Examples with VTDUtils net.heartsome.xml.vtdimpl.VTDUtils used on opensource projects

Example 71 with VTDUtils

use of net.heartsome.xml.vtdimpl.VTDUtils in project translationstudio8 by heartsome.

the class XLFHandler method getSrcContent.

/**
	 * 得到源的完整内容
	 * @param rowId
	 *            行的唯一标识
	 * @return 源的完整内容;
	 */
public String getSrcContent(String rowId) {
    String tuXPath = RowIdUtil.parseRowIdToXPath(rowId);
    VTDNav vn = getVTDNavByRowId(rowId);
    try {
        VTDUtils vu = new VTDUtils(vn);
        return vu.getValue(tuXPath + "/source/text()");
    } catch (NavException e) {
        LOGGER.error("", e);
        e.printStackTrace();
    }
    return null;
}

Example 72 with VTDUtils

use of net.heartsome.xml.vtdimpl.VTDUtils in project translationstudio8 by heartsome.

the class XLFHandler method getAutoQAFilteredTUText.

/**
	 * 针对自动品质检查，根据rowId获取当前tu节点的所有过滤后的数据，过滤条件为不包括上下文匹配，不包括完全匹配，不包括已锁文本，过滤条件在首选项中设置.包括源文本，源纯文本，目标文本，目标纯文本，源语言，目标语言 robert
	 * 2011-02-14
	 * @return ;
	 */
public QATUDataBean getAutoQAFilteredTUText(String rowId, Map<String, Boolean> filterMap) {
    QATUDataBean tuDataBean = new QATUDataBean();
    VTDNav vn = getVTDNavByRowId(rowId).duplicateNav();
    String tuXpath = RowIdUtil.parseRowIdToXPath(rowId);
    Assert.isNotNull(vn, Messages.getString("file.XLFHandler.msg4") + RowIdUtil.getFileNameByRowId(rowId));
    try {
        AutoPilot ap = new AutoPilot(vn);
        AutoPilot childAp = new AutoPilot(vn);
        AutoPilot langAp = new AutoPilot(vn);
        VTDUtils vUtils = new VTDUtils(vn);
        ap.selectXPath(tuXpath);
        if (ap.evalXPath() != -1) {
            // 首先过滤，如果有不应包括的文本段，则返回一个空对象
            if (!filterTheTU(vn, filterMap)) {
                tuDataBean.setPassFilter(false);
                return tuDataBean;
            }
            vn.push();
            String srcLang = "";
            // 取出源文本的纯文本之前，先查看其内容是否为空，若为空，则返回null，没有source节点，也返回null
            childAp.selectXPath("./source");
            if (childAp.evalXPath() != -1) {
                // 因为标准里面只有一个source，因此此处用if不用while
                String srcContent = vUtils.getElementContent();
                // 如果源文本为空或无值，则返回null
                if (srcContent == null || "".equals(srcContent)) {
                    return null;
                } else {
                    tuDataBean.setSrcPureText(getTUPureText(vn));
                    tuDataBean.setSrcContent(srcContent);
                }
                // 获取源语言
                int langIdx;
                if ((langIdx = vn.getAttrVal("xml:lang")) != -1) {
                    srcLang = vn.toString(langIdx);
                } else {
                    // 若该节点没有源语言，那么向上去查找file节点的源语言
                    langAp.selectXPath("ancestor::file");
                    if (langAp.evalXPath() != -1) {
                        if ((langIdx = vn.getAttrVal("source-language")) != -1) {
                            srcLang = vn.toString(langIdx);
                        }
                        langAp.resetXPath();
                    }
                }
                tuDataBean.setSrcLang(srcLang);
            } else {
                return null;
            }
            childAp.resetXPath();
            vn.pop();
            // 下面获取目标文本的纯文本，在之前先检查目标文本是否为空或为空值，若是，则返回null，若没有target节点，也返回空
            childAp.selectXPath("./target");
            if (childAp.evalXPath() != -1) {
                // 因为标准里面只有一个target，因此此处用if不用while
                String tgtContent = vUtils.getElementContent();
                // 如果源文本为空或无值，则返回空对象
                if (tgtContent == null || "".equals(tgtContent)) {
                    return tuDataBean;
                } else {
                    tuDataBean.setTgtContent(tgtContent);
                    tuDataBean.setTgtPureText(getTUPureText(vn));
                }
                // 获取目标语言
                String tgtLang = "";
                int langIdx;
                if ((langIdx = vn.getAttrVal("xml:lang")) != -1) {
                    tgtLang = vn.toString(langIdx);
                } else {
                    // 若该节点没有源语言，那么向上去查找file节点的源语言
                    langAp.selectXPath("ancestor::file");
                    if (langAp.evalXPath() != -1) {
                        if ((langIdx = vn.getAttrVal("target-language")) != -1) {
                            tgtLang = vn.toString(langIdx);
                        }
                    }
                }
                tuDataBean.setTgtLang(tgtLang);
            } else {
                return tuDataBean;
            }
        }
    } catch (Exception e) {
        LOGGER.error("", e);
        e.printStackTrace();
    }
    return tuDataBean;
}

Example 73 with VTDUtils

use of net.heartsome.xml.vtdimpl.VTDUtils in project translationstudio8 by heartsome.

the class XLFHandler method mergeSegment.

/**
	 * 合并两个文本段（注意：rowId1 和 rowId2 有可能不相邻，中间可能会间隔几个源文为空的文本段）
	 * @param rowId1
	 *            第一个文本段的 rowId
	 * @param rowId2
	 *            第二个文本段的 rowId;
	 */
public String mergeSegment(String rowId1, String rowId2) {
    // if (rowIndex + 1 >= rowIds.size()) {
    // return "";
    // }
    // String rowId1 = getRowId(rowIndex);
    // String rowId2 = getRowId(rowIndex + 1);
    String fileName1 = RowIdUtil.getFileNameByRowId(rowId1);
    String fileName2 = RowIdUtil.getFileNameByRowId(rowId2);
    if (fileName1 == null || fileName2 == null || !fileName1.equals(fileName2)) {
        return "";
    }
    VTDNav vn = vnMap.get(fileName1);
    // TransUnitBean tuTop = getTransUnit(rowIndex);
    // TransUnitBean tuBottom = getTransUnit(rowIndex + 1);
    // String srcContent = tuTop.getSrcContent() + "<ph id=\"hs-merge" +
    // tuTop.getId() + "~" + tuBottom.getId()
    // + "\"/>" + tuBottom.getSrcContent();
    // tuTop.setSrcContent(srcContent);
    // String tgtContent = tuTop.getTgtContent() + "<ph id=\"hs-merge" +
    // tuTop.getId() + "~" + tuBottom.getId()
    // + "\"/>" + tuBottom.getTgtContent();
    // tuTop.setTgtContent(tgtContent);
    // tuTop.setMatches(null);
    //
    // String tuXPath1 = RowIdUtil.parseRowIdToXPath(rowId1);
    // String tuXPath2 = RowIdUtil.parseRowIdToXPath(rowId2);
    // AutoPilot ap = new AutoPilot(vn);
    //
    // XMLModifier xm = new XMLModifier(vn);
    // delete(ap, xm, fileName2, tuXPath2, true);
    // update(ap, xm, fileName1, tuXPath2, tuTop.toXMLString(), true);
    // ap.selectXPath(tuXPath2);
    //
    // saveAndReparse(xm, fileName1);
    String tuXPath1 = RowIdUtil.parseRowIdToXPath(rowId1);
    String tuXPath2 = RowIdUtil.parseRowIdToXPath(rowId2);
    String tuid1 = RowIdUtil.getTUIdByRowId(rowId1);
    String tuid2 = RowIdUtil.getTUIdByRowId(rowId2);
    // 不需加数据库
    boolean isNotSendToTM = false;
    // 是否是疑问行
    boolean isNeeds_review = false;
    // 是否添加属性“不需添加到数据库”
    boolean isAddNotSendToTm = false;
    // 是否添加属性“疑问行”
    boolean isAddNeeds_review = false;
    StringBuffer nodeSB = new StringBuffer();
    AutoPilot ap = new AutoPilot(vn);
    try {
        VTDUtils vu = new VTDUtils(vn);
        String oldElementFragment = "";
        ap.selectXPath(tuXPath1);
        String xmlSpace = null;
        if (ap.evalXPath() != -1) {
            xmlSpace = vu.getCurrentElementAttribut("xml:space", "preserve");
            oldElementFragment += vu.getElementFragment();
            // 获取当前tu节点的属性 hs:send-to-tm="no" hs:needs-review="yes"
            isNotSendToTM = "no".equals(vu.getCurrentElementAttribut("hs:send-to-tm", ""));
            isNeeds_review = "yes".equals(vu.getCurrentElementAttribut("hs:needs-review", ""));
        }
        ap.selectXPath(tuXPath2);
        if (ap.evalXPath() != -1) {
            oldElementFragment += vu.getElementFragment();
            if (!isNotSendToTM) {
                isAddNotSendToTm = "no".equals(vu.getCurrentElementAttribut("hs:send-to-tm", ""));
            }
            if (!isNeeds_review) {
                isAddNeeds_review = "yes".equals(vu.getCurrentElementAttribut("hs:needs-review", ""));
            }
            // 开始获取批注
            ap.selectXPath("./note");
            while (ap.evalXPath() != -1) {
                nodeSB.append(vu.getElementFragment());
            }
        }
        XMLModifier xm = new XMLModifier(vn);
        String sourceContent2 = "";
        String targetContent2 = "";
        ap.selectXPath(tuXPath2 + "/source");
        String srcLang = null;
        if (ap.evalXPath() != -1) {
            srcLang = vu.getCurrentElementAttribut("xml:lang", null);
            sourceContent2 = vu.getElementContent();
            sourceContent2 = sourceContent2 == null ? "" : sourceContent2;
        }
        ap.selectXPath(tuXPath2 + "/target");
        if (ap.evalXPath() != -1) {
            targetContent2 = vu.getElementContent();
            targetContent2 = targetContent2 == null ? "" : targetContent2;
        }
        String curTime = "" + System.nanoTime();
        ap.selectXPath(tuXPath1 + "/source");
        if (ap.evalXPath() != -1) {
            String sourceContent1 = vu.getElementContent();
            sourceContent1 = sourceContent1 == null ? "" : sourceContent1;
            String newValue = sourceContent1 + "<ph id=\"hs-merge" + tuid1 + "~" + tuid2 + "\" splitMergeIndex=\"" + curTime + "\"/>" + sourceContent2;
            vu.update(ap, xm, tuXPath1 + "/source/text()", newValue);
        }
        ap.selectXPath(tuXPath1 + "/target");
        if (ap.evalXPath() != -1) {
            String targetContent1 = vu.getElementContent();
            targetContent1 = targetContent1 == null ? "" : targetContent1;
            String newValue = targetContent1 + "<ph id=\"hs-merge" + tuid1 + "~" + tuid2 + "\" splitMergeIndex=\"" + curTime + "\"/>" + targetContent2;
            vu.update(ap, xm, tuXPath1 + "/target/text()", newValue);
        } else {
            String newValue = "<ph id=\"hs-merge" + tuid1 + "~" + tuid2 + "\" splitMergeIndex=\"" + curTime + "\"/>" + targetContent2;
            vu.insert(ap, xm, tuXPath1 + "/target/text()", newValue);
        }
        // 移除alt-trans节点
        ap.selectXPath(tuXPath1 + "/alt-trans");
        while (ap.evalXPath() != -1) {
            xm.remove();
        }
        ap.selectXPath(tuXPath2);
        if (ap.evalXPath() != -1) {
            // xm.remove();
            // Bug #1054：合并文本段时不应丢失第二个文本段的 trans-unit 节点
            String transUnit = "<trans-unit id=\"" + tuid2 + "\" xml:space=\"" + xmlSpace + "\"><target/>" + "<source xml:lang=\"" + srcLang + "\"/></trans-unit>";
            vu.update(ap, xm, tuXPath2, transUnit);
        }
        // 整合并集中的属性以及批注
        ap.selectXPath(tuXPath1);
        if (ap.evalXPath() != -1) {
            // 注意添加属性时必须保持应有的空格
            String insertAttri = isAddNotSendToTm ? " hs:send-to-tm=\"no\" " : "";
            insertAttri += isAddNeeds_review ? " hs:needs-review=\"yes\" " : "";
            if (insertAttri.length() > 0) {
                xm.insertAttribute(insertAttri);
            }
            // 开始添加批注
            if (nodeSB.length() > 0) {
                xm.insertBeforeTail(nodeSB.toString().getBytes("UTF-8"));
            }
            vn.push();
            // 开始处理合并后的状态，保持草稿状态
            ap.selectXPath("./target");
            if (ap.evalXPath() != -1) {
                int index = vn.getAttrVal("state");
                if (index != -1) {
                    xm.updateToken(index, "new");
                } else {
                    xm.insertAttribute(" state=\"new\" ");
                }
            }
            vn.pop();
            // 删除批准状态
            int index = vn.getAttrVal("approved");
            if (index != -1) {
                xm.removeAttribute(index - 1);
            }
        }
        saveAndReparse(xm, fileName1);
        // 删除下面的行
        rowIds.remove(rowId2);
        tuSizeMap.put(fileName1, tuSizeMap.get(fileName1) - 1);
        return oldElementFragment;
    } catch (XPathParseException e) {
        LOGGER.error("", e);
        e.printStackTrace();
    } catch (XPathEvalException e) {
        LOGGER.error("", e);
        e.printStackTrace();
    } catch (NavException e) {
        LOGGER.error("", e);
        e.printStackTrace();
    } catch (ModifyException e) {
        LOGGER.error("", e);
        e.printStackTrace();
    } catch (UnsupportedEncodingException e) {
        LOGGER.error("", e);
        e.printStackTrace();
    }
    return "";
}

Example 74 with VTDUtils

use of net.heartsome.xml.vtdimpl.VTDUtils in project translationstudio8 by heartsome.

the class PreMachineTranslation method executeTranslation.

/**
	 * 根据构建参数执行预翻译 ;
	 * @throws InterruptedException
	 */
public List<PreMachineTranslationCounter> executeTranslation(IProgressMonitor monitor) throws InterruptedException {
    if (monitor == null) {
        monitor = new NullProgressMonitor();
    }
    simpleMatchers = getMacthers();
    if (null == simpleMatchers || simpleMatchers.isEmpty()) {
        return this.transCounters;
    }
    monitor.beginTask("", this.xlfFiles.size());
    monitor.setTaskName(Messages.getString("pretranslation.PreTranslation.task1"));
    try {
        for (String xlfPath : xlfFiles) {
            if (monitor != null && monitor.isCanceled()) {
                throw new InterruptedException();
            }
            currentCounter = new PreMachineTranslationCounter(xlfPath);
            this.transCounters.add(currentCounter);
            VTDNav vn = xlfHandler.getVnMap().get(xlfPath);
            VTDUtils vu = new VTDUtils(vn);
            AutoPilot ap = new AutoPilot(vu.getVTDNav());
            int tuNumber = xlfHandler.getNodeCount(xlfPath, "/xliff/file//descendant::trans-unit[(source/text()!='' or source/*)]");
            currentCounter.setTuNumber(tuNumber);
            ap.selectXPath("/xliff/file");
            String srcLang = "";
            String tgtLang = "";
            XMLModifier xm = new XMLModifier(vn);
            IProgressMonitor monitor2 = new SubProgressMonitor(monitor, 1);
            monitor2.beginTask(Messages.getString("pretranslation.PreTranslation.task2"), tuNumber);
            while (ap.evalXPath() != -1) {
                // 循环 file 节点
                String _srcLang = vu.getCurrentElementAttribut("source-language", "");
                String _tgtLang = vu.getCurrentElementAttribut("target-language", "");
                if (!_srcLang.equals("")) {
                    srcLang = _srcLang;
                }
                if (!_tgtLang.equals("")) {
                    tgtLang = _tgtLang;
                }
                if (srcLang.equals("") || tgtLang.equals("")) {
                    continue;
                }
                keepCurrentMatchs(vu, _srcLang, _tgtLang, xm, monitor2);
            }
            monitor2.done();
            FileOutputStream fos = new FileOutputStream(xlfPath);
            BufferedOutputStream bos = new BufferedOutputStream(fos);
            // 写入文件
            xm.output(bos);
            bos.close();
            fos.close();
        }
    } catch (XPathParseException e) {
        logger.error("", e);
        e.printStackTrace();
    } catch (NavException e) {
        logger.error("", e);
        e.printStackTrace();
    } catch (ModifyException e) {
        logger.error("", e);
        e.printStackTrace();
    } catch (XPathEvalException e) {
        logger.error("", e);
        e.printStackTrace();
    } catch (UnsupportedEncodingException e) {
        logger.error("", e);
        e.printStackTrace();
    } catch (FileNotFoundException e) {
        logger.error("", e);
        e.printStackTrace();
    } catch (TranscodeException e) {
        logger.error("", e);
        e.printStackTrace();
    } catch (IOException e) {
        logger.error("", e);
        e.printStackTrace();
    }
    monitor.done();
    return this.transCounters;
}

Example 75 with VTDUtils

use of net.heartsome.xml.vtdimpl.VTDUtils in project translationstudio8 by heartsome.

the class VTDUtilsTest method setUp.

@BeforeClass
public static void setUp() throws Exception {
    VTDGen vg = new VTDGen();
    if (vg.parseFile(testFile, true)) {
        vn = vg.getNav();
        vu = new VTDUtils(vn);
    }
}