| Index: trunk/tools/analysis/StreamingXMLHistory.php |
| — | — | @@ -77,7 +77,11 @@ |
| 78 | 78 | public $pagelog; |
| 79 | 79 | public $pagelogFile; |
| 80 | 80 | public $nextLog; |
| | 81 | + public $articleName; |
| 81 | 82 | |
| | 83 | + //page log reader |
| | 84 | + public $logreader; |
| | 85 | + |
| 82 | 86 | //md5 hashes of the revision texts |
| 83 | 87 | public $md5History; |
| 84 | 88 | |
| — | — | @@ -95,7 +99,8 @@ |
| 96 | 100 | $this->revTypes = array(); |
| 97 | 101 | $this->pagelog = $pagelog; |
| 98 | 102 | if($pagelog){ |
| 99 | | - $this->pagelogFile = fopen($this->pagelog, "r"); |
| | 103 | + $this->logreader = new XMLReader(); |
| | 104 | + $this->logreader->open($this->pagelog); |
| 100 | 105 | } |
| 101 | 106 | $this->oldSize = 0; |
| 102 | 107 | } |
| — | — | @@ -140,7 +145,8 @@ |
| 141 | 146 | "new?", |
| 142 | 147 | "edit size", |
| 143 | 148 | "net size change", |
| 144 | | - "anonymous?" |
| | 149 | + "anonymous?", |
| | 150 | + "log action" |
| 145 | 151 | ); |
| 146 | 152 | fputcsv($this->outputFile, $csvData); |
| 147 | 153 | } |
| — | — | @@ -150,9 +156,22 @@ |
| 151 | 157 | $reader->open($this->inputFileName); |
| 152 | 158 | $this->writeCSVHeader(); |
| 153 | 159 | $current_rev = 0; |
| | 160 | + |
| | 161 | + //get article title |
| | 162 | + while($reader->read()){ |
| | 163 | + if($reader->nodeType == XMLREADER::ELEMENT |
| | 164 | + && $reader->localName == "title"){ |
| | 165 | + $this->articleName = $reader->readInnerXml(); |
| | 166 | + echo "Reading ".$this->articleName."\n"; |
| | 167 | + break; |
| | 168 | + } |
| | 169 | + } |
| | 170 | + |
| | 171 | + //get first log |
| 154 | 172 | if($this->pagelog){ |
| 155 | 173 | $this->nextLog = $this->getNextLogDataLine(); |
| 156 | 174 | } |
| | 175 | + |
| 157 | 176 | //read each revision |
| 158 | 177 | while ( $reader->read()){ |
| 159 | 178 | if ( $reader->nodeType == XMLREADER::ELEMENT |
| — | — | @@ -178,14 +197,36 @@ |
| 179 | 198 | } |
| 180 | 199 | |
| 181 | 200 | public function getNextLogDataLine(){ |
| 182 | | - $csvArray = null; |
| 183 | | - // a CSV array for writing, make sure $csvArray[1] = timestamp |
| 184 | | - return $csvArray; |
| | 201 | + while($this->logreader->read()){ |
| | 202 | + if ($this->logreader->nodeType == XMLREADER::ELEMENT |
| | 203 | + && $this->logreader->localName == "logitem") { |
| | 204 | + $logitem = new SimpleXMLElement($this->logreader->readOuterXml()); |
| | 205 | + if(strcmp($logitem->logtitle, $this->articleName)==0){ |
| | 206 | + return array( |
| | 207 | + $logitem->id, |
| | 208 | + strtotime($logitem->timestamp), |
| | 209 | + $logitem->contributor->username, |
| | 210 | + "", |
| | 211 | + "", |
| | 212 | + "", |
| | 213 | + "", |
| | 214 | + "", |
| | 215 | + "", |
| | 216 | + $logitem->action |
| | 217 | + ); |
| | 218 | + } |
| | 219 | + } |
| | 220 | + } |
| 185 | 221 | } |
| 186 | 222 | |
| 187 | 223 | |
| 188 | 224 | //foreach revision... |
| 189 | 225 | public function parseRev($xmlTEXT){ |
| | 226 | + |
| | 227 | + if(!$xmlTEXT){ |
| | 228 | + return; |
| | 229 | + } |
| | 230 | + |
| 190 | 231 | $revision = new SimpleXMLElement($xmlTEXT); |
| 191 | 232 | $textSize = strlen($revision->text); |
| 192 | 233 | |
| — | — | @@ -194,17 +235,6 @@ |
| 195 | 236 | |
| 196 | 237 | $revertIndex = array_search($md5, $this->md5History); |
| 197 | 238 | |
| 198 | | - if($revertIndex === FALSE ){ |
| 199 | | - $isNew = 'yes'; |
| 200 | | - $this->revTypes[] = new Edit(true); |
| 201 | | - } |
| 202 | | - else{ |
| 203 | | - $revert = new Revert(count($this->revTypes), $this->revTypes, true, $revertIndex); |
| 204 | | - $this->revTypes[] = $revert; |
| 205 | | - $revert->updateHistory(); |
| 206 | | - } |
| 207 | | - $this->md5History[] = $md5; |
| 208 | | - |
| 209 | 239 | $csvData = array( |
| 210 | 240 | $revision->id, |
| 211 | 241 | strtotime($revision->timestamp), |
| — | — | @@ -216,7 +246,8 @@ |
| 217 | 247 | $isNew, |
| 218 | 248 | $textSize, |
| 219 | 249 | $textSize - $this->oldSize, |
| 220 | | - isset($revision->contributor->username)? "no":"yes" |
| | 250 | + isset($revision->contributor->username)? "no":"yes", |
| | 251 | + "" |
| 221 | 252 | ); |
| 222 | 253 | $this->oldSize = $textSize; |
| 223 | 254 | |
| — | — | @@ -230,6 +261,18 @@ |
| 231 | 262 | } |
| 232 | 263 | } |
| 233 | 264 | fputcsv($this->outputFile, $csvData); |
| | 265 | + |
| | 266 | + if($revertIndex === FALSE ){ |
| | 267 | + $isNew = 'yes'; |
| | 268 | + $this->revTypes[] = new Edit(true); |
| | 269 | + } |
| | 270 | + else{ |
| | 271 | + $revert = new Revert(count($this->revTypes), $this->revTypes, true, $revertIndex); |
| | 272 | + $this->revTypes[] = $revert; |
| | 273 | + $revert->updateHistory(); |
| | 274 | + } |
| | 275 | + $this->md5History[] = $md5; |
| | 276 | + |
| 234 | 277 | } |
| 235 | 278 | |
| 236 | 279 | |