_client_cell_list)

+                    website_dic['hs']['cell'].append(conn_hs_cell_list)

+                    website_dic['client']['time'].append(conn_client_time_list)

+                    website_dic['hs']['time'].append(conn_hs_time_list)

+            log_file.write('website is {} and size of client is {} and size of hs is {}\n'.format(website,len(website_dic['client']),len(website_dic['hs'])))

+            total_dic[website]=website_dic

+        else:

+            log_file.write("ERROR: client path is {} and hs path is {}\n".format(os.path.exists(webclient_path),os.path.exists(webhs_path)))

+    # return total_dic

+    np.savez_compressed('{}/results_relayx.npz'.format(save_path),data=total_dic)

+    log_file.close()

+    

+def main():

+    client_log_path='./client'

+    hs_log_path='./hs'

+    save_path='./'

+    process(client_log_path,hs_log_path,save_path)

+

+def main_test():

+    client_log_path=r'G:\桌面\selfrun1_data1\test_616\client_rp'

+    hs_log_path=r'G:\桌面\selfrun1_data1\test_616\hs_rp'

+    save_path=r'G:\桌面\selfrun1_data1\test_616'

+    process(client_log_path,hs_log_path,save_path)

+    

+if __name__=="__main__":

+    # print('process start...')

+    main()