Read the logs line by line instead of loading them into an array

4 files changed, 31 insertions, 22 deletions

diff --git a/common.inc.php b/common.inc.php
index a54fa0f..9b1b465 100644
--- a/common.inc.php
+++ b/common.inc.php
@@ -86,6 +86,33 @@ function validDate($date)
 }
 
 /**
+ * Read a log file line by line
+ *
+ * Reading the whole log using file() keeps every single line in memory at once,
+ * which for a busy day is several megabytes. All callers only walk the lines
+ * once and in order, so hand them out one at a time instead. A missing file
+ * yields nothing, the callers render their usual empty result in that case.
+ *
+ * @param string $filename the log file to read
+ *
+ * @return Generator the lines of the log, including their line endings
+ */
+function readLog($filename)
+{
+    $handle = @fopen($filename, 'r');
+    if ($handle === false) {
+        return;
+    }
+    try {
+        while (($line = fgets($handle)) !== false) {
+            yield $line;
+        }
+    } finally {
+        fclose($handle);
+    }
+}
+
+/**
  * Get the date from a filename
  *
  * @param string $filename the filename to extract the month from (e.g.
diff --git a/details.php b/details.php
index c7fc387..1d93516 100644
--- a/details.php
+++ b/details.php
@@ -34,14 +34,8 @@ $filename = 'dri-devel-'.$date.'.log';
 $pattern = '#(^|[^\"=]{1})(http://|ftp://|https://|mailto:|news:)([^\s<>]+)([\s\n<>]|$)#sm';
 
 $timehashes = array();
-// Guard the read, a missing log would otherwise leak the full path on disk
-if (file_exists(LOG_DIR.$filename)) {
-    $lines = file(LOG_DIR.$filename);
-} else {
-    $lines = array();
-}
 $anything_displayed = false;
-foreach ($lines as $line_num => $line) {
+foreach (readLog(LOG_DIR.$filename) as $line) {
     if (isFromChannel($line, $channel)) {
         if (!filterJoin($line)) {
             $user = getUser($line);
diff --git a/json.php b/json.php
index 0329e4e..462a8cc 100755
--- a/json.php
+++ b/json.php
@@ -39,13 +39,7 @@ if (!isset($_GET['mode'])) {
 
 require_once 'common.inc.php';
 
-// Read all lines of the log into an array
 $filename = 'dri-devel-'.$date.'.log';
-if (file_exists(LOG_DIR.$filename)) {
-    $lines = file(LOG_DIR.$filename);
-} else {
-    $lines = array();
-}
 
 header("Content-type: application/json;");
 
@@ -80,7 +74,7 @@ case 'watch':
     }
 
     // Assign all lines to a user
-    foreach ($lines as $line_num => $line) {
+    foreach (readLog(LOG_DIR.$filename) as $line) {
         if (isFromChannel($line, $channel)) {
             if (!filterJoin($line)) {
                 $user = substr(getUser($line), 2, -1);
@@ -124,7 +118,7 @@ case 'watch':
 case 'show':
     $retval['users'] = array();
     $retval['data'] = array();
-    foreach ($lines as $line_num => $line) {
+    foreach (readLog(LOG_DIR.$filename) as $line) {
         if (isFromChannel($line, $channel)) {
             if (!filterJoin($line)) {
                 $user = substr(getUser($line), 2, -1);
diff --git a/stats.php b/stats.php
index 76bc776..18bc6b6 100644
--- a/stats.php
+++ b/stats.php
@@ -35,18 +35,12 @@ if (!isset($_GET['date'])) {
 
 require_once 'common.inc.php';
 
-// Read all lines of the log into an array
 $filename = 'dri-devel-'.$date.'.log';
-if (file_exists(LOG_DIR.$filename)) {
-    $lines = file(LOG_DIR.$filename);
-} else {
-    $lines = array();
-}
 
 $retval = array();
 
 echo '<pre>';
-foreach ($lines as $line) {
+foreach (readLog(LOG_DIR.$filename) as $line) {
     $user = substr(getUser($line), 2, -1);
     if ($user) {
         if (in_array($user, $ignore)) {