Ускорен поиск по логу

3 files changed, 36 insertions, 41 deletions

diff --git a/day4/task3/client.py b/day4/task3/client.py
index 7f5c298..49663c8 100644
--- a/day4/task3/client.py
+++ b/day4/task3/client.py
@@ -28,14 +28,10 @@ def get_data(connection):
 
 
 def parse_data(data):
-    data = json.loads(data.decode())
-    arr = [f'Найдено {len(data)} совпадений:']
-    for error in data:
-        arr.append(f"[{error['date']}][{error['type']}][{error['module']}][{error['id']}] {error['text']}")
-    return '\n'.join(arr)
+    return json.loads(data.decode())
 
 
-HOST, PORT = ADDR = 'localhost', 6001
+HOST, PORT = ADDR = 'localhost', 6000
 
 cnt = 0
 sock = socket.socket()
@@ -67,6 +63,10 @@ while True:
     else:
         sock.sendall(pack_data(data))
         response = get_data(sock)
-        print(parse_data(response))
+        arr = parse_data(response)
+
+        print(f'Найдено {len(arr)} совпадений:')
+        for line in arr:
+            print(line)
 
 sock.close()
diff --git a/day4/task3/db.py b/day4/task3/db.py
index 15eba02..f4cbd0b 100644
--- a/day4/task3/db.py
+++ b/day4/task3/db.py
@@ -1,4 +1,5 @@
 import re
+import bisect
 from datetime import datetime
 
 f = open('otrs_error.log')
@@ -17,41 +18,30 @@ for line in lines:
         date = datetime.strptime(match.group('date'), '%c')
         module = match.group('module')
 
-        db.append({
-            'date': date,
-            'type': match.group('type'),
-            'module': module,
-            'id': match.group('id'),
-            'text': match.group('text')
-        })
+        db.append((date.hour, date.minute, date.second, module, line))
 
     else:
         print(f'Ошибка при разборе строки:\n{line}')
 
 
+db.sort()
+
+
 def search_by_date(hour, minute=None, second=None, data=None):
-    arr = []
-    print(f'filtering by h={hour}, m={minute}, s={second}, data="{data}"')
-    for err in db:
-        if err['date'].hour == int(hour):
-            if minute is not None:
-                if second is not None:
-                    if err['date'].minute == int(minute) and err['date'].second == int(second):
-                        if data in err['module']:
-                            _ = err.copy()
-                            _['date'] = _['date'].ctime()
-                            arr.append(_)
-                else:
-                    if err['date'].minute == int(minute):
-                        if data in err['module']:
-                            _ = err.copy()
-                            _['date'] = _['date'].ctime()
-                            arr.append(_)
-
-            else:
-                if data in err['module']:
-                    _ = err.copy()
-                    _['date'] = _['date'].ctime()
-                    arr.append(_)
-
-    return arr
+    low = [hour]
+    high = [hour]
+    if minute is not None:
+        low.append(minute)
+        high.append(minute)
+
+        if second is not None:
+            low.append(second)
+            high.append(second + 1)
+        else:
+            high.append(60)
+    else:
+        high.append(60)
+
+    left = bisect.bisect_left(db, tuple(low))
+    right = bisect.bisect_left(db, tuple(high))
+    return [t[4] for t in db[left:right] if data in t[3]]
diff --git a/day4/task3/server.py b/day4/task3/server.py
index b3cb619..f4e5ebe 100644
--- a/day4/task3/server.py
+++ b/day4/task3/server.py
@@ -3,6 +3,7 @@ import struct
 from db import search_by_date
 import re
 import json
+import time
 
 
 def pack_data(data):
@@ -14,7 +15,11 @@ def handle_request(data):
     match = re.match(r'(\d+):?(\d+)?:?(\d+)?-(.*)', data)
     if match:
         *date, msg = match.groups()
-        return pack_data(json.dumps(search_by_date(*date, msg)))
+
+        t = time.time()
+        found = search_by_date(*map(lambda i: int(i) if i is not None else i, date), msg)
+        print(f'Search took {time.time() - t} seconds')
+        return pack_data(json.dumps(found))
 
     else:
         print('String not matched.')
@@ -63,7 +68,7 @@ def handle_connection(connection, address):
             connection.sendall(response)
 
 
-HOST, PORT = ADDR = '0.0.0.0', 6001
+HOST, PORT = ADDR = '0.0.0.0', 6000
 sock = socket.socket()
 sock.bind(ADDR)
 sock.listen(5)