Better search algorithm and unicode handling

- Use a line based search (20% faster)
- Parse file as bytes to avoid offset by unicode characters

main.py

@@ -92,20 +92,19 @@ def search_log(name: str, query: str, start: int = 0):
     if resp:
         return resp
 
+    query_b = query.encode()
     matches = []
-    buffer_size = max(len(query), 1024)
     break_outer = False
-    with open(path) as f:
+    with open(path, "rb") as f:
         f.seek(start)
-        head = start
+        count = start
-        tail = start + 2 * buffer_size
+
-        buffer = f.read(tail)
+        while line := f.readline():
-        while True:
             offset = 0
             while True:
-                index = buffer.find(query, offset)
+                index = line.find(query_b, offset)
                 if index != -1:
-                    matches.append(head + index)
+                    matches.append(count + index)
                     offset = index + 1
                     if len(matches) >= 100:
                         break_outer = True
@@ -114,11 +113,6 @@ def search_log(name: str, query: str, start: int = 0):
                     break
             if break_outer:
                 break
+            count += len(line)
 
-            data =  f.read(buffer_size)
-            if not data:
-                break
-            buffer = buffer[-len(query):] + data
-            head = tail - len(query)
-            tail += len(data)
     return {"matches": matches}