Difference between revisions of "Crontab importer with metadata"

Latest revision as of 07:30, 13 June 2013

Import files with metadata values in the filename from server file system. This is a contrab task based on scripting ( bsh ).

Description:

Files stored in OpenKM server have a filename format like some_text - (CUPS,REC).pdf
grpName indicates the metadata group to be inserted.
contractUUID indicates the UUID of the OpenKM where will be imported the files.
systemFolder indicates from where OpenKM will try exporting files.
Verifies paths and extracts metadata from text between ();
Files are stored at basePath/year/month ( is year or month are not present are automatically created )
Is sended a mail with importation results

Metadata ( property groups )

<?xml version="1.0" encoding="UTF-8"?>
<!DOCTYPE property-groups PUBLIC "-//OpenKM//DTD Property Groups 2.1//EN"
"http://www.openkm.com/dtd/property-groups-2.1.dtd">
<property-groups>
    <property-group label="Contract" name="okg:contract">
    <input label="Cups" type="text" name="okp:contract.cups" width="200px"/>
    <select label="Año" name="okp:contract.year" type="simple">
        <option label="2012" value="2012"/>
    	<option label="2011" value="2011"/>
    	<option label="2010" value="2010"/>
        <option label="2009" value="2009"/>
        <option label="2008" value="2008"/>
    </select>
    <select label="Mes" name="okp:contract.month" type="simple">
        <option label="Enero" value="enero"/>
    	<option label="Febrero" value="febrero"/>
    	<option label="Marzo" value="marzo"/>
        <option label="Abril" value="abril"/>
        <option label="Mayo" value="mayo"/>
        <option label="Junio" value="junio"/>
        <option label="Julio" value="julio"/>
        <option label="Agosto" value="agosto"/>
        <option label="Septiembre" value="septiembre"/>
        <option label="Octubre" value="octubre"/>
        <option label="Noviembre" value="noviembre"/>
        <option label="Diciembre" value="diciembre"/>
    </select>
  </property-group>
</property-groups>

Scripting file

import java.io.*;
import java.util.*;
import com.openkm.bean.Document;
import com.openkm.module.db.DbDocumentModule;
import com.openkm.dao.DatabaseMetadataDAO;
import com.openkm.util.DatabaseMetadataUtils;
import com.openkm.dao.bean.DatabaseMetadataValue;
import com.openkm.api.OKMRepository;
import com.openkm.api.OKMFolder;
import com.openkm.bean.Folder;
import com.google.gson.Gson;
import com.openkm.dao.NodeBaseDAO;
import java.lang.StringBuffer;
import com.openkm.core.*;
import java.net.URLEncoder;
import com.openkm.util.MailUtils;
import java.util.Date;
import com.openkm.module.db.stuff.DbSessionManager;
import com.openkm.extension.core.*;
import com.openkm.api.OKMDocument;
import com.openkm.api.OKMPropertyGroup;

class Status {
	public String fileName = "";
	public String dstPath = "";
	public String cups = "";
	public String month = "";
	public String year = "";
	public String error = "";
}

// Months
List month = new ArrayList();
month.add("ENERO");
month.add("FEBRERO");
month.add("MARZO");
month.add("ABRIL");
month.add("MAYO");
month.add("JUNIO");
month.add("JULIO");
month.add("AGOSTO");
month.add("SEPTIEMBRE");
month.add("OCTUBRE");
month.add("NOVIEMBRE");
month.add("DICIEMBRE");

String token = DbSessionManager.getInstance().getSystemToken();
String grpName = "okg:contract";
String contractUUID = "56aabd01-eeeb-47b3-bb10-86ee609541cf";
String systemFolder = "/home/openkm/pending_to_import_folder";
String basePath = OKMRepository.getInstance().getNodePath(token, contractUUID);
Gson gson = new Gson();
List bad = new ArrayList();
List good = new ArrayList();
List toAddress = new ArrayList();

// Loading files
File folder = new File(systemFolder);
File[] listOfFiles = folder.listFiles(); 
for (int i = 0; i < listOfFiles.length; i++)  {
	File file = listOfFiles[i];
	if (file.isFile() && file.getName().toLowerCase().endsWith(".pdf")) {
		Status status = new Status();
		fileName = file.getName();
		status.fileName = fileName;
		if (fileName.indexOf("(")>0 && fileName.indexOf("(")<fileName.indexOf(")")) {
			// Get metadata
			String metadata = fileName.substring((fileName.indexOf("(")+1), fileName.lastIndexOf(")"));
			// get data
			String[] data = metadata.split(",");
			if (data.length!=2) {
				status.error = "Incorrect format -> name (cups,REN)<br>";
			} else {
				status.cups = data[0].toUpperCase().replaceAll(" ","");
				Date date = new Date(file.lastModified());
				status.year = String.valueOf(1900 + date.getYear());
				status.month = month.get(date.getMonth());
				if (!data[1].toUpperCase().equals("REN")) {
					status.error = "Error type REN not found<br>";
				}
				if (status.error.equals("")) {
					// test if folder year exists otherside create it 
					path = basePath + "/" + status.year;
					if (!OKMRepository.getInstance().hasNode(token, path)) {
						Folder folder = new Folder();
						folder.setPath(path);
						OKMFolder.getInstance().create(token, folder);
					}
					// test if folder month exists otherside create it 
					path = path + "/" + status.month;
					if (!OKMRepository.getInstance().hasNode(token, path)) {
						Folder folder = new Folder();
						folder.setPath(path);
						OKMFolder.getInstance().create(token, folder);
					}
					// Create document and adding metadata
					try {
						// Removing extra ( contents )
						fileName = fileName.substring(0,fileName.indexOf("("));
						// Removing - and spaces at ends
						while (fileName.substring(fileName.length()-1).equals(" ") || fileName.substring(fileName.length()-1).equals("-")) {
							fileName = fileName.substring(0,fileName.length()-1);
						}
						fileName = fileName + ".pdf";
						Document doc = new Document();
						path = path + "/" + fileName;
						doc.setPath(path);
						FileInputStream fis = new FileInputStream(file);
						doc = OKMDocument.getInstance().create(token, doc, fis);
						status.dstPath = path;
						
						// Create new metadata
						OKMPropertyGroup.getInstance().addGroup(token, doc.getPath(), grpName);
						Map properties = new HashMap();
						properties.put("okp:contract.cups",status.cups);
						properties.put("okp:contract.year",gson.toJson(new String[] {status.year}));
						properties.put("okp:contract.month",gson.toJson(new String[] {status.month.toLowerCase()}));
						OKMPropertyGroup.getInstance().setPropertiesSimple(token, doc.getPath(), grpName, properties);
						
						// Delete file
						boolean success = file.delete();
						if (!success) {
							status.error += "File can not been deleted";
						}
						
					} catch (PathNotFoundException e) {
						status.error += "PathNotFoundException";
					} catch (ItemExistsException e) {
						status.error += "ItemExistsException";
					} catch (UnsupportedMimeTypeException e) {
						status.error += "UnsupportedMimeTypeException";
					} catch (FileSizeExceededException e) {
						status.error += "FileSizeExceededException";
					} catch (VirusDetectedException e) {
						status.error += "VirusDetectedException";
					} catch (RepositoryException e) {
						status.error += "RepositoryException";
					} catch (DatabaseException e) {
						status.error += "DatabaseException";
					} catch (ExtensionException e) {
						status.error += "ExtensionException";
					} catch (IOException e) {
						status.error += "IOException";
					} catch (Exception e) {
						status.error += e.getMessage();
					} 
				}
			}
		} else {
			status.error = "Document format incorrect -> nombre - (cups,REN)<br>";
		}
		if (!status.error.equals("")) {
			bad.add(status);
		} else {
			good.add(status);
		}
   	}
}

StringBuffer result = new StringBuffer();
result.append("<h1>Import report</h1>");
result.append("</br></br>");
result.append("<table boder=\"0\" cellpadding=\"2\" cellspacing=\"0\" width=\"100%\">");
result.append("<tr>");
result.append("<td bgcolor=\"silver\"><b>Name</b></td>");
result.append("<td bgcolor=\"silver\"><b>CUPS</b></td>");
result.append("<td bgcolor=\"silver\"><b>Año</b></td>");
result.append("<td bgcolor=\"silver\"><b>Mes</b></td>");
result.append("<td bgcolor=\"silver\"><b>Error</b></td>");
result.append("</tr>");
result.append("<tr>");
result.append("<td colspan=\"6\" bgcolor=\"silver\"><b>Errors:"+bad.size()+"</b></td>");
result.append("</tr>");
for (Status status : bad) {
	result.append("<tr>");
	result.append("<td>"+status.fileName+"</td>");
	result.append("<td>"+status.cups+"</td>");
	result.append("<td>"+status.year+"</td>");
	result.append("<td>"+status.month+"</td>");
	result.append("<td><font color=\"red\">"+status.error+"</font></td>");
	result.append("</tr>");
}
result.append("</table>");
result.append("</br></br>");
result.append("<table boder=\"0\" cellpadding=\"2\" cellspacing=\"0\" width=\"100%\">");
result.append("<tr>");
result.append("<td bgcolor=\"silver\"><b>Name</b></td>");
result.append("<td bgcolor=\"silver\"><b>CUPS</b></td>");
result.append("<td bgcolor=\"silver\"><b>Year</b></td>");
result.append("<td bgcolor=\"silver\"><b>Month</b></td>");
result.append("<td bgcolor=\"silver\"><b>Destination</b></td>");
result.append("</tr>");
result.append("<tr>");
result.append("<td colspan=\"6\" bgcolor=\"silver\"><b>Imported correctly:"+good.size()+"</b></td>");
result.append("</tr>");
for (Status status : good) {
	result.append("<tr>");
	result.append("<td>"+status.fileName+"</td>");
	result.append("<td>"+status.cups+"</td>");
	result.append("<td>"+status.year+"</td>");
	result.append("<td>"+status.month+"</td>");
	result.append("<td><a href=\""+Config.APPLICATION_URL+"?docPath="+URLEncoder.encode(status.dstPath, "UTF-8")+"\">"+status.dstPath+"</font></td>");
	result.append("</tr>");
}
result.append("</table>");

// Sending mails
toAddress.add("some@mail.com");
MailUtils.sendMessage(toAddress, "Importing report",result.toString());

print(result.toString());