From 923f267c91e885940a574b0deda0c82e6a2c4d19 Mon Sep 17 00:00:00 2001
From: Aastha Mehta <aasthakm@mpi-sws.org>
Date: Tue, 1 Mar 2022 22:31:21 +0100
Subject: [PATCH] commit pending changes
---
sme_extn/sme_helper.php | 40 ++++++++++++++++++++++++++++++++++------
1 file changed, 34 insertions(+), 6 deletions(-)
diff --git a/sme_extn/sme_helper.php b/sme_extn/sme_helper.php
index da3e09b5..77eaddc4 100644
--- a/sme_extn/sme_helper.php
+++ b/sme_extn/sme_helper.php
@@ -477,6 +477,23 @@ function filesize_binning_wiktionary($file_size)
return "extra-large";
}
+function filesize_binning_wiktionary_unzipped($file_size)
+{
+ /*
+ * #frames in profiles: S - 21, M - 21, L - 64, XL - 3879 --> 4363
+ */
+# if (intval($file_size) <= 14600)
+# return "small";
+
+ if (intval($file_size) <= 24820)
+ return "medium";
+
+ if (intval($file_size) <= 81760)
+ return "large";
+
+ return "extra-large";
+}
+
function filesize_binning_medical_wiktionary($file_size)
{
/*
@@ -499,7 +516,7 @@ function filesize_binning_medical_wiktionary($file_size)
return "large";
if ((intval($file_size) > 12000 && intval($file_size) <= 49348)
- || (intval($file_size) > 85432 && intval($file_size) <= 600000))
+ || (intval($file_size) > 85432))
return "extra-large";
}
@@ -546,13 +563,24 @@ function set_request_id($pubinput, $privinput)
#print_r($privinput);
for ($i = 0; $i < $num_pub; $i++) {
- $cluster_name= $cpu_load . $pubinput[$i];
- $pubinput[$i]= $cpu_load . $pubinput[$i];
+ $cluster_name = $cpu_load . $pubinput[$i];
+ $pubinput[$i] = $cpu_load . $pubinput[$i];
if (isset($wgSMEPrivBinning) && $wgSMEPrivBinning == 1) {
- $hash=md5($privinput[$i]);
- $file_size = filesize($wgFileCacheDirectory.'/'.substr($hash,0,1).'/'.substr($hash,0,2).'/'.urlencode($privinput[$i]).'.html.gz');
- $cluster_suff = filesize_binning_medical_wiktionary(intval($file_size));
+ $hash = md5($privinput[$i]);
+ if (str_contains($wgFileCacheDirectory, "padded") == True) {
+ $sz_fpath = $wgFileCacheDirectory;
+ $sz_fpath .= '/'.substr($hash,0,1).'/'.substr($hash,0,2);
+ $sz_fpath .= '/'.urlencode($privinput[$i]).'.html';
+ $file_size = filesize($sz_fpath);
+ } else {
+ $sz_fpath = $wgFileCacheDirectory;
+ $sz_fpath .= '/'.substr($hash,0,1).'/'.substr($hash,0,2);
+ $sz_fpath .= '/'.urlencode($privinput[$i]).'.html.gz';
+ $file_size = filesize($sz_fpath);
+ }
+ $cluster_suff = filesize_binning_wiktionary(intval($file_size));
+# $cluster_suff = filesize_binning_wiktionary_unzipped(intval($file_size));
$cluster_name = $cluster_name . $cluster_suff;
$public_input_hash[] = md5($cluster_name, True);
} else {
--
GitLab