Revision 786696e3
Added by Andreas Müller over 10 years ago
| cdmlib-io/src/main/java/eu/etaxonomy/cdm/io/markup/MarkupDocumentImportNoComponent.java | ||
|---|---|---|
| 11 | 11 |
|
| 12 | 12 |
import java.net.MalformedURLException; |
| 13 | 13 |
import java.net.URL; |
| 14 |
import java.util.ArrayList; |
|
| 15 |
import java.util.Arrays; |
|
| 16 |
import java.util.HashMap; |
|
| 17 | 14 |
import java.util.HashSet; |
| 18 |
import java.util.List; |
|
| 19 | 15 |
import java.util.Map; |
| 20 | 16 |
import java.util.Set; |
| 21 | 17 |
import java.util.UUID; |
| 22 |
import java.util.regex.Matcher; |
|
| 23 |
import java.util.regex.Pattern; |
|
| 24 | 18 |
|
| 25 | 19 |
import javax.xml.stream.Location; |
| 26 | 20 |
import javax.xml.stream.XMLEventReader; |
| ... | ... | |
| 32 | 26 |
import org.apache.commons.lang.StringUtils; |
| 33 | 27 |
import org.apache.log4j.Logger; |
| 34 | 28 |
|
| 35 |
import eu.etaxonomy.cdm.common.CdmUtils; |
|
| 36 | 29 |
import eu.etaxonomy.cdm.io.common.mapping.UndefinedTransformerMethodException; |
| 37 | 30 |
import eu.etaxonomy.cdm.model.agent.TeamOrPersonBase; |
| 38 |
import eu.etaxonomy.cdm.model.common.AnnotatableEntity; |
|
| 39 |
import eu.etaxonomy.cdm.model.common.Annotation; |
|
| 40 |
import eu.etaxonomy.cdm.model.common.AnnotationType; |
|
| 41 |
import eu.etaxonomy.cdm.model.common.CdmBase; |
|
| 42 |
import eu.etaxonomy.cdm.model.common.Extension; |
|
| 43 | 31 |
import eu.etaxonomy.cdm.model.common.ExtensionType; |
| 44 | 32 |
import eu.etaxonomy.cdm.model.common.Language; |
| 45 | 33 |
import eu.etaxonomy.cdm.model.common.OriginalSourceType; |
| ... | ... | |
| 50 | 38 |
import eu.etaxonomy.cdm.model.description.Feature; |
| 51 | 39 |
import eu.etaxonomy.cdm.model.description.PolytomousKey; |
| 52 | 40 |
import eu.etaxonomy.cdm.model.description.PolytomousKeyNode; |
| 53 |
import eu.etaxonomy.cdm.model.description.PresenceAbsenceTermBase; |
|
| 54 |
import eu.etaxonomy.cdm.model.description.PresenceTerm; |
|
| 55 | 41 |
import eu.etaxonomy.cdm.model.description.TaxonDescription; |
| 56 | 42 |
import eu.etaxonomy.cdm.model.description.TextData; |
| 57 |
import eu.etaxonomy.cdm.model.location.NamedArea; |
|
| 58 |
import eu.etaxonomy.cdm.model.location.NamedAreaLevel; |
|
| 59 |
import eu.etaxonomy.cdm.model.media.IdentifiableMediaEntity; |
|
| 60 |
import eu.etaxonomy.cdm.model.media.Media; |
|
| 61 | 43 |
import eu.etaxonomy.cdm.model.name.CultivarPlantName; |
| 62 | 44 |
import eu.etaxonomy.cdm.model.name.NonViralName; |
| 63 | 45 |
import eu.etaxonomy.cdm.model.name.Rank; |
| 64 |
import eu.etaxonomy.cdm.model.occurrence.SpecimenOrObservationBase; |
|
| 65 | 46 |
import eu.etaxonomy.cdm.model.reference.Reference; |
| 66 | 47 |
import eu.etaxonomy.cdm.model.reference.ReferenceFactory; |
| 67 | 48 |
import eu.etaxonomy.cdm.model.taxon.Classification; |
| 68 | 49 |
import eu.etaxonomy.cdm.model.taxon.Taxon; |
| 69 | 50 |
import eu.etaxonomy.cdm.model.taxon.TaxonNode; |
| 70 |
import eu.etaxonomy.cdm.strategy.parser.NonViralNameParserImpl; |
|
| 71 | 51 |
|
| 72 | 52 |
|
| 73 | 53 |
/** |
| ... | ... | |
| 75 | 55 |
* |
| 76 | 56 |
*/ |
| 77 | 57 |
public class MarkupDocumentImportNoComponent extends MarkupImportBase {
|
| 58 |
@SuppressWarnings("unused")
|
|
| 78 | 59 |
private static final Logger logger = Logger.getLogger(MarkupDocumentImportNoComponent.class); |
| 79 | 60 |
|
| 80 | 61 |
private MarkupKeyImport keyImport; |
| 81 |
private MarkupSpecimenImport specimenImport; |
|
| 82 | 62 |
|
| 63 |
private MarkupModsImport modsImport; |
|
| 64 |
private MarkupFeatureImport featureImport; |
|
| 65 |
private MarkupSpecimenImport specimenImport; |
|
| 83 | 66 |
private MarkupNomenclatureImport nomenclatureImport; |
| 84 |
|
|
| 67 |
|
|
| 85 | 68 |
public MarkupDocumentImportNoComponent(MarkupDocumentImport docImport) {
|
| 86 | 69 |
super(docImport); |
| 87 | 70 |
this.keyImport = new MarkupKeyImport(docImport); |
| 88 | 71 |
this.specimenImport = new MarkupSpecimenImport(docImport); |
| 89 |
nomenclatureImport = new MarkupNomenclatureImport(docImport, keyImport, specimenImport); |
|
| 72 |
this.nomenclatureImport = new MarkupNomenclatureImport(docImport, specimenImport); |
|
| 73 |
this.modsImport = new MarkupModsImport(docImport); |
|
| 74 |
this.featureImport = new MarkupFeatureImport(docImport, specimenImport, nomenclatureImport); |
|
| 90 | 75 |
} |
| 91 | 76 |
|
| 92 | 77 |
public void doInvoke(MarkupImportState state) throws XMLStreamException {
|
| ... | ... | |
| 193 | 178 |
message = String.format(message, baseUrl); |
| 194 | 179 |
fireWarningEvent(message, next, 8); |
| 195 | 180 |
} |
| 196 |
} else if (isStartingElement(next, MODS)){
|
|
| 197 |
handleNotYetImplementedElement(next);
|
|
| 181 |
} else if (isStartingElement(next, MODS)){
|
|
| 182 |
modsImport.handleMods(state, reader, next);
|
|
| 198 | 183 |
} else {
|
| 199 | 184 |
handleUnexpectedElement(next); |
| 200 | 185 |
} |
| ... | ... | |
| 398 | 383 |
nomenclatureImport.handleNomenclature(state, reader, next); |
| 399 | 384 |
hasNomenclature = true; |
| 400 | 385 |
} else if (isStartingElement(next, FEATURE)) {
|
| 401 |
handleFeature(state, reader, next); |
|
| 386 |
featureImport.handleFeature(state, reader, next);
|
|
| 402 | 387 |
} else if (isStartingElement(next, NOTES)) {
|
| 403 | 388 |
// TODO is this the correct way to handle notes? |
| 404 | 389 |
String note = handleNotes(state, reader, next); |
| ... | ... | |
| 425 | 410 |
desc.addElement(textData); |
| 426 | 411 |
} |
| 427 | 412 |
textData = (TextData)desc.getElements().iterator().next(); |
| 428 |
makeFeatureFigureRef(state, reader, desc, false, textData, next); |
|
| 413 |
featureImport.makeFeatureFigureRef(state, reader, desc, false, textData, next);
|
|
| 429 | 414 |
} else if (isStartingElement(next, FIGURE)) {
|
| 430 |
handleFigure(state, reader, next); |
|
| 415 |
handleFigure(state, reader, next, specimenImport, nomenclatureImport);
|
|
| 431 | 416 |
} else if (isStartingElement(next, FOOTNOTE)) {
|
| 432 |
FootnoteDataHolder footnote = handleFootnote(state, reader, next); |
|
| 417 |
FootnoteDataHolder footnote = handleFootnote(state, reader, next, specimenImport, nomenclatureImport);
|
|
| 433 | 418 |
if (footnote.isRef()) {
|
| 434 | 419 |
String message = "Ref footnote not implemented here"; |
| 435 | 420 |
fireWarningEvent(message, next, 4); |
| ... | ... | |
| 458 | 443 |
WriterDataHolder writer = handleWriter(state, reader, next); |
| 459 | 444 |
taxon.addExtension(writer.extension); |
| 460 | 445 |
// TODO what if taxonTitle comes later |
| 461 |
if (StringUtils.isNotBlank(taxonTitle) |
|
| 462 |
&& writer.extension != null) {
|
|
| 446 |
if (StringUtils.isNotBlank(taxonTitle) && writer.extension != null) {
|
|
| 463 | 447 |
Reference<?> sec = ReferenceFactory.newBookSection(); |
| 464 | 448 |
sec.setTitle(taxonTitle); |
| 465 | 449 |
TeamOrPersonBase<?> author = createAuthor(writer.writer); |
| 466 | 450 |
sec.setAuthorTeam(author); |
| 467 |
sec.setInReference(state.getConfig() |
|
| 468 |
.getSourceReference()); |
|
| 451 |
sec.setInReference(state.getConfig().getSourceReference()); |
|
| 469 | 452 |
taxon.setSec(sec); |
| 470 | 453 |
registerFootnotes(state, sec, writer.footnotes); |
| 471 | 454 |
} else {
|
| ... | ... | |
| 645 | 628 |
|
| 646 | 629 |
} |
| 647 | 630 |
|
| 648 |
private WriterDataHolder handleWriter(MarkupImportState state, XMLEventReader reader, XMLEvent parentEvent) throws XMLStreamException {
|
|
| 649 |
String text = ""; |
|
| 650 |
checkNoAttributes(parentEvent); |
|
| 651 |
WriterDataHolder dataHolder = new WriterDataHolder(); |
|
| 652 |
List<FootnoteDataHolder> footnotes = new ArrayList<FootnoteDataHolder>(); |
|
| 653 |
|
|
| 654 |
// TODO handle attributes |
|
| 655 |
while (reader.hasNext()) {
|
|
| 656 |
XMLEvent next = readNoWhitespace(reader); |
|
| 657 |
if (isMyEndingElement(next, parentEvent)) {
|
|
| 658 |
text = CdmUtils.removeBrackets(text); |
|
| 659 |
if (checkMandatoryText(text, parentEvent)) {
|
|
| 660 |
text = normalize(text); |
|
| 661 |
dataHolder.writer = text; |
|
| 662 |
dataHolder.footnotes = footnotes; |
|
| 663 |
|
|
| 664 |
// Extension |
|
| 665 |
UUID uuidWriterExtension = MarkupTransformer.uuidWriterExtension; |
|
| 666 |
ExtensionType writerExtensionType = this |
|
| 667 |
.getExtensionType(state, uuidWriterExtension, |
|
| 668 |
"Writer", "writer", "writer"); |
|
| 669 |
Extension extension = Extension.NewInstance(); |
|
| 670 |
extension.setType(writerExtensionType); |
|
| 671 |
extension.setValue(text); |
|
| 672 |
dataHolder.extension = extension; |
|
| 673 |
|
|
| 674 |
// Annotation |
|
| 675 |
UUID uuidWriterAnnotation = MarkupTransformer.uuidWriterAnnotation; |
|
| 676 |
AnnotationType writerAnnotationType = this.getAnnotationType(state, uuidWriterAnnotation, "Writer", "writer", "writer", null); |
|
| 677 |
Annotation annotation = Annotation.NewInstance(text, writerAnnotationType, getDefaultLanguage(state)); |
|
| 678 |
dataHolder.annotation = annotation; |
|
| 679 |
|
|
| 680 |
return dataHolder; |
|
| 681 |
} else {
|
|
| 682 |
return null; |
|
| 683 |
} |
|
| 684 |
} else if (isStartingElement(next, FOOTNOTE_REF)) {
|
|
| 685 |
FootnoteDataHolder footNote = handleFootnoteRef(state, reader, next); |
|
| 686 |
if (footNote.isRef()) {
|
|
| 687 |
footnotes.add(footNote); |
|
| 688 |
} else {
|
|
| 689 |
logger.warn("Non ref footnotes not yet impelemnted");
|
|
| 690 |
} |
|
| 691 |
} else if (next.isCharacters()) {
|
|
| 692 |
text += next.asCharacters().getData(); |
|
| 693 |
|
|
| 694 |
} else {
|
|
| 695 |
handleUnexpectedElement(next); |
|
| 696 |
state.setUnsuccessfull(); |
|
| 697 |
} |
|
| 698 |
} |
|
| 699 |
throw new IllegalStateException("<writer> has no end tag");
|
|
| 700 |
} |
|
| 701 |
|
|
| 702 |
private void registerFootnotes(MarkupImportState state, AnnotatableEntity entity, List<FootnoteDataHolder> footnotes) {
|
|
| 703 |
for (FootnoteDataHolder footNote : footnotes) {
|
|
| 704 |
registerFootnoteDemand(state, entity, footNote); |
|
| 705 |
} |
|
| 706 |
} |
|
| 707 |
|
|
| 708 |
private void registerGivenFootnote(MarkupImportState state, FootnoteDataHolder footnote) {
|
|
| 709 |
state.registerFootnote(footnote); |
|
| 710 |
Set<AnnotatableEntity> demands = state.getFootnoteDemands(footnote.id); |
|
| 711 |
if (demands != null) {
|
|
| 712 |
for (AnnotatableEntity entity : demands) {
|
|
| 713 |
attachFootnote(state, entity, footnote); |
|
| 714 |
} |
|
| 715 |
} |
|
| 716 |
} |
|
| 717 |
|
|
| 718 |
private void registerGivenFigure(MarkupImportState state, XMLEvent next, String id, Media figure) {
|
|
| 719 |
state.registerFigure(id, figure); |
|
| 720 |
Set<AnnotatableEntity> demands = state.getFigureDemands(id); |
|
| 721 |
if (demands != null) {
|
|
| 722 |
for (AnnotatableEntity entity : demands) {
|
|
| 723 |
attachFigure(state, next, entity, figure); |
|
| 724 |
} |
|
| 725 |
} |
|
| 726 |
save(figure, state); |
|
| 727 |
} |
|
| 728 |
|
|
| 729 |
private void registerFootnoteDemand(MarkupImportState state, AnnotatableEntity entity, FootnoteDataHolder footnote) {
|
|
| 730 |
FootnoteDataHolder existingFootnote = state.getFootnote(footnote.ref); |
|
| 731 |
if (existingFootnote != null) {
|
|
| 732 |
attachFootnote(state, entity, existingFootnote); |
|
| 733 |
} else {
|
|
| 734 |
Set<AnnotatableEntity> demands = state.getFootnoteDemands(footnote.ref); |
|
| 735 |
if (demands == null) {
|
|
| 736 |
demands = new HashSet<AnnotatableEntity>(); |
|
| 737 |
state.putFootnoteDemands(footnote.ref, demands); |
|
| 738 |
} |
|
| 739 |
demands.add(entity); |
|
| 740 |
} |
|
| 741 |
} |
|
| 742 |
|
|
| 743 |
private void registerFigureDemand(MarkupImportState state, XMLEvent next, AnnotatableEntity entity, String figureRef) {
|
|
| 744 |
Media existingFigure = state.getFigure(figureRef); |
|
| 745 |
if (existingFigure != null) {
|
|
| 746 |
attachFigure(state, next, entity, existingFigure); |
|
| 747 |
} else {
|
|
| 748 |
Set<AnnotatableEntity> demands = state.getFigureDemands(figureRef); |
|
| 749 |
if (demands == null) {
|
|
| 750 |
demands = new HashSet<AnnotatableEntity>(); |
|
| 751 |
state.putFigureDemands(figureRef, demands); |
|
| 752 |
} |
|
| 753 |
demands.add(entity); |
|
| 754 |
} |
|
| 755 |
} |
|
| 756 |
|
|
| 757 |
private void attachFootnote(MarkupImportState state, AnnotatableEntity entity, FootnoteDataHolder footnote) {
|
|
| 758 |
AnnotationType annotationType = this.getAnnotationType(state, MarkupTransformer.uuidFootnote, "Footnote", "An e-flora footnote", "fn", null); |
|
| 759 |
Annotation annotation = Annotation.NewInstance(footnote.string, annotationType, getDefaultLanguage(state)); |
|
| 760 |
// TODO transient objects |
|
| 761 |
entity.addAnnotation(annotation); |
|
| 762 |
save(entity, state); |
|
| 763 |
} |
|
| 764 |
|
|
| 765 |
private void attachFigure(MarkupImportState state, XMLEvent next, AnnotatableEntity entity, Media figure) {
|
|
| 766 |
// IdentifiableEntity<?> toSave; |
|
| 767 |
if (entity.isInstanceOf(TextData.class)) {
|
|
| 768 |
TextData deb = CdmBase.deproxy(entity, TextData.class); |
|
| 769 |
deb.addMedia(figure); |
|
| 770 |
// toSave = ((TaxonDescription)deb.getInDescription()).getTaxon(); |
|
| 771 |
} else if (entity.isInstanceOf(SpecimenOrObservationBase.class)) {
|
|
| 772 |
String message = "figures for specimen should be handled as Textdata"; |
|
| 773 |
fireWarningEvent(message, next, 4); |
|
| 774 |
// toSave = ime; |
|
| 775 |
} else if (entity.isInstanceOf(IdentifiableMediaEntity.class)) {
|
|
| 776 |
IdentifiableMediaEntity<?> ime = CdmBase.deproxy(entity, IdentifiableMediaEntity.class); |
|
| 777 |
ime.addMedia(figure); |
|
| 778 |
// toSave = ime; |
|
| 779 |
} else {
|
|
| 780 |
String message = "Unsupported entity to attach media: %s"; |
|
| 781 |
message = String.format(message, entity.getClass().getName()); |
|
| 782 |
// toSave = null; |
|
| 783 |
} |
|
| 784 |
save(entity, state); |
|
| 785 |
} |
|
| 786 |
|
|
| 787 |
private Media handleFigure(MarkupImportState state, XMLEventReader reader, XMLEvent parentEvent) throws XMLStreamException {
|
|
| 788 |
// FigureDataHolder result = new FigureDataHolder(); |
|
| 789 |
|
|
| 790 |
Map<String, Attribute> attributes = getAttributes(parentEvent); |
|
| 791 |
String id = getAndRemoveAttributeValue(attributes, ID); |
|
| 792 |
String type = getAndRemoveAttributeValue(attributes, TYPE); |
|
| 793 |
String urlAttr = getAndRemoveAttributeValue(attributes, URL); |
|
| 794 |
checkNoAttributes(attributes, parentEvent); |
|
| 795 |
|
|
| 796 |
String urlString = null; |
|
| 797 |
String legendString = null; |
|
| 798 |
String titleString = null; |
|
| 799 |
String numString = null; |
|
| 800 |
String text = null; |
|
| 801 |
if (isNotBlank(urlAttr)){
|
|
| 802 |
urlString = CdmUtils.Nz(state.getBaseMediaUrl()) + urlAttr; |
|
| 803 |
} |
|
| 804 |
while (reader.hasNext()) {
|
|
| 805 |
XMLEvent next = readNoWhitespace(reader); |
|
| 806 |
if (isMyEndingElement(next, parentEvent)) {
|
|
| 807 |
if (isNotBlank(text)){
|
|
| 808 |
fireWarningEvent("Text not yet handled for figures: " + text, next, 4);
|
|
| 809 |
} |
|
| 810 |
Media media = makeFigure(state, id, type, urlString, legendString, titleString, numString, next); |
|
| 811 |
return media; |
|
| 812 |
} else if (isStartingElement(next, FIGURE_LEGEND)) {
|
|
| 813 |
// TODO same as figure string ? |
|
| 814 |
legendString = handleFootnoteString(state, reader, next); |
|
| 815 |
} else if (isStartingElement(next, FIGURE_TITLE)) {
|
|
| 816 |
titleString = getCData(state, reader, next); |
|
| 817 |
} else if (isStartingElement(next, URL)) {
|
|
| 818 |
String localUrl = getCData(state, reader, next); |
|
| 819 |
String url = CdmUtils.Nz(state.getBaseMediaUrl()) + localUrl; |
|
| 820 |
if (isBlank(urlString)){
|
|
| 821 |
urlString = url; |
|
| 822 |
} |
|
| 823 |
if (! url.equals(urlString)){
|
|
| 824 |
String message = "URL attribute and URL element differ. Attribute: %s, Element: %s"; |
|
| 825 |
fireWarningEvent(String.format(message, urlString, url), next, 2); |
|
| 826 |
} |
|
| 827 |
} else if (isStartingElement(next, NUM)) {
|
|
| 828 |
numString = getCData(state, reader, next); |
|
| 829 |
} else if (next.isCharacters()) {
|
|
| 830 |
text += CdmUtils.concat("", text, next.asCharacters().getData());
|
|
| 831 |
} else {
|
|
| 832 |
fireUnexpectedEvent(next, 0); |
|
| 833 |
} |
|
| 834 |
} |
|
| 835 |
throw new IllegalStateException("<figure> has no end tag");
|
|
| 836 |
} |
|
| 837 |
|
|
| 838 |
/** |
|
| 839 |
* @param state |
|
| 840 |
* @param id |
|
| 841 |
* @param type |
|
| 842 |
* @param urlString |
|
| 843 |
* @param legendString |
|
| 844 |
* @param titleString |
|
| 845 |
* @param numString |
|
| 846 |
* @param next |
|
| 847 |
*/ |
|
| 848 |
private Media makeFigure(MarkupImportState state, String id, String type, String urlString, |
|
| 849 |
String legendString, String titleString, String numString, XMLEvent next) {
|
|
| 850 |
Media media = null; |
|
| 851 |
boolean isFigure = false; |
|
| 852 |
try {
|
|
| 853 |
//TODO maybe everything is a figure as it is all taken from a book |
|
| 854 |
if ("lineart".equals(type)) {
|
|
| 855 |
isFigure = true; |
|
| 856 |
// media = Figure.NewInstance(url.toURI(), null, null, null); |
|
| 857 |
} else if (type == null || "photo".equals(type) |
|
| 858 |
|| "signature".equals(type) |
|
| 859 |
|| "others".equals(type)) {
|
|
| 860 |
//TODO |
|
| 861 |
} else {
|
|
| 862 |
String message = "Unknown figure type '%s'"; |
|
| 863 |
message = String.format(message, type); |
|
| 864 |
fireWarningEvent(message, next, 2); |
|
| 865 |
} |
|
| 866 |
media = docImport.getImageMedia(urlString, docImport.getReadMediaData(), isFigure); |
|
| 867 |
|
|
| 868 |
if (media != null){
|
|
| 869 |
// title |
|
| 870 |
if (StringUtils.isNotBlank(titleString)) {
|
|
| 871 |
media.putTitle(getDefaultLanguage(state), titleString); |
|
| 872 |
} |
|
| 873 |
// legend |
|
| 874 |
if (StringUtils.isNotBlank(legendString)) {
|
|
| 875 |
media.putDescription(getDefaultLanguage(state), legendString); |
|
| 876 |
} |
|
| 877 |
if (StringUtils.isNotBlank(numString)) {
|
|
| 878 |
// TODO use concrete source (e.g. DAPHNIPHYLLACEAE in FM |
|
| 879 |
// vol.13) |
|
| 880 |
Reference<?> citation = state.getConfig().getSourceReference(); |
|
| 881 |
media.addImportSource( numString, "num", citation, null); |
|
| 882 |
// TODO name used in source if available |
|
| 883 |
} |
|
| 884 |
// TODO which citation |
|
| 885 |
if (StringUtils.isNotBlank(id)) {
|
|
| 886 |
media.addImportSource(id, null, state.getConfig().getSourceReference(), null); |
|
| 887 |
} else {
|
|
| 888 |
String message = "Figure id should never be empty or null"; |
|
| 889 |
fireWarningEvent(message, next, 6); |
|
| 890 |
} |
|
| 891 |
|
|
| 892 |
// text |
|
| 893 |
// do nothing |
|
| 894 |
registerGivenFigure(state, next, id, media); |
|
| 895 |
|
|
| 896 |
}else{
|
|
| 897 |
String message = "No media found: "; |
|
| 898 |
fireWarningEvent(message, next, 4); |
|
| 899 |
} |
|
| 900 |
} catch (MalformedURLException e) {
|
|
| 901 |
String message = "Media uri has incorrect syntax: %s"; |
|
| 902 |
message = String.format(message, urlString); |
|
| 903 |
fireWarningEvent(message, next, 4); |
|
| 904 |
// } catch (URISyntaxException e) {
|
|
| 905 |
// String message = "Media uri has incorrect syntax: %s"; |
|
| 906 |
// message = String.format(message, urlString); |
|
| 907 |
// fireWarningEvent(message, next, 4); |
|
| 908 |
} |
|
| 909 |
|
|
| 910 |
return media; |
|
| 911 |
} |
|
| 912 |
|
|
| 913 |
private FigureDataHolder handleFigureRef(MarkupImportState state, XMLEventReader reader, XMLEvent parentEvent) |
|
| 914 |
throws XMLStreamException {
|
|
| 915 |
FigureDataHolder result = new FigureDataHolder(); |
|
| 916 |
Map<String, Attribute> attributes = getAttributes(parentEvent); |
|
| 917 |
result.ref = getAndRemoveAttributeValue(attributes, REF); |
|
| 918 |
checkNoAttributes(attributes, parentEvent); |
|
| 919 |
|
|
| 920 |
// text is not handled, needed only for debugging purposes |
|
| 921 |
String text = ""; |
|
| 922 |
while (reader.hasNext()) {
|
|
| 923 |
XMLEvent next = readNoWhitespace(reader); |
|
| 924 |
if (isMyEndingElement(next, parentEvent)) {
|
|
| 925 |
return result; |
|
| 926 |
} else if (isStartingElement(next, NUM)) {
|
|
| 927 |
String num = getCData(state, reader, next); |
|
| 928 |
result.num = num; // num is not handled during import |
|
| 929 |
} else if (isStartingElement(next, FIGURE_PART)) {
|
|
| 930 |
result.figurePart = getCData(state, reader, next); |
|
| 931 |
} else if (next.isCharacters()) {
|
|
| 932 |
text += next.asCharacters().getData(); |
|
| 933 |
} else {
|
|
| 934 |
fireUnexpectedEvent(next, 0); |
|
| 935 |
} |
|
| 936 |
} |
|
| 937 |
throw new IllegalStateException("<figureRef> has no end tag");
|
|
| 938 |
} |
|
| 939 |
|
|
| 940 |
private FootnoteDataHolder handleFootnote(MarkupImportState state, XMLEventReader reader, XMLEvent parentEvent) throws XMLStreamException {
|
|
| 941 |
FootnoteDataHolder result = new FootnoteDataHolder(); |
|
| 942 |
Map<String, Attribute> attributes = getAttributes(parentEvent); |
|
| 943 |
result.id = getAndRemoveAttributeValue(attributes, ID); |
|
| 944 |
// result.ref = getAndRemoveAttributeValue(attributes, REF); |
|
| 945 |
checkNoAttributes(attributes, parentEvent); |
|
| 946 |
|
|
| 947 |
while (reader.hasNext()) {
|
|
| 948 |
XMLEvent next = readNoWhitespace(reader); |
|
| 949 |
if (isStartingElement(next, FOOTNOTE_STRING)) {
|
|
| 950 |
String string = handleFootnoteString(state, reader, next); |
|
| 951 |
result.string = string; |
|
| 952 |
} else if (isMyEndingElement(next, parentEvent)) {
|
|
| 953 |
return result; |
|
| 954 |
} else {
|
|
| 955 |
fireUnexpectedEvent(next, 0); |
|
| 956 |
} |
|
| 957 |
} |
|
| 958 |
return result; |
|
| 959 |
} |
|
| 960 |
|
|
| 961 |
private FootnoteDataHolder handleFootnoteRef(MarkupImportState state, |
|
| 962 |
XMLEventReader reader, XMLEvent parentEvent) |
|
| 963 |
throws XMLStreamException {
|
|
| 964 |
FootnoteDataHolder result = new FootnoteDataHolder(); |
|
| 965 |
Map<String, Attribute> attributes = getAttributes(parentEvent); |
|
| 966 |
result.ref = getAndRemoveAttributeValue(attributes, REF); |
|
| 967 |
checkNoAttributes(attributes, parentEvent); |
|
| 968 |
|
|
| 969 |
// text is not handled, needed only for debugging purposes |
|
| 970 |
String text = ""; |
|
| 971 |
while (reader.hasNext()) {
|
|
| 972 |
XMLEvent next = readNoWhitespace(reader); |
|
| 973 |
// if (isStartingElement(next, FOOTNOTE_STRING)){
|
|
| 974 |
// String string = handleFootnoteString(state, reader, next); |
|
| 975 |
// result.string = string; |
|
| 976 |
// }else |
|
| 977 |
if (isMyEndingElement(next, parentEvent)) {
|
|
| 978 |
return result; |
|
| 979 |
} else if (next.isCharacters()) {
|
|
| 980 |
text += next.asCharacters().getData(); |
|
| 981 |
|
|
| 982 |
} else {
|
|
| 983 |
fireUnexpectedEvent(next, 0); |
|
| 984 |
} |
|
| 985 |
} |
|
| 986 |
return result; |
|
| 987 |
} |
|
| 988 |
|
|
| 989 |
|
|
| 990 |
|
|
| 991 |
private String handleFootnoteString(MarkupImportState state, XMLEventReader reader, XMLEvent parentEvent) throws XMLStreamException {
|
|
| 992 |
boolean isTextMode = true; |
|
| 993 |
String text = ""; |
|
| 994 |
while (reader.hasNext()) {
|
|
| 995 |
XMLEvent next = readNoWhitespace(reader); |
|
| 996 |
if (isMyEndingElement(next, parentEvent)) {
|
|
| 997 |
return text; |
|
| 998 |
} else if (next.isEndElement()) {
|
|
| 999 |
if (isEndingElement(next, FULL_NAME)) {
|
|
| 1000 |
popUnimplemented(next.asEndElement()); |
|
| 1001 |
} else if (isEndingElement(next, BR)) {
|
|
| 1002 |
isTextMode = true; |
|
| 1003 |
} else if (isHtml(next)) {
|
|
| 1004 |
text += getXmlTag(next); |
|
| 1005 |
} else {
|
|
| 1006 |
handleUnexpectedEndElement(next.asEndElement()); |
|
| 1007 |
} |
|
| 1008 |
} else if (next.isStartElement()) {
|
|
| 1009 |
if (isStartingElement(next, FULL_NAME)) {
|
|
| 1010 |
handleNotYetImplementedElement(next); |
|
| 1011 |
} else if (isStartingElement(next, GATHERING)) {
|
|
| 1012 |
text += specimenImport.handleInLineGathering(state, reader, next); |
|
| 1013 |
} else if (isStartingElement(next, REFERENCES)) {
|
|
| 1014 |
text += " " + handleInLineReferences(state, reader, next)+ " "; |
|
| 1015 |
} else if (isStartingElement(next, BR)) {
|
|
| 1016 |
text += "<br/>"; |
|
| 1017 |
isTextMode = false; |
|
| 1018 |
} else if (isStartingElement(next, NOMENCLATURE)) {
|
|
| 1019 |
handleNotYetImplementedElement(next); |
|
| 1020 |
} else if (isHtml(next)) {
|
|
| 1021 |
text += getXmlTag(next); |
|
| 1022 |
} else {
|
|
| 1023 |
handleUnexpectedStartElement(next.asStartElement()); |
|
| 1024 |
} |
|
| 1025 |
} else if (next.isCharacters()) {
|
|
| 1026 |
if (!isTextMode) {
|
|
| 1027 |
String message = "footnoteString is not in text mode"; |
|
| 1028 |
fireWarningEvent(message, next, 6); |
|
| 1029 |
} else {
|
|
| 1030 |
text += next.asCharacters().getData().trim(); |
|
| 1031 |
// getCData(state, reader, next); does not work as we have inner tags like <references> |
|
| 1032 |
} |
|
| 1033 |
} else {
|
|
| 1034 |
handleUnexpectedEndElement(next.asEndElement()); |
|
| 1035 |
} |
|
| 1036 |
} |
|
| 1037 |
throw new IllegalStateException("<footnoteString> has no closing tag");
|
|
| 1038 |
|
|
| 1039 |
} |
|
| 1040 |
|
|
| 1041 |
private String handleInLineReferences(MarkupImportState state,XMLEventReader reader, XMLEvent parentEvent) throws XMLStreamException {
|
|
| 1042 |
checkNoAttributes(parentEvent); |
|
| 1043 |
|
|
| 1044 |
boolean hasReference = false; |
|
| 1045 |
String text = ""; |
|
| 1046 |
while (reader.hasNext()) {
|
|
| 1047 |
XMLEvent next = readNoWhitespace(reader); |
|
| 1048 |
if (isMyEndingElement(next, parentEvent)) {
|
|
| 1049 |
checkMandatoryElement(hasReference, parentEvent.asStartElement(), REFERENCE); |
|
| 1050 |
return text; |
|
| 1051 |
} else if (isStartingElement(next, REFERENCE)) {
|
|
| 1052 |
text += handleInLineReference(state, reader, next); |
|
| 1053 |
hasReference = true; |
|
| 1054 |
} else {
|
|
| 1055 |
handleUnexpectedElement(next); |
|
| 1056 |
} |
|
| 1057 |
} |
|
| 1058 |
throw new IllegalStateException("<References> has no closing tag");
|
|
| 1059 |
} |
|
| 1060 |
|
|
| 1061 |
private String handleInLineReference(MarkupImportState state,XMLEventReader reader, XMLEvent parentEvent)throws XMLStreamException {
|
|
| 1062 |
Reference<?> reference = nomenclatureImport.handleReference(state, reader, parentEvent); |
|
| 1063 |
String result = "<cdm:ref uuid='%s'>%s</ref>"; |
|
| 1064 |
result = String.format(result, reference.getUuid(), reference.getTitleCache()); |
|
| 1065 |
save(reference, state); |
|
| 1066 |
return result; |
|
| 1067 |
} |
|
| 1068 |
|
|
| 1069 |
private void handleFeature(MarkupImportState state, XMLEventReader reader, XMLEvent parentEvent) throws XMLStreamException {
|
|
| 1070 |
Map<String, Attribute> attrs = getAttributes(parentEvent); |
|
| 1071 |
Boolean isFreetext = getAndRemoveBooleanAttributeValue(parentEvent, attrs, IS_FREETEXT, false); |
|
| 1072 |
String classValue =getAndRemoveRequiredAttributeValue(parentEvent, attrs, CLASS); |
|
| 1073 |
checkNoAttributes(attrs, parentEvent); |
|
| 1074 |
|
|
| 1075 |
|
|
| 1076 |
Feature feature = makeFeature(classValue, state, parentEvent, null); |
|
| 1077 |
Taxon taxon = state.getCurrentTaxon(); |
|
| 1078 |
TaxonDescription taxonDescription = getTaxonDescription(taxon, state.getConfig().getSourceReference(), NO_IMAGE_GALLERY, CREATE_NEW); |
|
| 1079 |
// TextData figureHolderTextData = null; //for use with one TextData for |
|
| 1080 |
// all figure only |
|
| 1081 |
|
|
| 1082 |
boolean isDescription = feature.equals(Feature.DESCRIPTION()); |
|
| 1083 |
DescriptionElementBase lastDescriptionElement = null; |
|
| 1084 |
|
|
| 1085 |
while (reader.hasNext()) {
|
|
| 1086 |
XMLEvent next = readNoWhitespace(reader); |
|
| 1087 |
if (isMyEndingElement(next, parentEvent)) {
|
|
| 1088 |
state.putFeatureToGeneralSorterList(feature); |
|
| 1089 |
return; |
|
| 1090 |
} else if (isEndingElement(next, DISTRIBUTION_LIST) || isEndingElement(next, HABITAT_LIST)) {
|
|
| 1091 |
// only handle list elements |
|
| 1092 |
} else if (isStartingElement(next, HEADING)) {
|
|
| 1093 |
makeFeatureHeading(state, reader, classValue, feature, next); |
|
| 1094 |
} else if (isStartingElement(next, WRITER)) {
|
|
| 1095 |
makeFeatureWriter(state, reader, feature, taxon, next); |
|
| 1096 |
// } else if (isStartingElement(next, DISTRIBUTION_LOCALITY)) {
|
|
| 1097 |
// if (!feature.equals(Feature.DISTRIBUTION())) {
|
|
| 1098 |
// String message = "Distribution locality only allowed for feature of type 'distribution'"; |
|
| 1099 |
// fireWarningEvent(message, next, 4); |
|
| 1100 |
// } |
|
| 1101 |
// handleDistributionLocality(state, reader, next); |
|
| 1102 |
} else if (isStartingElement(next, DISTRIBUTION_LIST) || isStartingElement(next, HABITAT_LIST)) {
|
|
| 1103 |
// only handle single list elements |
|
| 1104 |
} else if (isStartingElement(next, HABITAT)) {
|
|
| 1105 |
if (!(feature.equals(Feature.HABITAT()) |
|
| 1106 |
|| feature.equals(Feature.HABITAT_ECOLOGY()) |
|
| 1107 |
|| feature.equals(Feature.ECOLOGY()))) {
|
|
| 1108 |
String message = "Habitat only allowed for feature of type 'habitat','habitat ecology' or 'ecology'"; |
|
| 1109 |
fireWarningEvent(message, next, 4); |
|
| 1110 |
} |
|
| 1111 |
handleHabitat(state, reader, next); |
|
| 1112 |
} else if (isStartingElement(next, CHAR)) {
|
|
| 1113 |
List<TextData> textDataList = handleChar(state, reader, next, null); |
|
| 1114 |
for (TextData textData : textDataList){
|
|
| 1115 |
taxonDescription.addElement(textData); |
|
| 1116 |
} |
|
| 1117 |
} else if (isStartingElement(next, STRING)) {
|
|
| 1118 |
lastDescriptionElement = makeFeatureString(state, reader,feature, taxonDescription, lastDescriptionElement,next, isFreetext); |
|
| 1119 |
} else if (isStartingElement(next, FIGURE_REF)) {
|
|
| 1120 |
lastDescriptionElement = makeFeatureFigureRef(state, reader, taxonDescription, isDescription, lastDescriptionElement, next); |
|
| 1121 |
} else if (isStartingElement(next, REFERENCES)) {
|
|
| 1122 |
// TODO details/microcitation ?? |
|
| 1123 |
|
|
| 1124 |
List<Reference<?>> refs = handleReferences(state, reader, next); |
|
| 1125 |
if (!refs.isEmpty()) {
|
|
| 1126 |
// TODO |
|
| 1127 |
Reference<?> descriptionRef = state.getConfig().getSourceReference(); |
|
| 1128 |
TaxonDescription description = getTaxonDescription(taxon, descriptionRef, false, true); |
|
| 1129 |
TextData featurePlaceholder = docImport.getFeaturePlaceholder(state, description, feature, true); |
|
| 1130 |
for (Reference<?> citation : refs) {
|
|
| 1131 |
featurePlaceholder.addSource(OriginalSourceType.PrimaryTaxonomicSource, |
|
| 1132 |
null, null, citation, null); |
|
| 1133 |
} |
|
| 1134 |
} else {
|
|
| 1135 |
String message = "No reference found in references"; |
|
| 1136 |
fireWarningEvent(message, next, 6); |
|
| 1137 |
} |
|
| 1138 |
} else if (isStartingElement(next, NUM)) {
|
|
| 1139 |
//TODO |
|
| 1140 |
handleNotYetImplementedElement(next); |
|
| 1141 |
} else if (isEndingElement(next, NUM)) {
|
|
| 1142 |
//TODO |
|
| 1143 |
popUnimplemented(next.asEndElement()); |
|
| 1144 |
} else {
|
|
| 1145 |
handleUnexpectedElement(next); |
|
| 1146 |
} |
|
| 1147 |
} |
|
| 1148 |
throw new IllegalStateException("<Feature> has no closing tag");
|
|
| 1149 |
} |
|
| 1150 |
|
|
| 1151 |
/** |
|
| 1152 |
* @param state |
|
| 1153 |
* @param reader |
|
| 1154 |
* @param taxonDescription |
|
| 1155 |
* @param isDescription |
|
| 1156 |
* @param lastDescriptionElement |
|
| 1157 |
* @param next |
|
| 1158 |
* @return |
|
| 1159 |
* @throws XMLStreamException |
|
| 1160 |
*/ |
|
| 1161 |
private DescriptionElementBase makeFeatureFigureRef(MarkupImportState state, XMLEventReader reader,TaxonDescription taxonDescription, |
|
| 1162 |
boolean isDescription, DescriptionElementBase lastDescriptionElement, XMLEvent next) throws XMLStreamException {
|
|
| 1163 |
FigureDataHolder figureHolder = handleFigureRef(state, reader, next); |
|
| 1164 |
Feature figureFeature = getFeature(state, MarkupTransformer.uuidFigures, "Figures", "Figures", "Fig.",null); |
|
| 1165 |
if (isDescription) {
|
|
| 1166 |
TextData figureHolderTextData = null; |
|
| 1167 |
// if (figureHolderTextData == null){
|
|
| 1168 |
figureHolderTextData = TextData.NewInstance(figureFeature); |
|
| 1169 |
if (StringUtils.isNotBlank(figureHolder.num)) {
|
|
| 1170 |
String annotationText = "<num>" + figureHolder.num.trim() + "</num>"; |
|
| 1171 |
Annotation annotation = Annotation.NewInstance(annotationText, AnnotationType.TECHNICAL(), getDefaultLanguage(state)); |
|
| 1172 |
figureHolderTextData.addAnnotation(annotation); |
|
| 1173 |
} |
|
| 1174 |
if (StringUtils.isNotBlank(figureHolder.figurePart)) {
|
|
| 1175 |
String annotationText = "<figurePart>"+ figureHolder.figurePart.trim() + "</figurePart>"; |
|
| 1176 |
Annotation annotation = Annotation.NewInstance(annotationText,AnnotationType.EDITORIAL(), getDefaultLanguage(state)); |
|
| 1177 |
figureHolderTextData.addAnnotation(annotation); |
|
| 1178 |
} |
|
| 1179 |
// if (StringUtils.isNotBlank(figureText)){
|
|
| 1180 |
// figureHolderTextData.putText(language, figureText); |
|
| 1181 |
// } |
|
| 1182 |
taxonDescription.addElement(figureHolderTextData); |
|
| 1183 |
// } |
|
| 1184 |
registerFigureDemand(state, next, figureHolderTextData, figureHolder.ref); |
|
| 1185 |
} else {
|
|
| 1186 |
if (lastDescriptionElement == null) {
|
|
| 1187 |
String message = "No description element created yet that can be referred by figure. Create new TextData instead"; |
|
| 1188 |
fireWarningEvent(message, next, 4); |
|
| 1189 |
lastDescriptionElement = TextData.NewInstance(figureFeature); |
|
| 1190 |
taxonDescription.addElement(lastDescriptionElement); |
|
| 1191 |
} |
|
| 1192 |
registerFigureDemand(state, next, lastDescriptionElement, figureHolder.ref); |
|
| 1193 |
} |
|
| 1194 |
return lastDescriptionElement; |
|
| 1195 |
} |
|
| 1196 |
|
|
| 1197 |
/** |
|
| 1198 |
* @param state |
|
| 1199 |
* @param reader |
|
| 1200 |
* @param feature |
|
| 1201 |
* @param taxonDescription |
|
| 1202 |
* @param lastDescriptionElement |
|
| 1203 |
* @param distributionList |
|
| 1204 |
* @param next |
|
| 1205 |
* @return |
|
| 1206 |
* @throws XMLStreamException |
|
| 1207 |
*/ |
|
| 1208 |
private DescriptionElementBase makeFeatureString(MarkupImportState state,XMLEventReader reader, Feature feature, |
|
| 1209 |
TaxonDescription taxonDescription, DescriptionElementBase lastDescriptionElement, XMLEvent next, Boolean isFreetext) throws XMLStreamException {
|
|
| 1210 |
|
|
| 1211 |
//for specimen only |
|
| 1212 |
if (feature.equals(Feature.SPECIMEN()) || feature.equals(Feature.MATERIALS_EXAMINED())){
|
|
| 1213 |
|
|
| 1214 |
List<DescriptionElementBase> specimens = specimenImport.handleMaterialsExamined(state, reader, next, feature); |
|
| 1215 |
for (DescriptionElementBase specimen : specimens){
|
|
| 1216 |
taxonDescription.addElement(specimen); |
|
| 1217 |
lastDescriptionElement = specimen; |
|
| 1218 |
} |
|
| 1219 |
state.setCurrentCollector(null); |
|
| 1220 |
|
|
| 1221 |
return lastDescriptionElement; |
|
| 1222 |
}else{
|
|
| 1223 |
|
|
| 1224 |
//others |
|
| 1225 |
Map<String, String> subheadingMap = handleString(state, reader, next, feature); |
|
| 1226 |
for (String subheading : subheadingMap.keySet()) {
|
|
| 1227 |
Feature subheadingFeature = feature; |
|
| 1228 |
if (StringUtils.isNotBlank(subheading) && subheadingMap.size() > 1) {
|
|
| 1229 |
subheadingFeature = makeFeature(subheading, state, next, null); |
|
| 1230 |
} |
|
| 1231 |
if (feature.equals(Feature.COMMON_NAME()) && (isFreetext == null || !isFreetext)){
|
|
| 1232 |
List<DescriptionElementBase> commonNames = makeVernacular(state, subheading, subheadingMap.get(subheading)); |
|
| 1233 |
for (DescriptionElementBase commonName : commonNames){
|
|
| 1234 |
taxonDescription.addElement(commonName); |
|
| 1235 |
lastDescriptionElement = commonName; |
|
| 1236 |
} |
|
| 1237 |
}else {
|
|
| 1238 |
TextData textData = TextData.NewInstance(subheadingFeature); |
|
| 1239 |
textData.putText(getDefaultLanguage(state), subheadingMap.get(subheading)); |
|
| 1240 |
taxonDescription.addElement(textData); |
|
| 1241 |
lastDescriptionElement = textData; |
|
| 1242 |
// TODO how to handle figures when these data are split in |
|
| 1243 |
// subheadings |
|
| 1244 |
} |
|
| 1245 |
} |
|
| 1246 |
return lastDescriptionElement; |
|
| 1247 |
} |
|
| 1248 |
} |
|
| 1249 |
|
|
| 1250 |
private List<DescriptionElementBase> makeVernacular(MarkupImportState state, String subheading, String commonNameString) throws XMLStreamException {
|
|
| 1251 |
List<DescriptionElementBase> result = new ArrayList<DescriptionElementBase>(); |
|
| 1252 |
String[] splits = commonNameString.split(",");
|
|
| 1253 |
for (String split : splits){
|
|
| 1254 |
split = split.trim(); |
|
| 1255 |
if (! split.matches(".*\\(.*\\)\\.?")){
|
|
| 1256 |
fireWarningEvent("Common name string '"+split+"' does not match given pattern", state.getReader().peek(), 4);
|
|
| 1257 |
} |
|
| 1258 |
|
|
| 1259 |
String name = split.replaceAll("\\(.*\\)", "").replace(".", "").trim();
|
|
| 1260 |
String languageStr = split.replaceFirst(".*\\(", "").replaceAll("\\)\\.?", "").trim();
|
|
| 1261 |
|
|
| 1262 |
Language language = null; |
|
| 1263 |
if (StringUtils.isNotBlank(languageStr)){
|
|
| 1264 |
try {
|
|
| 1265 |
UUID langUuid = state.getTransformer().getLanguageUuid(languageStr); |
|
| 1266 |
TermVocabulary<?> voc = null; |
|
| 1267 |
language = getLanguage(state, langUuid, languageStr, languageStr, null, voc); |
|
| 1268 |
if (language == null){
|
|
| 1269 |
logger.warn("Language " + languageStr + " not recognized by transformer");
|
|
| 1270 |
} |
|
| 1271 |
} catch (UndefinedTransformerMethodException e) {
|
|
| 1272 |
throw new RuntimeException(e); |
|
| 1273 |
} |
|
| 1274 |
} |
|
| 1275 |
NamedArea area = null; |
|
| 1276 |
CommonTaxonName commonTaxonName = CommonTaxonName.NewInstance(name, language, area); |
|
| 1277 |
result.add(commonTaxonName); |
|
| 1278 |
} |
|
| 1279 |
|
|
| 1280 |
return result; |
|
| 1281 |
} |
|
| 1282 |
|
|
| 1283 |
/** |
|
| 1284 |
* @param state |
|
| 1285 |
* @param reader |
|
| 1286 |
* @param feature |
|
| 1287 |
* @param taxon |
|
| 1288 |
* @param next |
|
| 1289 |
* @throws XMLStreamException |
|
| 1290 |
*/ |
|
| 1291 |
private void makeFeatureWriter(MarkupImportState state,XMLEventReader reader, Feature feature, Taxon taxon, XMLEvent next) throws XMLStreamException {
|
|
| 1292 |
WriterDataHolder writer = handleWriter(state, reader, next); |
|
| 1293 |
if (isNotBlank(writer.writer)) {
|
|
| 1294 |
// TODO |
|
| 1295 |
Reference<?> ref = state.getConfig().getSourceReference(); |
|
| 1296 |
TaxonDescription description = getTaxonDescription(taxon, ref, |
|
| 1297 |
false, true); |
|
| 1298 |
TextData featurePlaceholder = docImport.getFeaturePlaceholder(state, |
|
| 1299 |
description, feature, true); |
|
| 1300 |
featurePlaceholder.addAnnotation(writer.annotation); |
|
| 1301 |
registerFootnotes(state, featurePlaceholder, writer.footnotes); |
|
| 1302 |
} else {
|
|
| 1303 |
String message = "Writer element is empty"; |
|
| 1304 |
fireWarningEvent(message, next, 4); |
|
| 1305 |
} |
|
| 1306 |
} |
|
| 1307 |
|
|
| 1308 |
/** |
|
| 1309 |
* @param state |
|
| 1310 |
* @param reader |
|
| 1311 |
* @param classValue |
|
| 1312 |
* @param feature |
|
| 1313 |
* @param next |
|
| 1314 |
* @throws XMLStreamException |
|
| 1315 |
*/ |
|
| 1316 |
private void makeFeatureHeading(MarkupImportState state, XMLEventReader reader, String classValue, Feature feature, XMLEvent next) throws XMLStreamException {
|
|
| 1317 |
String heading = handleHeading(state, reader, next); |
|
| 1318 |
if (StringUtils.isNotBlank(heading)) {
|
|
| 1319 |
if (!heading.equalsIgnoreCase(classValue)) {
|
|
| 1320 |
try {
|
|
| 1321 |
if (!feature.equals(state.getTransformer().getFeatureByKey( |
|
| 1322 |
heading))) {
|
|
| 1323 |
UUID headerFeatureUuid = state.getTransformer() |
|
| 1324 |
.getFeatureUuid(heading); |
|
| 1325 |
if (!feature.getUuid().equals(headerFeatureUuid)) {
|
|
| 1326 |
String message = "Feature heading '%s' differs from feature class '%s' and can not be transformed to feature"; |
|
| 1327 |
message = String.format(message, heading, |
|
| 1328 |
classValue); |
|
| 1329 |
fireWarningEvent(message, next, 1); |
|
| 1330 |
} |
|
| 1331 |
} |
|
| 1332 |
} catch (UndefinedTransformerMethodException e) {
|
|
| 1333 |
throw new RuntimeException(e); |
|
| 1334 |
} |
|
| 1335 |
} else {
|
|
| 1336 |
// do nothing |
|
| 1337 |
} |
|
| 1338 |
} |
|
| 1339 |
} |
|
| 1340 |
|
|
| 1341 |
private List<Reference<?>> handleReferences(MarkupImportState state, XMLEventReader reader, XMLEvent parentEvent) throws XMLStreamException {
|
|
| 1342 |
// attributes |
|
| 1343 |
Map<String, Attribute> attributes = getAttributes(parentEvent); |
|
| 1344 |
String bibliography = getAndRemoveAttributeValue(attributes, |
|
| 1345 |
BIBLIOGRAPHY); |
|
| 1346 |
String serialsAbbreviations = getAndRemoveAttributeValue(attributes, |
|
| 1347 |
SERIALS_ABBREVIATIONS); |
|
| 1348 |
if (isNotBlank(bibliography) || isNotBlank(serialsAbbreviations)) {
|
|
| 1349 |
String message = "Attributes not yet implemented for <references>"; |
|
| 1350 |
fireWarningEvent(message, parentEvent, 4); |
|
| 1351 |
} |
|
| 1352 |
|
|
| 1353 |
List<Reference<?>> result = new ArrayList<Reference<?>>(); |
|
| 1354 |
|
|
| 1355 |
// elements |
|
| 1356 |
while (reader.hasNext()) {
|
|
| 1357 |
XMLEvent next = readNoWhitespace(reader); |
|
| 1358 |
if (next.isEndElement()) {
|
|
| 1359 |
if (isMyEndingElement(next, parentEvent)) {
|
|
| 1360 |
return result; |
|
| 1361 |
} else {
|
|
| 1362 |
if (isEndingElement(next, HEADING)) {
|
|
| 1363 |
// NOT YET IMPLEMENTED |
|
| 1364 |
popUnimplemented(next.asEndElement()); |
|
| 1365 |
} else if (isEndingElement(next, WRITER)) {
|
|
| 1366 |
// NOT YET IMPLEMENTED |
|
| 1367 |
popUnimplemented(next.asEndElement()); |
|
| 1368 |
} else if (isEndingElement(next, FOOTNOTE)) {
|
|
| 1369 |
// NOT YET IMPLEMENTED |
|
| 1370 |
popUnimplemented(next.asEndElement()); |
|
| 1371 |
} else if (isEndingElement(next, STRING)) {
|
|
| 1372 |
// NOT YET IMPLEMENTED |
|
| 1373 |
popUnimplemented(next.asEndElement()); |
|
| 1374 |
} else if (isEndingElement(next, REF_NUM)) {
|
|
| 1375 |
// NOT YET IMPLEMENTED |
|
| 1376 |
popUnimplemented(next.asEndElement()); |
|
| 1377 |
} else {
|
|
| 1378 |
handleUnexpectedEndElement(next.asEndElement()); |
|
| 1379 |
} |
|
| 1380 |
} |
|
| 1381 |
} else if (next.isStartElement()) {
|
|
| 1382 |
if (isStartingElement(next, HEADING)) {
|
|
| 1383 |
handleNotYetImplementedElement(next); |
|
| 1384 |
} else if (isStartingElement(next, SUB_HEADING)) {
|
|
| 1385 |
String subheading = getCData(state, reader, next).trim(); |
|
| 1386 |
String excludePattern = "(i?)(References?|Literature):?"; |
|
| 1387 |
if (!subheading.matches(excludePattern)) {
|
|
| 1388 |
fireNotYetImplementedElement(next.getLocation(), next.asStartElement().getName(), 0); |
|
| 1389 |
} |
|
| 1390 |
} else if (isStartingElement(next, WRITER)) {
|
|
| 1391 |
handleNotYetImplementedElement(next); |
|
| 1392 |
} else if (isStartingElement(next, FOOTNOTE)) {
|
|
| 1393 |
handleNotYetImplementedElement(next); |
|
| 1394 |
} else if (isStartingElement(next, STRING)) {
|
|
| 1395 |
handleNotYetImplementedElement(next); |
|
| 1396 |
} else if (isStartingElement(next, REF_NUM)) {
|
|
| 1397 |
handleNotYetImplementedElement(next); |
|
| 1398 |
} else if (isStartingElement(next, REFERENCE)) {
|
|
| 1399 |
Reference<?> ref = nomenclatureImport.handleReference(state, reader, next); |
|
| 1400 |
result.add(ref); |
|
| 1401 |
} else {
|
|
| 1402 |
handleUnexpectedStartElement(next); |
|
| 1403 |
} |
|
| 1404 |
} else {
|
|
| 1405 |
handleUnexpectedElement(next); |
|
| 1406 |
} |
|
| 1407 |
} |
|
| 1408 |
throw new IllegalStateException("<References> has no closing tag");
|
|
| 1409 |
} |
|
| 1410 |
|
|
| 1411 |
private void handleHabitat(MarkupImportState state, XMLEventReader reader, XMLEvent parentEvent) throws XMLStreamException {
|
|
| 1412 |
checkNoAttributes(parentEvent); |
|
| 1413 |
Taxon taxon = state.getCurrentTaxon(); |
|
| 1414 |
// TODO which ref to take? |
|
| 1415 |
Reference<?> ref = state.getConfig().getSourceReference(); |
|
| 1416 |
|
|
| 1417 |
String text = ""; |
|
| 1418 |
while (reader.hasNext()) {
|
|
| 1419 |
XMLEvent next = readNoWhitespace(reader); |
|
| 1420 |
if (isMyEndingElement(next, parentEvent)) {
|
|
| 1421 |
TaxonDescription description = getTaxonDescription(taxon, ref, |
|
| 1422 |
false, true); |
|
| 1423 |
UUID uuidExtractedHabitat = MarkupTransformer.uuidExtractedHabitat; |
|
| 1424 |
Feature feature = getFeature( |
|
| 1425 |
state, |
|
| 1426 |
uuidExtractedHabitat, |
|
| 1427 |
"Extracted Habitat", |
|
| 1428 |
"An structured habitat that was extracted from a habitat text", |
|
| 1429 |
"extr. habit.", null); |
|
| 1430 |
TextData habitat = TextData.NewInstance(feature); |
|
| 1431 |
habitat.putText(getDefaultLanguage(state), text); |
|
| 1432 |
description.addElement(habitat); |
|
| 1433 |
|
|
| 1434 |
return; |
|
| 1435 |
} else if (next.isStartElement()) {
|
|
| 1436 |
if (isStartingElement(next, ALTITUDE)) {
|
|
| 1437 |
text = text.trim() + getTaggedCData(state, reader, next); |
|
| 1438 |
} else if (isStartingElement(next, LIFE_CYCLE_PERIODS)) {
|
|
| 1439 |
handleNotYetImplementedElement(next); |
|
| 1440 |
} else {
|
|
| 1441 |
handleUnexpectedStartElement(next.asStartElement()); |
|
| 1442 |
} |
|
| 1443 |
} else if (next.isCharacters()) {
|
|
| 1444 |
text += next.asCharacters().getData(); |
|
| 1445 |
} else {
|
|
| 1446 |
handleUnexpectedElement(next); |
|
| 1447 |
} |
|
| 1448 |
} |
|
| 1449 |
throw new IllegalStateException("<Habitat> has no closing tag");
|
|
| 1450 |
} |
|
| 1451 |
|
|
| 1452 |
private String getTaggedCData(MarkupImportState state, XMLEventReader reader, XMLEvent parentEvent) throws XMLStreamException {
|
|
| 1453 |
checkNoAttributes(parentEvent); |
|
| 1454 |
|
|
| 1455 |
String text = getXmlTag(parentEvent); |
|
| 1456 |
while (reader.hasNext()) {
|
|
| 1457 |
XMLEvent next = readNoWhitespace(reader); |
|
| 1458 |
if (isMyEndingElement(next, parentEvent)) {
|
|
| 1459 |
text += getXmlTag(next); |
|
| 1460 |
return text; |
|
| 1461 |
} else if (next.isStartElement()) {
|
|
| 1462 |
text += getTaggedCData(state, reader, next); |
|
| 1463 |
} else if (next.isEndElement()) {
|
|
| 1464 |
text += getTaggedCData(state, reader, next); |
|
| 1465 |
} else if (next.isCharacters()) {
|
|
| 1466 |
text += next.asCharacters().getData(); |
|
| 1467 |
} else {
|
|
| 1468 |
handleUnexpectedEndElement(next.asEndElement()); |
|
| 1469 |
} |
|
| 1470 |
} |
|
| 1471 |
throw new IllegalStateException("Some tag has no closing tag");
|
|
| 1472 |
} |
|
| 1473 |
|
|
| 1474 |
private String handleDistributionLocality(MarkupImportState state,XMLEventReader reader, XMLEvent parentEvent)throws XMLStreamException {
|
|
| 1475 |
Map<String, Attribute> attributes = getAttributes(parentEvent); |
|
| 1476 |
String classValue = getAndRemoveRequiredAttributeValue(parentEvent, attributes, CLASS); |
|
| 1477 |
String statusValue =getAndRemoveAttributeValue(attributes, STATUS); |
|
| 1478 |
String frequencyValue =getAndRemoveAttributeValue(attributes, FREQUENCY); |
|
| 1479 |
|
|
| 1480 |
|
|
| 1481 |
Taxon taxon = state.getCurrentTaxon(); |
|
| 1482 |
// TODO which ref to take? |
|
| 1483 |
Reference<?> ref = state.getConfig().getSourceReference(); |
|
| 1484 |
|
|
| 1485 |
String text = ""; |
|
| 1486 |
while (reader.hasNext()) {
|
|
| 1487 |
XMLEvent next = readNoWhitespace(reader); |
|
| 1488 |
if (isMyEndingElement(next, parentEvent)) {
|
|
| 1489 |
if (StringUtils.isNotBlank(text)) {
|
|
| 1490 |
String label = CdmUtils.removeTrailingDot(normalize(text)); |
|
| 1491 |
TaxonDescription description = getTaxonDescription(taxon, ref, false, true); |
|
| 1492 |
NamedAreaLevel level = makeNamedAreaLevel(state,classValue, next); |
|
| 1493 |
|
|
| 1494 |
//status |
|
| 1495 |
PresenceAbsenceTermBase<?> status = null; |
|
| 1496 |
if (isNotBlank(statusValue)){
|
|
| 1497 |
try {
|
|
| 1498 |
status = state.getTransformer().getPresenceTermByKey(statusValue); |
|
| 1499 |
if (status == null){
|
|
| 1500 |
//TODO |
|
| 1501 |
String message = "The presence/absence status '%s' could not be transformed to an CDM status"; |
|
| 1502 |
fireWarningEvent(String.format(message, statusValue), next, 4); |
|
| 1503 |
} |
|
| 1504 |
} catch (UndefinedTransformerMethodException e) {
|
|
| 1505 |
throw new RuntimeException(e); |
|
| 1506 |
} |
|
| 1507 |
}else{
|
|
| 1508 |
status = PresenceTerm.PRESENT(); |
|
| 1509 |
} |
|
| 1510 |
//frequency |
|
| 1511 |
if (isNotBlank(frequencyValue)){
|
|
| 1512 |
String message = "The frequency attribute is currently not yet available in CDM"; |
|
| 1513 |
fireWarningEvent(message, parentEvent, 6); |
|
| 1514 |
} |
|
| 1515 |
|
|
| 1516 |
NamedArea higherArea = null; |
|
| 1517 |
List<NamedArea> areas = new ArrayList<NamedArea>(); |
|
| 1518 |
|
|
| 1519 |
String patSingleArea = "([^,\\(]{3,})";
|
|
| 1520 |
String patSeparator = "(,|\\sand\\s)"; |
|
| 1521 |
String hierarchiePattern = String.format("%s\\((%s(%s%s)*)\\)",patSingleArea, patSingleArea, patSeparator, patSingleArea);
|
|
| 1522 |
Pattern patHierarchie = Pattern.compile(hierarchiePattern, Pattern.CASE_INSENSITIVE); |
|
| 1523 |
Matcher matcher = patHierarchie.matcher(label); |
|
| 1524 |
if (matcher.matches()){
|
|
| 1525 |
String higherAreaStr = matcher.group(1).trim(); |
|
| 1526 |
higherArea = makeArea(state, higherAreaStr, level); |
|
| 1527 |
String[] innerAreas = matcher.group(2).split(patSeparator); |
|
| 1528 |
for (String innerArea : innerAreas){
|
|
| 1529 |
if (isNotBlank(innerArea)){
|
|
| 1530 |
NamedArea singleArea = makeArea(state, innerArea.trim(), level); |
|
| 1531 |
areas.add(singleArea); |
|
| 1532 |
NamedArea partOf = singleArea.getPartOf(); |
|
| 1533 |
// if (partOf == null){
|
|
| 1534 |
// singleArea.setPartOf(higherArea); |
|
| 1535 |
// } |
|
| 1536 |
} |
|
| 1537 |
} |
|
| 1538 |
}else{
|
|
| 1539 |
NamedArea singleArea = makeArea(state, label, level); |
|
| 1540 |
areas.add(singleArea); |
|
| 1541 |
} |
|
| 1542 |
|
|
| 1543 |
for (NamedArea area : areas){
|
|
| 1544 |
//create distribution |
|
| 1545 |
Distribution distribution = Distribution.NewInstance(area,status); |
|
| 1546 |
description.addElement(distribution); |
|
| 1547 |
} |
|
| 1548 |
} else {
|
|
| 1549 |
String message = "Empty distribution locality"; |
|
| 1550 |
fireWarningEvent(message, next, 4); |
|
| 1551 |
} |
|
| 1552 |
return text; |
|
| 1553 |
} else if (isStartingElement(next, COORDINATES)) {
|
|
| 1554 |
//TODO |
|
| 1555 |
handleNotYetImplementedElement(next); |
|
| 1556 |
} else if (isEndingElement(next, COORDINATES)) {
|
|
| 1557 |
//TODO |
|
| 1558 |
popUnimplemented(next.asEndElement()); |
|
| 1559 |
} else if (next.isCharacters()) {
|
|
| 1560 |
text += next.asCharacters().getData(); |
|
| 1561 |
} else {
|
|
| 1562 |
handleUnexpectedElement(next); |
|
| 1563 |
} |
|
| 1564 |
} |
|
| 1565 |
throw new IllegalStateException("<DistributionLocality> has no closing tag");
|
|
| 1566 |
} |
|
| 1567 |
|
|
| 1568 |
private String handleHeading(MarkupImportState state,XMLEventReader reader, XMLEvent parentEvent)throws XMLStreamException {
|
|
| 1569 |
checkNoAttributes(parentEvent); |
|
| 1570 |
|
|
| 1571 |
String text = ""; |
|
| 1572 |
while (reader.hasNext()) {
|
|
| 1573 |
XMLEvent next = readNoWhitespace(reader); |
|
| 1574 |
if (isMyEndingElement(next, parentEvent)) {
|
|
| 1575 |
return text; |
|
| 1576 |
} else if (next.isStartElement()) {
|
|
| 1577 |
if (isStartingElement(next, FOOTNOTE)) {
|
|
| 1578 |
handleNotYetImplementedElement(next); |
|
| 1579 |
} else {
|
|
| 1580 |
handleUnexpectedStartElement(next.asStartElement()); |
|
| 1581 |
} |
|
| 1582 |
} else if (next.isCharacters()) {
|
|
| 1583 |
text += next.asCharacters().getData(); |
|
| 1584 |
} else {
|
|
| 1585 |
handleUnexpectedEndElement(next.asEndElement()); |
|
| 1586 |
} |
|
| 1587 |
} |
|
| 1588 |
throw new IllegalStateException("<String> has no closing tag");
|
|
| 1589 |
|
|
| 1590 |
} |
|
| 1591 |
|
|
| 1592 |
/** |
|
| 1593 |
* Handle string |
|
| 1594 |
* @param state |
|
| 1595 |
* @param reader |
|
| 1596 |
* @param parentEvent |
|
| 1597 |
* @param feature only needed for distributionLocalities |
|
| 1598 |
* @return |
|
| 1599 |
* @throws XMLStreamException |
|
| 1600 |
*/ |
|
| 1601 |
private Map<String, String> handleString(MarkupImportState state, XMLEventReader reader, XMLEvent parentEvent, Feature feature)throws XMLStreamException {
|
|
| 1602 |
// attributes |
|
| 1603 |
String classValue = getClassOnlyAttribute(parentEvent, false); |
|
| 1604 |
if (StringUtils.isNotBlank(classValue)) {
|
|
| 1605 |
String message = "class attribute for <string> not yet implemented"; |
|
| 1606 |
fireWarningEvent(message, parentEvent, 2); |
|
| 1607 |
} |
|
| 1608 |
|
|
| 1609 |
// subheadings |
|
| 1610 |
Map<String, String> subHeadingMap = new HashMap<String, String>(); |
|
| 1611 |
String currentSubheading = null; |
|
| 1612 |
|
|
| 1613 |
boolean isTextMode = true; |
|
| 1614 |
String text = ""; |
|
| 1615 |
while (reader.hasNext()) {
|
|
| 1616 |
XMLEvent next = readNoWhitespace(reader); |
|
| 1617 |
if (isMyEndingElement(next, parentEvent)) {
|
|
| 1618 |
putCurrentSubheading(subHeadingMap, currentSubheading, text); |
|
| 1619 |
return subHeadingMap; |
|
| 1620 |
} else if (isStartingElement(next, BR)) {
|
|
| 1621 |
text += "<br/>"; |
|
| 1622 |
isTextMode = false; |
|
| 1623 |
} else if (isEndingElement(next, BR)) {
|
|
| 1624 |
isTextMode = true; |
|
| 1625 |
} else if (isHtml(next)) {
|
|
| 1626 |
text += getXmlTag(next); |
|
| 1627 |
} else if (isStartingElement(next, SUB_HEADING)) {
|
|
| 1628 |
text = putCurrentSubheading(subHeadingMap,currentSubheading, text); |
|
| 1629 |
// TODO footnotes |
|
| 1630 |
currentSubheading = getCData(state, reader, next).trim(); |
|
| 1631 |
} else if (isStartingElement(next, DISTRIBUTION_LOCALITY)) {
|
|
| 1632 |
if (feature != null && !feature.equals(Feature.DISTRIBUTION())) {
|
|
| 1633 |
String message = "Distribution locality only allowed for feature of type 'distribution'"; |
|
| 1634 |
fireWarningEvent(message, next, 4); |
|
| 1635 |
} |
|
| 1636 |
text += handleDistributionLocality(state, reader, next); |
|
| 1637 |
} else if (next.isCharacters()) {
|
|
| 1638 |
if (! isTextMode) {
|
|
| 1639 |
String message = "String is not in text mode"; |
|
| 1640 |
fireWarningEvent(message, next, 6); |
|
| 1641 |
} else {
|
|
| 1642 |
text += next.asCharacters().getData(); |
|
| 1643 |
} |
|
| 1644 |
} else if (isStartingElement(next, HEADING)) {
|
|
| 1645 |
//TODO |
|
| 1646 |
handleNotYetImplementedElement(next); |
|
| 1647 |
} else if (isStartingElement(next, VERNACULAR_NAMES)) {
|
|
| 1648 |
//TODO |
|
| 1649 |
handleNotYetImplementedElement(next); |
|
| 1650 |
} else if (isEndingElement(next, HEADING)) {
|
|
| 1651 |
//TODO |
|
| 1652 |
popUnimplemented(next.asEndElement()); |
|
| 1653 |
} else if (isStartingElement(next, QUOTE)) {
|
|
| 1654 |
//TODO |
|
| 1655 |
handleNotYetImplementedElement(next); |
|
| 1656 |
} else if (isEndingElement(next, QUOTE)) {
|
|
| 1657 |
//TODO |
|
| 1658 |
popUnimplemented(next.asEndElement()); |
|
| 1659 |
} else if (isStartingElement(next, DEDICATION)) {
|
|
| 1660 |
//TODO |
|
| 1661 |
handleNotYetImplementedElement(next); |
|
| 1662 |
} else if (isEndingElement(next, DEDICATION)) {
|
|
| 1663 |
//TODO |
|
| 1664 |
popUnimplemented(next.asEndElement()); |
|
| 1665 |
} else if (isStartingElement(next, TAXONTYPE)) {
|
|
| 1666 |
//TODO |
|
| 1667 |
handleNotYetImplementedElement(next); |
|
| 1668 |
} else if (isEndingElement(next, TAXONTYPE)) {
|
|
| 1669 |
//TODO |
|
| 1670 |
popUnimplemented(next.asEndElement()); |
|
| 1671 |
} else if (isStartingElement(next, FULL_NAME)) {
|
|
| 1672 |
//TODO |
|
| 1673 |
handleNotYetImplementedElement(next); |
|
| 1674 |
} else if (isEndingElement(next, FULL_NAME)) {
|
|
| 1675 |
//TODO |
|
| 1676 |
popUnimplemented(next.asEndElement()); |
|
| 1677 |
}else if (isStartingElement(next, REFERENCES)) {
|
|
| 1678 |
//TODO |
|
| 1679 |
handleNotYetImplementedElement(next); |
|
| 1680 |
} else if (isEndingElement(next, REFERENCES)) {
|
|
| 1681 |
//TODO |
|
| 1682 |
popUnimplemented(next.asEndElement()); |
|
| 1683 |
} else if (isStartingElement(next, GATHERING)) {
|
|
| 1684 |
//TODO |
|
| 1685 |
handleNotYetImplementedElement(next); |
|
| 1686 |
} else if (isEndingElement(next, GATHERING)) {
|
|
| 1687 |
//TODO |
|
| 1688 |
popUnimplemented(next.asEndElement()); |
|
| 1689 |
} else if (isStartingElement(next, ANNOTATION)) {
|
|
| 1690 |
//TODO //TODO test handleSimpleAnnotation |
|
| 1691 |
handleNotYetImplementedElement(next); |
|
| 1692 |
} else if (isEndingElement(next, ANNOTATION)) {
|
|
| 1693 |
//TODO |
|
| 1694 |
popUnimplemented(next.asEndElement()); |
|
| 1695 |
} else if (isStartingElement(next, HABITAT)) {
|
|
| 1696 |
//TODO |
|
| 1697 |
handleNotYetImplementedElement(next); |
|
| 1698 |
} else if (isEndingElement(next, HABITAT)) {
|
|
| 1699 |
//TODO |
|
| 1700 |
popUnimplemented(next.asEndElement()); |
|
| 1701 |
} else if (isStartingElement(next, FIGURE_REF)) {
|
|
| 1702 |
//TODO |
|
| 1703 |
handleNotYetImplementedElement(next); |
|
| 1704 |
} else if (isEndingElement(next, FIGURE_REF)) {
|
|
| 1705 |
//TODO |
|
| 1706 |
popUnimplemented(next.asEndElement()); |
|
| 1707 |
} else if (isStartingElement(next, FIGURE)) {
|
|
| 1708 |
//TODO |
|
| 1709 |
handleNotYetImplementedElement(next); |
|
| 1710 |
} else if (isEndingElement(next, FIGURE)) {
|
|
| 1711 |
//TODO |
|
| 1712 |
popUnimplemented(next.asEndElement()); |
|
| 1713 |
} else if (isStartingElement(next, FOOTNOTE_REF)) {
|
|
| 1714 |
//TODO |
|
| 1715 |
handleNotYetImplementedElement(next); |
|
| 1716 |
} else if (isEndingElement(next, FOOTNOTE_REF)) {
|
|
| 1717 |
//TODO |
|
| 1718 |
popUnimplemented(next.asEndElement()); |
|
| 1719 |
} else if (isStartingElement(next, FOOTNOTE)) {
|
|
| 1720 |
//TODO |
|
| 1721 |
handleNotYetImplementedElement(next); |
|
| 1722 |
} else if (isEndingElement(next, FOOTNOTE)) {
|
|
| 1723 |
//TODO |
|
| 1724 |
popUnimplemented(next.asEndElement()); |
|
| 1725 |
} else if (isStartingElement(next, WRITER)) {
|
|
| 1726 |
//TODO |
|
| 1727 |
handleNotYetImplementedElement(next); |
|
| 1728 |
} else if (isEndingElement(next, WRITER)) {
|
|
| 1729 |
//TODO |
|
| 1730 |
popUnimplemented(next.asEndElement()); |
|
| 1731 |
} else if (isStartingElement(next, DATES)) {
|
|
| 1732 |
//TODO |
|
| 1733 |
handleNotYetImplementedElement(next); |
|
| 1734 |
} else if (isEndingElement(next, DATES)) {
|
|
| 1735 |
//TODO |
|
| 1736 |
popUnimplemented(next.asEndElement()); |
|
| 1737 |
} else {
|
|
| 1738 |
handleUnexpectedElement(next); |
|
| 1739 |
} |
|
| 1740 |
} |
|
| 1741 |
throw new IllegalStateException("<String> has no closing tag");
|
|
| 1742 |
} |
|
| 1743 |
|
|
| 1744 |
/** |
|
| 1745 |
* @param subHeadingMap |
|
| 1746 |
* @param currentSubheading |
|
| 1747 |
* @param text |
|
| 1748 |
* @return |
|
| 1749 |
*/ |
|
| 1750 |
private String putCurrentSubheading(Map<String, String> subHeadingMap, String currentSubheading, String text) {
|
|
| 1751 |
if (StringUtils.isNotBlank(text)) {
|
|
| 1752 |
text = removeStartingMinus(text); |
|
| 1753 |
subHeadingMap.put(currentSubheading, text.trim()); |
|
| 1754 |
} |
|
| 1755 |
return ""; |
|
| 1756 |
} |
|
| 1757 |
|
|
| 1758 |
private String removeStartingMinus(String string) {
|
|
| 1759 |
string = replaceStart(string, "-"); |
|
| 1760 |
string = replaceStart(string, "\u002d"); |
|
| 1761 |
string = replaceStart(string, "\u2013"); |
|
| 1762 |
string = replaceStart(string, "\u2014"); |
|
| 1763 |
string = replaceStart(string, "--"); |
|
| 1764 |
return string; |
|
| 1765 |
} |
|
| 1766 |
|
|
| 1767 |
/** |
|
| 1768 |
* @param value |
|
| 1769 |
* @param replacementString |
|
| 1770 |
*/ |
|
| 1771 |
private String replaceStart(String value, String replacementString) {
|
|
| 1772 |
if (value.startsWith(replacementString) ){
|
|
| 1773 |
value = value.substring(replacementString.length()).trim(); |
|
| 1774 |
} |
|
| 1775 |
while (value.startsWith("-") || value.startsWith("\u2014") ){
|
|
| 1776 |
value = value.substring("-".length()).trim();
|
|
| 1777 |
} |
|
| 1778 |
return value; |
|
| 1779 |
} |
|
| 1780 |
|
|
| 1781 |
private String getXmlTag(XMLEvent event) {
|
|
| 1782 |
String result; |
|
| 1783 |
if (event.isStartElement()) {
|
|
| 1784 |
result = "<" + event.asStartElement().getName().getLocalPart() |
|
| 1785 |
+ ">"; |
|
| 1786 |
} else if (event.isEndElement()) {
|
|
| 1787 |
result = "</" + event.asEndElement().getName().getLocalPart() + ">"; |
|
| 1788 |
} else {
|
|
| 1789 |
String message = "Only start or end elements are allowed as Html tags"; |
|
| 1790 |
throw new IllegalStateException(message); |
|
| 1791 |
} |
|
| 1792 |
return result; |
|
| 1793 |
} |
|
| 1794 |
|
|
| 1795 |
protected static final List<String> htmlList = Arrays.asList("sub", "sup",
|
|
| 1796 |
"ol", "ul", "li", "i", "b", "table", "br","tr","td"); |
|
| 1797 |
|
|
| 1798 |
private boolean isHtml(XMLEvent event) {
|
|
| 1799 |
if (event.isStartElement()) {
|
|
| 1800 |
String tag = event.asStartElement().getName().getLocalPart(); |
|
| 1801 |
return htmlList.contains(tag); |
|
| 1802 |
} else if (event.isEndElement()) {
|
|
| 1803 |
String tag = event.asEndElement().getName().getLocalPart(); |
|
| 1804 |
return htmlList.contains(tag); |
|
| 1805 |
} else {
|
|
| 1806 |
return false; |
|
| 1807 |
} |
|
| 1808 |
|
|
| 1809 |
} |
|
| 1810 |
|
|
| 1811 |
/** |
|
| 1812 |
* Handle the char or subchar element. As |
|
| 1813 |
* @param state the import state |
|
| 1814 |
* @param reader |
|
| 1815 |
* @param parentEvent |
|
| 1816 |
* @param parentFeature in case of subchars we need to attache the newly created feature to a parent feature, should be <code>null</code> |
|
| 1817 |
* for top level chars. |
|
| 1818 |
* @return List of TextData. Not a single one as the recursive TextData will also be returned |
|
| 1819 |
* @throws XMLStreamException |
|
| 1820 |
*/ |
|
| 1821 |
private List<TextData> handleChar(MarkupImportState state, XMLEventReader reader, XMLEvent parentEvent, Feature parentFeature) throws XMLStreamException {
|
|
| 1822 |
List<TextData> result = new ArrayList<TextData>(); |
|
| 1823 |
String classValue = getClassOnlyAttribute(parentEvent); |
|
| 1824 |
Feature feature = makeFeature(classValue, state, parentEvent, parentFeature); |
|
| 1825 |
|
|
| 1826 |
boolean isTextMode = true; |
|
| 1827 |
String text = ""; |
|
| 1828 |
while (reader.hasNext()) {
|
|
| 1829 |
XMLEvent next = readNoWhitespace(reader); |
|
| 1830 |
if (isMyEndingElement(next, parentEvent)) {
|
|
| 1831 |
state.putFeatureToCharSorterList(feature); |
|
| 1832 |
TextData textData = TextData.NewInstance(feature); |
|
| 1833 |
textData.putText(getDefaultLanguage(state), text); |
|
| 1834 |
result.add(textData); |
|
| 1835 |
return result; |
|
| 1836 |
} else if (isStartingElement(next, FIGURE_REF)) {
|
|
| 1837 |
//TODO |
|
| 1838 |
handleNotYetImplementedElement(next); |
|
| 1839 |
} else if (isStartingElement(next, FOOTNOTE_REF)) {
|
|
| 1840 |
//TODO |
|
| 1841 |
handleNotYetImplementedElement(next); |
|
| 1842 |
} else if (isStartingElement(next, BR)) {
|
|
| 1843 |
text += "<br/>"; |
|
| 1844 |
isTextMode = false; |
|
| 1845 |
} else if (isEndingElement(next, BR)) {
|
|
| 1846 |
isTextMode = true; |
|
| 1847 |
} else if (isHtml(next)) {
|
|
| 1848 |
text += getXmlTag(next); |
|
| 1849 |
} else if (next.isStartElement()) {
|
|
| 1850 |
if (isStartingElement(next, ANNOTATION)) {
|
|
| 1851 |
handleNotYetImplementedElement(next); //TODO test handleSimpleAnnotation |
|
| 1852 |
} else if (isStartingElement(next, ITALICS)) {
|
|
| 1853 |
handleNotYetImplementedElement(next); |
|
| 1854 |
} else if (isStartingElement(next, BOLD)) {
|
|
| 1855 |
handleNotYetImplementedElement(next); |
|
| 1856 |
} else if (isStartingElement(next, FIGURE)) {
|
|
| 1857 |
handleFigure(state, reader, next); |
|
| 1858 |
} else if (isStartingElement(next, SUB_CHAR)) {
|
|
| 1859 |
List<TextData> textData = handleChar(state, reader, next, feature); |
|
| 1860 |
result.addAll(textData); |
|
| 1861 |
} else if (isStartingElement(next, FOOTNOTE)) {
|
|
| 1862 |
FootnoteDataHolder footnote = handleFootnote(state, reader, next); |
|
| 1863 |
if (footnote.isRef()) {
|
|
| 1864 |
String message = "Ref footnote not implemented here"; |
|
| 1865 |
fireWarningEvent(message, next, 4); |
|
| 1866 |
} else {
|
|
| 1867 |
registerGivenFootnote(state, footnote); |
|
| 1868 |
} |
|
| 1869 |
} else {
|
|
| 1870 |
handleUnexpectedStartElement(next.asStartElement()); |
|
| 1871 |
} |
|
| 1872 |
} else if (next.isCharacters()) {
|
|
| 1873 |
if (!isTextMode) {
|
|
| 1874 |
String message = "String is not in text mode"; |
|
| 1875 |
fireWarningEvent(message, next, 6); |
|
| 1876 |
} else {
|
|
| 1877 |
text += next.asCharacters().getData(); |
|
| 1878 |
} |
|
| 1879 |
} else {
|
|
| 1880 |
handleUnexpectedEndElement(next.asEndElement()); |
|
| 1881 |
} |
|
| 1882 |
} |
|
| 1883 |
throw new IllegalStateException("RefPart has no closing tag");
|
|
| 1884 |
} |
|
| 1885 |
|
|
| 1886 |
/** |
|
| 1887 |
* @param classValue |
|
| 1888 |
* @param state |
|
| 1889 |
* @param parentEvent |
|
| 1890 |
* @param parentFeature |
|
| 1891 |
* @return |
|
| 1892 |
* @throws UndefinedTransformerMethodException |
|
| 1893 |
*/ |
|
| 1894 |
private Feature makeFeature(String classValue, MarkupImportState state, XMLEvent parentEvent, Feature parentFeature) {
|
|
| 1895 |
UUID uuid; |
|
| 1896 |
try {
|
|
| 1897 |
String featureText = StringUtils.capitalize(classValue); |
|
| 1898 |
if (parentFeature != null){
|
|
| 1899 |
featureText = "<%s>" + featureText; |
|
| 1900 |
featureText = String.format(featureText, parentFeature.getTitleCache()); |
|
| 1901 |
classValue = "<%s>" + classValue; |
|
| 1902 |
classValue = String.format(classValue, parentFeature.getTitleCache()); |
|
| 1903 |
} |
|
| 1904 |
|
|
| 1905 |
|
|
| 1906 |
Feature feature = state.getTransformer().getFeatureByKey(classValue); |
|
| 1907 |
if (feature != null) {
|
|
| 1908 |
return feature; |
|
| 1909 |
} |
|
| 1910 |
uuid = state.getTransformer().getFeatureUuid(classValue); |
|
| 1911 |
|
|
| 1912 |
if (uuid == null){
|
|
| 1913 |
uuid = state.getUnknownFeatureUuid(classValue); |
|
| 1914 |
} |
|
| 1915 |
|
|
| 1916 |
if (uuid == null) {
|
|
| 1917 |
// TODO |
|
| 1918 |
String message = "Uuid is not defined for '%s'"; |
|
| 1919 |
message = String.format(message, classValue); |
|
| 1920 |
fireWarningEvent(message, parentEvent, 8); |
|
| 1921 |
uuid = UUID.randomUUID(); |
|
| 1922 |
state.putUnknownFeatureUuid(classValue, uuid); |
|
| 1923 |
} |
|
| 1924 |
|
|
| 1925 |
// TODO eFlora vocabulary |
|
| 1926 |
TermVocabulary<Feature> voc = null; |
|
| 1927 |
feature = getFeature(state, uuid, featureText, featureText, classValue, voc); |
|
| 1928 |
if (parentFeature != null){
|
|
| 1929 |
parentFeature.addIncludes(feature); |
|
| 1930 |
save(parentFeature, state); |
|
| 1931 |
} |
|
| 1932 |
save(feature, state); |
|
| 1933 |
|
|
| 1934 |
if (feature == null) {
|
|
| 1935 |
throw new NullPointerException(classValue + " not recognized as a feature"); |
|
| 1936 |
} |
|
| 1937 |
// state.putFeatureToCurrentList(feature); |
|
| 1938 |
return feature; |
|
| 1939 |
} catch (Exception e) {
|
|
| 1940 |
String message = "Could not create feature for %s: %s"; |
|
| 1941 |
message = String.format(message, classValue, e.getMessage()); |
|
| 1942 |
fireWarningEvent(message, parentEvent, 4); |
|
| 1943 |
state.putUnknownFeatureUuid(classValue, null); |
|
| 1944 |
// e.printStackTrace(); |
|
| 1945 |
return Feature.UNKNOWN(); |
|
| 1946 |
} |
|
| 1947 |
} |
|
| 1948 | 631 |
|
| 1949 | 632 |
} |
Also available in: Unified diff
merge trunk into cdm3.3 branch