عرض عملي: لماذا ينهار ملف CSV لجهات الاتصال مع زيادة الحجم

Examples

Example 1: generate and search one million rows (simple)

import java.io.IOException;
import java.nio.file.*;
import java.util.*;

public class L05a {
    public static void main(String[] args) throws IOException {
        Path file = Path.of("contacts.csv");
        List<String> lines = new ArrayList<>();
        for (int i = 1; i <= 1_000_000; i++) lines.add(i + ",Name" + i + ",user" + i + "@example.com");
        Files.write(file, lines);
        System.out.println("File size in MB: " + Files.size(file) / 1_000_000);

        String wanted = "user999999@example.com";
        long start = System.nanoTime();
        int found = -1;
        for (String line : Files.readAllLines(file)) {
            if (line.endsWith("," + wanted)) { found = Integer.parseInt(line.substring(0, line.indexOf(','))); break; }
        }
        long ms = (System.nanoTime() - start) / 1_000_000;
        System.out.println("Found id " + found + " after reading the file; took about " + ms + " ms");
    }
}

Output from our run with JDK 21 on one laptop (your size will match, your time will not):

File size in MB: 41
Found id 999999 after reading the file; took about 393 ms

Roughly 0.4 seconds is not alarming for one search. Now picture 100 users each doing searches all day: each search re-reads the file, so the work per search grows with the data, and the machine's memory has to hold the whole file each time (readAllLines loads all lines). Our timing is one measurement on one machine, so treat it as an illustration, not a benchmark. A database that has an index on email reaches the row without reading the whole table; you will meet indexes in a later module.

Example 2: duplicates, invalid rows, and a comma (practical)

The program below appends a duplicate row to the file from Example 1, counts the rows with that email, and then splits a line containing a comma in the name. (Add this code at the end of main in L05a, after the search.)

Files.writeString(file, "5,Fake,user1@example.com" + System.lineSeparator(), StandardOpenOption.APPEND);
long dup = Files.readAllLines(file).stream().filter(l -> l.endsWith(",user1@example.com")).count();
System.out.println("Rows with user1@example.com: " + dup);

String bad = "6,Smith, John,john@example.com";
String[] f = bad.split(",");
System.out.println("fields = " + f.length + ", email column holds: '" + f[2] + "'");

Added output:

Rows with user1@example.com: 2
fields = 4, email column holds: ' John'

(The filter uses a stream. If streams are new to you, read it as "keep the lines that end with this email, then count them".) The file accepted a second row for the same email, with no warning. The comma in Smith, John produced four fields instead of three, so every later column moved: ' John' landed in the email column. That is the planned mistake of this demo. The fix is to quote the field in the file and to read it with a quote-aware reader, which narrows the problem for names but does not give the file the other protections: it still would not stop duplicates or a missing name.

Example 3: the problem list for 1 million contacts and 100 users (realistic)

Now put it together. Here are the problems of the Contacts CSV at one million contacts and 100 users at once, each paired with what a database offers. This is a worked answer for the style of the lab; your own list should use your own observations.

Problem 7 is a reasoned point, not something this demo showed: a plain file has only the permissions of the operating system on the whole file, while a DBMS usually lets you grant different rights to different users. Hedge such claims for your own system: say what you observed and what you inferred.