pdfreader.java

来自「有关对pdf操作的代码」· Java 代码 · 共 1,757 行 · 第 1/5 页

        }
    }

    /**
     * @return the percentage of the cross reference table that has been read
     */
    public double dumpPerc() {
        int total = 0;
        for (int k = 0; k < xrefObj.size(); ++k) {
            if (xrefObj.get(k) != null)
                ++total;
        }
        return (total * 100.0 / xrefObj.size());
    }

    protected void readDocObj() throws IOException {
        ArrayList streams = new ArrayList();
        xrefObj = new ArrayList(xref.length / 2);
        xrefObj.addAll(Collections.nCopies(xref.length / 2, null));
        for (int k = 2; k < xref.length; k += 2) {
            int pos = xref[k];
            if (pos <= 0 || xref[k + 1] > 0)
                continue;
            tokens.seek(pos);
            tokens.nextValidToken();
            if (tokens.getTokenType() != PRTokeniser.TK_NUMBER)
                tokens.throwError("Invalid object number.");
            objNum = tokens.intValue();
            tokens.nextValidToken();
            if (tokens.getTokenType() != PRTokeniser.TK_NUMBER)
                tokens.throwError("Invalid generation number.");
            objGen = tokens.intValue();
            tokens.nextValidToken();
            if (!tokens.getStringValue().equals("obj"))
                tokens.throwError("Token 'obj' expected.");
            PdfObject obj;
            try {
                obj = readPRObject();
                if (obj.isStream()) {
                    streams.add(obj);
                }
            }
            catch (Exception e) {
                obj = null;
            }
            xrefObj.set(k / 2, obj);
        }
        for (int k = 0; k < streams.size(); ++k) {
            checkPRStreamLength((PRStream)streams.get(k));
        }
        readDecryptedDocObj();
        if (objStmMark != null) {
            for (Iterator i = objStmMark.entrySet().iterator(); i.hasNext();) {
                Map.Entry entry = (Map.Entry)i.next();
                int n = ((Integer)entry.getKey()).intValue();
                IntHashtable h = (IntHashtable)entry.getValue();
                readObjStm((PRStream)xrefObj.get(n), h);
                xrefObj.set(n, null);
            }
            objStmMark = null;
        }
        xref = null;
    }

    private void checkPRStreamLength(PRStream stream) throws IOException {
        int fileLength = tokens.length();
        int start = stream.getOffset();
        boolean calc = false;
        int streamLength = 0;
        PdfObject obj = getPdfObjectRelease(stream.get(PdfName.LENGTH));
        if (obj != null && obj.type() == PdfObject.NUMBER) {
            streamLength = ((PdfNumber)obj).intValue();
            if (streamLength + start > fileLength - 20)
                calc = true;
            else {
                tokens.seek(start + streamLength);
                String line = tokens.readString(20);
                if (!line.startsWith("\nendstream") &&
                !line.startsWith("\r\nendstream") &&
                !line.startsWith("\rendstream") &&
                !line.startsWith("endstream"))
                    calc = true;
            }
        }
        else
            calc = true;
        if (calc) {
            byte tline[] = new byte[16];
            tokens.seek(start);
            while (true) {
                int pos = tokens.getFilePointer();
                if (!tokens.readLineSegment(tline))
                    break;
                if (equalsn(tline, endstream)) {
                    streamLength = pos - start;
                    break;
                }
                if (equalsn(tline, endobj)) {
                    tokens.seek(pos - 16);
                    String s = tokens.readString(16);
                    int index = s.indexOf("endstream");
                    if (index >= 0)
                        pos = pos - 16 + index;
                    streamLength = pos - start;
                    break;
                }
            }
        }
        stream.setLength(streamLength);
    }

    protected void readObjStm(PRStream stream, IntHashtable map) throws IOException {
        int first = ((PdfNumber)getPdfObject(stream.get(PdfName.FIRST))).intValue();
        int n = ((PdfNumber)getPdfObject(stream.get(PdfName.N))).intValue();
        byte b[] = getStreamBytes(stream, tokens.getFile());
        PRTokeniser saveTokens = tokens;
        tokens = new PRTokeniser(b);
        try {
            int address[] = new int[n];
            int objNumber[] = new int[n];
            boolean ok = true;
            for (int k = 0; k < n; ++k) {
                ok = tokens.nextToken();
                if (!ok)
                    break;
                if (tokens.getTokenType() != PRTokeniser.TK_NUMBER) {
                    ok = false;
                    break;
                }
                objNumber[k] = tokens.intValue();
                ok = tokens.nextToken();
                if (!ok)
                    break;
                if (tokens.getTokenType() != PRTokeniser.TK_NUMBER) {
                    ok = false;
                    break;
                }
                address[k] = tokens.intValue() + first;
            }
            if (!ok)
                throw new IOException("Error reading ObjStm");
            for (int k = 0; k < n; ++k) {
                if (map.containsKey(k)) {
                    tokens.seek(address[k]);
                    PdfObject obj = readPRObject();
                    xrefObj.set(objNumber[k], obj);
                }
            }
        }
        finally {
            tokens = saveTokens;
        }
    }

    /**
     * Eliminates the reference to the object freeing the memory used by it and clearing
     * the xref entry.
     * @param obj the object. If it's an indirect reference it will be eliminated
     * @return the object or the already erased dereferenced object
     */
    public static PdfObject killIndirect(PdfObject obj) {
        if (obj == null || obj.isNull())
            return null;
        PdfObject ret = getPdfObjectRelease(obj);
        if (obj.isIndirect()) {
            PRIndirectReference ref = (PRIndirectReference)obj;
            PdfReader reader = ref.getReader();
            int n = ref.getNumber();
            reader.xrefObj.set(n, null);
            if (reader.partial)
                reader.xref[n * 2] = -1;
        }
        return ret;
    }

    private void ensureXrefSize(int size) {
        if (size == 0)
            return;
        if (xref == null)
            xref = new int[size];
        else {
            if (xref.length < size) {
                int xref2[] = new int[size];
                System.arraycopy(xref, 0, xref2, 0, xref.length);
                xref = xref2;
            }
        }
    }

    protected void readXref() throws IOException {
        hybridXref = false;
        newXrefType = false;
        tokens.seek(tokens.getStartxref());
        tokens.nextToken();
        if (!tokens.getStringValue().equals("startxref"))
            throw new IOException("startxref not found.");
        tokens.nextToken();
        if (tokens.getTokenType() != PRTokeniser.TK_NUMBER)
            throw new IOException("startxref is not followed by a number.");
        int startxref = tokens.intValue();
        lastXref = startxref;
        eofPos = tokens.getFilePointer();
        try {
            if (readXRefStream(startxref)) {
                newXrefType = true;
                return;
            }
        }
        catch (Exception e) {}
        xref = null;
        tokens.seek(startxref);
        trailer = readXrefSection();
        PdfDictionary trailer2 = trailer;
        while (true) {
            PdfNumber prev = (PdfNumber)trailer2.get(PdfName.PREV);
            if (prev == null)
                break;
            tokens.seek(prev.intValue());
            trailer2 = readXrefSection();
        }
    }

    protected PdfDictionary readXrefSection() throws IOException {
        tokens.nextValidToken();
        if (!tokens.getStringValue().equals("xref"))
            tokens.throwError("xref subsection not found");
        int start = 0;
        int end = 0;
        int pos = 0;
        int gen = 0;
        while (true) {
            tokens.nextValidToken();
            if (tokens.getStringValue().equals("trailer"))
                break;
            if (tokens.getTokenType() != PRTokeniser.TK_NUMBER)
                tokens.throwError("Object number of the first object in this xref subsection not found");
            start = tokens.intValue();
            tokens.nextValidToken();
            if (tokens.getTokenType() != PRTokeniser.TK_NUMBER)
                tokens.throwError("Number of entries in this xref subsection not found");
            end = tokens.intValue() + start;
            if (start == 1) { // fix incorrect start number
                int back = tokens.getFilePointer();
                tokens.nextValidToken();
                pos = tokens.intValue();
                tokens.nextValidToken();
                gen = tokens.intValue();
                if (pos == 0 && gen == 65535) {
                    --start;
                    --end;
                }
                tokens.seek(back);
            }
            ensureXrefSize(end * 2);
            for (int k = start; k < end; ++k) {
                tokens.nextValidToken();
                pos = tokens.intValue();
                tokens.nextValidToken();
                gen = tokens.intValue();
                tokens.nextValidToken();
                int p = k * 2;
                if (tokens.getStringValue().equals("n")) {
                    if (xref[p] == 0 && xref[p + 1] == 0) {
//                        if (pos == 0)
//                            tokens.throwError("File position 0 cross-reference entry in this xref subsection");
                        xref[p] = pos;
                    }
                }
                else if (tokens.getStringValue().equals("f")) {
                    if (xref[p] == 0 && xref[p + 1] == 0)
                        xref[p] = -1;
                }
                else
                    tokens.throwError("Invalid cross-reference entry in this xref subsection");
            }
        }
        PdfDictionary trailer = (PdfDictionary)readPRObject();
        PdfNumber xrefSize = (PdfNumber)trailer.get(PdfName.SIZE);
        ensureXrefSize(xrefSize.intValue() * 2);
        PdfObject xrs = trailer.get(PdfName.XREFSTM);
        if (xrs != null && xrs.isNumber()) {
            int loc = ((PdfNumber)xrs).intValue();
            try {
                readXRefStream(loc);
                newXrefType = true;
                hybridXref = true;
            }
            catch (IOException e) {
                xref = null;
                throw e;
            }
        }
        return trailer;
    }

    protected boolean readXRefStream(int ptr) throws IOException {
        tokens.seek(ptr);
        int thisStream = 0;
        if (!tokens.nextToken())
            return false;
        if (tokens.getTokenType() != PRTokeniser.TK_NUMBER)
            return false;
        thisStream = tokens.intValue();
        if (!tokens.nextToken() || tokens.getTokenType() != PRTokeniser.TK_NUMBER)
            return false;
        if (!tokens.nextToken() || !tokens.getStringValue().equals("obj"))
            return false;
        PdfObject object = readPRObject();
        PRStream stm = null;
        if (object.isStream()) {
            stm = (PRStream)object;
            if (!PdfName.XREF.equals(stm.get(PdfName.TYPE)))
                return false;
        }
        else
            return false;
        if (trailer == null) {
            trailer = new PdfDictionary();
            trailer.putAll(stm);
        }
        stm.setLength(((PdfNumber)stm.get(PdfName.LENGTH)).intValue());
        int size = ((PdfNumber)stm.get(PdfName.SIZE)).intValue();
        PdfArray index;
        PdfObject obj = stm.get(PdfName.INDEX);
        if (obj == null) {
            index = new PdfArray();
            index.add(new int[]{0, size});
        }
        else
            index = (PdfArray)obj;
        PdfArray w = (PdfArray)stm.get(PdfName.W);
        int prev = -1;
        obj = stm.get(PdfName.PREV);
        if (obj != null)
            prev = ((PdfNumber)obj).intValue();
        // Each xref pair is a position
        // type 0 -> -1, 0
        // type 1 -> offset, 0
        // type 2 -> index, obj num
        ensureXrefSize(size * 2);
        if (objStmMark == null && !partial)
            objStmMark = new HashMap();
        if (objStmToOffset == null && partial)
            objStmToOffset = new IntHashtable();
        byte b[] = getStreamBytes(stm, tokens.getFile());
        int bptr = 0;
        ArrayList wa = w.getArrayList();
        int wc[] = new int[3];
        for (int k = 0; k < 3; ++k)
            wc[k] = ((PdfNumber)wa.get(k)).intValue();
        ArrayList sections = index.getArrayList();
        for (int idx = 0; idx < sections.size(); idx += 2) {