check if query is valid unicode string

[nominatim.git] / wikidata / pull_wikipedia_logs.php

diff --git a/wikidata/pull_wikipedia_logs.php b/wikidata/pull_wikipedia_logs.php
index 8dfabbb1c1a0fc588c0e78379444a1119fb96f87..a7f6a9a3e21c3cb8b8a70ecd16b06fe97fdfb959 100644 (file)
--- a/wikidata/pull_wikipedia_logs.php
+++ b/wikidata/pull_wikipedia_logs.php
@@ -1,91 +1,73 @@
 <?php
 
-       for($iTimestamp = mktime(0, 0, 0, 5, 1, 2013); $iTimestamp < mktime(0, 0, 0, 6, 15, 2013); $iTimestamp += 24*60*60)
-       {
-               $sYear = date("Y", $iTimestamp);
-               $sMonth = date("Y-m", $iTimestamp);
-               $sDay = date("Ymd", $iTimestamp);
+for ($iTimestamp = mktime(0, 0, 0, 5, 1, 2013); $iTimestamp < mktime(0, 0, 0, 6, 15, 2013); $iTimestamp += 24*60*60) {
+    $sYear = date("Y", $iTimestamp);
+    $sMonth = date("Y-m", $iTimestamp);
+    $sDay = date("Ymd", $iTimestamp);
 
-               for($iHour = 0; $iHour < 24; $iHour++)
-               {
-                       $sFilename = sprintf("pagecounts-".$sDay."-%02d0000", $iHour);
-                       echo $sFilename."\n";
-                       if (!file_exists($sFilename.'.gz'))
-                       {
-                               exec('wget http://dumps.wikimedia.org/other/pagecounts-raw/'.$sYear.'/'.$sMonth.'/'.$sFilename.'.gz');
-                       }
+    for ($iHour = 0; $iHour < 24; $iHour++) {
+        $sFilename = sprintf("pagecounts-".$sDay."-%02d0000", $iHour);
+        echo $sFilename."\n";
+        if (!file_exists($sFilename.'.gz')) {
+            exec('wget http://dumps.wikimedia.org/other/pagecounts-raw/'.$sYear.'/'.$sMonth.'/'.$sFilename.'.gz');
+        }
 
-                       exec('gzip -dc '.$sFilename.'.gz'.' | grep -e "^[a-z]\{2\} [^ :]\+ [0-9]\+" > hour.txt');
+        exec('gzip -dc '.$sFilename.'.gz'.' | grep -e "^[a-z]\{2\} [^ :]\+ [0-9]\+" > hour.txt');
 
-                       $hPrevTotals = @fopen("totals.txt", "r");
-                       $hDayTotals = @fopen("hour.txt", "r");
-                       $hNewTotals = @fopen("newtotals.txt", "w");
+        $hPrevTotals = @fopen("totals.txt", "r");
+        $hDayTotals = @fopen("hour.txt", "r");
+        $hNewTotals = @fopen("newtotals.txt", "w");
 
-                       $sPrevKey = $sDayKey = true;
-                       $sPrevLine = true;
-                       $sDayLine = true;
+        $sPrevKey = $sDayKey = true;
+        $sPrevLine = true;
+        $sDayLine = true;
 
-                       do
-                       {
-                               if ($sPrevKey === $sDayKey)
-                               {
-                                       if ($sPrevLine !== true) fputs($hNewTotals, "$sPrevKey ".($iPrevValue+$iDayValue)."\n");
-                                       $sPrevLine = true;
-                                       $sDayLine = true;
-                               }
-                               else if ($sDayKey !== false && ($sPrevKey > $sDayKey || $sPrevKey === false))
-                               {
-                                       fputs($hNewTotals, "$sDayKey ".($iDayValue)."\n");
-                                       $sDayLine = true;
-                               }
-                               else if ($sPrevKey !== false && ($sDayKey > $sPrevKey || $sDayKey === false))
-                               {
-                                       fputs($hNewTotals, "$sPrevKey ".($iPrevValue)."\n");
-                                       $sPrevLine = true;
-                               }
+        do {
+            if ($sPrevKey === $sDayKey) {
+                if ($sPrevLine !== true) fputs($hNewTotals, "$sPrevKey ".($iPrevValue+$iDayValue)."\n");
+                $sPrevLine = true;
+                $sDayLine = true;
+            } elseif ($sDayKey !== false && ($sPrevKey > $sDayKey || $sPrevKey === false)) {
+                fputs($hNewTotals, "$sDayKey ".($iDayValue)."\n");
+                $sDayLine = true;
+            } elseif ($sPrevKey !== false && ($sDayKey > $sPrevKey || $sDayKey === false)) {
+                fputs($hNewTotals, "$sPrevKey ".($iPrevValue)."\n");
+                $sPrevLine = true;
+            }
 
-                               if ($sPrevLine === true)
-                               {
-                                       $sPrevLine = $hPrevTotals?fgets($hPrevTotals, 4096):false;
-                                       if ($sPrevLine !== false)
-                                       {
-                                               $aPrevLine = explode(' ', $sPrevLine);
-                                               $sPrevKey = $aPrevLine[0].' '.$aPrevLine[1];
-                                               $iPrevValue = (int)$aPrevLine[2];
-                                       }
-                                       else
-                                       {
-                                               $sPrevKey = false;
-                                               $iPrevValue =  0;
-                                       }
-                               }
+            if ($sPrevLine === true) {
+                $sPrevLine = $hPrevTotals?fgets($hPrevTotals, 4096):false;
+                if ($sPrevLine !== false) {
+                    $aPrevLine = explode(' ', $sPrevLine);
+                    $sPrevKey = $aPrevLine[0].' '.$aPrevLine[1];
+                    $iPrevValue = (int)$aPrevLine[2];
+                } else {
+                    $sPrevKey = false;
+                    $iPrevValue =  0;
+                }
+            }
 
-                               if ($sDayLine === true)
-                               {
-                                       $sDayLine = $hDayTotals?fgets($hDayTotals, 4096):false;
-                                       if ($sDayLine !== false)
-                                       {
-                                               preg_match('#^([a-z]{2}) ([^ :]+) ([0-9]+) [0-9]+$#', $sDayLine, $aMatch);
-                                               $sDayKey = $aMatch[1].' '.$aMatch[2];
-                                               $iDayValue = (int)$aMatch[3];
-                                       }
-                                       else
-                                       {
-                                               $sDayKey = false;
-                                               $iDayValue = 0;
-                                       }
-                               }
+            if ($sDayLine === true) {
+                $sDayLine = $hDayTotals?fgets($hDayTotals, 4096):false;
+                if ($sDayLine !== false) {
+                    preg_match('#^([a-z]{2}) ([^ :]+) ([0-9]+) [0-9]+$#', $sDayLine, $aMatch);
+                    $sDayKey = $aMatch[1].' '.$aMatch[2];
+                    $iDayValue = (int)$aMatch[3];
+                } else {
+                    $sDayKey = false;
+                    $iDayValue = 0;
+                }
+            }
+        } while ($sPrevLine !== false || $sDayLine !== false);
 
-                       } while ($sPrevLine !== false || $sDayLine !== false);
+        @fclose($hPrevTotals);
+        @fclose($hDayTotals);
+        @fclose($hNewTotals);
 
-                       @fclose($hPrevTotals);
-                       @fclose($hDayTotals);
-                       @fclose($hNewTotals);
-
-                       @unlink("totals.txt");
-                       rename("newtotals.txt", "totals.txt");
-               }
-       }
+        @unlink("totals.txt");
+        rename("newtotals.txt", "totals.txt");
+    }
+}
 
 // Notes:
 /*