void lt_XMLParser::Impl::ChangeText( const int width, const int height, DjVuFile &dfile, const lt_XMLTags &tags ) { dfile.resume_decode(true); GP<DjVuText> text = DjVuText::create(); GP<DjVuTXT> txt = text->txt = DjVuTXT::create(); // to store the new text GP<ByteStream> textbs = ByteStream::create(); GP<DjVuInfo> info=(dfile.info); if(info) { const int h=info->height; const int w=info->width; txt->page_zone.text_start = 0; DjVuTXT::Zone &parent=txt->page_zone; parent.rect.xmin=0; parent.rect.ymin=0; parent.rect.ymax=h; parent.rect.xmax=w; double ws=1.0; if(width && width != w) { ws=((double)w)/((double)width); } double hs=1.0; if(height && height != h) { hs=((double)h)/((double)height); } make_child_layer(parent, tags, *textbs, h, ws,hs); textbs->write8(0); long len = textbs->tell(); txt->page_zone.text_length = len; textbs->seek(0,SEEK_SET); textbs->read(txt->textUTF8.getbuf(len), len); dfile.change_text(txt,false); } }
PageInfo DjVuSource::ReadPageInfo(int nPage, bool bNeedText, bool bNeedAnno) { ASSERT(nPage >= 0 && nPage < m_nPageCount); PageInfo pageInfo; pageInfo.szPage.cx = 100; pageInfo.szPage.cy = 100; pageInfo.nDPI = 100; pageInfo.bDecoded = true; GP<ByteStream> pAnnoStream; if (bNeedAnno) pAnnoStream = ByteStream::create(); GP<ByteStream> pTextStream; if (bNeedText) pTextStream = ByteStream::create(); try { // Get raw data from the document and decode only requested chunks // DjVuFile is not used to ensure that we do not wait for a lock // to be released and thus do not block the UI thread GURL url = m_pDjVuDoc->page_to_url(nPage); GP<DataPool> pool = m_pDjVuDoc->request_data(NULL, url); GP<ByteStream> stream = pool->get_stream(); GP<IFFByteStream> iff(IFFByteStream::create(stream)); // Check file format GUTF8String chkid; if (!iff->get_chunk(chkid) || (chkid != "FORM:DJVI" && chkid != "FORM:DJVU" && chkid != "FORM:PM44" && chkid != "FORM:BM44")) { return pageInfo; } bool bHasIW44 = false; // Find chunk with page info while (iff->get_chunk(chkid) != 0) { GP<ByteStream> chunk_stream = iff->get_bytestream(); if (chkid == "INFO") { // Get page dimensions and resolution from info chunk GP<DjVuInfo> pInfo = DjVuInfo::create(); pInfo->decode(*chunk_stream); // Check data for consistency pageInfo.szPage.cx = max(pInfo->width, 0); pageInfo.szPage.cy = max(pInfo->height, 0); pageInfo.nInitialRotate = pInfo->orientation; pageInfo.nDPI = max(pInfo->dpi, 0); if ((pInfo->orientation & 1) != 0) swap(pageInfo.szPage.cx, pageInfo.szPage.cy); } else if (!bHasIW44 && (chkid == "PM44" || chkid == "BM44")) { bHasIW44 = true; // Get image dimensions and resolution from bitmap chunk UINT serial = chunk_stream->read8(); UINT slices = chunk_stream->read8(); UINT major = chunk_stream->read8(); UINT minor = chunk_stream->read8(); UINT xhi = chunk_stream->read8(); UINT xlo = chunk_stream->read8(); UINT yhi = chunk_stream->read8(); UINT ylo = chunk_stream->read8(); pageInfo.szPage.cx = (xhi << 8) | xlo; pageInfo.szPage.cy = (yhi << 8) | ylo; pageInfo.nDPI = 100; } else if (chkid == "TXTa" || chkid == "TXTz") { pageInfo.bHasText = true; if (bNeedText) { const GP<IFFByteStream> iffout = IFFByteStream::create(pTextStream); iffout->put_chunk(chkid); iffout->copy(*chunk_stream); iffout->close_chunk(); } } else if (bNeedAnno && chkid == "FORM:ANNO") { pAnnoStream->copy(*chunk_stream); } else if (bNeedAnno && (chkid == "ANTa" || chkid == "ANTz")) { const GP<IFFByteStream> iffout = IFFByteStream::create(pAnnoStream); iffout->put_chunk(chkid); iffout->copy(*chunk_stream); iffout->close_chunk(); } else if (bNeedAnno && chkid == "INCL") { set<GUTF8String> processed; ReadAnnotations(chunk_stream, processed, pAnnoStream); } iff->seek_close_chunk(); } if (bNeedText && pTextStream->tell()) pageInfo.DecodeText(pTextStream); if (bNeedAnno && pAnnoStream->tell()) pageInfo.DecodeAnno(pAnnoStream); } catch (GException&) { } catch (...) { if (pApplication != NULL) pApplication->ReportFatalError(); } return pageInfo; }