/**
* Extract content from the document.
*
* @return void
*/
public function convert()
{
// In the future, we might break this into multiple different
// kinds of actions supported by CERMINE, but for now it only
// performs extraction.
$this->logger->infoTranslate('cermine.cermine.startExtraction');
$command = new Command();
// Run Java or JRE...
$command->setCommand($this->config['jre']);
// ... with CERMINE in the classpath, ...
$command->addSwitch('-cp', $this->config['cerminejar']);
// ... the content extraction command, ...
$command->addArgument('pl.edu.icm.cermine.PdfNLMContentExtractor');
// ... the input file, ...
$command->addSwitch('-path', $this->inputFile);
// Send STDERR to STDOUT, so we can capture it, but send
// STDOUT to our destination.
$command->addRedirect('2>&1 >' . $this->outputFile);
$this->logger->debugTranslate('cermine.cermine.executeCommandLog', $command->getCommand());
// Execute the conversion
$command->execute();
$this->status = $command->isSuccess();
$this->output = $command->getOutputString();
$this->logger->debugTranslate('cermine.cermine.executeCommandOutputLog', $this->getOutput());
}
/**
* Convert the document
*
* @return void
*/
public function convert()
{
$command = new Command();
// Set the base command (Python fails with unicode issues if
// PYTHONIOENCODING is not set)
$command->setCommand('export PYTHONIOENCODING=UTF-8; ' . $this->config['command']);
// Set the debug switch
$command->addSwitch('-d');
// Disable git debug filesystem
$command->addSwitch('--nogit');
// Set the input document type
$command->addArgument('docx');
// Add the input file
if (!$this->inputFile) {
throw new \Exception('No input file given');
}
$command->addArgument($this->inputFile);
// Add the output directory
if (!$this->outputDirectory) {
throw new \Exception('No output directory given');
}
$command->addArgument($this->outputDirectory);
// Redirect STDERR to STDOUT to captue it in $this->output
$command->addRedirect('2>&1 >/dev/null');
$this->logger->debugTranslate('nlmxmlconversion.metypeset.executeCommandLog', $command->getCommand());
// Execute the conversion
$command->execute();
$this->status = $command->isSuccess();
$this->output = $command->getOutputString();
$this->logger->debugTranslate('nlmxmlconversion.metypeset.executeCommandOutputLog', $this->output);
}
/**
* Convert the references xml to Bibtex
*
* @return void
*/
public function convert()
{
$this->logger->debugTranslate('bibtexconversion.converter.startLog');
$command = new Command();
$command->setCommand($this->config['command']);
$command->addArgument($this->inputFile);
$command->addRedirect('2> /dev/null');
$this->logger->debugTranslate('bibtexconversion.converter.xml2bib.commandLog', $command->getCommand());
// Run the xml2bib conversion
$command->execute();
$this->logger->debugTranslate('bibtexconversion.converter.OutputLog', $command->getOutputString());
if ($this->status = $command->isSuccess()) {
file_put_contents($this->outputFile, $command->getOutputString());
}
}
/**
* Convert the document
*/
public function convert()
{
// load article body text
$filepath = $this->extractArticleBodyTextIntoTempFile();
$this->logger->infoTranslate('ner.extractor.startExtraction');
$command = new Command();
$command->setCommand("cat {$filepath} | {$this->config['ner']} {$this->config['model']}");
$this->logger->debugTranslate('ner.extractor.executeCommandLog', $command->getCommand());
// Execute the conversion
$command->execute();
$this->status = $command->isSuccess();
$this->output = $command->getOutputString();
if (!$this->status) {
throw new \Exception("NER command did not run successfully");
}
// extract named entities from command output
$entities = $this->parseCommandOutput($this->output);
// serialization to json document
file_put_contents(utf8_encode($this->outputFile), json_encode($entities, JSON_PRETTY_PRINT));
// clean up
if (file_exists($filepath)) {
unlink($filepath);
}
}
/**
* Converts the output from the Bibtex conversion into a temporary XML
* document using bibtex2xml
*
* @return DOMDocument Document containing a parsed reference list
*/
protected function biblatex2xmlConvert()
{
$command = new Command();
$command->setCommand($this->config['command']);
$command->addArgument($this->inputFileBibtex);
$command->addRedirect('2> /dev/null');
$this->logger->debugTranslate('bibtexreferencesconversion.converter.biblatex2xml.commandLog', $command->getCommand());
// Run the xml2bib conversion
$command->execute();
$this->logger->debugTranslate('bibtexreferencesconversion.converter.OutputLog', $command->getOutputString());
if (!($this->status = $command->isSuccess())) {
return false;
}
$dom = new DOMDocument();
if (!$dom->loadXML($command->getOutputString())) {
$this->logger->debugTranslate('bibtexreferencesconversion.converter.biblatex2xml.noDOMLog', $this->libxmlErrors());
return false;
}
return $dom;
}
/**
* Do the wkhtmltopdf conversion
*
* @return void
*/
protected function execute()
{
$command = new Command();
// Set the base command
$command->setCommand($this->config['wkhtmltopdf']['command']);
// Add the input file
$inputFile = $this->outputTmpPath . '/document.html';
if (!$inputFile) {
throw new \Exception('No input file given');
}
$command->addArgument($inputFile);
// Add the output directory
if (!$this->outputFile) {
throw new \Exception('No output file given');
}
$command->addArgument($this->outputFile);
// Redirect STDERR to STDOUT to captue it in $this->output
$command->addRedirect('2>&1');
$this->logger->debugTranslate('pdfconversion.wkhtmltopdf.executeCommandLog', $command->getCommand());
// Execute the conversion
$command->execute();
$this->status = $command->isSuccess();
$this->output = $command->getOutputString();
$this->logger->debugTranslate('pdfconversion.wkhtmltopdf.executeCommandOutputLog', $this->output);
}
/**
* Runs the citation parser
*
* @param string $referencesFile Reference file to parse
*
* @return void
*/
protected function parsCitExecute($referencesFile)
{
// Build the shell command
$command = new Command();
$command->setCommand($this->config['command']);
$command->addSwitch('-m', 'extract_citations');
$command->addArgument($referencesFile);
$this->logger->debugTranslate('referencesconversion.converter.parsCit.commandLog', $command->getCommand());
// Run the ParsCit conversion
$command->execute();
$this->status = $command->isSuccess();
$this->output = $command->getOutputString();
// Remove the temporary files
$this->parsCitCleanup($referencesFile);
}
/**
* Convert the NLM XML document to Epub
*
* @return void
*/
public function convert()
{
$this->logger->debugTranslate('epubconversion.converter.startLog');
$cmdStr = $this->config['command'];
// This mktemp code should probably be factored out.
$this->logger->debugTranslate('epubconversion.converter.startMktemp');
// The jats2epub script uses hardcoded directories for
// output. We should make our own temp directories for those
// to live in, so we can clean them up without stepping on
// other conversions’ toes. UNIX ONLY. d-:
$sysTmp = sys_get_temp_dir();
if (substr($sysTmp, -1, 1) == '/') {
$sysTmp = substr($sysTmp, 0, -1);
}
$mktemp = new Command();
$mktemp->setCommand('mktemp');
$mktemp->addSwitch('-d');
$mktemp->addArgument($sysTmp . '/jats2epub.XXXXX');
$mktemp->addRedirect('2>&1');
$mktemp->execute();
if (!$mktemp->isSuccess()) {
$this->logger->infoTranslate('epubconversion.converter.errorMktemp', $mktemp->getOutputString());
$this->status = false;
return;
}
$thisTmp = $mktemp->getOutputString();
// We’re going to cd to the working directory, so we need an
// absolute path to the command.
$cmdStr = realpath($cmdStr);
$command = new Command();
// Do our conversion work in /tmp (or other appropriate
// place). Only argument is the input file.
$command->setCommand('cd ' . $thisTmp . ' && ' . $cmdStr);
$command->addArgument($this->inputFile);
// Look for a media directory.
$mediaDir = dirname($this->inputFile) . '/metypeset/media';
// If it exists, copy it to a subdirectory of our temp work
// space, then point jats2epub at the parent.
if (file_exists($mediaDir)) {
$jatsMediaDir = $thisTmp . "/extras";
@mkdir($jatsMediaDir);
$this->copy_dir($mediaDir, $jatsMediaDir);
$command->addArgument($jatsMediaDir);
}
// Redirect STDERR to STDOUT to captue it in $this->output
$command->addRedirect('2>&1');
$this->logger->debugTranslate('epubconversion.converter.startJats2epub');
// Execute the conversion
$command->execute();
$this->status = $command->isSuccess();
$this->output = $command->getOutputString();
if (!$this->status) {
$this->logger->infoTranslate('epubconversion.converter.errorJats2epub', $this->output);
return;
}
$this->logger->debugTranslate('epubconversion.converter.executeCommandOutputLog', $this->output);
// Find the output file(s).
$outfiles = glob($thisTmp . '/output_final/*.epub');
if (sizeof($outfiles) != 1) {
$this->status = false;
$this->logger->infoTranslate('epubconversion.converter.errorGlob');
return;
}
// If there was only one candidate file, move it to the
// target, and clean up after ourselves.
rename($outfiles[0], $this->outputFile);
$this->del_dir($thisTmp);
$this->logger->debugTranslate('epubconversion.converter.endLog');
return;
}
/**
* Execute the pandoc conversion
*
* This step takes a text file that lists the reference id's (prefixed with
* @, one line each) used in the NLMXML document and a bibtex file
* containing the bibliography and formats them according to a provided
* citation style file
*
* @return void
*/
protected function execute()
{
$command = new Command();
// Pandoc expects $HOME to be set
$commandPrefix = 'HOME=' . $this->outputPath . ' ';
$command->setCommand($commandPrefix . $this->config['command']);
// Produce typographically correct output
$command->addSwitch('--smart');
// Add the bibliography file
$command->addSwitch('--bibliography', $this->inputFileBibtex);
// Add the citation style file
$command->addSwitch('--csl', $this->citationStyleFile);
// Add the reference file
$command->addArgument($this->referencesFile);
$this->logger->debugTranslate('citationstyleconversion.converter.pandoc.commandLog', $command->getCommand());
// Run the pandoc conversion
$command->execute();
$this->status = $command->isSuccess();
$this->output = $command->getOutputString();
$this->logger->debugTranslate('citationstyleconversion.converter.OutputLog', $command->getOutputString());
}
/**
* Add the XMP sidecar to the PDF document
*
* @return void
*/
protected function addXmpSidecar()
{
$command = new Command();
// Set the base command
$command->setCommand($this->config['exiftool']['command']);
// Allow duplicates to be extracted
$command->addSwitch('-duplicates');
// Be verbose
$command->addSwitch('-verbose');
// Read tags from XMP sidecar
$command->addSwitch('-TagsFromFile');
// The XMP file
$command->addArgument($this->outputFileXmp);
// The PDF file
$command->addArgument($this->inputFilePdf);
// Redirect STDERR to STDOUT to captue it in $this->output
$command->addRedirect('2>&1');
$this->logger->debugTranslate('xmpconversion.exiftool.executePdfCommandLog', $command->getCommand());
// Add the XMP sidecar
$command->execute();
$this->status = $command->isSuccess();
$this->output = $command->getOutputString();
$this->logger->debugTranslate('xmpconversion.exiftool.executePdfCommandOutputLog', $this->output);
}
/**
* Convert the document
*
* @return void
*/
public function convert()
{
$command = new Command();
// Set the base command. If HOME is not set to a writeable
// directory, unoconv won’t work.
$command->setCommand($this->config['command']);
// Add verbosity switch
if ($this->verbose) {
$command->addSwitch('-vvv');
}
// Add the filter
if ($this->filter) {
$command->addSwitch('-f', $this->filter);
}
// Add the output file
if (!$this->outputFile) {
throw new \Exception('No output file given');
}
$command->addSwitch('-o', $this->outputFile);
// Add the input file
if (!$this->inputFile) {
throw new \Exception('No input file given');
}
$command->addArgument($this->inputFile);
// Redirect STDERR to STDOUT to captue it in $this->output
$command->addRedirect('2>&1');
$this->logger->debugTranslate('docxconversion.unoconv.executeCommandLog', $command->getCommand());
// Execute the conversion
$command->execute();
$this->status = $command->isSuccess();
$this->output = $command->getOutputString();
$this->logger->debugTranslate('docxconversion.unoconv.executeCommandOutputLog', $this->getOutput());
}
