roeyc

app/Parser.php

@@ -2,12 +2,48 @@
 
 namespace App;
 
-use Exception;
-
 final class Parser
 {
-    public function parse(string $inputPath, string $outputPath): void
-    {
-        throw new Exception('TODO');
-    }
-}
+	public function parse(string $inputPath, string $outputPath): void
+	{
+		gc_disable();
+		$posts = [];
+		$chunk_size = 16 * 1024;
+
+		$handle = fopen($inputPath, "r");
+		stream_set_read_buffer($handle, 0);
+		$last_line = '';
+		while (!feof($handle))
+		{
+			$chunk = fread($handle, $chunk_size);
+			$lines = explode("\n", $chunk);
+			$count = count($lines);
+			for ($i = 0; $i < $count; $i++)
+			{
+				if ($i === $count - 1)
+				{
+					$last_line = $lines[$i];
+				}
+				else
+				{
+					if ($i === 0)
+						$line = $last_line . $lines[$i];
+					else
+						$line = $lines[$i];
+
+					$path = substr($line, 19, -26);
+					$date = substr($line, -25, 10);
+					$posts[$path][] = $date;
+				}
+			}
+		}
+		fclose($handle);
+		foreach ($posts as &$dates)
+		{
+			$dates = array_count_values($dates);
+			ksort($dates);
+		}
+
+		file_put_contents($outputPath, json_encode($posts, JSON_PRETTY_PRINT));
+	}
+}

index.php

@@ -0,0 +1,16 @@
+<?php
+
+require_once 'app/Parser.php';
+
+$script_start = microtime(true);
+
+//file_put_contents('data/numbers.csv', implode("\n", range(100, 199)));
+$p = new app\Parser();
+//$p->parse('data/numbers.csv', 'data/data.json');
+//$p->parse('data/test-data.csv', 'data/test-data-actual.json');
+//$p->parse('/home/roey/data.csv', 'data/data.json');
+//$p->parse('data/data.csv', 'data/data.json');
+$p->parse('data/data-10M.csv', 'data/data.json');
+//$p->parse('data/data-100M.csv', 'data/data.json');
+
+echo "Total time: " . (microtime(true) - $script_start) . "\n";