You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
tdegraphics/kviewshell/plugins/djvu/libdjvu/DjVuDocEditor.cpp

2193 lines
68 KiB

//C- -------------------------------------------------------------------
//C- DjVuLibre-3.5
//C- Copyright (c) 2002 Leon Bottou and Yann Le Cun.
//C- Copyright (c) 2001 AT&T
//C-
//C- This software is subject to, and may be distributed under, the
//C- GNU General Public License, Version 2. The license should have
//C- accompanied the software or you may obtain a copy of the license
//C- from the Free Software Foundation at http://www.fsf.org .
//C-
//C- This program is distributed in the hope that it will be useful,
//C- but WITHOUT ANY WARRANTY; without even the implied warranty of
//C- MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
//C- GNU General Public License for more details.
//C-
//C- DjVuLibre-3.5 is derived from the DjVu(r) Reference Library
//C- distributed by Lizardtech Software. On July 19th 2002, Lizardtech
//C- Software authorized us to replace the original DjVu(r) Reference
//C- Library notice by the following text (see doc/lizard2002.djvu):
//C-
//C- ------------------------------------------------------------------
//C- | DjVu (r) Reference Library (v. 3.5)
//C- | Copyright (c) 1999-2001 LizardTech, Inc. All Rights Reserved.
//C- | The DjVu Reference Library is protected by U.S. Pat. No.
//C- | 6,058,214 and patents pending.
//C- |
//C- | This software is subject to, and may be distributed under, the
//C- | GNU General Public License, Version 2. The license should have
//C- | accompanied the software or you may obtain a copy of the license
//C- | from the Free Software Foundation at http://www.fsf.org .
//C- |
//C- | The computer code originally released by LizardTech under this
//C- | license and unmodified by other parties is deemed "the LIZARDTECH
//C- | ORIGINAL CODE." Subject to any third party intellectual property
//C- | claims, LizardTech grants recipient a worldwide, royalty-free,
//C- | non-exclusive license to make, use, sell, or otherwise dispose of
//C- | the LIZARDTECH ORIGINAL CODE or of programs derived from the
//C- | LIZARDTECH ORIGINAL CODE in compliance with the terms of the GNU
//C- | General Public License. This grant only confers the right to
//C- | infringe patent claims underlying the LIZARDTECH ORIGINAL CODE to
//C- | the extent such infringement is reasonably necessary to enable
//C- | recipient to make, have made, practice, sell, or otherwise dispose
//C- | of the LIZARDTECH ORIGINAL CODE (or portions thereof) and not to
//C- | any greater extent that may be necessary to utilize further
//C- | modifications or combinations.
//C- |
//C- | The LIZARDTECH ORIGINAL CODE is provided "AS IS" WITHOUT WARRANTY
//C- | OF ANY KIND, EITHER EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED
//C- | TO ANY WARRANTY OF NON-INFRINGEMENT, OR ANY IMPLIED WARRANTY OF
//C- | MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE.
//C- +------------------------------------------------------------------
//
// $Id: DjVuDocEditor.cpp,v 1.13 2005/05/25 20:24:52 leonb Exp $
// $Name: release_3_5_15 $
#ifdef HAVE_CONFIG_H
# include "config.h"
#endif
#if NEED_GNUG_PRAGMAS
# pragma implementation
#endif
#include "DjVuDocEditor.h"
#include "DjVuImage.h"
#include "IFFByteStream.h"
#include "DataPool.h"
#include "IW44Image.h"
#include "GOS.h"
#include "GURL.h"
#include "DjVuAnno.h"
#include "GRect.h"
#include "DjVmNav.h"
#include "debug.h"
#include <ctype.h>
#ifdef HAVE_NAMESPACES
namespace DJVU {
# ifdef NOT_DEFINED // Just to fool emacs c++ mode
}
#endif
#endif
static const char octets[4]={0x41,0x54,0x26,0x54};
int DjVuDocEditor::thumbnails_per_file=10;
// This is a structure for active files and DataPools. It may contain
// a DjVuFile, which is currently being used by someone (I check the list
// and get rid of hanging files from time to time) or a DataPool,
// which is "custom" with respect to the document (was modified or
// inserted), or both.
//
// DjVuFile is set to smth!=0 when it's created using url_to_file().
// It's reset back to ZERO in clean_files_map() when
// it sees, that a given file is not used by anyone.
// DataPool is updated when a file is inserted
class DjVuDocEditor::File : public GPEnabled
{
public:
// 'pool' below may be non-zero only if it cannot be retrieved
// by the DjVuDocument, that is it either corresponds to a
// modified DjVuFile or it has been inserted. Otherwise it's ZERO
// Once someone assigns a non-zero DataPool, it remains non-ZERO
// (may be updated if the file gets modified) and may be reset
// only by save() or save_as() functions.
GP<DataPool> pool;
// If 'file' is non-zero, it means, that it's being used by someone
// We check for unused files from time to time and ZERO them.
// But before we do it, we may save the DataPool in the case if
// file has been modified.
GP<DjVuFile> file;
};
void
DjVuDocEditor::check(void)
{
if (!initialized) G_THROW( ERR_MSG("DjVuDocEditor.not_init") );
}
DjVuDocEditor::DjVuDocEditor(void)
{
initialized=false;
refresh_cb=0;
refresh_cl_data=0;
}
DjVuDocEditor::~DjVuDocEditor(void)
{
if (!tmp_doc_url.is_empty())
{
tmp_doc_url.deletefile();
}
GCriticalSectionLock lock(&thumb_lock);
thumb_map.empty();
DataPool::close_all();
}
void
DjVuDocEditor::init(void)
{
DEBUG_MSG("DjVuDocEditor::init() called\n");
DEBUG_MAKE_INDENT(3);
// If you remove this check be sure to delete thumb_map
if (initialized) G_THROW( ERR_MSG("DjVuDocEditor.init") );
doc_url=GURL::Filename::UTF8("noname.djvu");
const GP<DjVmDoc> doc(DjVmDoc::create());
const GP<ByteStream> gstr(ByteStream::create());
doc->write(gstr);
gstr->seek(0, SEEK_SET);
doc_pool=DataPool::create(gstr);
orig_doc_type=UNKNOWN_TYPE;
orig_doc_pages=0;
initialized=true;
DjVuDocument::init(doc_url, this);
}
void
DjVuDocEditor::init(const GURL &url)
{
DEBUG_MSG("DjVuDocEditor::init() called: url='" << url << "'\n");
DEBUG_MAKE_INDENT(3);
// If you remove this check be sure to delete thumb_map
if (initialized)
G_THROW( ERR_MSG("DjVuDocEditor.init") );
// First - create a temporary DjVuDocument and check its type
doc_pool=DataPool::create(url);
doc_url=url;
const GP<DjVuDocument> tmp_doc(DjVuDocument::create_wait(doc_url,this));
if (!tmp_doc->is_init_ok())
G_THROW( ERR_MSG("DjVuDocEditor.open_fail") "\t" +url.get_string());
orig_doc_type=tmp_doc->get_doc_type();
orig_doc_pages=tmp_doc->get_pages_num();
if (orig_doc_type==OLD_BUNDLED ||
orig_doc_type==OLD_INDEXED ||
orig_doc_type==SINGLE_PAGE)
{
// Suxx. I need to convert it NOW.
// We will unlink this file in the destructor
tmp_doc_url=GURL::Filename::Native(tmpnam(0));
const GP<ByteStream> gstr(ByteStream::create(tmp_doc_url, "wb"));
tmp_doc->write(gstr, true); // Force DJVM format
gstr->flush();
doc_pool=DataPool::create(tmp_doc_url);
}
// OK. Now doc_pool contains data of the document in one of the
// new formats. It will be a lot easier to insert/delete pages now.
// 'doc_url' below of course doesn't refer to the file with the converted
// data, but we will take care of it by redirecting the request_data().
initialized=true;
DjVuDocument::init(doc_url, this);
// Cool. Now extract the thumbnails...
GCriticalSectionLock lock(&thumb_lock);
int pages_num=get_pages_num();
for(int page_num=0;page_num<pages_num;page_num++)
{
// Call DjVuDocument::get_thumbnail() here to bypass logic
// of DjVuDocEditor::get_thumbnail(). init() is the only safe
// place where we can still call DjVuDocument::get_thumbnail();
const GP<DataPool> pool(DjVuDocument::get_thumbnail(page_num, true));
if (pool)
{
thumb_map[page_to_id(page_num)]=pool;
}
}
// And remove then from DjVmDir so that DjVuDocument
// does not try to use them
unfile_thumbnails();
}
GP<DataPool>
DjVuDocEditor::request_data(const DjVuPort * source, const GURL & url)
{
DEBUG_MSG("DjVuDocEditor::request_data(): url='" << url << "'\n");
DEBUG_MAKE_INDENT(3);
// Check if we have either original data or converted (to new format),
// if all the story is about the DjVuDocument's data
if (url==doc_url)
return doc_pool;
// Now see if we have any file matching the url
const GP<DjVmDir::File> frec(djvm_dir->name_to_file(url.fname()));
if (frec)
{
GCriticalSectionLock lock(&files_lock);
GPosition pos;
if (files_map.contains(frec->get_load_name(), pos))
{
const GP<File> f(files_map[pos]);
if (f->file && f->file->get_init_data_pool())
return f->file->get_init_data_pool();// Favor DjVuFile's knowledge
else if (f->pool) return f->pool;
}
}
// Finally let DjVuDocument cope with it. It may be a connected DataPool
// for a BUNDLED format. Or it may be a file. Anyway, it was not
// manually included, so it should be in the document.
const GP<DataPool> pool(DjVuDocument::request_data(source, url));
// We do NOT update the 'File' structure, because our rule is that
// we keep a separate copy of DataPool in 'File' only if it cannot
// be retrieved from DjVuDocument (like it has been "inserted" or
// corresponds to a modified file).
return pool;
}
void
DjVuDocEditor::clean_files_map(void)
// Will go thru the map of files looking for unreferenced
// files or records w/o DjVuFile and DataPool.
// These will be modified and/or removed.
{
DEBUG_MSG("DjVuDocEditor::clean_files_map() called\n");
DEBUG_MAKE_INDENT(3);
GCriticalSectionLock lock(&files_lock);
// See if there are too old items in the "cache", which are
// not referenced by anyone. If the corresponding DjVuFile has been
// modified, obtain the new data and replace the 'pool'. Clear the
// DjVuFile anyway. If both DataPool and DjVuFile are zero, remove
// the entry.
for(GPosition pos=files_map;pos;)
{
const GP<File> f(files_map[pos]);
if (f->file && f->file->get_count()==1)
{
DEBUG_MSG("ZEROing file '" << f->file->get_url() << "'\n");
if (f->file->is_modified())
f->pool=f->file->get_djvu_data(false);
f->file=0;
}
if (!f->file && !f->pool)
{
DEBUG_MSG("Removing record '" << files_map.key(pos) << "'\n");
GPosition this_pos=pos;
++pos;
files_map.del(this_pos);
} else ++pos;
}
}
GP<DjVuFile>
DjVuDocEditor::url_to_file(const GURL & url, bool dont_create) const
{
DEBUG_MSG("DjVuDocEditor::url_to_file(): url='" << url << "'\n");
DEBUG_MAKE_INDENT(3);
// Check if have a DjVuFile with this url cached (created before
// and either still active or left because it has been modified)
GP<DjVmDir::File> frec;
if((const DjVmDir *)djvm_dir)
frec=djvm_dir->name_to_file(url.fname());
if (frec)
{
GCriticalSectionLock lock(&(const_cast<DjVuDocEditor *>(this)->files_lock));
GPosition pos;
if (files_map.contains(frec->get_load_name(), pos))
{
const GP<File> f(files_map[pos]);
if (f->file)
return f->file;
}
}
const_cast<DjVuDocEditor *>(this)->clean_files_map();
// We don't have the file cached. Let DjVuDocument create the file.
const GP<DjVuFile> file(DjVuDocument::url_to_file(url, dont_create));
// And add it to our private "cache"
if (file && frec)
{
GCriticalSectionLock lock(&(const_cast<DjVuDocEditor *>(this)->files_lock));
GPosition pos;
if (files_map.contains(frec->get_load_name(), pos))
{
files_map[frec->get_load_name()]->file=file;
}else
{
const GP<File> f(new File());
f->file=file;
const_cast<DjVuDocEditor *>(this)->files_map[frec->get_load_name()]=f;
}
}
return file;
}
GUTF8String
DjVuDocEditor::page_to_id(int page_num) const
{
if (page_num<0 || page_num>=get_pages_num())
G_THROW( ERR_MSG("DjVuDocEditor.page_num") "\t"+GUTF8String(page_num));
const GP<DjVmDir::File> f(djvm_dir->page_to_file(page_num));
if (! f)
G_THROW( ERR_MSG("DjVuDocEditor.page_num") "\t"+GUTF8String(page_num));
return f->get_load_name();
}
GUTF8String
DjVuDocEditor::find_unique_id(GUTF8String id)
{
const GP<DjVmDir> dir(get_djvm_dir());
GUTF8String base, ext;
const int dot=id.rsearch('.');
if(dot >= 0)
{
base=id.substr(0,dot);
ext=id.substr(dot+1,(unsigned int)-1);
}else
{
base=id;
}
int cnt=0;
while (!(!dir->id_to_file(id) &&
!dir->name_to_file(id) &&
!dir->title_to_file(id)))
{
cnt++;
id=base+"_"+GUTF8String(cnt);
if (ext.length())
id+="."+ext;
}
return id;
}
GP<DataPool>
DjVuDocEditor::strip_incl_chunks(const GP<DataPool> & pool_in)
{
DEBUG_MSG("DjVuDocEditor::strip_incl_chunks() called\n");
DEBUG_MAKE_INDENT(3);
const GP<IFFByteStream> giff_in(
IFFByteStream::create(pool_in->get_stream()));
const GP<ByteStream> gbs_out(ByteStream::create());
const GP<IFFByteStream> giff_out(IFFByteStream::create(gbs_out));
IFFByteStream &iff_in=*giff_in;
IFFByteStream &iff_out=*giff_out;
bool have_incl=false;
int chksize;
GUTF8String chkid;
if (iff_in.get_chunk(chkid))
{
iff_out.put_chunk(chkid);
while((chksize=iff_in.get_chunk(chkid)))
{
if (chkid!="INCL")
{
iff_out.put_chunk(chkid);
iff_out.copy(*iff_in.get_bytestream());
iff_out.close_chunk();
} else
{
have_incl=true;
}
iff_in.close_chunk();
}
iff_out.close_chunk();
}
if (have_incl)
{
gbs_out->seek(0,SEEK_SET);
return DataPool::create(gbs_out);
} else return pool_in;
}
GUTF8String
DjVuDocEditor::insert_file(const GURL &file_url, const GUTF8String &parent_id,
int chunk_num, DjVuPort *source)
// Will open the 'file_name' and insert it into an existing DjVuFile
// with ID 'parent_id'. Will insert the INCL chunk at position chunk_num
// Will NOT process ANY files included into the file being inserted.
// Moreover it will strip out any INCL chunks in that file...
{
DEBUG_MSG("DjVuDocEditor::insert_file(): fname='" << file_url <<
"', parent_id='" << parent_id << "'\n");
DEBUG_MAKE_INDENT(3);
const GP<DjVmDir> dir(get_djvm_dir());
if(!source)
source=this;
// Create DataPool and see if the file exists
GP<DataPool> file_pool;
if(file_url.is_empty()||file_url.is_local_file_url())
{
file_pool=DataPool::create(file_url);
}else
{
file_pool=source->request_data(source, file_url);
if(source != this)
{
file_pool=DataPool::create(file_pool->get_stream()->duplicate());
}
}
if(file_pool && file_url && DjVuDocument::djvu_import_codec)
{
(*DjVuDocument::djvu_import_codec)(file_pool,file_url,needs_compression_flag,can_compress_flag);
}
// Strip any INCL chunks
file_pool=strip_incl_chunks(file_pool);
// Check if parent ID is valid
GP<DjVmDir::File> parent_frec(dir->id_to_file(parent_id));
if (!parent_frec)
parent_frec=dir->name_to_file(parent_id);
if (!parent_frec)
parent_frec=dir->title_to_file(parent_id);
if (!parent_frec)
G_THROW( ERR_MSG("DjVuDocEditor.no_file") "\t" +parent_id);
const GP<DjVuFile> parent_file(get_djvu_file(parent_id));
if (!parent_file)
G_THROW( ERR_MSG("DjVuDocEditor.create_fail") "\t"+parent_id);
// Now obtain ID for the new file
const GUTF8String id(find_unique_id(file_url.fname()));
// Add it into the directory
const GP<DjVmDir::File> frec(
DjVmDir::File::create(id, id, id, DjVmDir::File::INCLUDE));
int pos=dir->get_file_pos(parent_frec);
if (pos>=0)
++pos;
dir->insert_file(frec, pos);
// Add it to our "cache"
{
const GP<File> f(new File);
f->pool=file_pool;
GCriticalSectionLock lock(&files_lock);
files_map[id]=f;
}
// And insert it into the parent DjVuFile
parent_file->insert_file(id, chunk_num);
return id;
}
// First it will insert the 'file_url' at position 'file_pos'.
//
// Then it will process all the INCL chunks in the file and try to do
// the same thing with the included files. If insertion of an included
// file fails, it will proceed with other INCL chunks until it does
// them all. In the very end we will throw exception to let the caller
// know about problems with included files.
//
// If the name of a file being inserted conflicts with some other
// name, which has been in DjVmDir prior to call to this function,
// it will be modified. name2id is the translation table to
// keep track of these modifications.
//
// Also, if a name is in name2id, we will not insert that file again.
//
// Will return TRUE if the file has been successfully inserted.
// FALSE, if the file contains NDIR chunk and has been skipped.
bool
DjVuDocEditor::insert_file(const GURL &file_url, bool is_page,
int & file_pos, GMap<GUTF8String, GUTF8String> & name2id,
DjVuPort *source)
{
DEBUG_MSG("DjVuDocEditor::insert_file(): file_url='" << file_url <<
"', is_page='" << is_page << "'\n");
DEBUG_MAKE_INDENT(3);
if (refresh_cb)
refresh_cb(refresh_cl_data);
// We do not want to insert the same file twice (important when
// we insert a group of files at the same time using insert_group())
// So we check if we already did that and return if so.
if (name2id.contains(file_url.fname()))
return true;
if(!source)
source=this;
GP<DataPool> file_pool;
if(file_url.is_empty()||file_url.is_local_file_url())
{
file_pool=DataPool::create(file_url);
}
else
{
file_pool=source->request_data(source, file_url);
if(source != this)
{
file_pool=DataPool::create(file_pool->get_stream());
}
}
// Create DataPool and see if the file exists
if(file_pool && !file_url.is_empty() && DjVuDocument::djvu_import_codec)
{
(*DjVuDocument::djvu_import_codec)(file_pool,file_url,
needs_compression_flag,
can_compress_flag);
}
// Oh. It does exist... Check that it has IFF structure
{
const GP<IFFByteStream> giff(
IFFByteStream::create(file_pool->get_stream()));
IFFByteStream &iff=*giff;
GUTF8String chkid;
int length;
length=iff.get_chunk(chkid);
if (chkid!="FORM:DJVI" && chkid!="FORM:DJVU" &&
chkid!="FORM:BM44" && chkid!="FORM:PM44")
G_THROW( ERR_MSG("DjVuDocEditor.not_1_page") "\t"+file_url.get_string());
// Wonderful. It's even a DjVu file. Scan for NDIR chunks.
// If NDIR chunk is found, ignore the file
while(iff.get_chunk(chkid))
{
if (chkid=="NDIR")
return false;
iff.close_chunk();
}
}
return insert_file(file_pool,file_url,is_page,file_pos,name2id,source);
}
bool
DjVuDocEditor::insert_file(const GP<DataPool> &file_pool,
const GURL &file_url, bool is_page,
int & file_pos, GMap<GUTF8String, GUTF8String> & name2id,
DjVuPort *source)
{
GUTF8String errors;
if(file_pool)
{
const GP<DjVmDir> dir(get_djvm_dir());
G_TRY
{
// Now get a unique name for this file.
// Check the name2id first...
const GUTF8String name=file_url.fname();
GUTF8String id;
if (name2id.contains(name))
{
id=name2id[name];
}else
{
// Check to see if this page exists with a different name.
if(!is_page)
{
GPList<DjVmDir::File> list(dir->get_files_list());
for(GPosition pos=list;pos;++pos)
{
DEBUG_MSG("include " << list[pos]->is_include()
<< " size=" << list[pos]->size << " length="
<< file_pool->get_length() << "\n");
if(list[pos]->is_include()
&& (!list[pos]->size
|| (list[pos]->size == file_pool->get_length())))
{
id=list[pos]->get_load_name();
GP<DjVuFile> file(get_djvu_file(id,false));
const GP<DataPool> pool(file->get_djvu_data(false));
if(file_pool->simple_compare(*pool))
{
// The files are the same, so just store the alias.
name2id[name]=id;
}
const GP<IFFByteStream> giff_old(IFFByteStream::create(pool->get_stream()));
const GP<IFFByteStream> giff_new(IFFByteStream::create(file_pool->get_stream()));
file=0;
if(giff_old->compare(*giff_new))
{
// The files are the same, so just store the alias.
name2id[name]=id;
return true;
}
}
}
}
// Otherwise create a new unique ID and remember the translation
id=find_unique_id(name);
name2id[name]=id;
}
// Good. Before we continue with the included files we want to
// complete insertion of this one. Notice, that insertion of
// children may fail, in which case we will have to modify
// data for this file to get rid of invalid INCL
// Create a file record with the chosen ID
const GP<DjVmDir::File> file(DjVmDir::File::create(id, id, id,
is_page ? DjVmDir::File::PAGE : DjVmDir::File::INCLUDE ));
// And insert it into the directory
file_pos=dir->insert_file(file, file_pos);
// And add the File record (containing the file URL and DataPool)
{
const GP<File> f(new File);
f->pool=file_pool;
GCriticalSectionLock lock(&files_lock);
files_map[id]=f;
}
// The file has been added. If it doesn't include anything else,
// that will be enough. Otherwise repeat what we just did for every
// included child. Don't forget to modify the contents of INCL
// chunks due to name2id translation.
// We also want to include here our file with shared annotations,
// if it exists.
GUTF8String chkid;
const GP<IFFByteStream> giff_in(
IFFByteStream::create(file_pool->get_stream()));
IFFByteStream &iff_in=*giff_in;
const GP<ByteStream> gstr_out(ByteStream::create());
const GP<IFFByteStream> giff_out(IFFByteStream::create(gstr_out));
IFFByteStream &iff_out=*giff_out;
const GP<DjVmDir::File> shared_frec(djvm_dir->get_shared_anno_file());
iff_in.get_chunk(chkid);
iff_out.put_chunk(chkid);
while(iff_in.get_chunk(chkid))
{
if (chkid!="INCL")
{
iff_out.put_chunk(chkid);
iff_out.copy(*iff_in.get_bytestream());
iff_in.close_chunk();
iff_out.close_chunk();
if (shared_frec && chkid=="INFO")
{
iff_out.put_chunk("INCL");
iff_out.get_bytestream()->writestring(shared_frec->get_load_name());
iff_out.close_chunk();
}
} else
{
GUTF8String name;
char buffer[1024];
int length;
while((length=iff_in.read(buffer, 1024)))
name+=GUTF8String(buffer, length);
while(isspace(name[0]))
{
name=name.substr(1,(unsigned int)-1);
}
while(isspace(name[(int)name.length()-1]))
{
name.setat(name.length()-1, 0);
}
const GURL::UTF8 full_url(name,file_url.base());
iff_in.close_chunk();
G_TRY {
if (insert_file(full_url, false, file_pos, name2id, source))
{
// If the child file has been inserted (doesn't
// contain NDIR chunk), add INCL chunk.
GUTF8String id=name2id[name];
iff_out.put_chunk("INCL");
iff_out.get_bytestream()->writestring(id);
iff_out.close_chunk();
}
} G_CATCH(exc) {
// Should an error occur, we move on. INCL chunk will
// not be copied.
if (errors.length())
errors+="\n\n";
errors+=exc.get_cause();
} G_ENDCATCH;
}
} // while(iff_in.get_chunk(chkid))
iff_out.close_chunk();
// Increment the file_pos past the page inserted.
if (file_pos>=0) file_pos++;
// We have just inserted every included file. We may have modified
// contents of the INCL chunks. So we need to update the DataPool...
gstr_out->seek(0);
const GP<DataPool> new_file_pool(DataPool::create(gstr_out));
{
// It's important that we replace the pool here anyway.
// By doing this we load the file into memory. And this is
// exactly what insert_group() wants us to do because
// it creates temporary files.
GCriticalSectionLock lock(&files_lock);
files_map[id]->pool=new_file_pool;
}
} G_CATCH(exc) {
if (errors.length())
errors+="\n\n";
errors+=exc.get_cause();
G_THROW(errors);
} G_ENDCATCH;
// The only place where we intercept exceptions is when we process
// included files. We want to process all of them even if we failed to
// process one. But here we need to let the exception propagate...
if (errors.length())
G_THROW(errors);
return true;
}
return false;
}
void
DjVuDocEditor::insert_group(const GList<GURL> & file_urls, int page_num,
void (* _refresh_cb)(void *), void * _cl_data)
// The function will insert every file from the list at position
// corresponding to page_num. If page_num is negative, concatenation
// will occur. Included files will be processed as well
{
refresh_cb=_refresh_cb;
refresh_cl_data=_cl_data;
G_TRY
{
// First translate the page_num to file_pos.
const GP<DjVmDir> dir(get_djvm_dir());
int file_pos;
if (page_num<0 || page_num>=dir->get_pages_num())
{
file_pos=-1;
}
else
{
file_pos=dir->get_page_pos(page_num);
}
// Now call the insert_file() for every page. We will remember the
// name2id translation table. Thus insert_file() will remember IDs
// it assigned to shared files
GMap<GUTF8String, GUTF8String> name2id;
GUTF8String errors;
for(GPosition pos=file_urls;pos;++pos)
{
const GURL &furl=file_urls[pos];
DEBUG_MSG( "Inserting file '" << furl << "'\n" );
G_TRY
{
// Check if it's a multipage document...
GP<DataPool> xdata_pool(DataPool::create(furl));
if(xdata_pool && furl.is_valid()
&& furl.is_local_file_url() && DjVuDocument::djvu_import_codec)
{
(*DjVuDocument::djvu_import_codec)(xdata_pool,furl,
needs_compression_flag,
can_compress_flag);
}
GUTF8String chkid;
IFFByteStream::create(xdata_pool->get_stream())->get_chunk(chkid);
if (name2id.contains(furl.fname())||(chkid=="FORM:DJVM"))
{
GMap<GUTF8String,void *> map;
map_ids(map);
DEBUG_MSG("Read DjVuDocument furl='" << furl << "'\n");
GP<ByteStream> gbs(ByteStream::create());
GP<DjVuDocument> doca(DjVuDocument::create_noinit());
doca->set_verbose_eof(verbose_eof);
doca->set_recover_errors(recover_errors);
doca->init(furl /* ,this */ );
doca->wait_for_complete_init();
get_portcaster()->add_route(doca,this);
DEBUG_MSG("Saving DjVuDocument url='" << furl << "' with unique names\n");
doca->write(gbs,map);
gbs->seek(0L);
DEBUG_MSG("Loading unique names\n");
GP<DjVuDocument> doc(DjVuDocument::create(gbs));
doc->set_verbose_eof(verbose_eof);
doc->set_recover_errors(recover_errors);
doc->wait_for_complete_init();
get_portcaster()->add_route(doc,this);
gbs=0;
DEBUG_MSG("Inserting pages\n");
int pages_num=doc->get_pages_num();
for(int page_num=0;page_num<pages_num;page_num++)
{
const GURL url(doc->page_to_url(page_num));
insert_file(url, true, file_pos, name2id, doc);
}
}
else
{
insert_file(furl, true, file_pos, name2id, this);
}
} G_CATCH(exc)
{
if (errors.length())
{
errors+="\n\n";
}
errors+=exc.get_cause();
}
G_ENDCATCH;
}
if (errors.length())
{
G_THROW(errors);
}
} G_CATCH_ALL
{
refresh_cb=0;
refresh_cl_data=0;
G_RETHROW;
} G_ENDCATCH;
refresh_cb=0;
refresh_cl_data=0;
}
void
DjVuDocEditor::insert_page(const GURL &file_url, int page_num)
{
DEBUG_MSG("DjVuDocEditor::insert_page(): furl='" << file_url << "'\n");
DEBUG_MAKE_INDENT(3);
GList<GURL> list;
list.append(file_url);
insert_group(list, page_num);
}
void
DjVuDocEditor::insert_page(GP<DataPool> & _file_pool,
const GURL & file_url, int page_num)
// Use _file_pool as source of data, create a new DjVuFile
// with name file_name, and insert it as page number page_num
{
DEBUG_MSG("DjVuDocEditor::insert_page(): pool size='" <<
_file_pool->get_size() << "'\n");
DEBUG_MAKE_INDENT(3);
const GP<DjVmDir> dir(get_djvm_dir());
// Strip any INCL chunks (we do not allow to insert hierarchies
// using this function)
const GP<DataPool> file_pool(strip_incl_chunks(_file_pool));
// Now obtain ID for the new file
const GUTF8String id(find_unique_id(file_url.fname()));
// Add it into the directory
const GP<DjVmDir::File> frec(DjVmDir::File::create(
id, id, id, DjVmDir::File::PAGE));
int pos=dir->get_page_pos(page_num);
dir->insert_file(frec, pos);
// Add it to our "cache"
{
GP<File> f=new File;
f->pool=file_pool;
GCriticalSectionLock lock(&files_lock);
files_map[id]=f;
}
}
void
DjVuDocEditor::generate_ref_map(const GP<DjVuFile> & file,
GMap<GUTF8String, void *> & ref_map,
GMap<GURL, void *> & visit_map)
// This private function is used to generate a list (implemented as map)
// of files referencing the given file. To get list of all parents
// for file with ID 'id' iterate map obtained as
// *((GMap<GUTF8String, void *> *) ref_map[id])
{
const GURL url=file->get_url();
const GUTF8String id(djvm_dir->name_to_file(url.fname())->get_load_name());
if (!visit_map.contains(url))
{
visit_map[url]=0;
GPList<DjVuFile> files_list=file->get_included_files(false);
for(GPosition pos=files_list;pos;++pos)
{
GP<DjVuFile> child_file=files_list[pos];
// First: add the current file to the list of parents for
// the child being processed
GURL child_url=child_file->get_url();
const GUTF8String child_id(
djvm_dir->name_to_file(child_url.fname())->get_load_name());
GMap<GUTF8String, void *> * parents=0;
if (ref_map.contains(child_id))
parents=(GMap<GUTF8String, void *> *) ref_map[child_id];
else
ref_map[child_id]=parents=new GMap<GUTF8String, void *>();
(*parents)[id]=0;
// Second: go recursively
generate_ref_map(child_file, ref_map, visit_map);
}
}
}
void
DjVuDocEditor::remove_file(const GUTF8String &id, bool remove_unref,
GMap<GUTF8String, void *> & ref_map)
// Private function, which will remove file with ID id.
//
// If will also remove all INCL chunks in parent files pointing
// to this one
//
// Finally, if remove_unref is TRUE, we will go down the files
// hierarchy removing every file, which becomes unreferenced.
//
// ref_map will be used to find out list of parents referencing
// this file (required when removing INCL chunks)
{
// First get rid of INCL chunks in parents
GMap<GUTF8String, void *> * parents=(GMap<GUTF8String, void *> *) ref_map[id];
if (parents)
{
for(GPosition pos=*parents;pos;++pos)
{
const GUTF8String parent_id((*parents).key(pos));
const GP<DjVuFile> parent(get_djvu_file(parent_id));
if (parent)
parent->unlink_file(id);
}
delete parents;
parents=0;
ref_map.del(id);
}
// We will accumulate errors here.
GUTF8String errors;
// Now modify the ref_map and process children if necessary
GP<DjVuFile> file=get_djvu_file(id);
if (file)
{
G_TRY {
GPList<DjVuFile> files_list=file->get_included_files(false);
for(GPosition pos=files_list;pos;++pos)
{
GP<DjVuFile> child_file=files_list[pos];
GURL child_url=child_file->get_url();
const GUTF8String child_id(
djvm_dir->name_to_file(child_url.fname())->get_load_name());
GMap<GUTF8String, void *> * parents=(GMap<GUTF8String, void *> *) ref_map[child_id];
if (parents) parents->del(id);
if (remove_unref && (!parents || !parents->size()))
remove_file(child_id, remove_unref, ref_map);
}
} G_CATCH(exc) {
if (errors.length()) errors+="\n\n";
errors+=exc.get_cause();
} G_ENDCATCH;
}
// Finally remove this file from the directory.
djvm_dir->delete_file(id);
// And get rid of its thumbnail, if any
GCriticalSectionLock lock(&thumb_lock);
GPosition pos(thumb_map.contains(id));
if (pos)
{
thumb_map.del(pos);
}
if (errors.length())
G_THROW(errors);
}
void
DjVuDocEditor::remove_file(const GUTF8String &id, bool remove_unref)
{
DEBUG_MSG("DjVuDocEditor::remove_file(): id='" << id << "'\n");
DEBUG_MAKE_INDENT(3);
if (!djvm_dir->id_to_file(id))
G_THROW( ERR_MSG("DjVuDocEditor.no_file") "\t"+id);
// First generate a map of references (containing the list of parents
// including this particular file. This will speed things up
// significatly.
GMap<GUTF8String, void *> ref_map; // GMap<GUTF8String, GMap<GUTF8String, void *> *> in fact
GMap<GURL, void *> visit_map; // To avoid loops
int pages_num=djvm_dir->get_pages_num();
for(int page_num=0;page_num<pages_num;page_num++)
generate_ref_map(get_djvu_file(page_num), ref_map, visit_map);
// Now call the function, which will do the removal recursively
remove_file(id, remove_unref, ref_map);
// And clear the ref_map
GPosition pos;
while((pos=ref_map))
{
GMap<GUTF8String, void *> * parents=(GMap<GUTF8String, void *> *) ref_map[pos];
delete parents;
ref_map.del(pos);
}
}
void
DjVuDocEditor::remove_page(int page_num, bool remove_unref)
{
DEBUG_MSG("DjVuDocEditor::remove_page(): page_num=" << page_num << "\n");
DEBUG_MAKE_INDENT(3);
// Translate the page_num to ID
GP<DjVmDir> djvm_dir=get_djvm_dir();
if (page_num<0 || page_num>=djvm_dir->get_pages_num())
G_THROW( ERR_MSG("DjVuDocEditor.bad_page") "\t"+GUTF8String(page_num));
// And call general remove_file()
remove_file(djvm_dir->page_to_file(page_num)->get_load_name(), remove_unref);
}
void
DjVuDocEditor::remove_pages(const GList<int> & page_list, bool remove_unref)
{
DEBUG_MSG("DjVuDocEditor::remove_pages() called\n");
DEBUG_MAKE_INDENT(3);
// First we need to translate page numbers to IDs (they will
// obviously be changing while we're removing pages one after another)
GP<DjVmDir> djvm_dir=get_djvm_dir();
GPosition pos ;
if (djvm_dir)
{
GList<GUTF8String> id_list;
for(pos=page_list;pos;++pos)
{
GP<DjVmDir::File> frec=djvm_dir->page_to_file(page_list[pos]);
if (frec)
id_list.append(frec->get_load_name());
}
for(pos=id_list;pos;++pos)
{
GP<DjVmDir::File> frec=djvm_dir->id_to_file(id_list[pos]);
if (frec)
remove_page(frec->get_page_num(), remove_unref);
}
}
}
void
DjVuDocEditor::move_file(const GUTF8String &id, int & file_pos,
GMap<GUTF8String, void *> & map)
// NOTE! file_pos here is the desired position in DjVmDir *after*
// the record with ID 'id' is removed.
{
if (!map.contains(id))
{
map[id]=0;
GP<DjVmDir::File> file_rec=djvm_dir->id_to_file(id);
if (file_rec)
{
file_rec=new DjVmDir::File(*file_rec);
djvm_dir->delete_file(id);
djvm_dir->insert_file(file_rec, file_pos);
if (file_pos>=0)
{
file_pos++;
// We care to move included files only if we do not append
// This is because the only reason why we move included
// files is to made them available sooner than they would
// be available if we didn't move them. By appending files
// we delay the moment when the data for the file becomes
// available, of course.
GP<DjVuFile> djvu_file=get_djvu_file(id);
if (djvu_file)
{
GPList<DjVuFile> files_list=djvu_file->get_included_files(false);
for(GPosition pos=files_list;pos;++pos)
{
const GUTF8String name(files_list[pos]->get_url().fname());
GP<DjVmDir::File> child_frec=djvm_dir->name_to_file(name);
// If the child is positioned in DjVmDir AFTER the
// file being processed (position is file_pos or greater),
// move it to file_pos position
if (child_frec)
if (djvm_dir->get_file_pos(child_frec)>file_pos)
move_file(child_frec->get_load_name(), file_pos, map);
}
}
}
}
}
}
void
DjVuDocEditor::move_page(int page_num, int new_page_num)
{
DEBUG_MSG("DjVuDocEditor::move_page(): page_num=" << page_num <<
", new_page_num=" << new_page_num << "\n");
DEBUG_MAKE_INDENT(3);
if (page_num==new_page_num) return;
int pages_num=get_pages_num();
if (page_num<0 || page_num>=pages_num)
G_THROW( ERR_MSG("DjVuDocEditor.bad_page") "\t"+GUTF8String(page_num));
const GUTF8String id(page_to_id(page_num));
int file_pos=-1;
if (new_page_num>=0 && new_page_num<pages_num)
if (new_page_num>page_num) // Moving toward the end
{
if (new_page_num<pages_num-1)
file_pos=djvm_dir->get_page_pos(new_page_num+1)-1;
} else
file_pos=djvm_dir->get_page_pos(new_page_num);
GMap<GUTF8String, void *> map;
move_file(id, file_pos, map);
}
#ifdef _WIN32_WCE_EMULATION // Work around odd behavior under WCE Emulation
#define CALLINGCONVENTION __cdecl
#else
#define CALLINGCONVENTION /* */
#endif
static int
CALLINGCONVENTION
cmp(const void * ptr1, const void * ptr2)
{
int num1=*(int *) ptr1;
int num2=*(int *) ptr2;
return num1<num2 ? -1 : num1>num2 ? 1 : 0;
}
static GList<int>
sortList(const GList<int> & list)
{
GArray<int> a(list.size()-1);
int cnt;
GPosition pos;
for(pos=list, cnt=0;pos;++pos, cnt++)
a[cnt]=list[pos];
qsort((int *) a, a.size(), sizeof(int), cmp);
GList<int> l;
for(int i=0;i<a.size();i++)
l.append(a[i]);
return l;
}
void
DjVuDocEditor::move_pages(const GList<int> & _page_list, int shift)
{
if (!shift) return;
GList<int> page_list=sortList(_page_list);
GList<GUTF8String> id_list;
for(GPosition pos=page_list;pos;++pos)
{
GP<DjVmDir::File> frec=djvm_dir->page_to_file(page_list[pos]);
if (frec)
id_list.append(frec->get_load_name());
}
if (shift<0)
{
// We have to start here from the smallest page number
// We will move it according to the 'shift', and all
// further moves are guaranteed not to affect its page number.
// We will be changing the 'min_page' to make sure that
// pages moved beyond the document will still be in correct order
int min_page=0;
for(GPosition pos=id_list;pos;++pos)
{
GP<DjVmDir::File> frec=djvm_dir->id_to_file(id_list[pos]);
if (frec)
{
int page_num=frec->get_page_num();
int new_page_num=page_num+shift;
if (new_page_num<min_page)
new_page_num=min_page++;
move_page(page_num, new_page_num);
}
}
} else
{
// We have to start here from the biggest page number
// We will move it according to the 'shift', and all
// further moves will not affect its page number.
// We will be changing the 'max_page' to make sure that
// pages moved beyond the document will still be in correct order
int max_page=djvm_dir->get_pages_num()-1;
for(GPosition pos=id_list.lastpos();pos;--pos)
{
GP<DjVmDir::File> frec=djvm_dir->id_to_file(id_list[pos]);
if (frec)
{
int page_num=frec->get_page_num();
int new_page_num=page_num+shift;
if (new_page_num>max_page)
new_page_num=max_page--;
move_page(page_num, new_page_num);
}
}
}
}
void
DjVuDocEditor::set_file_name(const GUTF8String &id, const GUTF8String &name)
{
DEBUG_MSG("DjVuDocEditor::set_file_name(), id='" << id << "', name='" << name << "'\n");
DEBUG_MAKE_INDENT(3);
// It's important to get the URL now, because later (after we
// change DjVmDir) id_to_url() will be returning a modified value
GURL url=id_to_url(id);
// Change DjVmDir. It will check if the name is unique
djvm_dir->set_file_name(id, name);
// Now find DjVuFile (if any) and rename it
GPosition pos;
if (files_map.contains(id, pos))
{
GP<File> file=files_map[pos];
GP<DataPool> pool=file->pool;
if (pool) pool->load_file();
GP<DjVuFile> djvu_file=file->file;
if (djvu_file) djvu_file->set_name(name);
}
}
void
DjVuDocEditor::set_page_name(int page_num, const GUTF8String &name)
{
DEBUG_MSG("DjVuDocEditor::set_page_name(), page_num='" << page_num << "'\n");
DEBUG_MAKE_INDENT(3);
if (page_num<0 || page_num>=get_pages_num())
G_THROW( ERR_MSG("DjVuDocEditor.bad_page") "\t"+GUTF8String(page_num));
set_file_name(page_to_id(page_num), name);
}
void
DjVuDocEditor::set_file_title(const GUTF8String &id, const GUTF8String &title)
{
DEBUG_MSG("DjVuDocEditor::set_file_title(), id='" << id << "', title='" << title << "'\n");
DEBUG_MAKE_INDENT(3);
// Just change DjVmDir. It will check if the title is unique
djvm_dir->set_file_title(id, title);
}
void
DjVuDocEditor::set_page_title(int page_num, const GUTF8String &title)
{
DEBUG_MSG("DjVuDocEditor::set_page_title(), page_num='" << page_num << "'\n");
DEBUG_MAKE_INDENT(3);
if (page_num<0 || page_num>=get_pages_num())
G_THROW( ERR_MSG("DjVuDocEditor.bad_page") "\t"+GUTF8String(page_num));
set_file_title(page_to_id(page_num), title);
}
//****************************************************************************
//************************** Shared annotations ******************************
//****************************************************************************
void
DjVuDocEditor::simplify_anno(void (* progress_cb)(float progress, void *),
void * cl_data)
// It's important that no decoding is done while this function
// is running. Otherwise the DjVuFile's decoding routines and
// this function may attempt to decode/modify a file's
// annotations at the same time.
{
// Get the name of the SHARED_ANNO file. We will not
// touch that file (will not move annotations from it)
GP<DjVmDir::File> shared_file=djvm_dir->get_shared_anno_file();
GUTF8String shared_id;
if (shared_file)
shared_id=shared_file->get_load_name();
GList<GURL> ignore_list;
if (shared_id.length())
ignore_list.append(id_to_url(shared_id));
// First, for every page get merged (or "flatten" or "projected")
// annotations and store them inside the top-level page file
int pages_num=djvm_dir->get_pages_num();
for(int page_num=0;page_num<pages_num;page_num++)
{
GP<DjVuFile> djvu_file=get_djvu_file(page_num);
if (!djvu_file)
G_THROW( ERR_MSG("DjVuDocEditor.page_fail") "\t"+page_num);
int max_level=0;
GP<ByteStream> anno;
anno=djvu_file->get_merged_anno(ignore_list, &max_level);
if (anno && max_level>0)
{
// This is the moment when we try to modify DjVuFile's annotations
// Make sure, that it's not being decoded
GSafeFlags & file_flags=djvu_file->get_safe_flags();
GMonitorLock lock(&file_flags);
while(file_flags & DjVuFile::DECODING)
file_flags.wait();
// Merge all chunks in one by decoding and encoding DjVuAnno
const GP<DjVuAnno> dec_anno(DjVuAnno::create());
dec_anno->decode(anno);
const GP<ByteStream> new_anno(ByteStream::create());
dec_anno->encode(new_anno);
new_anno->seek(0);
// And store it in the file
djvu_file->anno=new_anno;
djvu_file->rebuild_data_pool();
if ((file_flags & (DjVuFile::DECODE_OK |
DjVuFile::DECODE_FAILED |
DjVuFile::DECODE_STOPPED))==0)
djvu_file->anno=0;
}
if (progress_cb)
progress_cb((float)(page_num/2.0/pages_num), cl_data);
}
// Now remove annotations from every file except for
// the top-level page files and SHARED_ANNO file.
// Unlink empty files too.
GPList<DjVmDir::File> files_list=djvm_dir->get_files_list();
int cnt;
GPosition pos;
for(pos=files_list, cnt=0;pos;++pos, cnt++)
{
GP<DjVmDir::File> frec=files_list[pos];
if (!frec->is_page() && frec->get_load_name()!=shared_id)
{
GP<DjVuFile> djvu_file=get_djvu_file(frec->get_load_name());
if (djvu_file)
{
djvu_file->remove_anno();
if (djvu_file->get_chunks_number()==0)
remove_file(frec->get_load_name(), true);
}
}
if (progress_cb)
progress_cb((float)(0.5+cnt/2.0/files_list.size()), cl_data);
}
}
void
DjVuDocEditor::create_shared_anno_file(void (* progress_cb)(float progress, void *),
void * cl_data)
{
if (djvm_dir->get_shared_anno_file())
G_THROW( ERR_MSG("DjVuDocEditor.share_fail") );
// Prepare file with ANTa chunk inside
const GP<ByteStream> gstr(ByteStream::create());
const GP<IFFByteStream> giff(IFFByteStream::create(gstr));
IFFByteStream &iff=*giff;
iff.put_chunk("FORM:DJVI");
iff.put_chunk("ANTa");
iff.close_chunk();
iff.close_chunk();
ByteStream &str=*gstr;
str.flush();
str.seek(0);
const GP<DataPool> file_pool(DataPool::create(gstr));
// Get a unique ID for the new file
const GUTF8String id(find_unique_id("shared_anno.iff"));
// Add it into the directory
GP<DjVmDir::File> frec(DjVmDir::File::create(id, id, id,
DjVmDir::File::SHARED_ANNO));
djvm_dir->insert_file(frec, 1);
// Add it to our "cache"
{
GP<File> f=new File;
f->pool=file_pool;
GCriticalSectionLock lock(&files_lock);
files_map[id]=f;
}
// Now include this shared file into every top-level page file
int pages_num=djvm_dir->get_pages_num();
for(int page_num=0;page_num<pages_num;page_num++)
{
GP<DjVuFile> djvu_file=get_djvu_file(page_num);
djvu_file->insert_file(id, 1);
if (progress_cb)
progress_cb((float) page_num/pages_num, cl_data);
}
}
void
DjVuDocEditor::set_djvm_nav(GP<DjVmNav> n)
{
if (n && ! n->isValidBookmark())
G_THROW("Invalid bookmark data");
djvm_nav = n;
}
GP<DjVuFile>
DjVuDocEditor::get_shared_anno_file(void)
{
GP<DjVuFile> djvu_file;
GP<DjVmDir::File> frec=djvm_dir->get_shared_anno_file();
if (frec)
djvu_file=get_djvu_file(frec->get_load_name());
return djvu_file;
}
GP<DataPool>
DjVuDocEditor::get_thumbnail(int page_num, bool dont_decode)
// We override DjVuDocument::get_thumbnail() here because
// pages may have been shuffled and those "thumbnail file records"
// from the DjVmDir do not describe things correctly.
//
// So, first we will check the thumb_map[] if we have a predecoded
// thumbnail for the given page. If this is the case, we will
// return it. Otherwise we will ask DjVuDocument to generate
// this thumbnail for us.
{
const GUTF8String id(page_to_id(page_num));
GCriticalSectionLock lock(&thumb_lock);
const GPosition pos(thumb_map.contains(id));
if (pos)
{
// Get the image from the map
return thumb_map[pos];
} else
{
unfile_thumbnails();
return DjVuDocument::get_thumbnail(page_num, dont_decode);
}
}
int
DjVuDocEditor::get_thumbnails_num(void) const
{
GCriticalSectionLock lock((GCriticalSection *) &thumb_lock);
int cnt=0;
int pages_num=get_pages_num();
for(int page_num=0;page_num<pages_num;page_num++)
{
if (thumb_map.contains(page_to_id(page_num)))
cnt++;
}
return cnt;
}
int
DjVuDocEditor::get_thumbnails_size(void) const
{
DEBUG_MSG("DjVuDocEditor::remove_thumbnails(): doing it\n");
DEBUG_MAKE_INDENT(3);
GCriticalSectionLock lock((GCriticalSection *) &thumb_lock);
int pages_num=get_pages_num();
for(int page_num=0;page_num<pages_num;page_num++)
{
const GPosition pos(thumb_map.contains(page_to_id(page_num)));
if (pos)
{
const GP<ByteStream> gstr(thumb_map[pos]->get_stream());
GP<IW44Image> iwpix=IW44Image::create_decode(IW44Image::COLOR);
iwpix->decode_chunk(gstr);
int width=iwpix->get_width();
int height=iwpix->get_height();
return width<height ? width : height;
}
}
return -1;
}
void
DjVuDocEditor::remove_thumbnails(void)
{
DEBUG_MSG("DjVuDocEditor::remove_thumbnails(): doing it\n");
DEBUG_MAKE_INDENT(3);
unfile_thumbnails();
DEBUG_MSG("clearing thumb_map\n");
GCriticalSectionLock lock(&thumb_lock);
thumb_map.empty();
}
void
DjVuDocEditor::unfile_thumbnails(void)
// Will erase all "THUMBNAILS" files from DjVmDir.
// This function is useful when filing thumbnails (to get rid of
// those files, which currently exist: they need to be replaced
// anyway) and when calling DjVuDocument::get_thumbnail() to
// be sure, that it will not use wrong information from DjVmDir
{
DEBUG_MSG("DjVuDocEditor::unfile_thumbnails(): updating DjVmDir\n");
DEBUG_MAKE_INDENT(3);
{
GCriticalSectionLock lock(&threqs_lock);
threqs_list.empty();
}
if((const DjVmDir *)djvm_dir)
{
GPList<DjVmDir::File> xfiles_list=djvm_dir->get_files_list();
for(GPosition pos=xfiles_list;pos;++pos)
{
GP<DjVmDir::File> f=xfiles_list[pos];
if (f->is_thumbnails())
djvm_dir->delete_file(f->get_load_name());
}
}
}
void
DjVuDocEditor::file_thumbnails(void)
// The purpose of this function is to create files containing
// thumbnail images and register them in DjVmDir.
// If some of the thumbnail images are missing, they'll
// be generated with generate_thumbnails()
{
DEBUG_MSG("DjVuDocEditor::file_thumbnails(): updating DjVmDir\n");
DEBUG_MAKE_INDENT(3);
unfile_thumbnails();
// Generate thumbnails if they're missing due to some reason.
int thumb_num=get_thumbnails_num();
int size=thumb_num>0 ? get_thumbnails_size() : 128;
if (thumb_num!=get_pages_num())
{
generate_thumbnails(size);
}
DEBUG_MSG("filing thumbnails\n");
GCriticalSectionLock lock(&thumb_lock);
// The first thumbnail file always contains only one thumbnail
int ipf=1;
int image_num=0;
int page_num=0, pages_num=djvm_dir->get_pages_num();
GP<ByteStream> str(ByteStream::create());
GP<IFFByteStream> iff(IFFByteStream::create(str));
iff->put_chunk("FORM:THUM");
for(;;)
{
GUTF8String id(page_to_id(page_num));
const GPosition pos(thumb_map.contains(id));
if (! pos)
{
G_THROW( ERR_MSG("DjVuDocEditor.no_thumb") "\t"+GUTF8String(page_num));
}
iff->put_chunk("TH44");
iff->copy(*(thumb_map[pos]->get_stream()));
iff->close_chunk();
image_num++;
page_num++;
if (image_num>=ipf || page_num>=pages_num)
{
int i=id.rsearch('.');
if(i<=0)
{
i=id.length();
}
id=id.substr(0,i)+".thumb";
// Get unique ID for this file
id=find_unique_id(id);
// Create a file record with the chosen ID
GP<DjVmDir::File> file(DjVmDir::File::create(id, id, id,
DjVmDir::File::THUMBNAILS));
// Set correct file position (so that it will cover the next
// ipf pages)
int file_pos=djvm_dir->get_page_pos(page_num-image_num);
djvm_dir->insert_file(file, file_pos);
// Now add the File record (containing the file URL and DataPool)
// After we do it a simple save_as() will save the document
// with the thumbnails. This is because DjVuDocument will see
// the file in DjVmDir and will ask for data. We will intercept
// the request for data and will provide this DataPool
iff->close_chunk();
str->seek(0);
const GP<DataPool> file_pool(DataPool::create(str));
GP<File> f=new File;
f->pool=file_pool;
GCriticalSectionLock lock(&files_lock);
files_map[id]=f;
// And create new streams
str=ByteStream::create();
iff=IFFByteStream::create(str);
iff->put_chunk("FORM:THUM");
image_num=0;
// Reset ipf to correct value (after we stored first
// "exceptional" file with thumbnail for the first page)
if (page_num==1) ipf=thumbnails_per_file;
if (page_num>=pages_num) break;
}
}
}
int
DjVuDocEditor::generate_thumbnails(int thumb_size, int page_num)
{
DEBUG_MSG("DjVuDocEditor::generate_thumbnails(): doing it\n");
DEBUG_MAKE_INDENT(3);
if(page_num<(djvm_dir->get_pages_num()))
{
const GUTF8String id(page_to_id(page_num));
if (!thumb_map.contains(id))
{
const GP<DjVuImage> dimg(get_page(page_num, true));
GRect rect(0, 0, thumb_size, dimg->get_height()*thumb_size/dimg->get_width());
GP<GPixmap> pm=dimg->get_pixmap(rect, rect, get_thumbnails_gamma());
if (!pm)
{
const GP<GBitmap> bm(dimg->get_bitmap(rect, rect, sizeof(int)));
if (bm)
pm = GPixmap::create(*bm);
else
pm = GPixmap::create(rect.height(), rect.width(), &GPixel::WHITE);
}
// Store and compress the pixmap
const GP<IW44Image> iwpix(IW44Image::create_encode(*pm));
const GP<ByteStream> gstr(ByteStream::create());
IWEncoderParms parms;
parms.slices=97;
parms.bytes=0;
parms.decibels=0;
iwpix->encode_chunk(gstr, parms);
gstr->seek(0L);
thumb_map[id]=DataPool::create(gstr);
}
++page_num;
}
else
{
page_num = -1;
}
return page_num;
}
void
DjVuDocEditor::generate_thumbnails(int thumb_size,
bool (* cb)(int page_num, void *),
void * cl_data)
{
int page_num=0;
do
{
page_num=generate_thumbnails(thumb_size,page_num);
if (cb) if (cb(page_num, cl_data)) return;
} while(page_num>=0);
}
static void
store_file(const GP<DjVmDir> & src_djvm_dir, const GP<DjVmDoc> & djvm_doc,
GP<DjVuFile> & djvu_file, GMap<GURL, void *> & map)
{
GURL url=djvu_file->get_url();
if (!map.contains(url))
{
map[url]=0;
// Store included files first
GPList<DjVuFile> djvu_files_list=djvu_file->get_included_files(false);
for(GPosition pos=djvu_files_list;pos;++pos)
store_file(src_djvm_dir, djvm_doc, djvu_files_list[pos], map);
// Now store contents of this file
GP<DataPool> file_data=djvu_file->get_djvu_data(false);
GP<DjVmDir::File> frec=src_djvm_dir->name_to_file(url.name());
if (frec)
{
frec=new DjVmDir::File(*frec);
djvm_doc->insert_file(frec, file_data, -1);
}
}
}
void
DjVuDocEditor::save_pages_as(
const GP<ByteStream> &str, const GList<int> & _page_list)
{
GList<int> page_list=sortList(_page_list);
GP<DjVmDoc> djvm_doc=DjVmDoc::create();
GMap<GURL, void *> map;
for(GPosition pos=page_list;pos;++pos)
{
GP<DjVmDir::File> frec=djvm_dir->page_to_file(page_list[pos]);
if (frec)
{
GP<DjVuFile> djvu_file=get_djvu_file(frec->get_load_name());
if (djvu_file)
store_file(djvm_dir, djvm_doc, djvu_file, map);
}
}
djvm_doc->write(str);
}
void
DjVuDocEditor::save_file(const GUTF8String &file_id, const GURL &codebase,
const bool only_modified, GMap<GUTF8String,GUTF8String> & map)
{
if(only_modified)
{
for(GPosition pos=files_map;pos;++pos)
{
const GP<File> file_rec(files_map[pos]);
const bool file_modified=file_rec->pool ||
(file_rec->file && file_rec->file->is_modified());
if(!file_modified)
{
const GUTF8String id=files_map.key(pos);
const GUTF8String save_name(djvm_dir->id_to_file(id)->get_save_name());
if(id == save_name)
{
map[id]=id;
}
}
}
}
save_file(file_id,codebase,map);
}
void
DjVuDocEditor::save_file(
const GUTF8String &file_id, const GURL &codebase,
GMap<GUTF8String,GUTF8String> & map)
{
DEBUG_MSG("DjVuDocEditor::save_file(): ID='" << file_id << "'\n");
DEBUG_MAKE_INDENT(3);
if (!map.contains(file_id))
{
const GP<DjVmDir::File> file(djvm_dir->id_to_file(file_id));
GP<DataPool> file_pool;
const GPosition pos(files_map.contains(file_id));
if (pos)
{
const GP<File> file_rec(files_map[pos]);
if (file_rec->file)
file_pool=file_rec->file->get_djvu_data(false);
else
file_pool=file_rec->pool;
}
if (!file_pool)
{
DjVuPortcaster * pcaster=DjVuPort::get_portcaster();
file_pool=pcaster->request_data(this, id_to_url(file_id));
}
if (file_pool)
{
GMap<GUTF8String,GUTF8String> incl;
map[file_id]=get_djvm_doc()->save_file(codebase,*file,incl,file_pool);
for(GPosition pos=incl;pos;++pos)
{
save_file(incl.key(pos),codebase ,map);
}
}else
{
map[file_id]=file->get_save_name();
}
}
}
void
DjVuDocEditor::save(void)
{
DEBUG_MSG("DjVuDocEditor::save(): saving the file\n");
DEBUG_MAKE_INDENT(3);
if (!can_be_saved())
G_THROW( ERR_MSG("DjVuDocEditor.cant_save") );
save_as(GURL(), orig_doc_type!=INDIRECT);
}
void
DjVuDocEditor::write(const GP<ByteStream> &gbs, bool force_djvm)
{
DEBUG_MSG("DjVuDocEditor::write()\n");
DEBUG_MAKE_INDENT(3);
if (get_thumbnails_num()==get_pages_num())
{
file_thumbnails();
}else
{
remove_thumbnails();
}
clean_files_map();
DjVuDocument::write(gbs,force_djvm);
}
void
DjVuDocEditor::write(
const GP<ByteStream> &gbs,const GMap<GUTF8String,void *> &reserved)
{
DEBUG_MSG("DjVuDocEditor::write()\n");
DEBUG_MAKE_INDENT(3);
if (get_thumbnails_num()==get_pages_num())
{
file_thumbnails();
}else
{
remove_thumbnails();
}
clean_files_map();
DjVuDocument::write(gbs,reserved);
}
void
DjVuDocEditor::save_as(const GURL &where, bool bundled)
{
DEBUG_MSG("DjVuDocEditor::save_as(): where='" << where << "'\n");
DEBUG_MAKE_INDENT(3);
// First see if we need to generate (or just reshuffle) thumbnails...
// If we have an icon for every page, we will just call
// file_thumbnails(), which will update DjVmDir and will create
// the actual bundles with thumbnails (very fast)
// Otherwise we will remove the thumbnails completely because
// we really don't want to deal with documents, which have only
// some of their pages thumbnailed.
if (get_thumbnails_num()==get_pages_num())
{
file_thumbnails();
}else
{
remove_thumbnails();
}
GURL save_doc_url;
if (where.is_empty())
{
// Assume, that we just want to 'save'. Check, that it's possible
// and proceed.
bool can_be_saved_bundled=orig_doc_type==BUNDLED ||
orig_doc_type==OLD_BUNDLED ||
orig_doc_type==SINGLE_PAGE ||
orig_doc_type==OLD_INDEXED && orig_doc_pages==1;
if ((bundled ^ can_be_saved_bundled)!=0)
G_THROW( ERR_MSG("DjVuDocEditor.cant_save2") );
save_doc_url=doc_url;
} else
{
save_doc_url=where;
}
int save_doc_type=bundled ? BUNDLED : INDIRECT;
clean_files_map();
GCriticalSectionLock lock(&files_lock);
DjVuPortcaster * pcaster=DjVuPort::get_portcaster();
// First consider saving in SINGLE_FILE format (one file)
if(needs_compression())
{
DEBUG_MSG("Compressing on output\n");
remove_thumbnails();
if(! djvu_compress_codec)
{
G_THROW( ERR_MSG("DjVuDocEditor.no_codec") );
}
const GP<DjVmDoc> doc(get_djvm_doc());
GP<ByteStream> mbs(ByteStream::create());
doc->write(mbs);
mbs->flush();
mbs->seek(0,SEEK_SET);
djvu_compress_codec(mbs,save_doc_url,(!(const DjVmDir *)djvm_dir)||(djvm_dir->get_files_num()==1)||(save_doc_type!=INDIRECT));
files_map.empty();
doc_url=GURL();
}else
{
if (djvm_dir->get_files_num()==1)
{
// Here 'bundled' has no effect: we will save it as one page.
DEBUG_MSG("saving one file...\n");
GURL file_url=page_to_url(0);
const GUTF8String file_id(djvm_dir->page_to_file(0)->get_load_name());
GP<DataPool> file_pool;
GPosition pos=files_map.contains(file_id);
if (pos)
{
const GP<File> file_rec(files_map[pos]);
if (file_rec->pool && (!file_rec->file ||
!file_rec->file->is_modified()))
{
file_pool=file_rec->pool;
}else if (file_rec->file)
{
file_pool=file_rec->file->get_djvu_data(false);
}
}
// Even if file has not been modified (pool==0) we still want
// to save it.
if (!file_pool)
file_pool=pcaster->request_data(this, file_url);
if (file_pool)
{
DEBUG_MSG("Saving '" << file_url << "' to '" << save_doc_url << "'\n");
DataPool::load_file(save_doc_url);
const GP<ByteStream> gstr_out(ByteStream::create(save_doc_url, "wb"));
ByteStream &str_out=*gstr_out;
str_out.writall(octets, 4);
const GP<ByteStream> str_in(file_pool->get_stream());
str_out.copy(*str_in);
}
// Update the document's DataPool (to save memory)
const GP<DjVmDoc> doc(get_djvm_doc());
const GP<ByteStream> gstr=ByteStream::create();// One page: we can do it in the memory
doc->write(gstr);
gstr->seek(0, SEEK_SET);
const GP<DataPool> pool(DataPool::create(gstr));
doc_pool=pool;
init_data_pool=pool;
// Also update DjVmDir (to reflect changes in offsets)
djvm_dir=doc->get_djvm_dir();
} else if (save_doc_type==INDIRECT)
{
DEBUG_MSG("Saving in INDIRECT format to '" << save_doc_url << "'\n");
bool save_only_modified=!(save_doc_url!=doc_url || save_doc_type!=orig_doc_type);
GPList<DjVmDir::File> xfiles_list=djvm_dir->resolve_duplicates(false);
const GURL codebase=save_doc_url.base();
int pages_num=djvm_dir->get_pages_num();
GMap<GUTF8String, GUTF8String> map;
// First go thru the pages
for(int page_num=0;page_num<pages_num;page_num++)
{
const GUTF8String id(djvm_dir->page_to_file(page_num)->get_load_name());
save_file(id, codebase, save_only_modified, map);
}
// Next go thru thumbnails and similar stuff
GPosition pos;
for(pos=xfiles_list;pos;++pos)
save_file(xfiles_list[pos]->get_load_name(), codebase, save_only_modified, map);
// Finally - save the top-level index file
for(pos=xfiles_list;pos;++pos)
{
const GP<DjVmDir::File> file(xfiles_list[pos]);
file->offset=0;
file->size=0;
}
DataPool::load_file(save_doc_url);
const GP<ByteStream> gstr(ByteStream::create(save_doc_url, "wb"));
const GP<IFFByteStream> giff(IFFByteStream::create(gstr));
IFFByteStream &iff=*giff;
iff.put_chunk("FORM:DJVM", 1);
iff.put_chunk("DIRM");
djvm_dir->encode(giff->get_bytestream());
iff.close_chunk();
iff.close_chunk();
iff.flush();
// Update the document data pool (not required, but will save memory)
doc_pool=DataPool::create(save_doc_url);
init_data_pool=doc_pool;
// No reason to update DjVmDir as for this format it doesn't
// contain DJVM offsets
} else if (save_doc_type==BUNDLED || save_doc_type==OLD_BUNDLED)
{
DEBUG_MSG("Saving in BUNDLED format to '" << save_doc_url << "'\n");
// Can't be very smart here. Simply overwrite the file.
const GP<DjVmDoc> doc(get_djvm_doc());
DataPool::load_file(save_doc_url);
const GP<ByteStream> gstr(ByteStream::create(save_doc_url, "wb"));
doc->write(gstr);
gstr->flush();
// Update the document data pool (not required, but will save memory)
doc_pool=DataPool::create(save_doc_url);
init_data_pool=doc_pool;
// Also update DjVmDir (to reflect changes in offsets)
djvm_dir=doc->get_djvm_dir();
} else
{
G_THROW( ERR_MSG("DjVuDocEditor.cant_save") );
}
// Now, after we have saved the document w/o any error, detach DataPools,
// which are in the 'File's list to save memory. Detach everything.
// Even in the case when File->file is non-zero. If File->file is zero,
// remove the item from the list at all. If it's non-zero, it has
// to stay there because by definition files_map[] contains the list
// of all active files and customized DataPools
//
// In addition to it, look thru all active files and change their URLs
// to reflect changes in the document's URL (if there was a change)
// Another reason why file's URLs must be changed is that we may have
// saved the document in a different format, which changes the rules
// of file url composition.
for(GPosition pos=files_map;pos;)
{
const GP<File> file_rec(files_map[pos]);
file_rec->pool=0;
if (file_rec->file==0)
{
GPosition this_pos=pos;
++pos;
files_map.del(this_pos);
} else
{
// Change the file's url;
if (doc_url!=save_doc_url ||
orig_doc_type!=save_doc_type)
if (save_doc_type==BUNDLED)
file_rec->file->move(save_doc_url);
else file_rec->file->move(save_doc_url.base());
++pos;
}
}
}
orig_doc_type=save_doc_type;
doc_type=save_doc_type;
if (doc_url!=save_doc_url)
{
// Also update document's URL (we moved, didn't we?)
doc_url=save_doc_url;
init_url=save_doc_url;
}
}
GP<DjVuDocEditor>
DjVuDocEditor::create_wait(void)
{
DjVuDocEditor *doc=new DjVuDocEditor();
const GP<DjVuDocEditor> retval(doc);
doc->init();
return retval;
}
GP<DjVuDocEditor>
DjVuDocEditor::create_wait(const GURL &url)
{
DjVuDocEditor *doc=new DjVuDocEditor();
const GP<DjVuDocEditor> retval(doc);
doc->init(url);
return retval;
}
bool
DjVuDocEditor::inherits(const GUTF8String &class_name) const
{
return (class_name == "DjVuDocEditor")||DjVuDocument::inherits(class_name);
}
int
DjVuDocEditor::get_orig_doc_type(void) const
{
return orig_doc_type;
}
bool
DjVuDocEditor::can_be_saved(void) const
{
return !(needs_rename()||needs_compression()||orig_doc_type==UNKNOWN_TYPE ||
orig_doc_type==OLD_INDEXED);
}
int
DjVuDocEditor::get_save_doc_type(void) const
{
if (orig_doc_type==SINGLE_PAGE)
if (djvm_dir->get_files_num()==1)
return SINGLE_PAGE;
else
return BUNDLED;
else if (orig_doc_type==INDIRECT)
return INDIRECT;
else if (orig_doc_type==OLD_BUNDLED || orig_doc_type==BUNDLED)
return BUNDLED;
else
return UNKNOWN_TYPE;
}
GURL
DjVuDocEditor::get_doc_url(void) const
{
return doc_url.is_empty() ? init_url : doc_url;
}
#ifdef HAVE_NAMESPACES
}
# ifndef NOT_USING_DJVU_NAMESPACE
using namespace DJVU;
# endif
#endif