/// <summary>
        ///     The execute method.
        /// </summary>
        /// <returns>
        ///     The <see cref="bool" />.
        /// </returns>
        public override bool Execute()
        {
            int  skip       = 0;
            int  count      = 100;
            int  totalcount = 0;
            bool hascontent = true;

            TranslationServiceFacade.UserTranslation[] usertranslations = new TranslationServiceFacade.UserTranslation[100];
            usertranslations.Initialize();

            using (TmxWriter TmxDocument = new TmxWriter(this.TmxDocument.ValueString, this.sourceLanguage.ValueString, this.targetLanguage.ValueString, false))
            {
                do
                {
                    usertranslations = TranslationServiceFacade.GetUserTranslations(this.sourceLanguage.ValueString, this.targetLanguage.ValueString, skip, count);
                    skip            += count;
                    foreach (var usertrans in usertranslations)
                    {
                        if (usertrans.OriginalText == null)
                        {
                            hascontent = false;
                            break;
                        }
                        TmxDocument.TmxWriteSegment(usertrans.OriginalText, usertrans.TranslatedText, usertrans.From, usertrans.To, TmxWriter.TUError.good);
                        totalcount++;
                    }
                    Logger.WriteLine(LogLevel.None, "{0} translation units written.", totalcount);
                } while (hascontent);
            }


            return(true);
        }
        public ConvertToTmxCommand()
        {
            // Register the actual command with a simple (optional) description.
            IsCommand("ConvertToTmx", "Convert a valid XML file to a TMD or TMC file");

            // Add a longer description for the help on that specific command.
            //HasLongDescription("This can be used to quickly read a file's contents " +
            //"while optionally stripping out the ',' character.");

            // Required options/flags, append '=' to obtain the required value.
            HasRequiredOption("i|input=", "The full path of the input file.", i => InputFile = i);

            HasOption("o|output=", "The full path of the output file.", o => OutputFile = o);

            // Optional options/flags, append ':' to obtain an optional value, or null if not specified.
            //HasOption("s|strip:", "Strips ',' from the file before writing to output.",
            //    t => StripCommaCharacter = t == null ? true : Convert.ToBoolean(t));

            tmxWriter = new TmxWriter();
        }
Exemple #3
0
        /// <summary>
        ///     The execute method.
        /// </summary>
        /// <returns>
        ///     The <see cref="bool" />.
        /// </returns>
        public override bool Execute()
        {
            string SntFileName = Path.GetTempPath() + "_TmxUpload.snt";
            string uservalue   = user.ValueString;

            if (uservalue == string.Empty)
            {
                uservalue = "TmxUpload";
            }
            string ratingvalue = rating.ValueString;

            if (ratingvalue == string.Empty)
            {
                ratingvalue = "6";
            }

            TmxFile TmxIn = new TmxFile(this.TmxDocument.ValueString);

            string[] sntFilenames = TmxIn.WriteToSNTFiles(SntFileName);
            if (sntFilenames.Length != 2)
            {
                Logger.WriteLine(LogLevel.Error, "More than 2 languages in the TMX file. Must have exactly 2.");
                deleteSNTfiles(sntFilenames);
                return(false);
            }

            TranslationMemory TM = new TranslationMemory();

            TM.sourceLangID = this.sourceLanguage.ValueString.ToLowerInvariant();
            TM.targetLangID = this.targetLanguage.ValueString.ToLowerInvariant();

            // Read langauge names from Tmx
            string TmxSourceLanguage = Path.GetFileNameWithoutExtension(sntFilenames[0]);

            TmxSourceLanguage = TmxSourceLanguage.Substring(TmxSourceLanguage.LastIndexOf('_') + 1).ToLowerInvariant();
            string TmxTargetLanguage = Path.GetFileNameWithoutExtension(sntFilenames[1]);

            TmxTargetLanguage = TmxTargetLanguage.Substring(TmxTargetLanguage.LastIndexOf('_') + 1).ToLowerInvariant();

            if (TmxSourceLanguage.Substring(0, 2) != TM.sourceLangID)
            {
                Logger.WriteLine(LogLevel.Error, "Source language mismatch between command line {0} and TMX language {1}. Please edit TmxLangMap.csv to fix. Aborting.", TM.sourceLangID, TmxSourceLanguage);
                deleteSNTfiles(sntFilenames);
                return(false);
            }

            if (TmxTargetLanguage.Substring(0, 2) != TM.targetLangID)
            {
                Logger.WriteLine(LogLevel.Error, "Target language mismatch between command line {0} and TMX language {1}. Please edit TmxLangMap.csv to fix. Aborting.", TM.targetLangID, TmxTargetLanguage);
                deleteSNTfiles(sntFilenames);
                return(false);
            }


            string[] sntSource = File.ReadAllLines(sntFilenames[0]);
            string[] sntTarget = File.ReadAllLines(sntFilenames[1]);
            if (sntSource.Length != sntTarget.Length)
            {
                Logger.WriteLine(LogLevel.Error, "Unequal number of segments. The TMX must have the same number of segments in the two given languages.");
                deleteSNTfiles(sntFilenames);
                return(false);
            }

            Logger.WriteLine(LogLevel.None, "{0} translation units read.", sntSource.Length);

            TmxWriter ErrorTmx = new TmxWriter(Path.GetFileNameWithoutExtension(this.TmxDocument.ValueString) + ".errors." + TmxSourceLanguage + "_" + TmxTargetLanguage + "." + DateTime.Now.ToString("yyyyMMddThhmmssZ") + ".tmx", TmxSourceLanguage, TmxTargetLanguage);

            //Load into TM and perform error check on each line.
            int ratioViolationCount    = 0; //counts number of ratio violations
            int sntCountViolationCount = 0; //counts number of unequal sentence count violation.

            for (int sntLineIndex = 0; sntLineIndex < sntSource.Length; sntLineIndex++)
            {
                //show a progress message.
                if ((sntLineIndex % 10) == 0)
                {
                    Logger.WriteLine(LogLevel.Debug, "{0} of {1} sentences aligned and error checked.", sntLineIndex, sntSource.Length);
                }

                //Length discrepancy check
                float ratio = Math.Abs(sntSource[sntLineIndex].Length / sntTarget[sntLineIndex].Length);
                if ((ratio > 3) && ((sntSource.Length > 15) || (sntTarget.Length > 15))) //skip the segment, and add to error.tmx
                {
                    Logger.WriteLine(LogLevel.Debug, "Length ratio exceeded. Segment skipped: {0}", sntSource[sntLineIndex].Substring(0, (60 < sntSource[sntLineIndex].Length)?60:sntSource[sntLineIndex].Length));
                    ratioViolationCount++;
                    ErrorTmx.TmxWriteSegment(sntSource[sntLineIndex], sntTarget[sntLineIndex], TmxSourceLanguage, TmxTargetLanguage, TmxWriter.TUError.lengthratio);

                    if ((ratioViolationCount / sntSource.Length) > 0.10)
                    {
                        Logger.WriteLine(LogLevel.Error, "Length ratio exceeded for 10% of segments. Probably not a translation. Aborting.");
                        deleteSNTfiles(sntFilenames);
                        return(false);
                    }
                    continue;
                }

                //TODO: special handling of bpt/ept
                sntSource[sntLineIndex] = System.Net.WebUtility.HtmlDecode(sntSource[sntLineIndex]);
                sntTarget[sntLineIndex] = System.Net.WebUtility.HtmlDecode(sntTarget[sntLineIndex]);

                //throw away segments with tags
                if ((sntSource[sntLineIndex].Contains("<") && sntSource[sntLineIndex].Contains(">")) && (sntTarget[sntLineIndex].Contains("<") && sntTarget[sntLineIndex].Contains(">")))
                {
                    Logger.WriteLine(LogLevel.Debug, "Tagged segment. Segment skipped: {0}", sntSource[sntLineIndex].Substring(0, (60 < sntSource[sntLineIndex].Length) ? 60 : sntSource[sntLineIndex].Length));
                    ErrorTmx.TmxWriteSegment(sntSource[sntLineIndex], sntTarget[sntLineIndex], TmxSourceLanguage, TmxTargetLanguage, TmxWriter.TUError.tagging);
                    continue;
                }

                //Encode the remaining <>&
                sntSource[sntLineIndex] = System.Net.WebUtility.HtmlEncode(sntSource[sntLineIndex]);
                sntTarget[sntLineIndex] = System.Net.WebUtility.HtmlEncode(sntTarget[sntLineIndex]);



                int[] sourceSentLengths = TranslationServiceFacade.BreakSentences(sntSource[sntLineIndex], TM.sourceLangID);
                int[] targetSentLengths = TranslationServiceFacade.BreakSentences(sntTarget[sntLineIndex], TM.targetLangID);

                //unequal sentence count violation check
                if (sourceSentLengths.Length != targetSentLengths.Length)
                {
                    sntCountViolationCount++;
                    Logger.WriteLine(LogLevel.Debug, "Unequal number of sentences in segment. Segment skipped: {0}", sntSource[sntLineIndex].Substring(0, (60 < sntSource[sntLineIndex].Length)?60:sntSource[sntLineIndex].Length));
                    ErrorTmx.TmxWriteSegment(sntSource[sntLineIndex], sntTarget[sntLineIndex], TmxSourceLanguage, TmxTargetLanguage, TmxWriter.TUError.sentencecountmismatch);

                    if ((sntCountViolationCount / sntSource.Length) > 0.10)
                    {
                        Logger.WriteLine(LogLevel.Error, "Unequal sentence count exceeded for 10% of segments. Probably not a translation. Aborting.");
                        deleteSNTfiles(sntFilenames);
                        return(false);
                    }
                    continue;
                }


                //Split multiple sentences
                int startIndexSrc = 0;
                int startIndexTgt = 0;
                for (int j = 0; j < sourceSentLengths.Length; j++)
                {
                    TranslationUnit TU = new TranslationUnit();
                    TU.strSource  = sntSource[sntLineIndex].Substring(startIndexSrc, sourceSentLengths[j]);
                    TU.strTarget  = sntTarget[sntLineIndex].Substring(startIndexTgt, targetSentLengths[j]);
                    startIndexSrc = sourceSentLengths[j];
                    startIndexTgt = targetSentLengths[j];
                    TU.rating     = int.Parse(ratingvalue);
                    TU.user       = uservalue.ToUpperInvariant();
                    TM.Add(TU);
                }
            }
            ErrorTmx.Dispose();


            //Add the whole TM list to CTF, if a CTF write was requested.

            if (boolWrite.ValueString.ToLowerInvariant() == "true")
            {
                int SentenceCount = 0;
                foreach (TranslationUnit TU in TM)
                {
                    TranslationServiceFacade.AddTranslation(TU.strSource, TU.strTarget, TM.sourceLangID, TM.targetLangID, TU.rating, TU.user);
                    if ((SentenceCount % 10) == 0)
                    {
                        Logger.WriteLine(LogLevel.Debug, "{0} of {1} sentences written. Continuing...", SentenceCount, sntSource.Length);
                    }
                    //Do not change the sleep time. This is slow and needs to be slow - the AddTranslation method is designed for interactive use.
                    Thread.Sleep(500);
                    SentenceCount++;
                }
                Logger.WriteLine(LogLevel.Msg, "{0} sentences written to CTF. Write complete. ", SentenceCount);
            }
            else
            {
                //Just list the entire TM on screen.
                foreach (TranslationUnit TU in TM)
                {
                    Logger.WriteLine(LogLevel.None, "{0} || {1}", TU.strSource, TU.strTarget);
                }
            }


            return(true);
        }
Exemple #4
0
        /// <summary>
        ///     The execute method.
        /// </summary>
        /// <returns>
        ///     The <see cref="bool" />.
        /// </returns>
        public override bool Execute()
        {
            string SntFileName = Path.GetTempPath() + "_TmxUpload.snt";
            string uservalue   = user.ValueString;

            if (uservalue == string.Empty)
            {
                uservalue = "TmxUpload";
            }
            string ratingvalue = rating.ValueString;

            if (ratingvalue == string.Empty)
            {
                ratingvalue = "6";
            }
            if (!File.Exists(TmxDocument.ValueString))
            {
                Logger.WriteLine(LogLevel.Error, "File {0} does not exist.", TmxDocument.ValueString);
                return(false);
            }

            TmxFile TmxIn = new TmxFile(this.TmxDocument.ValueString);

            string[] sntFilenames = TmxIn.WriteToSNTFiles(SntFileName);
            if (sntFilenames != null)
            {
                if (sntFilenames.Length != 2)
                {
                    Logger.WriteLine(LogLevel.Error, "More than 2 languages in the TMX file. Must have exactly 2.");
                    deleteSNTfiles(sntFilenames);
                    return(false);
                }
            }
            else
            {
                Logger.WriteLine(LogLevel.Error, "Not a TMX file. Aborting.");
                return(false);
            }
            TranslationMemory TM = new TranslationMemory();

            TM.sourceLangID = this.sourceLanguage.ValueString.ToLowerInvariant();
            TM.targetLangID = this.targetLanguage.ValueString.ToLowerInvariant();


            // Read language names from Tmx
            string TmxSourceLanguage = Path.GetFileNameWithoutExtension(sntFilenames[0]);

            TmxSourceLanguage = TmxSourceLanguage.Substring(TmxSourceLanguage.LastIndexOf('_') + 1).ToLowerInvariant();
            string TmxTargetLanguage = Path.GetFileNameWithoutExtension(sntFilenames[1]);

            TmxTargetLanguage = TmxTargetLanguage.Substring(TmxTargetLanguage.LastIndexOf('_') + 1).ToLowerInvariant();

            if (TmxSourceLanguage.Substring(0, 2) != TM.sourceLangID)
            {
                Logger.WriteLine(LogLevel.Error, "Source language mismatch between command line {0} and TMX language {1}. Please edit TmxLangMap.csv to fix. Aborting.", TM.sourceLangID, TmxSourceLanguage);
                deleteSNTfiles(sntFilenames);
                return(false);
            }

            if (TmxTargetLanguage.Substring(0, 2) != TM.targetLangID)
            {
                Logger.WriteLine(LogLevel.Error, "Target language mismatch between command line {0} and TMX language {1}. Please edit TmxLangMap.csv to fix. Aborting.", TM.targetLangID, TmxTargetLanguage);
                deleteSNTfiles(sntFilenames);
                return(false);
            }

            TranslationMemory TMErrors = new TranslationMemory();

            TMErrors.sourceLangID = this.sourceLanguage.ValueString.ToLowerInvariant();
            TMErrors.targetLangID = this.targetLanguage.ValueString.ToLowerInvariant();


            string[] sntSource = File.ReadAllLines(sntFilenames[0]);
            string[] sntTarget = File.ReadAllLines(sntFilenames[1]);
            if (sntSource.Length != sntTarget.Length)
            {
                Logger.WriteLine(LogLevel.Error, "Unequal number of segments. The TMX must have the same number of segments in the two given languages.");
                deleteSNTfiles(sntFilenames);
                return(false);
            }

            TmxWriter ErrorTmx = new TmxWriter(Path.GetFileNameWithoutExtension(this.TmxDocument.ValueString) + ".errors." + TmxSourceLanguage + "_" + TmxTargetLanguage + "." + DateTime.Now.ToString("yyyyMMddThhmmssZ") + ".tmx", TmxSourceLanguage, TmxTargetLanguage, false);

            //Load into TM without error checks. The AddTranslationSegments method performs error checks.
            for (int sntLineIndex = 0; sntLineIndex < sntSource.Length; sntLineIndex++)
            {
                TranslationUnit TU = new TranslationUnit(sntSource[sntLineIndex], sntTarget[sntLineIndex], Convert.ToInt16(ratingvalue), uservalue, "", TUStatus.good, "");
                TM.Add(TU);
            }

            Logger.WriteLine(LogLevel.None, "{0} translation units read.", sntSource.Length);


            bool AddToCTF = false;

            if (boolWrite.ValueString.ToLowerInvariant() == "true")
            {
                AddToCTF = true;
            }
            TMFunctions.AddTranslationSegmentsResponse ATSResponse = TMFunctions.AddTranslationSegments(TM, TMErrors, TM.sourceLangID, TM.targetLangID, Convert.ToInt16(ratingvalue), uservalue, AddToCTF);

            Logger.WriteLine(LogLevel.Msg, "Error segments: {0}", ATSResponse.errorsegments);

            if (AddToCTF)
            {
                Logger.WriteLine(LogLevel.Msg, "{0} sentences written to CTF. Write complete. ", ATSResponse.sentencesadded);
            }
            else
            {
                Logger.WriteLine(LogLevel.None, "Nothing written to CTF.");
            }


            return(true);
        }