Examples with Headline biblemulticonverter.data.FormattedText.Headline used on opensource projects

Example 21 with Headline

use of biblemulticonverter.data.FormattedText.Headline in project BibleMultiConverter by schierlm.

the class MyBibleZone method doImport.

@Override
public Bible doImport(File inputFile) throws Exception {
    SqlJetDb db = SqlJetDb.open(inputFile, false);
    SqlJetDb footnoteDB = null;
    File footnoteFile = new File(inputFile.getParentFile(), inputFile.getName().replace(".SQLite3", ".commentaries.SQLite3"));
    if (inputFile.getName().endsWith(".SQLite3") && footnoteFile.exists()) {
        footnoteDB = SqlJetDb.open(footnoteFile, false);
        if (!footnoteDB.getTable("commentaries").getIndexesNames().contains("commentaries_index")) {
            footnoteDB.close();
            footnoteDB = SqlJetDb.open(footnoteFile, true);
            checkIndex(footnoteDB, "commentaries", "commentaries_index", "CREATE INDEX commentaries_index on commentaries(book_number, chapter_number_from, verse_number_from)");
        }
        footnoteDB.beginTransaction(SqlJetTransactionMode.READ_ONLY);
    }
    if (!db.getTable("verses").getIndexesNames().contains("versesIndex") || (db.getSchema().getTable("stories") != null && !db.getTable("stories").getIndexesNames().contains("stories_index"))) {
        db.close();
        db = SqlJetDb.open(inputFile, true);
        checkIndex(db, "verses", "verses_index", "CREATE UNIQUE INDEX verses_index on verses (book_number, chapter, verse)");
        if (db.getSchema().getTable("stories") != null)
            if (db.getSchema().getTable("stories").getColumn("order_if_several") == null)
                checkIndex(db, "stories", "stories_index", "CREATE UNIQUE INDEX stories_index on stories(book_number, chapter, verse)");
            else
                checkIndex(db, "stories", "stories_index", "CREATE UNIQUE INDEX stories_index on stories(book_number, chapter, verse, order_if_several)");
    }
    db.beginTransaction(SqlJetTransactionMode.READ_ONLY);
    String bibleName = null;
    MetadataBook mb = new MetadataBook();
    ISqlJetCursor cursor = db.getTable("info").open();
    while (!cursor.eof()) {
        String fn = cursor.getString("name");
        String fv = cursor.getString("value");
        if (fn.equals("description")) {
            bibleName = fv;
        } else if (!fv.isEmpty()) {
            fv = fv.replaceAll("[\r\n]+", "\n").replaceAll(" *\n *", "\n").replaceAll("\n$", "");
            try {
                mb.setValue("MyBible.zone@" + fn.replace('_', '.'), fv);
            } catch (IllegalArgumentException ex) {
                System.out.println("WARNING: Skipping malformed metadata property " + fn);
            }
        }
        cursor.next();
    }
    cursor.close();
    if (bibleName == null) {
        System.out.println("WARNING: No bible name in info table");
        bibleName = inputFile.getName();
    }
    Bible result = new Bible(bibleName.trim());
    if (!mb.getKeys().isEmpty()) {
        mb.finished();
        result.getBooks().add(mb.getBook());
    }
    Map<Integer, Book> bookIDMap = new HashMap<>();
    cursor = db.getTable("books").open();
    while (!cursor.eof()) {
        int num = (int) cursor.getInteger("book_number");
        String col = cursor.getString("book_color");
        String shortName = cursor.getString("short_name").trim().replace(" ", "").replaceAll("[^A-Z0-9a-zäöü]++", "");
        if (!shortName.isEmpty())
            shortName = shortName.substring(0, 1).toUpperCase() + shortName.substring(1);
        String longName = cursor.getString("long_name").trim();
        BookID bid = null;
        for (MyBibleZoneBook bi : BOOK_INFO) {
            if (bi.bookNumber == num) {
                bid = bi.bookID;
                if (!col.equals(bi.bookColor))
                    System.out.println("WARNING: Book " + bid.getOsisID() + " uses color " + col + " and not " + bi.bookColor);
            }
        }
        if (bid == null) {
            System.out.println("WARNING: Book number " + num + " unknown; skipping: " + shortName + "/" + longName);
            // generate dummy entry not stored in result object
            bookIDMap.put(num, new Book("Xxx", BookID.BOOK_Gen, "X", "X"));
        } else {
            if (shortName.length() < 2)
                shortName = bid.getOsisID().replaceAll("[^A-Z0-9a-zäöü]++", "");
            Book bk = new Book(shortName, bid, longName, longName);
            result.getBooks().add(bk);
            bookIDMap.put(num, bk);
        }
        cursor.next();
    }
    cursor.close();
    if (db.getSchema().getTable("introductions") != null) {
        cursor = db.getTable("introductions").open();
        while (!cursor.eof()) {
            int num = (int) cursor.getInteger("book_number");
            String intro = cursor.getString("introduction");
            Book bk;
            if (num == 0) {
                bk = new Book("Intro", BookID.INTRODUCTION, "_Introduction_", "_Introduction_");
                if (!result.getBooks().isEmpty() && result.getBooks().get(0).getId().equals(BookID.METADATA)) {
                    result.getBooks().add(1, bk);
                } else {
                    result.getBooks().add(0, bk);
                }
            } else {
                bk = bookIDMap.get(num);
            }
            if (bk == null) {
                System.out.println("WARNING: Skipping introduction for nonexisting book " + num);
            } else {
                FormattedText ft = new FormattedText();
                convertFromHTML(intro, ft.getAppendVisitor());
                ft.finished();
                if (bk.getChapters().isEmpty())
                    bk.getChapters().add(new Chapter());
                bk.getChapters().get(0).setProlog(ft);
            }
            cursor.next();
        }
        cursor.close();
    }
    cursor = db.getTable("verses").order("verses_index");
    while (!cursor.eof()) {
        int b = (int) cursor.getInteger("book_number");
        int c = (int) cursor.getInteger("chapter");
        int v = (int) cursor.getInteger("verse");
        String text = cursor.getString("text");
        if (text == null)
            text = "";
        text = text.trim();
        if (!text.isEmpty()) {
            Book bk = bookIDMap.get(b);
            if (bk == null) {
                System.out.println("WARNING: Verse for unknown book " + b + " skipped");
            } else {
                while (bk.getChapters().size() < c) bk.getChapters().add(new Chapter());
                Chapter ch = bk.getChapters().get(c - 1);
                Verse vv = new Verse("" + v);
                try {
                    String rest = convertFromVerse(text, vv.getAppendVisitor(), footnoteDB, new int[] { b, c, v });
                    if (!rest.isEmpty()) {
                        System.out.println("WARNING: Treating tags as plaintext: " + rest);
                        vv.getAppendVisitor().visitText(rest.replace('\t', ' ').replaceAll("  +", " "));
                    }
                } catch (RuntimeException ex) {
                    throw new RuntimeException(text, ex);
                }
                ch.getVerses().add(vv);
                vv.finished();
            }
        }
        cursor.next();
    }
    cursor.close();
    if (db.getSchema().getTable("stories") != null) {
        cursor = db.getTable("stories").order("stories_index");
        Map<Verse, List<FormattedText.Headline>> subheadings = new HashMap<>();
        Map<Verse, Chapter> subheadingChapters = new HashMap<>();
        while (!cursor.eof()) {
            int b = (int) cursor.getInteger("book_number");
            int c = (int) cursor.getInteger("chapter");
            int v = (int) cursor.getInteger("verse");
            String title = cursor.getString("title").trim();
            Book bk = bookIDMap.get(b);
            if (bk == null) {
                System.out.println("WARNING: Subheading for unknown book " + b + " skipped");
            } else if (bk.getChapters().size() < c) {
                System.out.println("WARNING: Subheading for unknown chapter " + b + " " + c + " skipped");
            } else {
                Chapter ch = bk.getChapters().get(c - 1);
                Verse vv = null;
                for (Verse vvv : ch.getVerses()) {
                    if (vvv.getNumber().equals("" + v))
                        vv = vvv;
                }
                if (vv == null) {
                    System.out.println("WARNING: Subheading for unknown verse " + b + " " + c + ":" + v + " skipped");
                } else {
                    List<FormattedText.Headline> hls = subheadings.get(vv);
                    if (hls == null) {
                        hls = new ArrayList<>();
                        subheadings.put(vv, hls);
                        subheadingChapters.put(vv, ch);
                    }
                    Headline hl = new Headline(1);
                    while (title.contains("<x>")) {
                        int pos = title.indexOf("<x>");
                        hl.getAppendVisitor().visitText(title.substring(0, pos));
                        title = title.substring(pos + 3);
                        pos = title.indexOf("</x>");
                        if (pos == -1)
                            System.out.println("WARNING: Unclosed cross reference: " + title);
                        else {
                            String ref = title.substring(0, pos);
                            title = title.substring(pos + 4);
                            hl.getAppendVisitor().visitFormattingInstruction(FormattingInstructionKind.BOLD).visitText(ref);
                        }
                    }
                    hl.getAppendVisitor().visitText(title);
                    hl.finished();
                    hls.add(hl);
                }
            }
            cursor.next();
        }
        cursor.close();
        for (Verse vv : subheadings.keySet()) {
            Chapter cc = subheadingChapters.get(vv);
            Verse vnew = new Verse(vv.getNumber());
            for (Headline hl : subheadings.get(vv)) {
                hl.accept(vnew.getAppendVisitor().visitHeadline(hl.getDepth()));
            }
            vv.accept(vnew.getAppendVisitor());
            vnew.finished();
            int pos = cc.getVerses().indexOf(vv);
            cc.getVerses().set(pos, vnew);
        }
    }
    if (footnoteDB != null) {
        footnoteDB.commit();
        footnoteDB.close();
    }
    db.commit();
    db.close();
    return result;
}

Example 22 with Headline

use of biblemulticonverter.data.FormattedText.Headline in project BibleMultiConverter by schierlm.

the class AbstractParatextFormat method doImport.

@Override
public Bible doImport(File inputFile) throws Exception {
    List<ParatextBook> books = doImportBooks(inputFile);
    String bibleName = null;
    for (ParatextBook book : books) {
        if (bibleName == null || book.getBibleName().isEmpty()) {
            bibleName = book.getBibleName();
        } else {
            String bookBibleName = book.getBibleName();
            // use common suffix
            if (bookBibleName.length() > bibleName.length()) {
                bookBibleName = bookBibleName.substring(bookBibleName.length() - bibleName.length());
            } else if (bibleName.length() > bookBibleName.length()) {
                bibleName = bibleName.substring(bibleName.length() - bookBibleName.length());
            }
            for (int i = bibleName.length() - 1; i >= 0; i--) {
                if (bibleName.charAt(i) != bookBibleName.charAt(i)) {
                    bibleName = bibleName.substring(i + 1);
                    break;
                }
            }
        }
    }
    Bible bible = new Bible((bibleName == null || bibleName.isEmpty()) ? "Imported Bible" : bibleName);
    for (ParatextBook book : books) {
        String longName = book.getAttributes().get("toc1");
        if (longName == null || longName.isEmpty())
            longName = book.getId().getEnglishName();
        String shortName = book.getAttributes().get("toc2");
        if (shortName == null || shortName.isEmpty())
            shortName = longName;
        String abbr = book.getAttributes().get("toc3"), fallbackAbbr = book.getId().getId().getOsisID().replace("x-", "").replace("-", "");
        if (abbr == null)
            abbr = fallbackAbbr;
        abbr = abbr.replace(" ", "");
        if (!Utils.compilePattern(Utils.BOOK_ABBR_REGEX).matcher(abbr).matches()) {
            System.out.println("WARNING: Unsupported book abbreviation " + abbr + ", using " + fallbackAbbr + " instead");
            abbr = fallbackAbbr;
        }
        final Book bk = new Book(abbr, book.getId().getId(), shortName, longName);
        bible.getBooks().add(bk);
        final boolean forceProlog = book.getId().getId().getZefID() < 0;
        final ParatextImportContext ctx = new ParatextImportContext();
        ctx.nt = book.getId().getId().isNT();
        book.accept(new ParatextBookContentVisitor<RuntimeException>() {

            @Override
            public void visitChapterStart(int newChapter) throws RuntimeException {
                if (ctx.cnum != -1 && !ctx.headlines.isEmpty()) {
                    System.out.println("WARNING: Ignoring unreferenced headlines");
                    ctx.headlines.clear();
                }
                if (ctx.cnum == 0 && newChapter == 1) {
                    // we are in prolog (chapter already exists)
                    ctx.cnum = newChapter;
                } else if (newChapter >= 1 && newChapter > ctx.cnum) {
                    if (ctx.cnum == -1)
                        ctx.cnum = 0;
                    while (ctx.cnum < newChapter - 1) {
                        bk.getChapters().add(new Chapter());
                        ctx.cnum++;
                    }
                    ctx.currentChapter = new Chapter();
                    bk.getChapters().add(ctx.currentChapter);
                    ctx.cnum = newChapter;
                } else {
                    System.out.println("WARNING: Ignoring chapter number " + newChapter + ", current chapter is " + ctx.cnum);
                }
                ctx.currentVisitor = null;
                ctx.currentVerse = null;
                ctx.currentParagraph = ParatextImportContext.CurrentParagraph.NONE;
            }

            @Override
            public void visitParagraphStart(ParagraphKind kind) throws RuntimeException {
                if (ctx.currentParagraph != ParatextImportContext.CurrentParagraph.NONE) {
                    if (ctx.currentParagraph == ParatextImportContext.CurrentParagraph.PROLOG || (ctx.currentParagraph == ParatextImportContext.CurrentParagraph.NORMAL && ctx.currentVisitor != null)) {
                        ctx.currentVisitor.visitLineBreak(LineBreakKind.PARAGRAPH);
                    }
                    ctx.currentParagraph = ParatextImportContext.CurrentParagraph.NONE;
                }
                if (kind.getCategory() == ParagraphKindCategory.SKIP) {
                // do nothing
                } else if (kind.getCategory() == ParagraphKindCategory.HEADLINE) {
                    Headline hl = null;
                    if (kind.isJoinHeadlines() && !ctx.headlines.isEmpty()) {
                        hl = ctx.headlines.get(ctx.headlines.size() - 1);
                        if (hl.getDepth() == kind.getHeadlineDepth() || kind.getHeadlineDepth() == 0) {
                            hl.getAppendVisitor().visitText(" ");
                        } else {
                            hl = null;
                        }
                    }
                    if (hl == null) {
                        hl = new Headline(kind.getHeadlineDepth());
                        ctx.headlines.add(hl);
                    }
                    ctx.currentParagraph = ParatextImportContext.CurrentParagraph.HEADLINE;
                    ctx.currentVisitor = hl.getAppendVisitor();
                    if (kind.getExtraFormatting() != null) {
                        ctx.currentVisitor = ctx.currentVisitor.visitFormattingInstruction(kind.getExtraFormatting());
                    }
                } else {
                    // BLANK_LINE, TABLE_ROW, TEXT
                    if (kind.isProlog() || forceProlog) {
                        if (ctx.cnum == -1) {
                            ctx.cnum = 0;
                            ctx.currentChapter = new Chapter();
                            bk.getChapters().add(ctx.currentChapter);
                        }
                        if (ctx.currentChapter.getProlog() == null) {
                            ctx.currentChapter.setProlog(new FormattedText());
                        }
                        if (!ctx.currentChapter.getVerses().isEmpty()) {
                            System.out.println("WARNING: Adding to prolog after verses have been added!");
                        }
                        ctx.currentVisitor = ctx.currentChapter.getProlog().getAppendVisitor();
                        ctx.currentParagraph = ParatextImportContext.CurrentParagraph.PROLOG;
                        ctx.flushHeadlines();
                    } else {
                        ctx.currentParagraph = ParatextImportContext.CurrentParagraph.NORMAL;
                    }
                }
            }

            @Override
            public void visitTableCellStart(String tag) throws RuntimeException {
                ctx.ensureParagraph();
                if (!tag.matches("t[hc]r?1") && ctx.currentParagraph != ParatextImportContext.CurrentParagraph.HEADLINE && ctx.currentVisitor != null) {
                    ctx.currentVisitor.visitLineBreak(LineBreakKind.NEWLINE_WITH_INDENT);
                }
            }

            @Override
            public void visitParatextCharacterContent(ParatextCharacterContent content) throws RuntimeException {
                ctx.ensureParagraph();
                content.accept(new ParatextImportVisitor(ctx));
            }
        });
        if (!ctx.headlines.isEmpty()) {
            System.out.println("WARNING: Ignoring unreferenced headlines");
            ctx.headlines.clear();
        }
        for (Chapter ch : bk.getChapters()) {
            if (ch.getProlog() != null)
                ch.getProlog().finished();
            for (Verse v : ch.getVerses()) v.finished();
        }
    }
    return bible;
}

Example 23 with Headline

use of biblemulticonverter.data.FormattedText.Headline in project BibleMultiConverter by schierlm.

the class VirtualVerse method validate.

public void validate(Bible bible, BookID book, String bookAbbr, int cnumber, List<String> danglingReferences, Map<String, Set<String>> dictionaryEntries) {
    int lastHeadlineDepth = 0;
    String location = bookAbbr + " " + cnumber + ":v" + getNumber();
    for (Headline headline : headlines) {
        if (headline.getDepth() <= lastHeadlineDepth)
            throw new IllegalStateException("Invalid headline depth order at " + location + ": " + headline.getDepth() + " after " + lastHeadlineDepth);
        lastHeadlineDepth = headline.getDepth() == 9 ? 8 : headline.getDepth();
        headline.validate(bible, book, location + ":Headline", danglingReferences, dictionaryEntries);
    }
    Set<String> verseNumbers = new HashSet<String>();
    for (Verse verse : verses) {
        if (!verseNumbers.add(verse.getNumber()))
            throw new IllegalStateException("Duplicate verse number");
        verse.validate(bible, book, location + ":" + verse.getNumber(), danglingReferences, dictionaryEntries);
    }
}

Example 24 with Headline

use of biblemulticonverter.data.FormattedText.Headline in project BibleMultiConverter by schierlm.

the class Chapter method createVirtualVerses.

public List<VirtualVerse> createVirtualVerses() {
    // split up verses to separate headlines
    final List<VirtualVerse> tempVerses = new ArrayList<VirtualVerse>();
    BitSet numericVerseNumbers = new BitSet(verses.size());
    for (final Verse verse : verses) {
        int num;
        try {
            num = Integer.parseInt(verse.getNumber());
            numericVerseNumbers.set(num);
        } catch (NumberFormatException ex) {
            // ignore nonnumeric verse numbers
            num = Integer.MAX_VALUE;
        }
        final int vnum = num;
        verse.accept(new VisitorAdapter<RuntimeException>(null) {

            VirtualVerse vv = new VirtualVerse(vnum);

            boolean hasContent = false;

            {
                tempVerses.add(vv);
                vv.getVerses().add(new Verse(verse.getNumber()));
            }

            @Override
            public Visitor<RuntimeException> visitHeadline(int depth) {
                Headline h = new Headline(depth);
                if (hasContent) {
                    vv = new VirtualVerse(vnum);
                    tempVerses.add(vv);
                    vv.getVerses().add(new Verse(verse.getNumber()));
                    hasContent = false;
                }
                vv.getHeadlines().add(h);
                return h.getAppendVisitor();
            }

            @Override
            public int visitElementTypes(String elementTypes) throws RuntimeException {
                return 0;
            }

            @Override
            public void visitStart() {
                hasContent = true;
            }

            @Override
            public boolean visitEnd() throws RuntimeException {
                hasContent = true;
                return false;
            }

            @Override
            protected void beforeVisit() {
                hasContent = true;
            }

            @Override
            protected Visitor<RuntimeException> getVisitor() {
                return vv.getVerses().get(0).getAppendVisitor();
            }
        });
    }
    // group verses sensibly
    List<VirtualVerse> result = new ArrayList<VirtualVerse>();
    VirtualVerse current = null;
    int nextverse = 1;
    for (VirtualVerse vv : tempVerses) {
        for (Headline h : vv.getHeadlines()) h.finished();
        for (Verse v : vv.getVerses()) v.finished();
        boolean makeNew;
        if (current == null || vv.getHeadlines().size() > 0) {
            makeNew = true;
            if (vv.getNumber() != Integer.MAX_VALUE && vv.getNumber() > nextverse) {
                nextverse = vv.getNumber();
            }
        } else if (vv.getNumber() == Integer.MAX_VALUE) {
            makeNew = false;
        } else {
            // numeric verse without headlines; may be both as new verse and
            // as appended one;
            // decide based on verse number
            int vnum = vv.getNumber();
            if (vnum < nextverse) {
                makeNew = false;
            } else if (vnum > nextverse + 1 && numericVerseNumbers.nextSetBit(nextverse) < vnum) {
                makeNew = false;
                numericVerseNumbers.clear(vnum);
            } else {
                makeNew = true;
                nextverse = vnum;
            }
        }
        if (makeNew) {
            current = new VirtualVerse(nextverse);
            current.getHeadlines().addAll(vv.getHeadlines());
            for (Verse vvv : vv.getVerses()) {
                if (vvv.getElementTypes(1).length() > 0)
                    current.getVerses().add(vvv);
            }
            result.add(current);
            nextverse++;
        } else {
            for (Verse vvv : vv.getVerses()) {
                if (vvv.getElementTypes(1).length() > 0)
                    current.getVerses().add(vvv);
            }
        }
    }
    return result;
}