Example 1 with FormattingInstructionKind
use of biblemulticonverter.data.FormattedText.FormattingInstructionKind in project BibleMultiConverter by schierlm.
the class TheWord method parseLine.
private int parseLine(Visitor<RuntimeException> visitor, String line, int pos, String endTag) {
Visitor<RuntimeException> garbageVisitor = new FormattedText().getAppendVisitor();
while (pos < line.length()) {
if (line.charAt(pos) != '<') {
int endPos = line.indexOf('<', pos);
if (endPos == -1)
endPos = line.length();
visitor.visitText(line.substring(pos, endPos).replaceAll("[\r\n\t ]+", " "));
pos = endPos;
continue;
}
if (endTag != null && line.startsWith(endTag, pos))
break;
if (pos + 2 < line.length() && line.charAt(pos + 2) == '>' && "bius".indexOf(line.charAt(pos + 1)) != -1) {
String newEndTag = "</" + line.charAt(pos + 1) + ">";
if (parseLine(garbageVisitor, line, pos + 3, newEndTag) != -1) {
FormattingInstructionKind kind;
switch(line.charAt(pos + 1)) {
case 'b':
kind = FormattingInstructionKind.BOLD;
break;
case 'i':
kind = FormattingInstructionKind.ITALIC;
break;
case 'u':
kind = FormattingInstructionKind.UNDERLINE;
break;
case 's':
kind = FormattingInstructionKind.STRIKE_THROUGH;
break;
default:
throw new RuntimeException("Cannot happen");
}
pos = parseLine(visitor.visitFormattingInstruction(kind), line, pos + 3, newEndTag);
continue;
}
} else if (line.startsWith("<sub>", pos) || line.startsWith("<sup>", pos)) {
String newEndTag = "</" + line.substring(pos + 1, pos + 5);
if (parseLine(garbageVisitor, line, pos + 5, newEndTag) != -1) {
FormattingInstructionKind kind = line.charAt(pos + 3) == 'p' ? FormattingInstructionKind.SUPERSCRIPT : FormattingInstructionKind.SUBSCRIPT;
pos = parseLine(visitor.visitFormattingInstruction(kind), line, pos + 5, newEndTag);
continue;
}
} else if (line.startsWith("<FR>", pos)) {
if (parseLine(garbageVisitor, line, pos + 4, "<Fr>") != -1) {
pos = parseLine(visitor.visitFormattingInstruction(FormattingInstructionKind.WORDS_OF_JESUS), line, pos + 4, "<Fr>");
continue;
}
} else if (line.startsWith("<FO>", pos)) {
if (parseLine(garbageVisitor, line, pos + 4, "<Fo>") != -1) {
pos = parseLine(visitor.visitFormattingInstruction(FormattingInstructionKind.LINK), line, pos + 4, "<Fo>");
continue;
}
} else if (line.startsWith("<font color=\"gray\">/</font>", pos)) {
visitor.visitVerseSeparator();
pos += 27;
continue;
} else if (line.startsWith("<CL>", pos)) {
visitor.visitLineBreak(LineBreakKind.NEWLINE);
pos += 4;
continue;
} else if (line.startsWith("<CM>", pos)) {
visitor.visitLineBreak(LineBreakKind.PARAGRAPH);
pos += 4;
continue;
} else if (line.startsWith("<CI><PI>", pos)) {
visitor.visitLineBreak(LineBreakKind.NEWLINE_WITH_INDENT);
pos += 8;
continue;
} else if (line.startsWith("<TS", pos) && pos + 3 < line.length()) {
char next = line.charAt(pos + 3);
int depth, len;
if (next == '>') {
depth = 1;
len = 4;
} else if (pos + 4 < line.length() && line.charAt(pos + 4) == '>' && next >= '1' && next <= '3') {
depth = next - '0';
len = 5;
} else {
depth = len = 0;
}
String end = "<Ts>", altEnd = len == 5 ? "<Ts" + next + ">" : "<Ts>";
if (line.indexOf(altEnd, pos) != -1 && (line.indexOf(end, pos) == -1 || line.indexOf(altEnd, pos) < line.indexOf(end, pos)))
end = altEnd;
if (len > 0 && parseLine(garbageVisitor, line, pos + len, end) != -1) {
pos = parseLine(visitor.visitHeadline(depth), line, pos + len, end);
continue;
}
} else if (line.startsWith("<RF", pos)) {
int closePos = line.indexOf('>', pos);
if (parseLine(garbageVisitor, line, closePos + 1, "<Rf>") != -1) {
pos = parseLine(visitor.visitFootnote(), line, closePos + 1, "<Rf>");
continue;
}
} else if (line.startsWith("<FI>", pos)) {
if (parseLine(garbageVisitor, line, pos + 4, "<Fi>") != -1) {
pos = parseLine(visitor.visitFormattingInstruction(FormattingInstructionKind.ITALIC), line, pos + 4, "<Fi>");
continue;
}
} else if (line.startsWith("<S%", pos)) {
int closePos = line.indexOf('>', pos);
if (parseLine(garbageVisitor, line, closePos + 1, "<s%>") != -1) {
String[] strongs = line.substring(pos + 3, closePos).split("%");
int[] strongNumbers = new int[strongs.length];
try {
for (int i = 0; i < strongs.length; i++) {
strongNumbers[i] = Integer.parseInt(strongs[i]);
}
pos = parseLine(visitor.visitGrammarInformation(strongNumbers, null, null), line, closePos + 1, "<s%>");
continue;
} catch (NumberFormatException ex) {
// malformed Strongs tag
}
}
} else if (line.startsWith("<XWG", pos) || line.startsWith("<XWH", pos)) {
int closePos = line.indexOf('>', pos);
try {
int number = Integer.parseInt(line.substring(pos + 4, closePos));
visitor.visitGrammarInformation(new int[] { number }, null, null);
pos = closePos + 1;
continue;
} catch (NumberFormatException ex) {
System.out.println("WARNING: Invalid Strong number in tag " + line.substring(pos, closePos + 1));
warningCount++;
}
} else if (line.startsWith("<WT", pos)) {
// TODO parse morph information
} else if (line.startsWith("<RX", pos)) {
// TODO parse cross references
} else if (line.startsWith("<CI>", pos) || line.startsWith("<PF", pos) || line.startsWith("<PI", pos)) {
// extra formatting not supported by BMC
} else if (warningCount < 100) {
System.out.println("WARNING: Skipping unknown tag " + line.substring(pos, Math.min(pos + 20, line.length())));
warningCount++;
}
// the tag is not supported (yet), skip the first character
visitor.visitText("<");
pos++;
}
if (endTag != null) {
if (line.startsWith(endTag, pos))
pos += endTag.length();
else
pos = -1;
}
return pos;
}
Also used :
FormattingInstructionKind(biblemulticonverter.data.FormattedText.FormattingInstructionKind)
FormattedText(biblemulticonverter.data.FormattedText)
Example 2 with FormattingInstructionKind
use of biblemulticonverter.data.FormattedText.FormattingInstructionKind in project BibleMultiConverter by schierlm.
the class ZefDic method createXMLBible.
protected Dictionary createXMLBible(Bible bible) throws Exception {
final ObjectFactory of = new ObjectFactory();
Dictionary doc = of.createDictionary();
doc.setDicversion("1");
doc.setRevision("1");
doc.setRefbible("any");
doc.setType(TEnumDicType.X_DICTIONARY);
String title = null;
if (bible.getName().matches("X_(DICTIONARY|COMMENTARY|STRONG|DAILY)@.*")) {
String[] parts = bible.getName().split("@", 2);
doc.setType(TEnumDicType.valueOf(parts[0]));
doc.setRefbible(parts[1]);
} else {
title = bible.getName();
}
doc.setINFORMATION(of.createTINFORMATION());
doc.getINFORMATION().getTitleOrCreatorOrDescription().add(new JAXBElement<String>(new QName("title"), String.class, title));
MetadataBook metadata = bible.getMetadataBook();
if (metadata != null) {
for (String key : metadata.getKeys()) {
String value = metadata.getValue(key);
if (value.equals("-empty-"))
value = "";
if (key.equals(MetadataBookKey.version.toString())) {
doc.setDicversion(value);
} else if (key.equals(MetadataBookKey.revision.toString())) {
doc.setRevision(value);
} else if (Arrays.asList(INFORMATION_KEYS).contains(key)) {
doc.getINFORMATION().getTitleOrCreatorOrDescription().add(new JAXBElement<String>(new QName(key), String.class, value));
}
}
}
for (Book bk : bible.getBooks()) {
if (bk.getId().equals(BookID.METADATA))
continue;
if (!bk.getId().equals(BookID.DICTIONARY_ENTRY)) {
System.out.println("WARNING: Unable to export book " + bk.getAbbr());
continue;
}
final TItem item = of.createTItem();
if (!bk.getLongName().equals(bk.getShortName())) {
TItem itm = of.createTItem();
itm.setId(bk.getShortName());
appendTextElement(itm, "title", bk.getLongName());
TParagraph para2 = of.createTParagraph();
SeeType see = of.createSeeType();
see.setContent(bk.getLongName());
para2.getContent().add(new JAXBElement<SeeType>(new QName("see"), SeeType.class, see));
itm.getContent().add(new JAXBElement<TParagraph>(new QName("description"), TParagraph.class, para2));
doc.getItem().add(itm);
}
item.setId(bk.getLongName());
doc.getItem().add(item);
class ZefState {
TParagraph para = of.createTParagraph();
boolean eatParagraph = false;
public void flushPara(TItem item) {
item.getContent().add(new JAXBElement<TParagraph>(new QName("description"), TParagraph.class, para));
para = of.createTParagraph();
}
}
final ZefState state = new ZefState();
FormattedText text = bk.getChapters().get(0).getProlog();
class LevelVisitor implements Visitor<RuntimeException> {
final List<Serializable> target;
private LevelVisitor(ZefState state) {
target = state.para.getContent();
}
private LevelVisitor(MyAnyType parent) {
target = parent.getContent();
}
private LevelVisitor(TStyle parent) {
target = parent.getContent();
}
@Override
public int visitElementTypes(String elementTypes) throws RuntimeException {
return 0;
}
@Override
public Visitor<RuntimeException> visitHeadline(int depth) throws RuntimeException {
System.out.println("WARNING: Nested headlines are not supported");
return null;
}
@Override
public void visitStart() throws RuntimeException {
}
@Override
public void visitText(String text) throws RuntimeException {
if (text.length() > 0)
target.add(text);
}
@Override
public Visitor<RuntimeException> visitFootnote() throws RuntimeException {
System.out.println("WARNING: footnotes are not supported");
return null;
}
@Override
public Visitor<RuntimeException> visitCrossReference(String bookAbbr, BookID book, int firstChapter, String firstVerse, int lastChapter, String lastVerse) throws RuntimeException {
if (firstChapter != lastChapter || !firstVerse.equals(lastVerse))
System.out.println("WARNING: Cross references to verse ranges are not supported");
BibLinkType b = of.createBibLinkType();
b.setBn("" + book.getZefID());
b.setCn1("" + firstChapter);
b.setVn1(firstVerse);
target.add(new JAXBElement<BibLinkType>(new QName("bib_link"), BibLinkType.class, b));
return null;
}
@Override
public Visitor<RuntimeException> visitFormattingInstruction(FormattingInstructionKind kind) throws RuntimeException {
String tag;
switch(kind) {
case BOLD:
tag = "strong";
break;
case ITALIC:
tag = "em";
break;
case SUPERSCRIPT:
tag = "sup";
break;
case SUBSCRIPT:
tag = "sub";
break;
default:
return visitCSSFormatting(kind.getCss());
}
MyAnyType mat = of.createMyAnyType();
target.add(new JAXBElement<MyAnyType>(new QName(tag), MyAnyType.class, mat));
return new LevelVisitor(mat);
}
@Override
public Visitor<RuntimeException> visitCSSFormatting(String css) throws RuntimeException {
TStyle style = of.createTStyle();
style.setCss(css);
target.add(of.createTStyleSTYLE(style));
return new LevelVisitor(style);
}
@Override
public void visitVerseSeparator() throws RuntimeException {
System.out.println("WARNING: Verse separators are not supported");
}
@Override
public void visitLineBreak(LineBreakKind kind) throws RuntimeException {
System.out.println("WARNING: Nested line breaks are not supported");
}
@Override
public Visitor<RuntimeException> visitGrammarInformation(int[] strongs, String[] rmac, int[] sourceIndices) throws RuntimeException {
System.out.println("WARNING: Grammar information is not supported");
return null;
}
@Override
public Visitor<RuntimeException> visitDictionaryEntry(String dictionary, String entry) throws RuntimeException {
if (dictionary.equals("reflink")) {
RefLinkType r = of.createRefLinkType();
r.setMscope(entry.substring(1).replace('-', ';'));
target.add(new JAXBElement<RefLinkType>(new QName("reflink"), RefLinkType.class, r));
} else {
SeeType see = of.createSeeType();
see.setTarget(dictionary.equals("dict") ? "x-self" : dictionary);
see.setContent(entry);
target.add(new JAXBElement<SeeType>(new QName("see"), SeeType.class, see));
}
return null;
}
@Override
public void visitRawHTML(RawHTMLMode mode, String raw) throws RuntimeException {
System.out.println("WARNING: Raw html output not supported");
}
@Override
public Visitor<RuntimeException> visitVariationText(String[] variations) throws RuntimeException {
throw new IllegalStateException("Variations not supported");
}
@Override
public Visitor<RuntimeException> visitExtraAttribute(ExtraAttributePriority prio, String category, String key, String value) throws RuntimeException {
return prio.handleVisitor(category, this);
}
@Override
public boolean visitEnd() throws RuntimeException {
return false;
}
}
;
text.accept(new Visitor<RuntimeException>() {
@Override
public int visitElementTypes(String elementTypes) throws RuntimeException {
return 0;
}
@Override
public Visitor<RuntimeException> visitHeadline(int depth) throws RuntimeException {
MyAnyType mat = of.createMyAnyType();
JAXBElement<MyAnyType> elem = new JAXBElement<>(new QName("title"), MyAnyType.class, mat);
if (depth == 1) {
state.flushPara(item);
item.getContent().add(elem);
} else {
state.para.getContent().add(elem);
}
return new LevelVisitor(mat);
}
@Override
public void visitStart() throws RuntimeException {
}
@Override
public void visitText(String text) throws RuntimeException {
new LevelVisitor(state).visitText(text);
}
@Override
public Visitor<RuntimeException> visitFootnote() throws RuntimeException {
System.out.println("WARNING: footnotes are not supported");
return null;
}
@Override
public Visitor<RuntimeException> visitCrossReference(String bookAbbr, BookID book, int firstChapter, String firstVerse, int lastChapter, String lastVerse) throws RuntimeException {
return new LevelVisitor(state).visitCrossReference(bookAbbr, book, firstChapter, firstVerse, lastChapter, lastVerse);
}
@Override
public Visitor<RuntimeException> visitFormattingInstruction(FormattingInstructionKind kind) throws RuntimeException {
return new LevelVisitor(state).visitFormattingInstruction(kind);
}
@Override
public Visitor<RuntimeException> visitCSSFormatting(String css) throws RuntimeException {
return new LevelVisitor(state).visitCSSFormatting(css);
}
@Override
public void visitVerseSeparator() throws RuntimeException {
System.out.println("WARNING: Verse separators are not supported");
}
@Override
public void visitLineBreak(LineBreakKind kind) throws RuntimeException {
if (state.eatParagraph) {
state.eatParagraph = false;
} else {
state.flushPara(item);
state.para = of.createTParagraph();
}
}
@Override
public Visitor<RuntimeException> visitGrammarInformation(int[] strongs, String[] rmac, int[] sourceIndices) throws RuntimeException {
System.out.println("WARNING: Grammar information is not supported");
return null;
}
@Override
public Visitor<RuntimeException> visitDictionaryEntry(String dictionary, String entry) throws RuntimeException {
return new LevelVisitor(state).visitDictionaryEntry(dictionary, entry);
}
@Override
public void visitRawHTML(RawHTMLMode mode, String raw) throws RuntimeException {
System.out.println("WARNING: Raw html output not supported");
}
@Override
public Visitor<RuntimeException> visitVariationText(String[] variations) throws RuntimeException {
throw new IllegalStateException("Variations not supported");
}
@Override
public Visitor<RuntimeException> visitExtraAttribute(ExtraAttributePriority prio, String category, String key, String value) throws RuntimeException {
if (prio == ExtraAttributePriority.KEEP_CONTENT && category.equals("zefdic")) {
// "zefdic", "field", "pronunciation");
return null;
} else {
return prio.handleVisitor(category, this);
}
}
@Override
public boolean visitEnd() throws RuntimeException {
return false;
}
});
state.flushPara(item);
}
return doc;
}
Also used :
Dictionary(biblemulticonverter.schema.zefdic1.Dictionary)
TParagraph(biblemulticonverter.schema.zefdic1.TParagraph)
ExtraAttributePriority(biblemulticonverter.data.FormattedText.ExtraAttributePriority)
Visitor(biblemulticonverter.data.FormattedText.Visitor)
TItem(biblemulticonverter.schema.zefdic1.TItem)
RawHTMLMode(biblemulticonverter.data.FormattedText.RawHTMLMode)
ObjectFactory(biblemulticonverter.schema.zefdic1.ObjectFactory)
BookID(biblemulticonverter.data.BookID)
MetadataBook(biblemulticonverter.data.MetadataBook)
Book(biblemulticonverter.data.Book)
List(java.util.List)
BibLinkType(biblemulticonverter.schema.zefdic1.BibLinkType)
RefLinkType(biblemulticonverter.schema.zefdic1.RefLinkType)
MetadataBook(biblemulticonverter.data.MetadataBook)
TStyle(biblemulticonverter.schema.zefdic1.TStyle)
MyAnyType(biblemulticonverter.schema.zefdic1.MyAnyType)
QName(javax.xml.namespace.QName)
FormattingInstructionKind(biblemulticonverter.data.FormattedText.FormattingInstructionKind)
FormattedText(biblemulticonverter.data.FormattedText)
JAXBElement(javax.xml.bind.JAXBElement)
LineBreakKind(biblemulticonverter.data.FormattedText.LineBreakKind)
SeeType(biblemulticonverter.schema.zefdic1.SeeType)
Example 3 with FormattingInstructionKind
use of biblemulticonverter.data.FormattedText.FormattingInstructionKind in project BibleMultiConverter by schierlm.
the class ZefaniaXMLMyBible method doExport.
@Override
public void doExport(Bible bible, String... exportArgs) throws Exception {
new StrippedDiffable().mergeIntroductionPrologs(bible);
final ObjectFactory f = new ObjectFactory();
XMLBIBLE doc = f.createXMLBIBLE();
doc.setBiblename(bible.getName());
doc.setType(EnumModtyp.X_BIBLE);
BigInteger revision = null;
MetadataBook metadata = bible.getMetadataBook();
if (metadata != null) {
for (MetadataBookKey key : Arrays.asList(MetadataBookKey.revision, MetadataBookKey.version, MetadataBookKey.date, MetadataBookKey.title)) {
String digits = metadata.getValue(key);
if (digits == null)
continue;
digits = digits.replaceAll("[^0-9]+", "");
if (!digits.isEmpty()) {
revision = new BigInteger(digits);
break;
}
}
}
if (revision == null) {
String digits = bible.getName().replaceAll("[^0-9]+", "");
if (!digits.isEmpty()) {
revision = new BigInteger(digits);
}
}
if (revision != null) {
doc.setRevision(revision);
}
doc.setINFORMATION(f.createINFORMATION());
List<DIV> prologs = new ArrayList<DIV>();
for (Book bk : bible.getBooks()) {
if (bk.getId().equals(BookID.METADATA))
continue;
int bsnumber = bk.getId().getZefID();
final BIBLEBOOK book = f.createBIBLEBOOK();
book.setBnumber(BigInteger.valueOf(bsnumber));
book.setBname(bk.getShortName());
book.setBsname(bk.getAbbr());
doc.getBIBLEBOOK().add(book);
int cnumber = 0;
for (Chapter cch : bk.getChapters()) {
cnumber++;
if (cch.getProlog() != null) {
DIV xx = f.createDIV();
prologs.add(xx);
NOTE xxx = f.createNOTE();
xx.setNOTE(xxx);
xxx.setType("x-studynote");
NOTE prolog = xxx;
DIV vers = f.createDIV();
prolog.getContent().add("<p>");
prolog.getContent().add(vers);
prolog.getContent().add("</p>");
vers.setNOTE(f.createNOTE());
final List<List<Object>> targetStack = new ArrayList<List<Object>>();
targetStack.add(vers.getNOTE().getContent());
cch.getProlog().accept(new Visitor<IOException>() {
@Override
public Visitor<IOException> visitHeadline(int depth) throws IOException {
if (depth > 6)
depth = 6;
STYLE s = f.createSTYLE();
s.setCss("-zef-dummy: true");
targetStack.get(0).add("<h" + depth + ">");
targetStack.get(0).add(new JAXBElement<STYLE>(new QName("STYLE"), STYLE.class, s));
targetStack.get(0).add("</h" + depth + ">");
targetStack.add(0, s.getContent());
return this;
}
@Override
public void visitVerseSeparator() throws IOException {
STYLE x = f.createSTYLE();
x.setCss("color:gray");
x.getContent().add("/");
targetStack.get(0).add(new JAXBElement<STYLE>(new QName("STYLE"), STYLE.class, x));
}
@Override
public void visitText(String text) throws IOException {
targetStack.get(0).add(text);
}
@Override
public Visitor<IOException> visitFormattingInstruction(FormattedText.FormattingInstructionKind kind) throws IOException {
String startTag, endTag;
if (kind.getHtmlTag() != null) {
startTag = "<" + kind.getHtmlTag() + ">";
endTag = "</" + kind.getHtmlTag() + ">";
} else {
startTag = "<span style=\"" + kind.getCss() + "\">";
endTag = "</span>";
}
STYLE s = f.createSTYLE();
s.setCss("-zef-dummy: true");
targetStack.get(0).add(startTag);
targetStack.get(0).add(new JAXBElement<STYLE>(new QName("STYLE"), STYLE.class, s));
targetStack.get(0).add(endTag);
targetStack.add(0, s.getContent());
return this;
}
@Override
public Visitor<IOException> visitFootnote() throws IOException {
System.out.println("WARNING: Footnotes in prolog are not supported");
return null;
}
@Override
public Visitor<IOException> visitCrossReference(String bookAbbr, BookID book, int firstChapter, String firstVerse, int lastChapter, String lastVerse) throws IOException {
System.out.println("WARNING: Cross references in prologs are not supported");
STYLE s = f.createSTYLE();
s.setCss("-zef-dummy: true");
targetStack.get(0).add(new JAXBElement<STYLE>(new QName("STYLE"), STYLE.class, s));
targetStack.add(0, s.getContent());
return this;
}
@Override
public Visitor<IOException> visitVariationText(String[] variations) throws IOException {
throw new RuntimeException("Variations not supported");
}
@Override
public void visitLineBreak(LineBreakKind kind) throws IOException {
BR br = f.createBR();
br.setArt(kind == LineBreakKind.PARAGRAPH ? EnumBreak.X_P : EnumBreak.X_NL);
targetStack.get(0).add(" ");
targetStack.get(0).add(kind == LineBreakKind.PARAGRAPH ? "<p>" : "<br>");
targetStack.get(0).add(br);
}
@Override
public Visitor<IOException> visitGrammarInformation(int[] strongs, String[] rmac, int[] sourceIndices) throws IOException {
throw new RuntimeException("Grammar tags in prologs not supported");
}
@Override
public FormattedText.Visitor<IOException> visitDictionaryEntry(String dictionary, String entry) throws IOException {
throw new RuntimeException("Dictionary entries in prologs not supported");
}
@Override
public void visitRawHTML(RawHTMLMode mode, String raw) throws IOException {
throw new RuntimeException("Raw HTML in prologs not supported");
}
@Override
public Visitor<IOException> visitCSSFormatting(String css) throws IOException {
STYLE s = f.createSTYLE();
s.setCss("-zef-dummy: true");
targetStack.get(0).add("<span style=\"" + css + "\">");
targetStack.get(0).add(new JAXBElement<STYLE>(new QName("STYLE"), STYLE.class, s));
targetStack.get(0).add("</span>");
targetStack.add(0, s.getContent());
return this;
}
@Override
public int visitElementTypes(String elementTypes) throws IOException {
return 0;
}
@Override
public Visitor<IOException> visitExtraAttribute(ExtraAttributePriority prio, String category, String key, String value) throws IOException {
if (prio == ExtraAttributePriority.KEEP_CONTENT)
return visitCSSFormatting("-zef-extra-attribute-" + category + "-" + key + ": " + value);
else if (prio == ExtraAttributePriority.SKIP)
return null;
throw new RuntimeException("Extra attributes not supported");
}
@Override
public void visitStart() throws IOException {
}
@Override
public boolean visitEnd() throws IOException {
targetStack.remove(0);
return false;
}
});
if (targetStack.size() != 0)
throw new RuntimeException();
}
if (cch.getVerses().size() == 0)
continue;
CHAPTER chapter = f.createCHAPTER();
chapter.setCnumber(BigInteger.valueOf(cnumber));
book.getCHAPTER().add(chapter);
for (VirtualVerse vv : cch.createVirtualVerses()) {
for (Headline h : vv.getHeadlines()) {
CAPTION caption = f.createCAPTION();
EnumCaptionType[] types = new EnumCaptionType[] { null, EnumCaptionType.X_H_1, EnumCaptionType.X_H_2, EnumCaptionType.X_H_3, EnumCaptionType.X_H_4, EnumCaptionType.X_H_5, EnumCaptionType.X_H_6, EnumCaptionType.X_H_6, EnumCaptionType.X_H_6, EnumCaptionType.X_H_6 };
caption.setType(types[h.getDepth()]);
caption.setVref(BigInteger.valueOf(vv.getNumber()));
final StringBuilder sb = new StringBuilder();
h.accept(new FormattedText.VisitorAdapter<RuntimeException>(null) {
@Override
protected void beforeVisit() throws RuntimeException {
throw new IllegalStateException();
}
@Override
public Visitor<RuntimeException> visitFormattingInstruction(FormattingInstructionKind kind) throws RuntimeException {
System.out.println("WARNING: Formatting instructions in captions are not supported (stripped)");
return this;
}
@Override
public Visitor<RuntimeException> visitFootnote() throws RuntimeException {
System.out.println("WARNING: Footnotes in captions are not supported (stripped)");
return null;
}
@Override
public Visitor<RuntimeException> visitCSSFormatting(String css) throws RuntimeException {
System.out.println("WARNING: CSS Formatting in captions are not supported (stripped)");
return this;
}
@Override
public Visitor<RuntimeException> visitExtraAttribute(ExtraAttributePriority prio, String category, String key, String value) throws RuntimeException {
return prio.handleVisitor(category, this);
}
public void visitText(String text) throws RuntimeException {
sb.append(text);
}
});
caption.getContent().add(sb.toString());
chapter.getPROLOGOrCAPTIONOrVERS().add(caption);
}
VERS vers = f.createVERS();
vers.setVnumber(BigInteger.valueOf(vv.getNumber()));
for (DIV prolog : prologs) {
vers.getContent().add(prolog);
}
prologs.clear();
chapter.getPROLOGOrCAPTIONOrVERS().add(vers);
boolean first = true;
for (Verse v : vv.getVerses()) {
if (!first || !v.getNumber().equals("" + vv.getNumber())) {
STYLE x = f.createSTYLE();
x.setCss("font-weight: bold");
x.getContent().add("(" + v.getNumber() + ")");
vers.getContent().add(new JAXBElement<STYLE>(new QName("STYLE"), STYLE.class, x));
vers.getContent().add(" ");
}
first = false;
final List<List<Object>> targetStack = new ArrayList<List<Object>>();
targetStack.add(vers.getContent());
v.accept(new FormattedText.Visitor<IOException>() {
@Override
public void visitVerseSeparator() throws IOException {
STYLE x = f.createSTYLE();
x.setCss("color:gray");
x.getContent().add("/");
targetStack.get(0).add(new JAXBElement<STYLE>(new QName("STYLE"), STYLE.class, x));
}
@Override
public void visitText(String text) throws IOException {
targetStack.get(0).add(text);
}
@Override
public FormattedText.Visitor<IOException> visitFormattingInstruction(biblemulticonverter.data.FormattedText.FormattingInstructionKind kind) throws IOException {
STYLE x = f.createSTYLE();
x.setCss(kind.getCss());
targetStack.get(0).add(new JAXBElement<STYLE>(new QName("STYLE"), STYLE.class, x));
targetStack.add(0, x.getContent());
return this;
}
@Override
public Visitor<IOException> visitFootnote() throws IOException {
DIV x = f.createDIV();
targetStack.get(0).add(x);
NOTE n = f.createNOTE();
x.setNOTE(n);
n.setType("x-studynote");
final List<List<Object>> footnoteStack = new ArrayList<List<Object>>();
footnoteStack.add(n.getContent());
return new Visitor<IOException>() {
@Override
public void visitStart() throws IOException {
}
@Override
public void visitVerseSeparator() throws IOException {
STYLE x = f.createSTYLE();
x.setCss("color:gray");
x.getContent().add("/");
footnoteStack.get(0).add(new JAXBElement<STYLE>(new QName("STYLE"), STYLE.class, x));
}
@Override
public void visitText(String text) throws IOException {
footnoteStack.get(0).add(text);
}
@Override
public void visitLineBreak(LineBreakKind kind) throws IOException {
BR br = f.createBR();
br.setArt(kind == LineBreakKind.PARAGRAPH ? EnumBreak.X_P : EnumBreak.X_NL);
footnoteStack.get(0).add(" ");
footnoteStack.get(0).add(br);
}
@Override
public Visitor<IOException> visitFormattingInstruction(FormattedText.FormattingInstructionKind kind) throws IOException {
String startTag, endTag;
if (kind.getHtmlTag() != null) {
startTag = "<" + kind.getHtmlTag() + ">";
endTag = "</" + kind.getHtmlTag() + ">";
} else {
startTag = "<span style=\"" + kind.getCss() + "\">";
endTag = "</span>";
}
STYLE s = f.createSTYLE();
s.setCss("-zef-dummy: true");
footnoteStack.get(0).add(startTag);
footnoteStack.get(0).add(new JAXBElement<STYLE>(new QName("STYLE"), STYLE.class, s));
footnoteStack.get(0).add(endTag);
footnoteStack.add(0, s.getContent());
return this;
}
@Override
public Visitor<IOException> visitFootnote() throws IOException {
throw new RuntimeException("Footnotes in footnotes are not supported");
}
@Override
public Visitor<IOException> visitGrammarInformation(int[] strongs, String[] rmac, int[] sourceIndices) throws IOException {
GRAM gram = f.createGRAM();
if (strongs != null) {
StringBuilder entryBuilder = new StringBuilder();
for (int i = 0; i < strongs.length; i++) {
entryBuilder.append((i > 0 ? " " : "") + strongs[i]);
}
String entry = entryBuilder.toString();
gram.setStr(entry);
}
if (rmac != null) {
StringBuilder entryBuilder = new StringBuilder();
for (int i = 0; i < rmac.length; i++) {
if (i > 0)
entryBuilder.append(' ');
entryBuilder.append(rmac[i]);
}
gram.setRmac(entryBuilder.toString());
}
footnoteStack.get(0).add(new JAXBElement<GRAM>(new QName("gr"), GRAM.class, gram));
footnoteStack.add(0, gram.getContent());
return this;
}
@Override
public FormattedText.Visitor<IOException> visitDictionaryEntry(String dictionary, String entry) throws IOException {
GRAM gram = f.createGRAM();
gram.setStr(entry);
footnoteStack.get(0).add(new JAXBElement<GRAM>(new QName("gr"), GRAM.class, gram));
footnoteStack.add(0, gram.getContent());
return this;
}
@Override
public void visitRawHTML(RawHTMLMode mode, String raw) throws IOException {
if (mode != RawHTMLMode.ONLINE)
footnoteStack.get(0).add(raw);
}
@Override
public Visitor<IOException> visitVariationText(String[] variations) throws IOException {
throw new RuntimeException("Variations not supported");
}
@Override
public FormattedText.Visitor<IOException> visitCrossReference(String bookAbbr, BookID book, int firstChapter, String firstVerse, int lastChapter, String lastVerse) throws IOException {
STYLE s = f.createSTYLE();
s.setCss("-zef-dummy: true");
int bookID = book.getZefID();
String mscope, xmscope;
try {
int start = firstVerse.equals("^") ? 1 : Integer.parseInt(firstVerse.replaceAll("[a-zG]|[,/][0-9]*", ""));
int end;
if (firstChapter == lastChapter && !lastVerse.equals("$")) {
end = Integer.parseInt(lastVerse.replaceAll("[a-z]|[,/][0-9]*", ""));
} else {
end = -1;
}
mscope = bookID + "," + firstChapter + "," + start + "," + end;
xmscope = bookID + ";" + firstChapter + ";" + start + "-" + end;
} catch (NumberFormatException ex) {
ex.printStackTrace();
mscope = bookID + ",1,1,999";
xmscope = bookID + ";1;1-999";
}
if (footnoteStack.size() == 1) {
List<Object> outerList = targetStack.get(0);
XREF xref = new XREF();
xref.setMscope(xmscope);
outerList.add(outerList.size() - 1, xref);
}
footnoteStack.get(0).add("<a href=\"mybible:content=location&locations=" + mscope + "\">");
footnoteStack.get(0).add(new JAXBElement<STYLE>(new QName("STYLE"), STYLE.class, s));
footnoteStack.get(0).add("</a>");
footnoteStack.add(0, s.getContent());
return this;
}
public boolean visitEnd() throws IOException {
footnoteStack.remove(0);
return false;
}
@Override
public int visitElementTypes(String elementTypes) throws IOException {
return 0;
}
@Override
public Visitor<IOException> visitHeadline(int depth) throws IOException {
throw new RuntimeException("Headlines in footnotes not supported");
}
@Override
public Visitor<IOException> visitCSSFormatting(String css) throws IOException {
STYLE s = f.createSTYLE();
s.setCss("-zef-dummy: true");
footnoteStack.get(0).add("<span style=\"" + css + "\">");
footnoteStack.get(0).add(new JAXBElement<STYLE>(new QName("STYLE"), STYLE.class, s));
footnoteStack.get(0).add("</span>");
footnoteStack.add(s.getContent());
return this;
}
@Override
public Visitor<IOException> visitExtraAttribute(ExtraAttributePriority prio, String category, String key, String value) throws IOException {
System.out.println("WARNING: Extra attributes not supported");
Visitor<IOException> result = prio.handleVisitor(category, this);
if (result != null)
footnoteStack.add(0, footnoteStack.get(0));
return result;
}
};
}
@Override
public FormattedText.Visitor<IOException> visitCrossReference(String bookAbbr, BookID book, int firstChapter, String firstVerse, int lastChapter, String lastVerse) throws IOException {
throw new RuntimeException("Xref outside of footnotes not supported!");
}
@Override
public void visitLineBreak(LineBreakKind kind) throws IOException {
BR br = f.createBR();
br.setArt(kind == LineBreakKind.PARAGRAPH ? EnumBreak.X_P : EnumBreak.X_NL);
targetStack.get(0).add(" ");
targetStack.get(0).add(br);
}
@Override
public Visitor<IOException> visitGrammarInformation(int[] strongs, String[] rmac, int[] sourceIndices) throws IOException {
GRAM gram = f.createGRAM();
if (strongs != null) {
StringBuilder entryBuilder = new StringBuilder();
for (int i = 0; i < strongs.length; i++) {
entryBuilder.append((i > 0 ? " " : "") + strongs[i]);
}
String entry = entryBuilder.toString();
gram.setStr(entry);
}
if (rmac != null) {
StringBuilder entryBuilder = new StringBuilder();
for (int i = 0; i < rmac.length; i++) {
if (i > 0)
entryBuilder.append(' ');
entryBuilder.append(rmac[i]);
}
gram.setRmac(entryBuilder.toString());
}
targetStack.get(0).add(new JAXBElement<GRAM>(new QName("gr"), GRAM.class, gram));
targetStack.add(0, gram.getContent());
return this;
}
@Override
public FormattedText.Visitor<IOException> visitDictionaryEntry(String dictionary, String entry) throws IOException {
GRAM gram = f.createGRAM();
gram.setStr(entry);
targetStack.get(0).add(new JAXBElement<GRAM>(new QName("gr"), GRAM.class, gram));
targetStack.add(0, gram.getContent());
return this;
}
@Override
public void visitRawHTML(RawHTMLMode mode, String raw) throws IOException {
throw new RuntimeException("Raw HTML is not supported");
}
@Override
public Visitor<IOException> visitVariationText(String[] variations) throws IOException {
throw new RuntimeException("Variations not supported");
}
@Override
public boolean visitEnd() throws IOException {
targetStack.remove(0);
return false;
}
@Override
public int visitElementTypes(String elementTypes) throws IOException {
return 0;
}
@Override
public Visitor<IOException> visitHeadline(int depth) throws IOException {
throw new RuntimeException("Headline in virtual verse is impossible");
}
@Override
public void visitStart() throws IOException {
}
@Override
public Visitor<IOException> visitCSSFormatting(String css) throws IOException {
STYLE x = f.createSTYLE();
x.setCss(css);
targetStack.get(0).add(new JAXBElement<STYLE>(new QName("STYLE"), STYLE.class, x));
targetStack.add(0, x.getContent());
return this;
}
@Override
public Visitor<IOException> visitExtraAttribute(ExtraAttributePriority prio, String category, String key, String value) throws IOException {
System.out.println("WARNING: Extra attributes not supported");
Visitor<IOException> result = prio.handleVisitor(category, this);
if (result != null)
targetStack.add(0, targetStack.get(0));
return result;
}
});
if (targetStack.size() != 0)
throw new RuntimeException();
}
}
}
if (book.getCHAPTER().size() == 0) {
doc.getBIBLEBOOK().remove(book);
}
}
final Document docc = DocumentBuilderFactory.newInstance().newDocumentBuilder().newDocument();
JAXBContext.newInstance(ObjectFactory.class.getPackage().getName()).createMarshaller().marshal(doc, docc);
docc.getDocumentElement().setAttribute("xmlns:xsi", "http://www.w3.org/2001/XMLSchema-instance");
docc.getDocumentElement().setAttribute("xsi:noNamespaceSchemaLocation", "zef2005.xsd");
docc.normalize();
maskWhitespaceNodes(docc.getDocumentElement());
try (FileOutputStream fos = new FileOutputStream(exportArgs[0])) {
Transformer transformer = TransformerFactory.newInstance().newTransformer();
transformer.setOutputProperty(OutputKeys.INDENT, "yes");
transformer.setOutputProperty("{http://xml.apache.org/xslt}indent-amount", "2");
transformer.transform(new DOMSource(docc), new StreamResult(fos));
}
}
Also used :
DOMSource(javax.xml.transform.dom.DOMSource)
VirtualVerse(biblemulticonverter.data.VirtualVerse)
ArrayList(java.util.ArrayList)
RawHTMLMode(biblemulticonverter.data.FormattedText.RawHTMLMode)
Document(org.w3c.dom.Document)
ObjectFactory(biblemulticonverter.schema.zef2005.ObjectFactory)
BookID(biblemulticonverter.data.BookID)
Book(biblemulticonverter.data.Book)
MetadataBook(biblemulticonverter.data.MetadataBook)
Headline(biblemulticonverter.data.FormattedText.Headline)
VERS(biblemulticonverter.schema.zef2005.VERS)
List(java.util.List)
ArrayList(java.util.ArrayList)
MetadataBook(biblemulticonverter.data.MetadataBook)
StreamResult(javax.xml.transform.stream.StreamResult)
STYLE(biblemulticonverter.schema.zef2005.STYLE)
JAXBElement(javax.xml.bind.JAXBElement)
FormattedText(biblemulticonverter.data.FormattedText)
BIBLEBOOK(biblemulticonverter.schema.zef2005.BIBLEBOOK)
LineBreakKind(biblemulticonverter.data.FormattedText.LineBreakKind)
XREF(biblemulticonverter.schema.zef2005.XREF)
FileOutputStream(java.io.FileOutputStream)
BigInteger(java.math.BigInteger)
FormattingInstructionKind(biblemulticonverter.data.FormattedText.FormattingInstructionKind)
ExtraAttributePriority(biblemulticonverter.data.FormattedText.ExtraAttributePriority)
Transformer(javax.xml.transform.Transformer)
Visitor(biblemulticonverter.data.FormattedText.Visitor)
BR(biblemulticonverter.schema.zef2005.BR)
NOTE(biblemulticonverter.schema.zef2005.NOTE)
MetadataBookKey(biblemulticonverter.data.MetadataBook.MetadataBookKey)
FormattingInstructionKind(biblemulticonverter.data.FormattedText.FormattingInstructionKind)
XMLBIBLE(biblemulticonverter.schema.zef2005.XMLBIBLE)
QName(javax.xml.namespace.QName)
Chapter(biblemulticonverter.data.Chapter)
FormattingInstructionKind(biblemulticonverter.data.FormattedText.FormattingInstructionKind)
IOException(java.io.IOException)
DIV(biblemulticonverter.schema.zef2005.DIV)
EnumCaptionType(biblemulticonverter.schema.zef2005.EnumCaptionType)
CHAPTER(biblemulticonverter.schema.zef2005.CHAPTER)
GRAM(biblemulticonverter.schema.zef2005.GRAM)
CAPTION(biblemulticonverter.schema.zef2005.CAPTION)
VirtualVerse(biblemulticonverter.data.VirtualVerse)
Verse(biblemulticonverter.data.Verse)
Example 4 with FormattingInstructionKind
use of biblemulticonverter.data.FormattedText.FormattingInstructionKind in project BibleMultiConverter by schierlm.
the class OSIS method parseStructuredTextElement.
private void parseStructuredTextElement(Visitor<RuntimeException> vv, Element elem) {
if (elem.getNodeName().equals("lb")) {
vv.visitLineBreak(LineBreakKind.NEWLINE);
} else if (elem.getNodeName().equals("brp")) {
vv.visitLineBreak(LineBreakKind.PARAGRAPH);
} else if (elem.getNodeName().equals("divineName")) {
parseStructuredTextChildren(vv.visitFormattingInstruction(FormattingInstructionKind.DIVINE_NAME), elem);
} else if (elem.getNodeName().equals("woj")) {
parseStructuredTextChildren(vv.visitFormattingInstruction(FormattingInstructionKind.WORDS_OF_JESUS), elem);
} else if (elem.getNodeName().equals("catchWord")) {
parseStructuredTextChildren(vv.visitCSSFormatting("osis-style: catchWord; font-style:italic;"), elem);
} else if (elem.getNodeName().equals("hi")) {
FormattingInstructionKind kind;
if (elem.getAttribute("type").equals("italic")) {
kind = FormattingInstructionKind.ITALIC;
} else if (elem.getAttribute("type").equals("bold")) {
kind = FormattingInstructionKind.BOLD;
} else {
kind = null;
printWarning("WARNING: Invalid hi type: " + elem.getAttribute("type"));
}
if (elem.getChildNodes().getLength() != 0) {
Visitor<RuntimeException> vv1 = kind == null ? vv : vv.visitFormattingInstruction(kind);
parseStructuredTextChildren(vv1, elem);
}
} else if (elem.getNodeName().equals("seg") || elem.getNodeName().equals("transChange") || elem.getNodeName().equals("rdg")) {
String css;
if (elem.getNodeName().equals("seg") && elem.getAttribute("type").equals("x-alternative")) {
css = "osis-style: alternative; color: gray;";
} else if (elem.getNodeName().equals("transChange") && elem.getAttribute("type").equals("added")) {
css = "osis-style: added; font-style:italic;";
} else if (elem.getNodeName().equals("transChange") && elem.getAttribute("type").equals("deleted")) {
css = "osis-style: deleted; text-decoration: line-through; color: gray;";
} else if (elem.getNodeName().equals("transChange") && elem.getAttribute("type").equals("amplified")) {
css = "osis-style: amplified; font-style: italic;";
} else if (elem.getNodeName().equals("transChange") && elem.getAttribute("type").isEmpty()) {
css = "osis-style: trans-change;";
} else if (elem.getNodeName().equals("rdg") && elem.getAttribute("type").equals("alternative")) {
css = "osis-style: alternative-reading; color: gray;";
} else if (elem.getNodeName().equals("rdg") && elem.getAttribute("type").equals("x-literal")) {
css = "osis-style: literal-reading; color: gray;";
} else if (elem.getNodeName().equals("rdg") && elem.getAttribute("type").equals("x-meaning")) {
css = "osis-style: meaning-reading; color: gray;";
} else if (elem.getNodeName().equals("rdg") && elem.getAttribute("type").equals("x-equivalent")) {
css = "osis-style: equivalent-reading; color: gray;";
} else if (elem.getNodeName().equals("rdg") && elem.getAttribute("type").equals("x-identity")) {
css = "osis-style: identity-reading; color: gray;";
} else if (elem.getNodeName().equals("rdg") && elem.getAttribute("type").isEmpty()) {
css = "osis-style: reading; color: gray;";
} else {
css = null;
printWarning("WARNING: Invalid " + elem.getNodeName() + " type: " + elem.getAttribute("type"));
}
if (elem.getChildNodes().getLength() != 0) {
Visitor<RuntimeException> vv1 = css == null ? vv : vv.visitCSSFormatting(css);
parseStructuredTextChildren(vv1, elem);
}
} else if (elem.getNodeName().equals("note")) {
if (elem.getAttribute("type").equals("crossReference")) {
Visitor<RuntimeException> fn = vv.visitFootnote();
fn.visitText(FormattedText.XREF_MARKER);
if (elem.getFirstChild() != null && elem.getFirstChild().getNodeName().equals("reference")) {
for (Node n = elem.getFirstChild(); n != null; n = n.getNextSibling()) {
if (n instanceof Text) {
fn.visitText(n.getTextContent());
continue;
}
Element e = (Element) n;
String[] ref = e.getAttribute("osisRef").split("\\.");
if (ref.length != 3) {
printWarning("WARNING: Invalid reference target: " + e.getAttribute("osisRef"));
fn.visitText(e.getTextContent());
continue;
}
BookID bookID = BookID.fromOsisId(ref[0]);
int ch = Integer.parseInt(ref[1]);
String vs = ref[2];
fn.visitCrossReference(ref[0], bookID, ch, vs, ch, vs).visitText(e.getTextContent());
}
} else if (elem.getTextContent().length() > 0) {
// OSIS converted from USFM contains a reference back to the verse itself
for (Node n = elem.getFirstChild(); n != null; n = n.getNextSibling()) {
if (n instanceof Element && n.getNodeName().equals("reference")) {
if (((Element) n).getAttribute("osisRef").equals(((Element) elem).getAttribute("osisRef"))) {
n = n.getPreviousSibling();
n.getParentNode().removeChild(n.getNextSibling());
}
}
}
boolean first = true;
for (String ref : elem.getTextContent().split("\\|")) {
Matcher m = XREF_PATTERN.matcher(ref);
if (!m.matches()) {
ref = ref.trim();
if (ref.startsWith("1 ") || ref.startsWith("2 ") || ref.startsWith("3 ")) {
ref = ref.substring(0, 1) + ref.substring(2);
}
m = XREF_PATTERN_2.matcher(ref);
if (m.matches()) {
try {
BookID.fromOsisId(m.group(1));
} catch (IllegalArgumentException ex) {
BookID bk = null;
for (BookID id : BookID.values()) {
if (id.getThreeLetterCode().equalsIgnoreCase(m.group(1)))
bk = id;
}
if (bk != null) {
m = XREF_PATTERN_2.matcher(bk.getOsisID() + " " + m.group(2) + "." + m.group(3));
} else {
m = XREF_PATTERN_2.matcher("");
}
}
}
}
if (!first)
fn.visitText("; ");
first = false;
if (m.matches()) {
String book = m.group(1);
BookID bookID = BookID.fromOsisId(book);
int ch = Integer.parseInt(m.group(2));
String vs = m.group(3);
fn.visitCrossReference(book, bookID, ch, vs, ch, vs).visitText(ref);
} else {
printWarning("WARNING: Malformed cross reference: " + ref);
fn.visitText(ref.replaceAll("[\r\n\t ]+", " ").trim());
}
}
} else {
printWarning("WARNING: crossReference without content");
fn.visitText("-");
}
} else if (elem.getFirstChild() != null) {
Visitor<RuntimeException> v = vv.visitFootnote();
parseStructuredTextChildren(v, elem);
}
} else if (elem.getNodeName().equals("w")) {
if (elem.getFirstChild() == null)
// skip empty w tags
return;
String src = elem.getAttribute("src");
Visitor<RuntimeException> v = vv;
int[] strong = null, idx = null;
List<Integer> strongList = new ArrayList<Integer>();
for (String lemma : elem.getAttribute("lemma").trim().split(" +")) {
if (!lemma.startsWith("strong:G") && !lemma.startsWith("strong:H"))
continue;
String rawStrong = lemma.substring(8);
if (!rawStrong.matches("0*[1-9][0-9]*(-0*[1-9][0-9]*)*")) {
printWarning("WARNING: Invalid strong dictionary entry: " + rawStrong);
continue;
}
String[] strs = rawStrong.split("-");
for (String str : strs) {
strongList.add(Integer.parseInt(str));
}
}
if (!strongList.isEmpty())
strong = strongList.stream().mapToInt(s -> s).toArray();
List<String> rmac = new ArrayList<>();
for (String morph : elem.getAttribute("morph").trim().split(" +")) {
if (morph.startsWith("robinson:")) {
String rmacCandidate = morph.substring(9);
if (Utils.compilePattern(Utils.RMAC_REGEX).matcher(rmacCandidate).matches()) {
rmac.add(rmacCandidate);
} else {
printWarning("WARNING: Invalid RMAC: " + rmacCandidate);
}
}
}
if (src.matches("[0-9]{2}( [0-9]{2})*")) {
String[] strs = src.split(" ");
idx = new int[strs.length];
for (int i = 0; i < strs.length; i++) {
idx[i] = Integer.parseInt(strs[i]);
}
}
if (strong == null && rmac.isEmpty() && idx == null) {
printWarning("INFO: Skipped <w> tag without any usable information");
} else {
v = v.visitGrammarInformation(strong, rmac.isEmpty() ? null : rmac.toArray(new String[rmac.size()]), idx);
}
parseStructuredTextChildren(v, elem);
} else if (elem.getNodeName().equals("reference")) {
String osisRef = elem.getAttribute("osisRef");
if (osisRef.contains("\u00A0")) {
printWarning("WARNING: osisRef contains Non-Breaking spaces: '" + osisRef + "'");
osisRef = osisRef.replace('\u00A0', ' ');
}
if (!osisRef.equals(osisRef.trim())) {
printWarning("WARNING: Removed whitespace from osisRef '" + osisRef + "' - replaced by '" + osisRef.trim() + "'");
osisRef = osisRef.trim();
}
Matcher fixupMatcher = Utils.compilePattern("([A-Z0-9][A-Z0-9a-z]+\\.[0-9]+\\.)([0-9]+)((?:[+-][0-9]+)+)").matcher(osisRef);
if (fixupMatcher.matches()) {
osisRef = fixupMatcher.group(1) + fixupMatcher.group(2);
for (String suffix : fixupMatcher.group(3).split("(?=[+-])")) {
if (suffix.isEmpty())
continue;
osisRef += suffix.substring(0, 1).replace('+', ' ') + fixupMatcher.group(1) + suffix.substring(1);
}
printWarning("INFO: Replaced osisRef " + elem.getAttribute("osisRef") + " by " + osisRef);
}
if (osisRefMap == null) {
osisRefMap = new Properties();
String path = System.getProperty("biblemulticonverter.osisrefmap", "");
if (!path.isEmpty()) {
try (FileInputStream fis = new FileInputStream(path)) {
osisRefMap.load(fis);
} catch (IOException ex) {
throw new RuntimeException(ex);
}
}
}
String mappedOsisRef = osisRefMap.getProperty(osisRef);
if (mappedOsisRef != null) {
printWarning("INFO: Replaced osisRef " + osisRef + " by " + mappedOsisRef + " (based on osisRef map)");
osisRef = mappedOsisRef;
}
if (osisRef.matches("[^ ]+ [^ ]+") && elem.getFirstChild() instanceof Text && elem.getFirstChild().getNextSibling() == null) {
String value = elem.getTextContent();
int lastPos = value.lastIndexOf('.');
if (lastPos != -1) {
Element newElem = elem.getOwnerDocument().createElement("reference");
newElem.setAttribute("osisRef", osisRef.split(" ")[0]);
newElem.appendChild(elem.getOwnerDocument().createTextNode(value.substring(0, lastPos)));
parseStructuredTextElement(vv, newElem);
vv.visitText(".");
newElem = elem.getOwnerDocument().createElement("reference");
newElem.setAttribute("osisRef", osisRef.split(" ")[1]);
newElem.appendChild(elem.getOwnerDocument().createTextNode(value.substring(lastPos + 1)));
parseStructuredTextElement(vv, newElem);
return;
}
}
Visitor<RuntimeException> v = vv;
if (osisRef.matches("[A-Z0-9][A-Z0-9a-z]+\\.[0-9]+\\.[0-9]+")) {
String[] osisRefParts = osisRef.split("\\.");
int chapter = Integer.parseInt(osisRefParts[1]);
try {
v = v.visitCrossReference(osisRefParts[0], BookID.fromOsisId(osisRefParts[0]), chapter, osisRefParts[2], chapter, osisRefParts[2]);
} catch (IllegalArgumentException ex) {
printWarning("WARNING: " + ex.getMessage());
}
} else if (osisRef.matches("([A-Z0-9][A-Z0-9a-z]+)\\.[0-9]+\\.[0-9]+-\\1\\.[0-9]+\\.[0-9]+")) {
String[] osisRefParts = osisRef.split("[.-]");
int firstChapter = Integer.parseInt(osisRefParts[1]);
int lastChapter = Integer.parseInt(osisRefParts[4]);
try {
v = v.visitCrossReference(osisRefParts[0], BookID.fromOsisId(osisRefParts[0]), firstChapter, osisRefParts[2], lastChapter, osisRefParts[5]);
} catch (IllegalArgumentException ex) {
printWarning("WARNING: " + ex.getMessage());
}
} else {
printWarning("WARNING: Unsupported osisRef: " + osisRef);
}
parseStructuredTextChildren(v, elem);
} else if (elem.getNodeName().equals("variation")) {
parseStructuredTextChildren(vv.visitVariationText(new String[] { elem.getAttribute("name") }), elem);
} else {
printWarning("WARNING: invalid structured element level tag: " + elem.getNodeName());
}
}
Also used :
ObjectFactory(biblemulticonverter.schema.roundtripxml.ObjectFactory)
DOMSource(javax.xml.transform.dom.DOMSource)
Text(org.w3c.dom.Text)
LineBreakKind(biblemulticonverter.data.FormattedText.LineBreakKind)
Arrays(java.util.Arrays)
XPath(javax.xml.xpath.XPath)
XPathConstants(javax.xml.xpath.XPathConstants)
StreamResult(javax.xml.transform.stream.StreamResult)
Bible(biblemulticonverter.data.Bible)
MetadataBook(biblemulticonverter.data.MetadataBook)
MetadataBookKey(biblemulticonverter.data.MetadataBook.MetadataBookKey)
ArrayList(java.util.ArrayList)
HashSet(java.util.HashSet)
FormattedText(biblemulticonverter.data.FormattedText)
Matcher(java.util.regex.Matcher)
ValidateXML(biblemulticonverter.tools.ValidateXML)
Document(org.w3c.dom.Document)
Node(org.w3c.dom.Node)
Utils(biblemulticonverter.data.Utils)
XMLConstants(javax.xml.XMLConstants)
Book(biblemulticonverter.data.Book)
VirtualVerse(biblemulticonverter.data.VirtualVerse)
RawHTMLMode(biblemulticonverter.data.FormattedText.RawHTMLMode)
Properties(java.util.Properties)
SchemaFactory(javax.xml.validation.SchemaFactory)
NodeList(org.w3c.dom.NodeList)
FormattingInstructionKind(biblemulticonverter.data.FormattedText.FormattingInstructionKind)
Verse(biblemulticonverter.data.Verse)
Headline(biblemulticonverter.data.FormattedText.Headline)
Set(java.util.Set)
IOException(java.io.IOException)
FileInputStream(java.io.FileInputStream)
Comment(org.w3c.dom.Comment)
File(java.io.File)
BookID(biblemulticonverter.data.BookID)
List(java.util.List)
Element(org.w3c.dom.Element)
DocumentBuilder(javax.xml.parsers.DocumentBuilder)
Pattern(java.util.regex.Pattern)
DocumentBuilderFactory(javax.xml.parsers.DocumentBuilderFactory)
TransformerFactory(javax.xml.transform.TransformerFactory)
ExtraAttributePriority(biblemulticonverter.data.FormattedText.ExtraAttributePriority)
Chapter(biblemulticonverter.data.Chapter)
Visitor(biblemulticonverter.data.FormattedText.Visitor)
Visitor(biblemulticonverter.data.FormattedText.Visitor)
Matcher(java.util.regex.Matcher)
Node(org.w3c.dom.Node)
Element(org.w3c.dom.Element)
ArrayList(java.util.ArrayList)
FormattingInstructionKind(biblemulticonverter.data.FormattedText.FormattingInstructionKind)
Text(org.w3c.dom.Text)
FormattedText(biblemulticonverter.data.FormattedText)
IOException(java.io.IOException)
Properties(java.util.Properties)
FileInputStream(java.io.FileInputStream)
BookID(biblemulticonverter.data.BookID)
Example 5 with FormattingInstructionKind
use of biblemulticonverter.data.FormattedText.FormattingInstructionKind in project BibleMultiConverter by schierlm.
the class YCHPalmBible method doExport.
@Override
public void doExport(Bible bible, String... exportArgs) throws Exception {
String filename = exportArgs[0];
String description = bible.getName();
MetadataBook metadata = bible.getMetadataBook();
if (metadata != null) {
String metaDescription = bible.getMetadataBook().getValue(MetadataBookKey.description);
if (metaDescription != null)
description = metaDescription;
}
try (final BufferedWriter bw = new BufferedWriter(new OutputStreamWriter(new FileOutputStream(filename), "windows-1252"))) {
bw.write("<PARSERINFO ENCODE=\"Cp1252\" WORDTYPE=\"SPCSEP\">");
bw.newLine();
bw.write("<BIBLE NAME=\"" + bible.getName() + "\" INFO=\"" + description + "\">");
bw.newLine();
Visitor<IOException> contentVisitor = new FormattedText.VisitorAdapter<IOException>(null) {
@Override
public void visitVerseSeparator() throws IOException {
// strip
}
@Override
public void visitText(String text) throws IOException {
bw.write(text);
}
@Override
public Visitor<IOException> visitFormattingInstruction(FormattingInstructionKind kind) throws IOException {
return this;
}
@Override
public Visitor<IOException> visitCSSFormatting(String css) throws IOException {
return this;
}
@Override
public void visitLineBreak(LineBreakKind kind) throws IOException {
bw.write(" ");
}
};
for (Book bk : bible.getBooks()) {
int zefID = bk.getId().getZefID();
if (zefID < 1 || zefID >= PALM_BOOK_NUMBERS.length || PALM_BOOK_NUMBERS[zefID] == 0) {
System.out.println("WARNING: Skipping unsupported book " + bk.getAbbr() + " (" + bk.getId().getOsisID() + ")");
continue;
}
bw.write("<BOOK NAME=\"" + bk.getShortName() + "\" NUMBER=\"" + PALM_BOOK_NUMBERS[bk.getId().getZefID()] + "\" SHORTCUT=\"" + bk.getAbbr() + "\">");
bw.newLine();
String longtitle = bk.getLongName();
int chapter = 0, verse;
for (Chapter chap : bk.getChapters()) {
chapter++;
if (chap.getProlog() != null)
System.out.println("WARNING: Skipping prolog (prologs not supported)!");
verse = 1;
String chaptext = CHAPTER_NAME + " " + chapter;
bw.write("<CHAPTER>");
bw.newLine();
for (VirtualVerse v : chap.createVirtualVerses()) {
while (v.getNumber() > verse) {
bw.write("<VERSE></VERSE>");
bw.newLine();
verse++;
}
if (v.getNumber() != verse)
throw new RuntimeException("Verse is " + v.getNumber() + ", should be " + verse);
boolean needVersText = false;
bw.write("<VERSE>");
if (longtitle != null) {
bw.write("<BOOKTEXT>" + longtitle);
longtitle = null;
needVersText = true;
}
if (chaptext != null) {
bw.write("<CHAPTEXT>" + chaptext);
chaptext = null;
needVersText = true;
}
for (Headline hl : v.getHeadlines()) {
bw.write("<DESCTEXT>");
hl.accept(contentVisitor);
needVersText = true;
}
if (needVersText)
bw.write("<VERSTEXT>");
for (Verse vv : v.getVerses()) {
if (!vv.getNumber().equals("" + v.getNumber())) {
bw.write("{" + vv.getNumber() + "} ");
}
vv.accept(contentVisitor);
}
bw.write("</VERSE>");
verse++;
bw.newLine();
}
bw.write("</CHAPTER>");
bw.newLine();
}
bw.write("</BOOK>");
bw.newLine();
}
bw.write("</BIBLE>");
bw.newLine();
}
}
Also used :
MetadataBook(biblemulticonverter.data.MetadataBook)
VirtualVerse(biblemulticonverter.data.VirtualVerse)
Chapter(biblemulticonverter.data.Chapter)
FormattingInstructionKind(biblemulticonverter.data.FormattedText.FormattingInstructionKind)
IOException(java.io.IOException)
BufferedWriter(java.io.BufferedWriter)
LineBreakKind(biblemulticonverter.data.FormattedText.LineBreakKind)
MetadataBook(biblemulticonverter.data.MetadataBook)
Book(biblemulticonverter.data.Book)
FileOutputStream(java.io.FileOutputStream)
Headline(biblemulticonverter.data.FormattedText.Headline)
OutputStreamWriter(java.io.OutputStreamWriter)
VirtualVerse(biblemulticonverter.data.VirtualVerse)
Verse(biblemulticonverter.data.Verse)
Aggregations
FormattingInstructionKind (biblemulticonverter.data.FormattedText.FormattingInstructionKind)8
Visitor (biblemulticonverter.data.FormattedText.Visitor)6
BookID (biblemulticonverter.data.BookID)5
IOException (java.io.IOException)5
List (java.util.List)5
JAXBElement (javax.xml.bind.JAXBElement)5
Book (biblemulticonverter.data.Book)4
FormattedText (biblemulticonverter.data.FormattedText)4
LineBreakKind (biblemulticonverter.data.FormattedText.LineBreakKind)4
MetadataBook (biblemulticonverter.data.MetadataBook)4
ArrayList (java.util.ArrayList)4
Chapter (biblemulticonverter.data.Chapter)3
ExtraAttributePriority (biblemulticonverter.data.FormattedText.ExtraAttributePriority)3
Headline (biblemulticonverter.data.FormattedText.Headline)3
RawHTMLMode (biblemulticonverter.data.FormattedText.RawHTMLMode)3
Verse (biblemulticonverter.data.Verse)3
VirtualVerse (biblemulticonverter.data.VirtualVerse)3
BR (biblemulticonverter.schema.zef2005.BR)3
DIV (biblemulticonverter.schema.zef2005.DIV)3
GRAM (biblemulticonverter.schema.zef2005.GRAM)3
