transcriber14parser.java

编辑视频文件

        tierNameSet.add(SECTION_TIER_NAME);
        tierNameSet.add(TURN_TIER_NAME);

        for (int i = 0; i < sectionArrayList.size(); i++) {
            SectionRecord sectionR = (SectionRecord) sectionArrayList.get(i);

            long sectionBegin = new Double((new Double(sectionR.startTime)).doubleValue() * 1000).longValue();
            long sectionEnd = new Double((new Double(sectionR.endTime)).doubleValue() * 1000).longValue();

            String sectionBeginSlotId = TS_PREFIX + timeSlotCounter++;
            String sectionEndSlotId = TS_PREFIX + timeSlotCounter++;

            String value = sectionR.type;

            if ((sectionR.topicId != null) && !sectionR.topicId.equals("")) {
                value += (" - " + topicHash.get(sectionR.topicId));
            }

            addAnnotRecordAndTimeSlots(sectionBegin, sectionEnd,
                sectionBeginSlotId, sectionEndSlotId, value, SECTION_TIER_NAME);

            // turns
            String turnBeginSlotId = "";
            String turnEndSlotId = "";

            for (int j = 0; j < sectionR.turnRecords.size(); j++) {
                TurnRecord turnR = (TurnRecord) sectionR.turnRecords.get(j);

                long turnBegin = new Double((new Double(turnR.startTime)).doubleValue() * 1000).longValue();
                long turnEnd = new Double((new Double(turnR.endTime)).doubleValue() * 1000).longValue();

                if (turnBegin == sectionBegin) {
                    turnBeginSlotId = sectionBeginSlotId;
                } else if (!turnEndSlotId.equals("")) {
                    turnBeginSlotId = turnEndSlotId;
                } else {
                    turnBeginSlotId = TS_PREFIX + timeSlotCounter++;
                }

                if (turnEnd == sectionEnd) {
                    turnEndSlotId = sectionEndSlotId;
                } else {
                    turnEndSlotId = TS_PREFIX + timeSlotCounter++;
                }

                String speakerString = "";

                if (turnR.speakers != null) {
                    StringTokenizer tokenizer = new StringTokenizer(turnR.speakers);

                    while (tokenizer.hasMoreTokens()) {
                        speakerString += (String) speakersHash.get(tokenizer.nextToken());

                        if (tokenizer.hasMoreTokens()) {
                            speakerString += " + ";
                        }
                    }
                } else {
                    speakerString = "(no speaker)";
                }

                String parentId = addAnnotRecordAndTimeSlots(turnBegin,
                        turnEnd, turnBeginSlotId, turnEndSlotId, speakerString,
                        TURN_TIER_NAME);

                if ((turnR.mode != null) && !turnR.mode.equals("")) {
                    tierNameSet.add(MODE_TIER_NAME);
                    addAnnotRecord(parentId, turnR.mode, MODE_TIER_NAME);
                }

                if ((turnR.fidelity != null) && !turnR.fidelity.equals("")) {
                    tierNameSet.add(FIDELITY_TIER_NAME);
                    addAnnotRecord(parentId, turnR.fidelity, FIDELITY_TIER_NAME);
                }

                if ((turnR.channel != null) && !turnR.channel.equals("")) {
                    tierNameSet.add(CHANNEL_TIER_NAME);
                    addAnnotRecord(parentId, turnR.channel, CHANNEL_TIER_NAME);
                }
            }
        }
    }

    private void processUtterances() {
        // add tier names to tierNameSet
        Iterator speakerNameIter = speakersHash.values().iterator();

        while (speakerNameIter.hasNext()) {
            tierNameSet.add(speakerNameIter.next());
        }

        // iterate over utteranceRecords
        for (int i = 0; i < utteranceRecords.size(); i++) {
            UtteranceRecord utteranceR = (UtteranceRecord) utteranceRecords.get(i);

            long uttBegin = new Double((new Double(utteranceR.startTime)).doubleValue() * 1000).longValue();
            long uttEnd = new Double((new Double(utteranceR.endTime)).doubleValue() * 1000).longValue();

            String beginSlotId = TS_PREFIX + timeSlotCounter++;
            String endSlotId = TS_PREFIX + timeSlotCounter++;

            addAnnotRecordAndTimeSlots(uttBegin, uttEnd, beginSlotId,
                endSlotId, utteranceR.text, utteranceR.speaker);
        }
    }

    private void processBackgrounds() {
        tierNameSet.add(BACKGROUND_TIER_NAME);

        long bgBeginTime = 0;
        long bgEndTime = 0;

        for (int i = 0; i < backgroundArrayList.size(); i++) {
            BackgroundRecord backgroundR = (BackgroundRecord) backgroundArrayList.get(i);

            bgEndTime = new Double((new Double(backgroundR.time)).doubleValue() * 1000).longValue();

            if ((lastBackgroundRecord != null) &&
                    !lastBackgroundRecord.level.equals("off")) { // not first background
                bgBeginTime = new Double((new Double(lastBackgroundRecord.time)).doubleValue() * 1000).longValue();

                String beginSlotId = TS_PREFIX + timeSlotCounter++;
                String endSlotId = TS_PREFIX + timeSlotCounter++;

                addAnnotRecordAndTimeSlots(bgBeginTime, bgEndTime, beginSlotId,
                    endSlotId, lastBackgroundRecord.type, BACKGROUND_TIER_NAME);
            }

            lastBackgroundRecord = backgroundR;
        }

        if ((lastBackgroundRecord != null) &&
                !lastBackgroundRecord.level.equals("off") &&
                (sectionArrayList.size() > 0)) { // handle last background record, end at last section end

            SectionRecord sectionR = (SectionRecord) sectionArrayList.get(sectionArrayList.size() -
                    1);

            if (sectionR != null) {
                bgBeginTime = new Double((new Double(lastBackgroundRecord.time)).doubleValue() * 1000).longValue();
                bgEndTime = new Double((new Double(sectionR.endTime)).doubleValue() * 1000).longValue();

                String beginSlotId = TS_PREFIX + timeSlotCounter++;
                String endSlotId = TS_PREFIX + timeSlotCounter++;

                addAnnotRecordAndTimeSlots(bgBeginTime, bgEndTime, beginSlotId,
                    endSlotId, lastBackgroundRecord.type, BACKGROUND_TIER_NAME);
            }
        }
    }

    private void processComments() {
        tierNameSet.add(COMMENT_TIER_NAME);

        // iterate over commentRecords
        for (int i = 0; i < commentRecords.size(); i++) {
            CommentRecord commentR = (CommentRecord) commentRecords.get(i);

            long commBegin = new Double((new Double(commentR.begin)).doubleValue() * 1000).longValue();
            long commEnd = new Double((new Double(commentR.end)).doubleValue() * 1000).longValue();

            String beginSlotId = TS_PREFIX + timeSlotCounter++;
            String endSlotId = TS_PREFIX + timeSlotCounter++;

            addAnnotRecordAndTimeSlots(commBegin, commEnd, beginSlotId,
                endSlotId, commentR.desc, COMMENT_TIER_NAME);
        }
    }

    private String addAnnotRecordAndTimeSlots(long begin, long end,
        String beginSlotId, String endSlotId, String value, String tierName) {
        timeSlots.put(beginSlotId, new Long(begin));
        timeSlots.put(endSlotId, new Long(end));

        String annId = ANN_PREFIX + annotationCounter++;

        AnnotationRecord annRec = new AnnotationRecord();
        annRec.setAnnotationId(annId);
        annRec.setAnnotationType(AnnotationRecord.ALIGNABLE);
        annRec.setBeginTimeSlotId(beginSlotId);
        annRec.setEndTimeSlotId(endSlotId);
        annRec.setValue(value);

        annotRecordToTierMap.put(annRec, tierName);

        return annId;
    }

    private void addAnnotRecord(String parentId, String value, String tierName) {
        AnnotationRecord annRec = new AnnotationRecord();
        annRec.setAnnotationId(ANN_PREFIX + annotationCounter++);
        annRec.setAnnotationType(AnnotationRecord.REFERENCE);
        annRec.setReferredAnnotId(parentId);
        annRec.setValue(value);

        annotRecordToTierMap.put(annRec, tierName);
    }

    private void println(String s) {
        if (verbose) {
            System.out.println(s);
        }
    }

    private void printErrorLocationInfo(String message) {
        System.out.println(message);
        System.out.println("Exception for " + currentFileName);
    }

    /*
     * This method should be in a Utility class or a URL class
     * Convert a path to a file URL string. Takes care of Samba related problems
     * file:///path works for all files except for samba file systems, there we need file://machine/path,
     * i.e. 2 slashes insteda of 3
     *
     * What's with relative paths?
     */
    private String pathToURLString(String path) {
        // replace all back slashes by forward slashes
        path = path.replace('\\', '/');

        // remove leading slashes and count them
        int n = 0;

        while (path.charAt(0) == '/') {
            path = path.substring(1);
            n++;
        }

        // add the file:// or file:/// prefix
        if (n == 2) {
            return "file://" + path;
        } else {
            return "file:///" + path;
        }
    }

    /**
     * @see mpi.eudico.server.corpora.clomimpl.abstr.Parser#setDecoderInfo(mpi.eudico.server.corpora.clom.DecoderInfo)
     */
    public void setDecoderInfo(DecoderInfo decoderInfo) {
        if (decoderInfo instanceof TranscriberDecoderInfo) {
            this.decoderInfo = (TranscriberDecoderInfo) decoderInfo;
            singleSpeechTier = this.decoderInfo.isSingleSpeakerTier();
        }
    }

    /**
     * DOCUMENT ME!
     */
    class Transcriber14Handler extends DefaultHandler {
        /**
         * ContentHandler method
         *
         * @throws SAXException DOCUMENT ME!
         */
        public void startDocument() throws SAXException {
            parseError = false;
        }

        /**
         * ContentHandler method
         *
         * @throws SAXException DOCUMENT ME!
         */
        public void endDocument() throws SAXException {
        }

        /**
         * ContentHandler method
         *
         * @param nameSpaceURI DOCUMENT ME!
         * @param name DOCUMENT ME!
         * @param rawName DOCUMENT ME!
         * @param attributes DOCUMENT ME!
         *
         * @throws SAXException DOCUMENT ME!
         */
        public void startElement(String nameSpaceURI, String name,
            String qName, Attributes attributes) throws SAXException {
            //	System.out.println("startElement called for name:" + name);
            if (qName.equals("Trans")) {
                audioFileName = attributes.getValue("audio_filename");
                scribe = attributes.getValue("scribe");
                language = attributes.getValue("xml:lang");
                date = attributes.getValue("version_date");
            } else if (qName.equals("Speaker")) {
                speakersHash.put(attributes.getValue("id"),
                    attributes.getValue("name"));

                if (!singleSpeechTier) {
                    currentSpeakerContents.put(attributes.getValue("id"), "");
                }
            } else if (qName.equals("Section")) {
                currentSectionRecord = new SectionRecord(attributes.getValue(
                            "type"), attributes.getValue("startTime"),
                        attributes.getValue("endTime"),
                        attributes.getValue("topic"));
                sectionArrayList.add(currentSectionRecord);
            } else if (qName.equals("Turn")) {
                currentTurnRecord = new TurnRecord(attributes.getValue(
                            "startTime"), attributes.getValue("endTime"),
                        attributes.getValue("speaker"),
                        attributes.getValue("mode"),
                        attributes.getValue("fidelity"),
                        attributes.getValue("channel"));

                if (currentSectionRecord != null) {
                    currentSectionRecord.turnRecords.add(currentTurnRecord);
                }

                speakersForCurrentTurn = attributes.getValue("speaker");

                if (!singleSpeechTier) {
                    if (speakersForCurrentTurn != null) {
                        StringTokenizer tokenizer = new StringTokenizer(speakersForCurrentTurn);

                        if (tokenizer.hasMoreTokens()) {
                            currentSpeakerId = tokenizer.nextToken();
                        } else {
                            currentSpeakerId = "";
                        }
                    } else {
                        currentSpeakerId = SPEAKER_UNSPECIFIED;
                        speakersHash.put(SPEAKER_UNSPECIFIED,
                            SPEAKER_UNSPECIFIED);

                        currentSpeakerContents.put(SPEAKER_UNSPECIFIED, "");
                    }
                }
            } else if (qName.equals("Sync")) {
                String time = attributes.getValue("time");
                storeUtterances(time);
                storeComments(time);

                // store
                lastSyncTime = time;
            } else if (qName.equals("Event")) {
                String desc = attributes.getValue("desc");
                String extent = attributes.getValue("extent");

                String eventString = TranscriberEvent.getEventString(desc,
                        extent);

                if ((currentSpeakerContents != null) &&
                        (currentSpeakerId != null)) {
                    String content = (String) currentSpeakerContents.get(currentSpeakerId);

                    if (content != null) {
                        content += eventString;
                    }

                    currentSpeakerContents.put(currentSpeakerId, content);
                }
            } else if (qName.equals("Who")) {
                String nb = attributes.getValue("nb");

                if (singleSpeechTier) {
                    if ((currentSpeakerContents != null) &&
                            (currentSpeakerId != null)) {
                        String content = (String) currentSpeakerContents.get(currentSpeakerId);
                        String whoStr = "(" + SP_PREFIX + ":" + nb + ")";

                        if (content != null) {
                            if (content.length() == 0) {
                                content = whoStr;
                            } else {
                                content += (" " + whoStr);
                            }
                        } else {
                            content = whoStr;
                        }

                        currentSpeakerContents.put(currentSpeakerId, content);
                    }