// ------------------------------------------------------------------------------------
// Name: AddFeature
// Goal: Add the feature [strFname] with value [strFvalue] to the node
// Result:
// <fs type='[strFStype]'>
// <f name='[strFname] value='[strFvalue]' />
// </fs>
// History:
// 26-05-2010 ERK Created
// ------------------------------------------------------------------------------------
public bool AddFeature(ref XmlDocument pdxThis, ref XmlNode ndxThis, string strFStype, string strFname, string strFvalue)
{
XmlNode ndxChild = null; // Child node of type <fs>
try {
// Validate
if (pdxThis == null)
{
return(false);
}
// Add this inside <fs type='NP'> to <f name='NPtype' value='...'/>
ndxChild = oXmlTools.SetXmlNodeChild(pdxThis, ref ndxThis, "fs", "type;" + strFStype, "type", strFStype);
// Validate
if (ndxChild == null)
{
// Show error
errHandle.Status("Could not add an appropriate <fs type='" + strFStype + "'> child");
// Return failure
return(false);
}
// Add or replace the <f> child
if (oXmlTools.SetXmlNodeChild(pdxThis, ref ndxChild, "f", "name;" + strFname, "value", strFvalue) == null)
{
// Show error
errHandle.Status("Could not add an appropriate <f name='" + strFname + "'> child");
// Return failure
return(false);
}
// Return success
return(true);
} catch (Exception ex) {
// Show error
errHandle.DoError("modAdapt/AddFeature", ex);
// Return failure
return(false);
}
}
// ================================ METHODS ===================================================
/* -------------------------------------------------------------------------------------
* Name: repairOneFolia
* Goal: Check and repair one folia file
* History:
* 21/mar/2016 ERK Created
* ------------------------------------------------------------------------------------- */
public bool repairOneFolia(String sFileFoliaGz)
{
try {
// Get the unzipped file name
String sFileFolia = sFileFoliaGz.Replace(".gz", "");
// Unzip the file
if (!General.DecompressFile(sFileFoliaGz, sFileFolia))
{
errHandle.DoError("cmbConv/repairOneFolia", "Could not decompress"); return(false);
}
// Check for empty begintime/endtime
String[] arLine = File.ReadAllLines(sFileFolia);
/*
* if (sFileFoliaGz.Contains("S-O_00214801")) {
* int iStop = 2;
* } */
bool bChanged = false; int iChanges = 0;
for (int i = 0; i < arLine.Length; i++)
{
String sLine = arLine[i];
// Check the syntax of the begintime and the endtime
if (adaptTime(ref sLine, sLine.IndexOf("begintime=")))
{
bChanged = true;
}
if (adaptTime(ref sLine, sLine.IndexOf("endtime=")))
{
bChanged = true;
}
/*
* if (sLine.Contains("begintime=\"\"")) {
* sLine = sLine.Replace("begintime=\"\"", "begintime=\"00:00:00.000\"");
* bChanged = true;
* }
* if (sLine.Contains("endtime=\"\"")) {
* sLine = sLine.Replace("endtime=\"\"", "endtime=\"00:00:00.000\"");
* bChanged = true;
* } */
if (bChanged)
{
arLine[i] = sLine; iChanges++; bChanged = false;
}
}
// Only save results if something changed
if (iChanges > 0)
{
File.WriteAllLines(sFileFolia, arLine);
// Show repair log
errHandle.Status("Repaired file: [" + sFileFoliaGz + "] (" + iChanges + " repairs)");
// And compress into .gz
if (!General.CompressFile(sFileFolia, sFileFoliaGz))
{
errHandle.DoError("cmbConv/repairOneFolia", "Could not compress"); return(false);
}
}
else
{
// Show repair log
errHandle.Status("Unchanged: [" + sFileFoliaGz + "]");
}
// Remove the unzipped file again
File.Delete(sFileFolia);
// Return positively
return(true);
} catch (Exception ex) {
errHandle.DoError("cmdConv/repairOneFolia", ex);
return(false);
}
}
static List <String> lstCmdi = null; // List of .cmdi.xml files in the [sCmdi] directory
// Command-line entry point + argument handling
static void Main(string[] args)
{
String sLanguage = ""; // Which language to take (two-letter code 'en' or 'nl')
String sFolia = ""; // Base directory to take (e.g. /vol/tensusers/ekomen)
String sSubtiel = ""; // Base directory for the output (e.g: /vol/tensusers/ekomen/subtiel)
String sCmdi = ""; // Base directory for CMDI files
String sAction = "combi"; // The particular type of action expected. Default "combi"
bool bIsDebug = false; // Debugging
try {
// Check command-line options
for (int i = 0; i < args.Length; i++)
{
// get this argument
String sArg = args[i];
if (sArg.StartsWith("-"))
{
// Check out the arguments
switch (sArg.Substring(1))
{
case "f": // Root directory under which the .folia.xml.gz files are located
sFolia = Path.GetFullPath(args[++i]);
break;
case "c": // Root directory where the CMDI files are located
sCmdi = Path.GetFullPath(args[++i]);
break;
case "o": // Root directory where the output files are going to be stored
sSubtiel = Path.GetFullPath(args[++i]);
break;
case "d": // Debugging
bIsDebug = true;
break;
case "l": // Language (three letter code)
sLanguage = args[++i];
break;
case "r": // INSTEAD of 'combi', perform 'REPAIR' action
sAction = "repair";
break;
}
}
else
{
// Throw syntax error and leave
SyntaxError("1 - i=" + i + " args=" + args.Length + " argCurrent=[" + sArg + "]"); return;
}
}
// Check presence of input/output
if (sFolia == "" || !Directory.Exists(sFolia))
{
SyntaxError("No (valid) base directory for FoLiA input"); return;
}
// Initialise the Treebank Xpath functions, which may make use of tb:matches()
XPathFunctions.conTb.AddNamespace("tb", XPathFunctions.TREEBANK_EXTENSIONS);
// Create a new instance of the combination class
cmbConv oConv = new cmbConv(errHandle);
// Other directories depend on ACTION
switch (sAction)
{
case "repair":
// No further directories need be present
oConv.output = sFolia;
// Process all the directories with this action
WalkDirectoryTree(sFolia, "*.folia.xml.gz", sAction, ref oConv);
break;
default:
if (sCmdi == "" || !Directory.Exists(sCmdi))
{
SyntaxError("No (valid) cmdi directory"); return;
}
if (sSubtiel == "")
{
SyntaxError("No subtiel directory"); return;
}
// If the target directory is not there, create it
if (!Directory.Exists(sSubtiel))
{
Directory.CreateDirectory(sSubtiel);
}
oConv.output = sSubtiel;
// find .cmdi.xml files
errHandle.Status("Finding .cmdi.xml files...");
lstCmdi = Directory.GetFiles(sCmdi, "*.cmdi.xml", SearchOption.AllDirectories).ToList();
// Convert the list of files into a dictionary
oConv.cmdi(lstCmdi);
// Find .folia.xml.gz files
errHandle.Status("Finding directories in " + sFolia);
// Walk all directories
errHandle.Status("Processing .folia.xml.gz files...");
// Output the header
oConv.doHeaderCsv();
// Resursive call for each subdirectory.
WalkDirectoryTree(sFolia, "*.folia.xml.gz", sAction, ref oConv);
// Save the harvested information to an xml file
if (!oConv.harvestSaveXml(sSubtiel + "/harvest.xml"))
{
errHandle.DoError("Main", "Could not create harvest summary xml file");
return;
}
break;
}
// Exit the program
errHandle.Status("Ready");
} catch (Exception ex) {
errHandle.DoError("Main", ex); // Provide standard error message
throw;
}
}
// =================== Local variables ===============================================
// Command-line entry point + argument handling
static void Main(string[] args)
{
String sDutch = ""; // Input directory for DUTCH subtitles
String sEnglish = ""; // Input directory for ENGLISH subtitles
String sOutput = ""; // Output directory
String sParallel = ""; // File in .xml format that contains the parallel between EN-NL
bool bIsDebug = false; // Debugging
bool bForce = false; // Force
String sAction = "english"; // Type of action to be taken
try {
// Check command-line options
for (int i = 0; i < args.Length; i++)
{
// get this argument
String sArg = args[i];
if (sArg.StartsWith("-"))
{
// Check out the arguments
switch (sArg.Substring(1))
{
case "p": // Xml file containing the parallels between Eng and NL
sParallel = args[++i];
break;
case "n": // Input directory for Dutch: .folia.xml and .cmdi.xml files
sDutch = args[++i];
break;
case "e": // Input directory for English: .folia.xml and .cmdi.xml files
sEnglish = args[++i];
break;
case "o": // Top output directory
sOutput = args[++i];
break;
case "f": // Force
bForce = true;
break;
case "a": // Action
sAction = args[++i].ToLower();
if (sAction != "english")
{
SyntaxError("The only action right now is 'english'");
return;
}
break;
case "d": // Debugging
bIsDebug = true;
break;
default:
// Throw syntax error and leave
SyntaxError("Unknown option: [" + sArg + "]");
return;
}
}
else
{
// Throw syntax error and leave
SyntaxError("1 - i=" + i + " args=" + args.Length + " argCurrent=[" + sArg + "]"); return;
}
}
// Check presence of input/output
if (sDutch == "" || sEnglish == "")
{
SyntaxError("Both dutch and english input must be specified"); return;
}
if (sParallel == "")
{
SyntaxError("The XML file containing the Dutch-English parallels must be specified"); return;
}
// Check input directory and parallels file
if (!Directory.Exists(sDutch))
{
errHandle.DoError("Main", "Cannot find Dutch input file(s) in: " + sDutch); return;
}
if (!Directory.Exists(sEnglish))
{
errHandle.DoError("Main", "Cannot find English input file(s) in: " + sEnglish); return;
}
if (!File.Exists(sParallel))
{
errHandle.DoError("Main", "Cannot find parallel file in: " + sParallel); return;
}
// Initialize the main entry point for the conversion
engConv objConv = new engConv(errHandle);
// Set directories where input is situated and output should come
objConv.dutch = sDutch;
objConv.english = sEnglish;
objConv.output = sOutput;
// Initialise the Treebank Xpath functions, which may make use of tb:matches()
opsub.util.XPathFunctions.conTb.AddNamespace("tb", opsub.util.XPathFunctions.TREEBANK_EXTENSIONS);
// Start reading the parallels XML
ParReader oParallel = new ParReader(sParallel, errHandle);
// Walk all the <linkGrp> elements
XmlReader rdLinkGrp = null;
while (oParallel.getNextLinkGrp(ref rdLinkGrp))
{
// Process this one
if (!objConv.ConvertOneEngToFolia(rdLinkGrp, bForce, bIsDebug))
{
errHandle.DoError("Main", "Could not convert English");
return;
}
}
/*
* XmlNode ndxLinkGrp = oParallel.getNextLinkGrp();
* while (ndxLinkGrp!= null) {
* // Process this one
* if (!objConv.ConvertOneEngToFolia(ndxLinkGrp, bForce, bIsDebug)) { errHandle.DoError("Main", "Could not convert English"); return; }
* // Go to the next one
* ndxLinkGrp = oParallel.getNextLinkGrp();
* }
*/
// Exit the program
errHandle.Status("Ready");
} catch (Exception ex) {
errHandle.DoError("Main", ex); // Provide standard error message
throw;
}
}