PDF metadata: Do not crash when reading malformed PDF files

session.vim

@@ -1,6 +1,9 @@
 " Project wide builtins
 let $PYFLAKES_BUILTINS = "_,dynamic_property,__,P,I,lopen,icu_lower,icu_upper,icu_title,ngettext"
 
+" Include directories for C modules
+let g:syntastic_c_include_dirs = [ '/usr/include/podofo']
+
 fun! CalibreLog()
     " Setup buffers to edit the calibre changelog and version info prior to
     " making a release.

setup/build_environment.py

@@ -6,7 +6,7 @@ __license__   = 'GPL v3'
 __copyright__ = '2009, Kovid Goyal <kovid@kovidgoyal.net>'
 __docformat__ = 'restructuredtext en'
 
-import os, socket, struct, subprocess, glob
+import os, socket, struct, subprocess
 from distutils.spawn import find_executable
 
 from PyQt4 import pyqtconfig
@@ -84,7 +84,6 @@ ft_lib_dirs = []
 ft_libs = []
 jpg_libs = []
 jpg_lib_dirs = []
-poppler_objs = []
 fc_inc = '/usr/include/fontconfig'
 fc_lib = '/usr/lib'
 podofo_inc = '/usr/include/podofo'
@@ -114,12 +113,7 @@ if iswindows:
     jpg_libs = ['jpeg']
     ft_lib_dirs = [sw_lib_dir]
     ft_libs = ['freetype']
-    poppler_inc_dirs = consolidate('POPPLER_INC_DIR',
-            r'%s\poppler;%s'%(sw_inc_dir, sw_inc_dir))
 
-    poppler_lib_dirs = consolidate('POPPLER_LIB_DIR', sw_lib_dir)
-    popplerqt4_lib_dirs = poppler_lib_dirs
-    poppler_libs = ['poppler']
     magick_inc_dirs = [os.path.join(prefix, 'build', 'ImageMagick-6.7.6')]
     magick_lib_dirs = [os.path.join(magick_inc_dirs[0], 'VisualMagick', 'lib')]
     magick_libs = ['CORE_RL_wand_', 'CORE_RL_magick_']
@@ -128,13 +122,6 @@ if iswindows:
 elif isosx:
     fc_inc = '/sw/include/fontconfig'
     fc_lib = '/sw/lib'
-    poppler = glob.glob('/sw/build/poppler-*')[-1]
-    poppler_inc_dirs = consolidate('POPPLER_INC_DIR',
-            '{0}/poppler:{0}'.format(poppler))
-    poppler_lib_dirs = consolidate('POPPLER_LIB_DIR',
-            '/sw/lib')
-    poppler_libs = ['poppler']
-    popplerqt4_lib_dirs = poppler_lib_dirs
     podofo_inc = '/sw/podofo'
     podofo_lib = '/sw/lib'
     magick_inc_dirs = consolidate('MAGICK_INC',
@@ -147,22 +134,15 @@ elif isosx:
     png_libs = ['png12']
 else:
     # Include directories
-    poppler_inc_dirs = pkgconfig_include_dirs('poppler',
-        'POPPLER_INC_DIR', '/usr/include/poppler')
     png_inc_dirs = pkgconfig_include_dirs('libpng', 'PNG_INC_DIR',
         '/usr/include')
     magick_inc_dirs = pkgconfig_include_dirs('MagickWand', 'MAGICK_INC', '/usr/include/ImageMagick')
 
     # Library directories
-    poppler_lib_dirs = popplerqt4_lib_dirs = pkgconfig_lib_dirs('poppler', 'POPPLER_LIB_DIR',
-        '/usr/lib')
     png_lib_dirs = pkgconfig_lib_dirs('libpng', 'PNG_LIB_DIR', '/usr/lib')
     magick_lib_dirs = pkgconfig_lib_dirs('MagickWand', 'MAGICK_LIB', '/usr/lib')
 
     # Libraries
-    poppler_libs = pkgconfig_libs('poppler', '', '')
-    if not poppler_libs:
-        poppler_libs = ['poppler']
     magick_libs = pkgconfig_libs('MagickWand', '', '')
     if not magick_libs:
         magick_libs = ['MagickWand', 'MagickCore']
@@ -176,26 +156,6 @@ fc_error = None if os.path.exists(os.path.join(fc_inc, 'fontconfig.h')) else \
             'Try setting the FC_INC_DIR and FC_LIB_DIR environment '
             'variables.')
 
-
-poppler_error = None
-poppler_cflags = ['-DPNG_SKIP_SETJMP_CHECK'] if islinux else []
-if not poppler_inc_dirs or not os.path.exists(
-        os.path.join(poppler_inc_dirs[0], 'OutputDev.h')):
-    poppler_error = \
-    ('Poppler not found on your system. Various PDF related',
-    ' functionality will not work. Use the POPPLER_INC_DIR and',
-    ' POPPLER_LIB_DIR environment variables. calibre requires '
-    ' the poppler XPDF headers. If your distro does not '
-    ' include them you will have to re-compile poppler '
-    ' by hand with --enable-xpdf-headers')
-else:
-    lh = os.path.join(poppler_inc_dirs[0], 'Link.h')
-    if 'class AnnotLink' not in open(lh, 'rb').read():
-        poppler_cflags.append('-DPOPPLER_OLD_LINK_TYPE')
-    ph = os.path.join(poppler_inc_dirs[0], 'Page.h')
-    if 'getLinks(Catalog' in open(ph, 'rb').read():
-        poppler_cflags.append('-DPOPPLER_PRE_20')
-
 magick_error = None
 if not magick_inc_dirs or not os.path.exists(os.path.join(magick_inc_dirs[0],
     'wand')):

setup/commands.py

@@ -8,7 +8,7 @@ __docformat__ = 'restructuredtext en'
 
 __all__ = [
         'pot', 'translations', 'get_translations', 'iso639',
-        'build', 'build_pdf2xml', 'server',
+        'build', 'server',
         'gui',
         'develop', 'install',
         'kakasi', 'coffee', 'resources',
@@ -31,9 +31,8 @@ translations = Translations()
 get_translations = GetTranslations()
 iso639 = ISO639()
 
-from setup.extensions import Build, BuildPDF2XML
+from setup.extensions import Build
 build = Build()
-build_pdf2xml = BuildPDF2XML()
 
 from setup.server import Server
 server = Server()

setup/extensions.py

@@ -12,14 +12,11 @@ from distutils import sysconfig
 from PyQt4.pyqtconfig import QtGuiModuleMakefile
 
 from setup import Command, islinux, isbsd, isosx, SRC, iswindows
-from setup.build_environment import (fc_inc, fc_lib, chmlib_inc_dirs,
+from setup.build_environment import (fc_inc, fc_lib, chmlib_inc_dirs, fc_error,
-        fc_error, poppler_libs, poppler_lib_dirs, poppler_inc_dirs, podofo_inc,
+        podofo_inc, podofo_lib, podofo_error, pyqt, OSX_SDK, NMAKE, QMAKE,
-        podofo_lib, podofo_error, poppler_error, pyqt, OSX_SDK, NMAKE,
+        msvc, MT, win_inc, win_lib, win_ddk, magick_inc_dirs, magick_lib_dirs,
-        QMAKE, msvc, MT, win_inc, win_lib, png_inc_dirs, win_ddk,
+        magick_libs, chmlib_lib_dirs, sqlite_inc_dirs, icu_inc_dirs,
-        magick_inc_dirs, magick_lib_dirs, png_lib_dirs, png_libs,
+        icu_lib_dirs)
-        magick_error, magick_libs, ft_lib_dirs, ft_libs, jpg_libs,
-        jpg_lib_dirs, chmlib_lib_dirs, sqlite_inc_dirs, icu_inc_dirs,
-        icu_lib_dirs, poppler_cflags)
 MT
 isunix = islinux or isosx or isbsd
 
@@ -51,7 +48,6 @@ class Extension(object):
 
 reflow_sources = glob.glob(os.path.join(SRC, 'calibre', 'ebooks', 'pdf', '*.cpp'))
 reflow_headers = glob.glob(os.path.join(SRC, 'calibre', 'ebooks', 'pdf', '*.h'))
-reflow_error = poppler_error if poppler_error else magick_error
 
 pdfreflow_libs = []
 if iswindows:
@@ -107,16 +103,6 @@ extensions = [
         inc_dirs=magick_inc_dirs
         ),
 
-    Extension('pdfreflow',
-                reflow_sources,
-                headers=reflow_headers,
-                libraries=poppler_libs+magick_libs+png_libs+ft_libs+jpg_libs+pdfreflow_libs,
-                lib_dirs=poppler_lib_dirs+magick_lib_dirs+png_lib_dirs+ft_lib_dirs+jpg_lib_dirs,
-                inc_dirs=poppler_inc_dirs+magick_inc_dirs+png_inc_dirs,
-                error=reflow_error,
-                cflags=poppler_cflags
-                ),
-
     Extension('lzx',
             ['calibre/utils/lzx/lzxmodule.c',
                     'calibre/utils/lzx/compressor.c',
@@ -445,48 +431,5 @@ class Build(Command):
             shutil.rmtree(build_dir)
 
 
-class BuildPDF2XML(Command):
-
-    description = 'Build command line pdf2xml utility'
-
-    def run(self, opts):
-        dest = os.path.expanduser('~/bin/pdf2xml')
-        if iswindows:
-            dest = r'C:\cygwin\home\kovid\sw\bin\pdf2xml.exe'
-        odest = self.j(self.d(self.SRC), 'build', 'objects', 'pdf2xml')
-        if not os.path.exists(odest):
-            os.makedirs(odest)
-
-        objects = []
-        for src in reflow_sources:
-            if src.endswith('python.cpp'):
-                continue
-            obj = self.j(odest, self.b(src+('.obj' if iswindows else '.o')))
-            if self.newer(obj, [src]+reflow_headers):
-                cmd = [cxx, '-pthread', '-pedantic', '-ggdb', '-c', '-Wall', '-I/usr/include/poppler',
-                        '-I/usr/include/ImageMagick',
-                        '-DPDF2XML', '-o', obj, src]
-                if iswindows:
-                    cmd = [cxx, '/c', '/MD', '/W3', '/EHsc', '/Zi', '/DPDF2XML']
-                    cmd += ['-I'+x for x in poppler_inc_dirs+magick_inc_dirs]
-                    cmd += ['/Fo'+obj, src]
-                self.info(*cmd)
-                self.check_call(cmd)
-            objects.append(obj)
-
-        if self.newer(dest, objects):
-            cmd = ['g++', '-ggdb', '-o', dest]+objects+['-lpoppler', '-lMagickWand',
-            '-lpng', '-lpthread']
-            if iswindows:
-                cmd = [msvc.linker] + '/INCREMENTAL:NO /DEBUG /NODEFAULTLIB:libcmt.lib'.split()
-                cmd += ['/LIBPATH:'+x for x in magick_lib_dirs+poppler_lib_dirs]
-                cmd += [x+'.lib' for x in
-                        png_libs+magick_libs+poppler_libs+ft_libs+jpg_libs+pdfreflow_libs]
-                cmd += ['/OUT:'+dest] + objects
-            self.info(*cmd)
-            self.check_call(cmd)
-
-        self.info('Binary installed as', dest)
-
 
 

setup/installer/linux/freeze2.py

@@ -22,6 +22,8 @@ QTDLLS         = ('QtCore', 'QtGui', 'QtNetwork', 'QtSvg', 'QtXml', 'QtWebKit',
 MAGICK_PREFIX = '/usr'
 binary_includes = [
                 '/usr/bin/pdftohtml',
+                '/usr/bin/pdfinfo',
+                '/usr/bin/pdftoppm',
                 '/usr/lib/libwmflite-0.2.so.7',
                 '/usr/lib/liblcms.so.1',
                 '/usr/lib/liblzma.so.0',

setup/installer/osx/app/main.py

@@ -387,7 +387,8 @@ class Py2App(object):
         info('\nAdding poppler')
         for x in ('libpoppler.25.dylib',):
             self.install_dylib(os.path.join(SW, 'lib', x))
-        self.install_dylib(os.path.join(SW, 'bin', 'pdftohtml'), False)
+        for x in ('pdftohtml', 'pdftoppm', 'pdfinfo'):
+            self.install_dylib(os.path.join(SW, 'bin', x), False)
 
     @flush
     def add_libjpeg(self):

setup/installer/windows/freeze.py

@@ -260,7 +260,8 @@ class Win32Freeze(Command, WixMixIn):
 
         print '\tAdding misc binary deps'
         bindir = os.path.join(SW, 'bin')
-        shutil.copy2(os.path.join(bindir, 'pdftohtml.exe'), self.base)
+        for x in ('pdftohtml', 'pdfinfo', 'pdftoppm'):
+            shutil.copy2(os.path.join(bindir, x+'.exe'), self.base)
         for pat in ('*.dll',):
             for f in glob.glob(os.path.join(bindir, pat)):
                 ok = True

setup/installer/windows/notes.rst

@@ -293,9 +293,7 @@ In Cmake: disable GTK, Qt, OPenjpeg, cpp, lcms, gtk_tests, qt_tests. Enable qt4,
 
 NOTE: poppler must be built as a static library, unless you build the qt4 bindings
 
-Now do the same for the pdftohtml project
+cp build/utils/Release/*.exe ../../bin/
-
-cp poppler/*.h ~/sw/include/poppler && cp goo/*.h ~/sw/include/poppler/goo && cp splash/*.h ~/sw/include/poppler/splash && cp build/Release/poppler.lib ../../lib/ && cp build/utils/Release/pdftohtml.exe ../../bin/
 
 
 podofo

src/calibre/constants.py

@@ -82,7 +82,6 @@ class Plugins(collections.Mapping):
                 'podofo',
                 'cPalmdoc',
                 'fontconfig',
-                'pdfreflow',
                 'progress_indicator',
                 'chmlib',
                 'chm_extra',

src/calibre/ebooks/metadata/pdf.py

@@ -4,25 +4,80 @@ __copyright__ = '2008, Kovid Goyal <kovid at kovidgoyal.net>'
 '''Read meta information from PDF files'''
 
 #import re
+import os, subprocess, shutil
 from functools import partial
 
 from calibre import prints
-from calibre.constants import plugins
+from calibre.constants import iswindows
+from calibre.ptempfile import TemporaryDirectory
 from calibre.ebooks.metadata import MetaInformation, string_to_authors
-
+from calibre.utils.ipc.simple_worker import fork_job
-pdfreflow, pdfreflow_error = plugins['pdfreflow']
 
 #_isbn_pat = re.compile(r'ISBN[: ]*([-0-9Xx]+)')
 
+def read_info(outputdir, get_cover):
+    ''' Read info dict and cover from a pdf file named src.pdf in outputdir.
+    Note that this function changes the cwd to outputdir and is therefore not
+    thread safe. Run it using fork_job. This is necessary as there is no safe
+    way to pass unicode paths via command line arguments. This also ensures
+    that if poppler crashes, no stale file handles are left for the original
+    file, only for src.pdf.'''
+
+    from calibre.ebooks.pdf.pdftohtml import PDFTOHTML
+    os.chdir(outputdir)
+    base = os.path.dirname(PDFTOHTML)
+    suffix = '.exe' if iswindows else ''
+    pdfinfo = os.path.join(base, 'pdfinfo') + suffix
+    pdftoppm = os.path.join(base, 'pdftoppm') + suffix
+
+    try:
+        raw = subprocess.check_output([pdfinfo, '-enc', 'UTF-8', 'src.pdf'])
+    except subprocess.CalledProcessError as e:
+        prints('pdfinfo errored out with return code: %d'%e.returncode)
+        return None
+    try:
+        raw = raw.decode('utf-8')
+    except UnicodeDecodeError:
+        prints('pdfinfo returned no UTF-8 data')
+        return None
+
+    ans = {}
+    for line in raw.splitlines():
+        if u':' not in line: continue
+        field, val = line.partition(u':')[::2]
+        val = val.strip()
+        if field and val:
+            ans[field] = val.strip()
+
+    if get_cover:
+        try:
+            subprocess.check_call([pdftoppm, '-singlefile', '-jpeg',
+                'src.pdf', 'cover'])
+        except subprocess.CalledProcessError as e:
+            prints('pdftoppm errored out with return code: %d'%e.returncode)
+
+    return ans
+
 def get_metadata(stream, cover=True):
-    if pdfreflow is None:
+    with TemporaryDirectory('_pdf_metadata_read') as pdfpath:
-        raise RuntimeError(pdfreflow_error)
+        stream.seek(0)
-    stream.seek(0)
+        with open(os.path.join(pdfpath, 'src.pdf'), 'wb') as f:
-    raw = stream.read()
+            shutil.copyfileobj(stream, f)
-    #isbn = _isbn_pat.search(raw)
+        res = fork_job('calibre.ebooks.metadata.pdf', 'read_info', (pdfpath,
-    #if isbn is not None:
+            bool(cover)))
-    #    isbn = isbn.group(1).replace('-', '').replace(' ', '')
+        info = res['result']
-    info = pdfreflow.get_metadata(raw, cover)
+        with open(res['stdout_stderr'], 'rb') as f:
+            raw = f.read().strip()
+            if raw:
+                prints(raw)
+        if not info:
+            raise ValueError('Could not read info dict from PDF')
+        covpath = os.path.join(pdfpath, 'cover.jpg')
+        cdata = None
+        if cover and os.path.exists(covpath):
+            with open(covpath, 'rb') as f:
+                cdata = f.read()
+
     title = info.get('Title', None)
     au = info.get('Author', None)
     if au is None:
@@ -46,12 +101,8 @@ def get_metadata(stream, cover=True):
     if subject:
         mi.tags.insert(0, subject)
 
-    if cover and 'cover' in info:
+    if cdata:
-        data = info['cover']
+        mi.cover_data = ('jpeg', cdata)
-        if data is None:
-            prints(title, 'has no pages, cover extraction impossible.')
-        else:
-            mi.cover_data = ('png', data)
 
     return mi
 

src/calibre/ebooks/pdf/fonts.cpp

@@ -1,151 +0,0 @@
-/**
- * Copyright 2009 Kovid Goyal <kovid@kovidgoyal.net>
- * License: GNU GPL v2+
- */
-
-
-
-#include "fonts.h"
-#include "utils.h"
-
-using namespace calibre_reflow;
-using namespace std;
-
-XMLColor::XMLColor(GfxRGB rgb) {
-  this->r = static_cast<int>(rgb.r/65535.0*255.0);
-  this->g = static_cast<int>(rgb.g/65535.0*255.0);
-  this->b = static_cast<int>(rgb.b/65535.0*255.0);
-  if (!(this->ok(this->r) && this->ok(this->b) && this->ok(this->g))) {
-    this->r = 0; this->g = 0; this->b = 0;
-  }
-}
-
-string XMLColor::str() const {
-    ostringstream oss;
-    oss << "rgb(" << this->r << "," << this->g << "," << this->b << ")";
-    return oss.str();
-}
-
-static const char *FONT_MODS[7] = {
-    "-bolditalic", "-boldoblique", "-bold", "-italic", "-oblique", "-roman",
-    NULL
-};
-
-#ifdef _WIN32
-#define ap_toupper(c) (toupper(((unsigned char)(c))))
-static inline
-const char *strcasestr(const char *h, const char *n )
-{ /* h="haystack", n="needle" */
-    const char *a=h, *e=n;
-
-    if( !h || !*h || !n || !*n ) { return 0; }
-
-    while( *a && *e ) {
-        if( ap_toupper(*a)!=ap_toupper(*e) ) {
-            ++h; a=h; e=n;
-        }
-        else {
-            ++a; ++e;
-        }
-    }
-    return *e ? 0 : h;
-}
-#endif
-
-static string* family_name(const string *font_name) {
-    if (!font_name) return NULL;
-    string *fn = new string(*font_name);
-    size_t pos;
-    const char *p;
-    for (size_t i = 0; FONT_MODS[i] != NULL; i++) {
-        p = strcasestr(fn->c_str(), FONT_MODS[i]);
-        if (p != NULL) {
-            pos = p - fn->c_str();
-            fn->replace(pos, strlen(FONT_MODS[i]), "");
-            break;
-        }
-    }
-    return fn;
-}
-
-XMLFont::XMLFont(string* font_name, double size, GfxRGB rgb) :
-        size(size-1), line_size(-1.0), italic(false), bold(false), font_name(font_name),
-        font_family(NULL), color(rgb)  {
-
-
-    if (!this->font_name) this->font_name = new string(DEFAULT_FONT_FAMILY);
-    this->font_family = family_name(this->font_name);
-    if (strcasestr(font_name->c_str(), "bold")) this->bold = true;
-
-    if (strcasestr(font_name->c_str(),"italic")||
-        strcasestr(font_name->c_str(),"oblique")) this->italic = true;
-
-
-}
-
-XMLFont& XMLFont::operator=(const XMLFont& x){
-   if (this==&x) return *this; 
-   this->size = x.size;
-   this->line_size = x.line_size;
-   this->italic = x.italic;
-   this->bold = x.bold;
-   this->color = x.color;
-   if (this->font_name) delete this->font_name;
-   this->font_name = new string(*x.font_name);
-   if (this->font_family) delete this->font_family;
-   this->font_family = new string(*x.font_family);
-   return *this;
-}
-
-bool XMLFont::operator==(const XMLFont &f) const {
-    return (fabs(this->size - f.size) < 0.1) && 
-        (fabs(this->line_size - f.line_size) < 0.1) &&
-        (this->italic == f.italic) &&
-        (this->bold == f.bold) &&
-        (this->color == f.color) &&
-        ((*this->font_family) == (*f.font_family));
-}
-
-bool XMLFont::eq_upto_inline(const XMLFont &f) const {
-    return (fabs(this->size - f.size) < 0.1) && 
-        (fabs(this->line_size - f.line_size) < 0.1) &&
-        (this->color == f.color) &&
-        ((*this->font_family) == (*f.font_family));
-}
-
-string XMLFont::str(Fonts::size_type id) const {
-    ostringstream oss;
-    oss << "<font id=\"" << id << "\" ";
-    oss << "family=\"" << encode_for_xml(*this->font_family) << "\" ";
-    oss << "color=\"" << this->color.str() << "\" ";
-    oss << setiosflags(ios::fixed) << setprecision(2) 
-        << "size=\"" << this->size << "\"";
-    oss << "/>";
-    return oss.str();
-}
-
-Fonts::size_type Fonts::add_font(XMLFont *f) {
-    Fonts::iterator it;
-    size_type i;
-    for ( i=0, it=this->begin(); it < this->end(); it++, i++ ) {
-        if (**it == *f) return i;
-    }
-    this->push_back(f);
-    return this->size()-1;
-}
-
-Fonts::size_type Fonts::add_font(string* font_name, double size, GfxRGB rgb) {
-    XMLFont *f = NULL;
-    if (font_name == NULL) 
-        font_name = new string("Unknown");
-        // font_name must not be deleted
-    f = new XMLFont(font_name, size, rgb);
-
-    return this->add_font(f);
-}
-
-Fonts::~Fonts() {
-    Fonts::iterator it;
-    for ( it=this->begin(); it < this->end(); it++ ) delete *it;
-    this->resize(0);
-}

src/calibre/ebooks/pdf/fonts.h

@@ -1,102 +0,0 @@
-/**
- * Copyright 2009 Kovid Goyal <kovid@kovidgoyal.net>
- * License: GNU GPL v2+
- */
-
-
-#pragma once
-
-#include <vector>
-#include <sstream>
-#include <iomanip>
-#include <ctype.h>
-#include <math.h>
-#include <GfxState.h>
-
-using namespace std;
-
-#define DEFAULT_FONT_FAMILY "Times New Roman"
-
-namespace calibre_reflow {
-
-class XMLColor {
-
- private:
-   unsigned int r;
-   unsigned int g;
-   unsigned int b;
-   inline bool ok(unsigned int xcol) const { 
-       return ( (xcol <= 255) && (xcol >= 0) );
-   }
-
- public:
-   XMLColor():r(0),g(0),b(0){}
-
-   XMLColor(GfxRGB rgb);
-   
-   XMLColor(const XMLColor& x) {
-       this->r=x.r; this->g=x.g; this->b=x.b;
-   }
-   
-   XMLColor& operator=(const XMLColor &x){
-     this->r=x.r; this->g=x.g; this->b=x.b;
-     return *this;
-   }
-   
-   ~XMLColor(){}
-   
-   string str() const; 
-
-   bool operator==(const XMLColor &col) const {
-     return ((r==col.r)&&(g==col.g)&&(b==col.b));
-   }
-
-};  
-
-
-class XMLFont {
-
-private:
-    double size;
-    double line_size;
-    bool italic;
-    bool bold;
-    string *font_name;
-    string *font_family;
-    XMLColor color;
-
-public:  
-    XMLFont(const char *font_family=DEFAULT_FONT_FAMILY, double size=12.0) : 
-        size(size), line_size(-1.0), italic(false), bold(false),
-        font_name(new string(font_family)), font_family(new string(font_family)),
-        color() {}
-
-    XMLFont(string* font_name, double size, GfxRGB rgb);    
-    XMLFont(const XMLFont& other) :
-        size(other.size), line_size(other.line_size), italic(other.italic),
-        bold(other.bold), font_name(new string(*other.font_name)),
-        font_family(other.font_family), color(other.color) {}
-
-    XMLColor get_color() { return this->color; }
-    string* get_font_name() { return this->font_name; }
-    double get_size() const { return this->size; }
-    double get_line_size() { return this->line_size; }
-    void set_line_size(double ls) { this->line_size = ls; } 
-    bool is_italic() const { return this->italic; }
-    bool is_bold() const { return this->bold; }
-    ~XMLFont() { delete this->font_name; delete this->font_family; }
-    XMLFont& operator=(const XMLFont& other);
-    bool operator==(const XMLFont &other) const;
-    bool eq_upto_inline(const XMLFont &f) const;
-    string str(vector<XMLFont*>::size_type id) const;
-};
-
-class Fonts : public vector<XMLFont*> {
-    public:
-        Fonts::size_type add_font(XMLFont *f);
-        Fonts::size_type add_font(string* font_name, double size, GfxRGB rgb); 
-        ~Fonts();
-};
-
-
-}

src/calibre/ebooks/pdf/images.cpp

@@ -1,433 +0,0 @@
-/**
- * Copyright 2009 Kovid Goyal <kovid@kovidgoyal.net>
- * License: GNU GPL v2+
- */
-
-
-
-#include <stdio.h>
-#include <errno.h>
-#include <sstream>
-#include <algorithm>
-#include <iomanip>
-#include <math.h>
-#include <iostream>
-#include <wand/MagickWand.h>
-#include <zlib.h>
-
-#include "images.h"
-#include "utils.h"
-
-#ifdef _WIN32
-inline double round(double x) { return (x-floor(x))>0.5 ? ceil(x) : floor(x); }
-#endif
-
-#define xoutRound(x) ( static_cast<int>(round(x)) )
-using namespace std;
-using namespace calibre_reflow;
-
-calibre_reflow::ImageInfo::ImageInfo(GfxState *state) {
-    // get image position and size
-    state->transform(0, 0, &xt, &yt);
-    state->transformDelta(1, 1, &wt, &ht);
-    if (wt > 0) {
-        x0 = xoutRound(xt);
-        w0 = xoutRound(wt);
-    } else {
-        x0 = xoutRound(xt + wt);
-        w0 = xoutRound(-wt);
-    }
-    if (ht > 0) {
-        y0 = xoutRound(yt);
-        h0 = xoutRound(ht);
-    } else {
-        y0 = xoutRound(yt + ht);
-        h0 = xoutRound(-ht);
-    }
-    state->transformDelta(1, 0, &xt, &yt);
-    rotate = fabs(xt) < fabs(yt);
-    if (rotate) {
-        w1 = h0;
-        h1 = w0;
-        x_flip = ht < 0;
-        y_flip = wt > 0;
-    } else {
-        w1 = w0;
-        h1 = h0;
-        x_flip = wt < 0;
-        y_flip = ht > 0;
-    }
-    //cout << x_flip << "|" << y_flip << endl;
-}
-
-void XMLImages::clear() {
-    vector<XMLImage*>::iterator it;
-    for (it = this->masks.begin(); it < this->masks.end(); it++)
-        delete *it;
-    for (it = this->images.begin(); it < this->images.end(); it++)
-        delete *it;
-    this->masks.clear();
-    this->images.clear();
-}
-
-void XMLImages::add_mask(GfxState *state, Object *ref, Stream *str,
-				  unsigned int width, unsigned int height, bool invert,
-				  bool interpolate, bool inline_img) {
-}
-
-static void throw_magick_exception(MagickWand *wand) {
-    ExceptionType severity;
-    char *description = MagickGetException(wand, &severity);
-    ostringstream oss;
-    oss << description << endl;
-    description=(char *) MagickRelinquishMemory(description);
-    wand = DestroyMagickWand(wand);
-    MagickWandTerminus();
-    throw ReflowException(oss.str().c_str());
-}
-
-
-static void flip_image(string file_name, bool x_flip, bool y_flip) {
-    MagickWand *magick_wand;
-    MagickBooleanType status;
-
-    MagickWandGenesis();
-    magick_wand = NewMagickWand();
-    status = MagickReadImage(magick_wand, file_name.c_str());
-    if (status == MagickFalse) throw_magick_exception(magick_wand);
-
-    if (y_flip) {
-        status = MagickFlipImage(magick_wand);
-        if (status == MagickFalse) throw_magick_exception(magick_wand);
-    }
-    if (x_flip) {
-        status = MagickFlopImage(magick_wand);
-        if (status == MagickFalse) throw_magick_exception(magick_wand);
-    }
-
-    status = MagickWriteImage(magick_wand, NULL);
-    if (status == MagickFalse) throw_magick_exception(magick_wand);
-
-    magick_wand = DestroyMagickWand(magick_wand);
-    MagickWandTerminus();
-}
-
-void XMLImages::add(GfxState *state, Object *ref, Stream *str,
-			      unsigned int width, unsigned int height, GfxImageColorMap *colorMap,
-			      bool interpolate, int *maskColors, bool inline_img) {
-    XMLImage *img = new XMLImage(state);
-    this->images.push_back(img); 
-    img->width = width; img->height = height;
-    img->type = (str->getKind() == strDCT) ? jpeg : png;
-    string file_name = this->file_name(img);
-
-    FILE *of = fopen(file_name.c_str(), "wb");
-    if (!of) throw ReflowException(strerror(errno));
-
-    if (img->type == jpeg) {
-        int c;
-        str = str->getNextStream();
-        str->reset();
-
-        // copy the stream
-        while ((c = str->getChar()) != EOF) fputc(c, of);
-    } else { //Render as PNG
-        Guchar *p;
-        GfxRGB rgb;
-        png_byte *row = (png_byte *) malloc(3 * width);   // 3 bytes/pixel: RGB
-        png_bytep *row_pointer= &row;
-
-        PNGWriter *writer = new PNGWriter();
-        writer->init(of, width, height); 
-
-        // Initialize the image stream
-        ImageStream *imgStr = new ImageStream(str, width,
-                            colorMap->getNumPixelComps(), colorMap->getBits());
-        imgStr->reset();
-
-        // For each line...
-        for (unsigned int y = 0; y < height; y++) {
-            // Convert into a PNG row
-            p = imgStr->getLine();
-            for (unsigned int x = 0; x < width; x++) {
-                colorMap->getRGB(p, &rgb);
-                // Write the RGB pixels into the row
-                row[3*x]= colToByte(rgb.r);
-                row[3*x+1]= colToByte(rgb.g);
-                row[3*x+2]= colToByte(rgb.b);
-                p += colorMap->getNumPixelComps();
-            }
-
-            writer->writeRow(row_pointer);
-        }
-
-        writer->close();
-        delete writer;
-
-        free(row);
-        imgStr->close();
-        delete imgStr;
-
-    }
-    fclose(of);
-    img->written = true;
-    if (img->info.x_flip || img->info.y_flip)
-        flip_image(file_name, img->info.x_flip, img->info.y_flip);
-}
-
-
-string XMLImages::file_name(const XMLImage *img) const {
-    vector<XMLImage*>::const_iterator ir, mr;
-    size_t idx = 0;
-    bool mask = false;
-
-    ir = find( this->images.begin(), this->images.end(), img);
-    if (ir == this->images.end()) {
-        mr = find( this->masks.begin(), this->masks.end(), img);
-        idx = mr - this->masks.begin();
-        mask = true;
-    } else idx = ir - this->images.begin();
-
-    ostringstream oss;
-    oss << ((mask) ? "mask" : "image") << "-" << idx+1 << '.';
-    oss << ((img->type == jpeg) ? "jpg" : "png");
-    return oss.str();
-}
-
-vector<string*> XMLImages::str() const {
-    vector<string*> ans;
-    vector <XMLImage*>::const_iterator it;
-    for (it = this->masks.begin(); it < this->masks.end(); it++) {
-        if ((*it)->written) 
-           ans.push_back(new string((*it)->str(it - this->masks.begin(), true,
-                           this->file_name(*it)))); 
-    }
-    for (it = this->images.begin(); it < this->images.end(); it++) {
-        if ((*it)->written) 
-           ans.push_back(new string((*it)->str(it - this->images.begin(), false,
-                           this->file_name(*it)))); 
-    }
-    return ans; 
-}
-
-string XMLImage::str(size_t num, bool mask, string file_name) const {
-    ostringstream oss;
-    oss << "<img type=\"" << ((mask) ? "mask" : "image") << "\" "
-        << "src=\"" << file_name << "\" "
-        << "iwidth=\"" << this->width << "\" iheight=\"" << this->height << "\" "
-        << "rwidth=\"" << this->info.w1 << "\" rheight=\"" << this->info.h1 << "\" "
-        << setiosflags(ios::fixed) << setprecision(2)
-        << "top=\"" << this->info.y0 << "\" left=\"" << this->info.x0 << "\"/>";
-    return oss.str();
-
-
-}
-PNGWriter::~PNGWriter()
-{
-	/* cleanup heap allocation */
-	png_destroy_write_struct(&png_ptr, &info_ptr);
-}
-
-void PNGWriter::init(FILE *f, int width, int height)
-{
-	/* initialize stuff */
-	png_ptr = png_create_write_struct(PNG_LIBPNG_VER_STRING, NULL, NULL, NULL);
-	if (!png_ptr) 
-        throw ReflowException("png_create_write_struct failed");
-
-	info_ptr = png_create_info_struct(png_ptr);
-	if (!info_ptr) 
-		throw ReflowException("png_create_info_struct failed");
-
-	if (setjmp(png_jmpbuf(png_ptr))) 
-		throw ReflowException("png_jmpbuf failed");
-
-	/* write header */
-	png_init_io(png_ptr, f);
-	if (setjmp(png_jmpbuf(png_ptr))) 
-		throw ReflowException("Error during writing header");
-	
-	// Set up the type of PNG image and the compression level
-	png_set_compression_level(png_ptr, Z_BEST_COMPRESSION);
-
-	png_byte bit_depth = 8;
-	png_byte color_type = PNG_COLOR_TYPE_RGB;
-	png_byte interlace_type = PNG_INTERLACE_NONE;
-
-	png_set_IHDR(png_ptr, info_ptr, width, height, bit_depth, color_type, interlace_type, PNG_COMPRESSION_TYPE_DEFAULT, PNG_FILTER_TYPE_DEFAULT);
-
-	png_write_info(png_ptr, info_ptr);
-	if (setjmp(png_jmpbuf(png_ptr))) 
-		throw ReflowException("error during writing png info bytes");
-	
-}
-
-void PNGWriter::writePointers(png_bytep *rowPointers)
-{
-	png_write_image(png_ptr, rowPointers);
-	/* write bytes */
-	if (setjmp(png_jmpbuf(png_ptr))) 
-        throw ReflowException("Error during writing bytes");
-}
-
-void PNGWriter::writeRow(png_bytep *row)
-{
-	// Write the row to the file
-	png_write_rows(png_ptr, row, 1);
-	if (setjmp(png_jmpbuf(png_ptr))) 
-		throw ReflowException("error during png row write");
-}
-
-void PNGWriter::close()
-{
-	/* end write */
-	png_write_end(png_ptr, info_ptr);
-	if (setjmp(png_jmpbuf(png_ptr))) 
-		throw ReflowException("Error during end of write");
-}
-
-void PNGWriter::write_splash_bitmap(SplashBitmap *bitmap) {
-    SplashColorPtr row = bitmap->getDataPtr();
-    int height = bitmap->getHeight();
-    int row_size = bitmap->getRowSize();
-    png_bytep *row_pointers = new png_bytep[height];
-
-    for (int y = 0; y < height; ++y) {
-        row_pointers[y] = row;
-        row += row_size;
-    }
-    this->writePointers(row_pointers);
-    delete[] row_pointers;
-}
-
-void calibre_png_mem_write(png_structp png_ptr, png_bytep data, png_size_t length) {
-    if (!png_ptr || length < 1) return;
-    vector<char> *buf = static_cast< vector<char>* >(png_get_io_ptr(png_ptr));
-    buf->reserve(buf->capacity() + length); 
-    do {
-        buf->push_back(static_cast<char>(*data));
-        data++; length--;
-    } while(length > 0);
-}
-
-void calibre_png_mem_flush(png_structp png_ptr) {}
-
-void PNGMemWriter::init(vector<char> *buf, int width, int height) {
-    /* initialize stuff */
-    this->png_ptr = png_create_write_struct(PNG_LIBPNG_VER_STRING, NULL, NULL, NULL);
-    if (!this->png_ptr) 
-        throw ReflowException("png_create_write_struct failed");
-
-    this->info_ptr = png_create_info_struct(png_ptr);
-    if (!this->info_ptr) 
-        throw ReflowException("png_create_info_struct failed");
-
-    if (setjmp(png_jmpbuf(this->png_ptr))) 
-        throw ReflowException("png_jmpbuf failed");
-
-    png_set_write_fn(this->png_ptr, static_cast<void *>(buf),
-            calibre_png_mem_write, calibre_png_mem_flush);
-    if (setjmp(png_jmpbuf(this->png_ptr))) 
-        throw ReflowException("png_set_write failed");
-
-
-    // Set up the type of PNG image and the compression level
-    png_set_compression_level(this->png_ptr, Z_BEST_COMPRESSION);
-
-    png_byte bit_depth = 8;
-    png_byte color_type = PNG_COLOR_TYPE_RGB;
-    png_byte interlace_type = PNG_INTERLACE_NONE;
-
-    png_set_IHDR(this->png_ptr, this->info_ptr, width, height,
-            bit_depth, color_type, interlace_type,
-            PNG_COMPRESSION_TYPE_DEFAULT, PNG_FILTER_TYPE_DEFAULT);
-
-    png_write_info(png_ptr, info_ptr);
-    if (setjmp(png_jmpbuf(png_ptr))) 
-        throw ReflowException("error during writing png info bytes");
-
-}
-
-/*
-void calibre_jpeg_error_exit (j_common_ptr cinfo)
-{
-    // cinfo->err really points to a my_error_mgr struct, so coerce pointer 
-    calibre_jpeg_err_mgr *err = (calibre_jpeg_err_mgr *)(cinfo->err);
-
-    // Always display the message. 
-    // We could postpone this until after returning, if we chose. 
-    //(*cinfo->err->output_message) (cinfo);
-
-    // Return control to the setjmp point 
-    longjmp(err->setjmp_buffer, 1);
-}
-
-
-JPEGWriter::JPEGWriter() {
-    this->cinfo.err = jpeg_std_error(&this->jerr.pub);
-    jpeg_create_compress(&this->cinfo);
-    this->jerr.pub.error_exit = calibre_jpeg_error_exit;
-    this->check();
-    this->outfile = NULL;
-}
-
-void JPEGWriter::init(int width, int height) {
-    cinfo.image_width = width; 
-    cinfo.image_height = height;
-    cinfo.input_components = 3;       // # of color components per pixel 
-    cinfo.in_color_space = JCS_RGB;
-    jpeg_set_defaults(&this->cinfo);
-    this->check();
-    jpeg_start_compress(&this->cinfo, TRUE);
-    this->check();
-}
-
-void JPEGWriter::init_io(FILE *f) {
-    jpeg_stdio_dest(&this->cinfo, f);
-    this->check();
-    this->outfile = f;
-}
-
-void JPEGWriter::check() {
-    if (setjmp(jerr.setjmp_buffer)) this->raise();
-}
-
-void JPEGWriter::raise() {
-    char buffer[JMSG_LENGTH_MAX];
-
-    // Create the message 
-    (*this->cinfo.err->format_message) ((jpeg_common_struct *)(&this->cinfo), buffer);
-    jpeg_destroy_compress(&this->cinfo);
-    throw ReflowException(buffer);
-}
-
-void JPEGWriter::write_image(JSAMPARRAY image_buffer, JDIMENSION num) {
-    size_t num_written = jpeg_write_scanlines(&this->cinfo, image_buffer, num);
-    this->check();
-    if (num_written != num) {
-        jpeg_destroy_compress(&this->cinfo);
-        throw ReflowException("Failed to write all JPEG scanlines.");
-    }
-}
-
-void JPEGWriter::write_splash_bitmap(SplashBitmap *bitmap) {
-    SplashColorPtr row = bitmap->getDataPtr();
-    int height = bitmap->getHeight();
-    int row_size = bitmap->getRowSize();
-    JSAMPARRAY row_pointers = new JSAMPLE*[height];
-
-    for (int y = 0; y < height; ++y) {
-        row_pointers[y] = row;
-        row += row_size;
-    }
-    this->write_image(row_pointers, height);
-    delete[] row_pointers;
-    jpeg_finish_compress(&this->cinfo);
-    this->check();
-    fclose(this->outfile);
-}
-
-JPEGWriter::~JPEGWriter() {
-    jpeg_destroy_compress(&this->cinfo);
-}
-*/

src/calibre/ebooks/pdf/images.h

@@ -1,135 +0,0 @@
-/**
- * Copyright 2009 Kovid Goyal <kovid@kovidgoyal.net>
- * License: GNU GPL v2+
- */
-
-
-
-#pragma once
-
-#include <vector>
-#include <GfxState.h>
-#include <splash/SplashBitmap.h>
-#include <png.h>
-#include <jpeglib.h>
-#include "utils.h"
-
-using namespace std;
-
-namespace calibre_reflow {
-
-    enum ImageType {
-        jpeg, png
-    };
-
-    class PNGWriter
-    {
-        public:
-            PNGWriter() {}
-            ~PNGWriter();
-
-            void init(FILE *f, int width, int height);
-
-            void writePointers(png_bytep *rowPointers);
-            void writeRow(png_bytep *row);
-            void write_splash_bitmap(SplashBitmap *bitmap);
-            void close();
-
-        protected:
-            png_structp png_ptr;
-            png_infop info_ptr;
-    };
-
-
-    class PNGMemWriter : public PNGWriter
-    {
-
-        public:
-            void init(vector<char> *buf, int width, int height);
-    };
-
-    class ImageInfo {
-        public:
-
-            ImageInfo(GfxState *state);
-
-        private:
-            int x0, y0;			// top left corner of image
-            int w0, h0, w1, h1;		// size of image
-            double xt, yt, wt, ht;
-            bool rotate, x_flip, y_flip;
-
-            friend class XMLImage; 
-            friend class XMLImages;
-
-    };
-
-    class XMLImage {
-        private:
-            double x, y;
-            unsigned int width, height;
-            ImageType type;
-            bool written;
-            ImageInfo info;
-
-            friend class XMLImages;
-
-        public:
-            XMLImage(GfxState *state) :
-                x(0.), y(0.), width(0), height(0), type(jpeg), written(false), info(state)
-            {}
-
-            ~XMLImage() {}
-
-            string str(size_t num, bool mask, string file_name) const;
-    };
-
-    class XMLImages {
-        private:
-            vector<XMLImage*> images;
-            vector<XMLImage*> masks;
-
-        public:
-
-            ~XMLImages() { this->clear(); }
-
-            void add_mask(GfxState *state, Object *ref, Stream *str,
-				  unsigned int width, unsigned int height, bool invert,
-				  bool interpolate, bool inline_img);
-
-            void add(GfxState *state, Object *ref, Stream *str,
-			      unsigned int width, unsigned int height, GfxImageColorMap *colorMap,
-			      bool interpolate, int *maskColors, bool inline_img);
-
-            string file_name(const XMLImage *img) const;
-            vector<string*> str() const;
-            void clear();
-    };
-/*
-    struct calibre_jpeg_err_mgr {
-        struct jpeg_error_mgr pub;    // "public" fields 
-
-        jmp_buf setjmp_buffer;    // for return to caller 
-    };
-
-    class JPEGWriter {
-        private:
-            FILE *outfile;
-            
-        protected:
-            struct jpeg_compress_struct cinfo;
-            struct calibre_jpeg_err_mgr jerr;
-
-            void raise();
-            void check();
-
-        public:
-            JPEGWriter();
-            ~JPEGWriter();
-            void init_io(FILE *f);
-            void init(int width, int height);
-            void write_image(JSAMPARRAY image_buffer, JDIMENSION number_of_scanlines);
-            void write_splash_bitmap(SplashBitmap *bitmap);
-    };
-*/
-}

src/calibre/ebooks/pdf/links.cpp

@@ -1,56 +0,0 @@
-/**
- * Copyright 2009 Kovid Goyal <kovid@kovidgoyal.net>
- * License: GNU GPL v2+
- */
-
-
-
-#include "links.h"
-#include "utils.h"
-
-using namespace std;
-using namespace calibre_reflow;
-
-XMLLink& XMLLink::operator=(const XMLLink &x) {
-    if (this==&x) return *this;
-    if (this->dest) {delete this->dest; this->dest=NULL;} 
-    this->x_min = x.x_min;
-    this->y_min = x.y_min;
-    this->x_max = x.x_max;
-    this->y_max = x.y_max;
-    this->dest = new string(*x.dest);
-    return *this;
-}
-
-bool XMLLink::in_link(double xmin,double ymin,double xmax,double ymax) const {
-    double y = (ymin + ymax)/2;
-    if (y > this->y_max) return false;
-    return (y > this->y_min) && (xmin < this->x_max) && (xmax > this->x_min);
-}
-
-string XMLLink::get_link_start() {
-    ostringstream oss;
-    oss << "<a href=\"";
-    if (this->dest) oss << encode_for_xml(*this->dest);
-    oss << "\">";
-    return oss.str();
-}
-
-XMLLinks::~XMLLinks() {
-    for(XMLLinks::iterator i = this->begin(); i != this->end(); i++)
-        delete *i;
-    this->clear();
-}
-
-bool XMLLinks::in_link(double xmin, double ymin, double xmax,
-        double ymax, XMLLinks::size_type &p) const {
-    for(XMLLinks::const_iterator i = this->begin(); i != this->end(); i++) {
-        if ( (*i)->in_link(xmin, ymin, xmax, ymax) ) {
-            p = (i - this->begin());
-            return true;
-        }
-    }
-    return false;
-}
-
-

src/calibre/ebooks/pdf/links.h

@@ -1,66 +0,0 @@
-/**
- * Copyright 2009 Kovid Goyal <kovid@kovidgoyal.net>
- * License: GNU GPL v2+
- */
-
-
-
-#pragma once
-#include <vector>
-#include <sstream>
-
-using namespace std;
-
-namespace calibre_reflow {
-
-class XMLLink {
-
-private:  
-  double x_min;
-  double y_min;
-  double x_max;
-  double y_max;
-  string* dest;
-
-public:
-  XMLLink() : dest(NULL) {}
-  XMLLink(const XMLLink& x) : 
-      x_min(x.x_min), y_min(x.y_min), x_max(x.x_max),
-      y_max(x.y_max), dest(new string(*x.dest)) {}
-  XMLLink(double x_min, double y_min, double x_max,
-          double y_max, const char *dest) :
-      x_min((x_min < x_max) ? x_min : x_max),
-      y_min((y_min < y_max) ? y_min : y_max),
-      x_max((x_max > x_min) ? x_max : x_min),
-      y_max((y_max > y_min) ? y_max : y_min),
-      dest(new string(dest)) {}
-
-  ~XMLLink() { delete this->dest; }
-
-  string* get_dest() { return this->dest; }
-  double  get_x1() const {return x_min;}
-  double  get_x2() const {return x_max;}
-  double  get_y1() const {return y_min;}
-  double  get_y2() const {return y_max;}
-
-  XMLLink& operator=(const XMLLink &x);
-  bool operator==(const XMLLink &x) const {
-      return (this->dest != NULL) && (x.dest != NULL) && 
-          this->dest->compare(*x.dest) == 0;
-  }
-  bool in_link(double xmin, double ymin, double xmax, double ymax) const;
-  string get_link_start();
-  
-};
-
-class XMLLinks : public vector<XMLLink*> {
-    public:
-        ~XMLLinks();
-
-        bool in_link(double xmin, double ymin, double xmax,
-                    double ymax, XMLLinks::size_type &p) const;
-};
-
-
-}
-   

src/calibre/ebooks/pdf/main.cpp

@@ -1,251 +0,0 @@
-/**
- * Copyright 2009 Kovid Goyal <kovid@kovidgoyal.net>
- * License: GNU GPL v2+
- */
-
-
-
-#ifndef PDF2XML
-#define UNICODE
-#define PY_SSIZE_T_CLEAN
-#include <Python.h>
-#endif
-
-#include "reflow.h"
-
-using namespace std;
-using namespace calibre_reflow;
-
-#ifndef PDF2XML
-
-extern "C" {
-
-    static PyObject *
-    pdfreflow_reflow(PyObject *self, PyObject *args) {
-        char *pdfdata;
-        Py_ssize_t size;
-        int first_page, last_page, num = 0;
-
-        if (!PyArg_ParseTuple(args, "s#ii", &pdfdata, &size, &first_page, &last_page))
-            return NULL;
-
-        try {
-            Reflow reflow(pdfdata, static_cast<std::ifstream::pos_type>(size));
-            num = reflow.render(first_page, last_page);
-        } catch (std::exception &e) {
-            PyErr_SetString(PyExc_RuntimeError, e.what()); return NULL;
-        } catch (...) {
-            PyErr_SetString(PyExc_RuntimeError,
-                    "Unknown exception raised while rendering PDF"); return NULL;
-        }
-
-        return Py_BuildValue("i", num);
-    }
-
-    static PyObject *
-    pdfreflow_get_metadata(PyObject *self, PyObject *args) {
-        char *pdfdata;
-        Py_ssize_t size;
-        map<string,string> info;
-        PyObject *cover;
-        PyObject *ans = PyDict_New();
-
-        if (!ans) return PyErr_NoMemory();
-
-        if (!PyArg_ParseTuple(args, "s#O", &pdfdata, &size, &cover))
-            return NULL;
-
-        Reflow *reflow = NULL;
-        try {
-            reflow = new Reflow(pdfdata, size);
-            info = reflow->get_info();
-            if (PyObject_IsTrue(cover)) {
-                if (reflow->numpages() > 0) {
-                    vector<char> *data = reflow->render_first_page();
-                    if (data && data->size() > 0) {
-                        PyObject *d = PyBytes_FromStringAndSize(&((*data)[0]), data->size());
-                        delete data;
-                        if (d == NULL) {delete reflow; return PyErr_NoMemory();}
-                        if (PyDict_SetItemString(ans, "cover", d) == -1) {delete reflow; return NULL;}
-                        Py_XDECREF(d);
-                    }
-                } else {
-                    if (PyDict_SetItemString(ans, "cover", Py_None) == -1) {delete reflow; return NULL;}
-                }
-            }
-        } catch (std::exception &e) {
-            PyErr_SetString(PyExc_RuntimeError, e.what()); delete reflow; return NULL;
-        } catch (...) {
-            PyErr_SetString(PyExc_RuntimeError,
-                    "Unknown exception raised while getting metadata from PDF"); delete reflow; return NULL;
-        }
-        delete reflow; reflow = NULL;
-
-
-        for (map<string,string>::const_iterator it = info.begin() ; it != info.end(); it++ ) {
-            PyObject *key = PyUnicode_Decode((*it).first.c_str(), (*it).first.size(), "UTF-8", "replace");
-            if (!key) return NULL;
-            PyObject *val = PyUnicode_Decode((*it).second.c_str(), (*it).second.size(), "UTF-8", "replace");
-            if (!val) return NULL;
-            if (PyDict_SetItem(ans, key, val) == -1) return NULL;
-            Py_XDECREF(key); Py_XDECREF(val);
-        }
-        return ans;
-    }
-
-    static PyObject *
-    pdfreflow_get_numpages(PyObject *self, PyObject *args) {
-        char *pdfdata;
-        int num = 0;
-        Py_ssize_t size;
-        map<string,string> info;
-
-        if (!PyArg_ParseTuple(args, "s#", &pdfdata, &size))
-            return NULL;
-
-        Reflow *reflow = NULL;
-        try {
-            reflow = new Reflow(pdfdata, size);
-            num = reflow->numpages();
-        } catch (std::exception &e) {
-            PyErr_SetString(PyExc_RuntimeError, e.what()); delete reflow; return NULL;
-        } catch (...) {
-            PyErr_SetString(PyExc_RuntimeError,
-                    "Unknown exception raised while getting metadata from PDF"); delete reflow; return NULL;
-        }
-
-        delete reflow; reflow = NULL;
-        return Py_BuildValue("i", num);
-    }
-
-
-    static PyObject *
-    pdfreflow_set_metadata(PyObject *self, PyObject *args) {
-        char *pdfdata;
-        Py_ssize_t size;
-        PyObject *info;
-
-        if (!PyArg_ParseTuple(args, "s#O", &pdfdata, &size, &info))
-            return NULL;
-
-        if (!PyDict_Check(info)) {
-            PyErr_SetString(PyExc_ValueError, "Info object must be a dictionary.");
-            return NULL;
-        }
-
-        char Title[10] = "Title", Author[10] = "Author", Keywords[10] = "Keywords";
-        char *keys[3] = { Title, Author, Keywords };
-        map<char *, char *> pinfo;
-        PyObject *val = NULL, *utf8 = NULL;
-
-        for (int i = 0; i < 3; i++) {
-            val = PyDict_GetItemString(info, keys[i]);
-            if (!val || !PyUnicode_Check(val)) continue;
-            utf8 = PyUnicode_AsUTF8String(val);
-            if (!utf8) continue;
-            pinfo[keys[i]] = PyString_AS_STRING(utf8);
-        }
-
-        PyObject *ans = NULL;
-        try {
-            Reflow reflow(pdfdata, static_cast<std::ifstream::pos_type>(size));
-            if (reflow.is_locked()) {
-                PyErr_SetString(PyExc_ValueError, "Setting metadata not possible in encrypeted PDFs");
-                return NULL;
-            }
-            string result = reflow.set_info(pinfo);
-            ans = PyString_FromStringAndSize(result.c_str(), result.size());
-        } catch (std::exception &e) {
-            PyErr_SetString(PyExc_RuntimeError, e.what()); return NULL;
-        } catch (...) {
-            PyErr_SetString(PyExc_RuntimeError,
-                    "Unknown exception raised while getting metadata from PDF"); return NULL;
-        }
-        return ans;
-    }
-
-    static 
-    PyMethodDef pdfreflow_methods[] = {
-        {"reflow", pdfreflow_reflow, METH_VARARGS,
-        "reflow(pdf_data, first_page, last_page)\n\n"
-                "Reflow the specified PDF. Returns the number of pages in the PDF. If last_page is -1 renders to end of document."
-        },
-        {"get_metadata", pdfreflow_get_metadata, METH_VARARGS,
-        "get_metadata(pdf_data, cover)\n\n"
-                "Get metadata and (optionally) cover from the specified PDF."
-        },
-        {"set_metadata", pdfreflow_set_metadata, METH_VARARGS,
-        "get_metadata(info_dict)\n\n"
-                "Set metadata in the specified PDF. Currently broken."
-        },
-        {"get_numpages", pdfreflow_get_numpages, METH_VARARGS,
-        "get_numpages(pdf_data)\n\n"
-                "Get number of pages in the PDF."
-        },
-
-        {NULL, NULL, 0, NULL}
-    };
-
-
-    PyMODINIT_FUNC
-    initpdfreflow(void) 
-    {
-        PyObject* m;
-
-        m = Py_InitModule3("pdfreflow", pdfreflow_methods,
-                        "Reflow a PDF file");
-
-        if (m == NULL) return;
-
-    }
-}
-
-
-#else
-
-int main(int argc, char **argv) {
-    char *memblock;
-    ifstream::pos_type size;
-    int ret = 0;
-    map<string,string> info;
-    Reflow *reflow = NULL;
-
-
-    if (argc != 2)  {
-        cerr << "Usage: " << argv[0] << " file.pdf" << endl;
-        return 1;
-    }
-
-    ifstream file (argv[1], ios::in|ios::binary|ios::ate);
-    if (file.is_open()) {
-        size = file.tellg();
-        memblock = new char[size];
-        file.seekg (0, ios::beg);
-        file.read (memblock, size);
-        file.close();
-    } else {
-        cerr << "Unable to read from: " << argv[1] << endl;
-        return 1;
-    }
-
-    try {
-        reflow = new Reflow(memblock, size);
-        info = reflow->get_info();
-        for (map<string,string>::const_iterator it = info.begin() ; it != info.end(); it++ ) {
-            cout << (*it).first << " : " << (*it).second << endl;
-        }
-        //reflow->render();
-        vector<char> *data = reflow->render_first_page();
-        ofstream file("cover.png", ios::binary);
-        file.write(&((*data)[0]), data->size());
-        delete data;
-        file.close();
-    } catch(exception &e) {
-        cerr << e.what() << endl;
-        ret = 1;
-    }
-    delete reflow;
-    delete[] memblock;
-    return ret;
-}
-#endif

src/calibre/ebooks/pdf/reflow.cpp

@@ -1,976 +0,0 @@
-/**
- * Copyright 2009 Kovid Goyal <kovid@kovidgoyal.net>
- * License: GNU GPL v2+
- */
-
-#include <Outline.h>
-#include <PDFDocEncoding.h>
-#include <poppler/ErrorCodes.h>
-#include <goo/GooList.h>
-#include <SplashOutputDev.h>
-#include <splash/SplashBitmap.h>
-#include <splash/SplashErrorCodes.h>
-#include "reflow.h"
-#include "utils.h"
-
-using namespace std;
-using namespace calibre_reflow;
-
-static const size_t num_info_keys = 8;
-static const char* info_keys[num_info_keys] = {
-    "Title", "Subject", "Keywords", "Author", "Creator", "Producer",
-    "CreationDate", "ModDate"
-};
-static char encoding[10] = "UTF-8";
-static char yes[10] = "yes";
-
-
-//------------------------------------------------------------------------
-// XMLString
-//------------------------------------------------------------------------
-
-XMLString::XMLString(GfxState *state, GooString *s, double current_font_size,
-        Fonts *fonts) : 
-    text(new vector<Unicode>(0)), x_right(new vector<double>(0)),
-    yx_next(NULL), xy_next(NULL), fonts(fonts), font_idx(0), xml_text(NULL),
-    link(NULL), x_min(0), x_max(0), y_min(0), y_max(0), col(0), dir(text_dir_unknown)
-{
-    double x = 0, y = 0;
-    GfxFont *font;
-
-    state->transform(state->getCurX(), state->getCurY(), &x, &y);
-
-    if ((font = state->getFont())) {
-        double ascent = font->getAscent();
-        double descent = font->getDescent();
-        if( ascent > 1.05 ){
-            //printf( "ascent=%.15g is too high, descent=%.15g\n", ascent, descent );
-            ascent = 1.05;
-        }
-        if( descent < -0.4 ){
-            //printf( "descent %.15g is too low, ascent=%.15g\n", descent, ascent );
-            descent = -0.4;
-        }
-        this->y_min = y - ascent * current_font_size;
-        this->y_max = y - descent * current_font_size;
-        GfxRGB rgb;
-        state->getFillRGB(&rgb);
-        GooString *name = state->getFont()->getName();
-        if (!name)
-            this->font_idx = this->fonts->add_font(NULL, current_font_size-1, rgb);
-        else
-            this->font_idx = this->fonts->add_font(
-                    new string(name->getCString()), current_font_size-1, rgb);
-
-    } else {
-        // this means that the PDF file draws text without a current font,
-        // which should never happen
-        this->y_min = y - 0.95 * current_font_size;
-        this->y_max = y + 0.35 * current_font_size;
-    }
-    if (this->y_min == this->y_max) {
-        // this is a sanity check for a case that shouldn't happen -- but
-        // if it does happen, we want to avoid dividing by zero later
-        this->y_min = y;
-        this->y_max = y + 1;
-    }
-}
-
-void XMLString::add_char(GfxState *state, double x, double y,
-            	       double dx, double dy, Unicode u) {
-    if (dir == text_dir_unknown) {
-        //dir = UnicodeMap::getDirection(u);
-        dir = text_dir_left_right;
-    } 
-
-    if (this->text->capacity() == this->text->size()) {
-        this->text->reserve(text->size()+16);
-        this->x_right->reserve(x_right->size()+16);
-    }
-    this->text->push_back(u);
-    if (this->length() == 1) {
-        this->x_min = x;
-    }
-    this->x_max = x + dx;
-    this->x_right->push_back(x_max);
-    //printf("added char: %f %f xright = %f\n", x, dx, x+dx);
-}
-
-void XMLString::end_string()
-{
-    if( this->dir == text_dir_right_left && this->length() > 1 )
-    {
-        //printf("will reverse!\n");
-        reverse(this->text->begin(), this->text->end());
-    }
-}
-
-static string encode_unicode_chars(const Unicode *u, size_t num) {
-    ostringstream oss;
-    UnicodeMap *uMap;
-    char buf[10];
-    int n;
-    if (!(uMap = globalParams->getTextEncoding())) {
-        throw ReflowException("Failed to allocate unicode map.");
-    }
-
-    for (size_t i = 0; i < num; i++) {
-        switch (u[i]) {
-            case '&': oss << "&amp;";  break;
-        	case '<': oss << "&lt;";  break;
-        	case '>': oss << "&gt;";  break;
-        	default:  
-	        {
-                // convert unicode to string
-                if ((n = uMap->mapUnicode(u[i], buf, sizeof(buf))) > 0) {
-                    buf[n] = 0;
-                    oss << buf;
-                }
-            }
-        }
-    }
-    uMap->decRefCnt();
-    return oss.str();
-}
-
-void XMLString::encode() {
-    delete this->xml_text;
-    this->xml_text = new string(encode_unicode_chars(&((*this->text)[0]), this->text->size()));
-}
-
-string XMLString::str() const {
-        ostringstream oss;
-        oss << "<text font=\"" << this->font_idx << "\" ";
-        oss << setiosflags(ios::fixed) << setprecision(2)
-            << "top=\"" << this->y_min << "\" left=\"" << this->x_min
-            << "\" width=\"" << this->x_max - this->x_min << "\" "
-            << "height=\"" << this->y_max - this->y_min << "\">";
-        oss << *this->xml_text << "</text>";
-        return oss.str();
-}
-
-XMLString::~XMLString() {
-    delete this->text; delete this->x_right; delete this->xml_text;
-}
-
-
-//------------------------------------------------------------------------
-// XMLPage
-//------------------------------------------------------------------------
-
-XMLPage::XMLPage(unsigned int num, GfxState *state, ofstream *output, Fonts* fonts) :
-    current_string(NULL), num(num), output(output), current_font_size(0.0),
-    yx_strings(NULL), xy_strings(NULL), yx_cur1(NULL), yx_cur2(NULL),
-    fonts(fonts), links(new XMLLinks())
-{
-    (*this->output) << setiosflags(ios::fixed) << setprecision(2) <<
-        "\t\t<page number=\"" << this->num << "\" width=\"" <<        
-        state->getPageWidth() << "\" height=\"" << state->getPageHeight() <<
-        "\">" << endl;
-    if (!(*this->output)) throw ReflowException(strerror(errno));
-}
-
-XMLPage::~XMLPage() {
-   (*this->output) << "\t\t</page>" << endl;
-   if (!(*this->output)) throw ReflowException(strerror(errno));
-   for (XMLString *tmp = this->yx_strings; tmp; tmp = tmp->yx_next)
-       delete tmp;
-
-   delete this->links;
-}
-
-void XMLPage::update_font(GfxState *state) {
-    GfxFont *font;
-    double *fm;
-    char *name;
-    int code;
-    double w;
-
-    current_font_size = state->getTransformedFontSize();
-
-    if ((font = state->getFont()) && font->getType() == fontType3) {
-        // This is a hack which makes it possible to deal with some Type 3
-        // fonts.  The problem is that it's impossible to know what the
-        // base coordinate system used in the font is without actually
-        // rendering the font.  This code tries to guess by looking at the
-        // width of the character 'm' (which breaks if the font is a
-        // subset that doesn't contain 'm').
-        for (code = 0; code < 256; ++code) {
-            if ((name = ((Gfx8BitFont *)font)->getCharName(code)) &&
-                name[0] == 'm' && name[1] == '\0')  break;
-            
-        }
-        if (code < 256) {
-            w = ((Gfx8BitFont *)font)->getWidth(code);
-            if (w != 0) {
-                // 600 is a generic average 'm' width -- yes, this is a hack
-                current_font_size *= w / 0.6;
-            }
-        }
-        fm = font->getFontMatrix();
-        if (fm[0] != 0) {
-            current_font_size *= fabs(fm[3] / fm[0]);
-        }
-    }
-
-}
-
-void XMLPage::draw_char(GfxState *state, double x, double y,
-                double dx, double dy,
-                double originX, double originY,
-                CharCode code, int nBytes, Unicode *u, int uLen) {
-    if ( (state->getRender() & 3) == 3)  return; //Hidden text
-    double x1, y1, w1, h1, dx2, dy2;
-    int i;
-    state->transform(x, y, &x1, &y1);
-    
-    // check that new character is in the same direction as current string
-    // and is not too far away from it before adding 
-    if (this->current_string->character_does_not_belong_to_string(state, x1)) {
-        this->end_string();
-        this->begin_string(state, NULL);
-    }
-    state->textTransformDelta(state->getCharSpace() * state->getHorizScaling(),
-                    0, &dx2, &dy2);
-    dx -= dx2;
-    dy -= dy2;
-    state->transformDelta(dx, dy, &w1, &h1);
-    if (uLen != 0) {
-        w1 /= uLen;
-        h1 /= uLen;
-    }
-    for (i = 0; i < uLen; ++i) {
-        this->current_string->add_char(state, x1 + i*w1, y1 + i*h1, w1, h1, u[i]);
-    }
-
-}
-
-void XMLPage::end_string() {
-  XMLString *p1 = NULL, *p2 = NULL;
-  double h, y1, y2;
-
-  // throw away zero-length strings -- they don't have valid xMin/xMax
-  // values, and they're useless anyway
-  if (this->current_string->length() == 0) {
-    delete this->current_string;
-    this->current_string = NULL;
-    return;
-  }
-
-  this->current_string->end_string();
-
-  // insert string in y-major list
-  h = this->current_string->height();
-  y1 = this->current_string->y_min + 0.5 * h;
-  y2 = this->current_string->y_min + 0.8 * h;
-  if (gFalse) { //rawOrder
-    p1 = this->yx_cur1;
-    p2 = NULL;
-  } else if (
-          (!this->yx_cur1 ||
-              (y1 >= this->yx_cur1->y_min &&
-               (y2 >= this->yx_cur1->y_max ||
-                this->current_string->x_max >= this->yx_cur1->x_min))) &&
-             (!this->yx_cur2 ||
-              (y1 < this->yx_cur2->y_min ||
-               (y2 < this->yx_cur2->y_max &&
-                this->current_string->x_max < this->yx_cur2->x_min)))
-             ) {
-    p1 = this->yx_cur1;
-    p2 = this->yx_cur2;
-  } else {
-    for (p1 = NULL, p2 = this->yx_strings; p2; p1 = p2, p2 = p2->yx_next) {
-      if (y1 < p2->y_min || (y2 < p2->y_max && this->current_string->x_max < p2->x_min))
-        break;
-    }
-    this->yx_cur2 = p2;
-  }
-  this->yx_cur1 = this->current_string;
-  if (p1)
-    p1->yx_next = this->current_string;
-  else
-    this->yx_strings = this->current_string;
-  this->current_string->yx_next = p2;
-  this->current_string = NULL;
-}
-
-void XMLPage::end() {
-  XMLLinks::size_type link_index = 0;
-  Fonts::size_type pos = 0;
-  XMLFont* h;
-
-  for (XMLString *tmp = this->yx_strings; tmp; tmp = tmp->yx_next) {
-     pos = tmp->font_idx;
-     h = this->fonts->at(pos);
-
-     tmp->encode();
-
-     if (this->links->in_link(
-            tmp->x_min, tmp->y_min, tmp->x_max, tmp->y_max, link_index)) {
-       tmp->link = links->at(link_index);
-     }
-  }
-
-  this->coalesce();
-
-  for (XMLString *tmp = yx_strings; tmp; tmp=tmp->yx_next) {
-    if (tmp->xml_text && tmp->xml_text->size() > 0) {
-        (*this->output) << "\t\t\t" << tmp->str() << endl;
-        if (!(*this->output)) throw ReflowException(strerror(errno));
-    }
-  }
-}
-
-static const char *strrstr( const char *s, const char *ss )
-{
-  const char *p = strstr( s, ss );
-  for( const char *pp = p; pp != NULL; pp = strstr( p+1, ss ) ){
-    p = pp;
-  }
-  return p;
-}
-
-
-static void close_tags( string *xml_text, bool &finish_a, bool &finish_italic, bool &finish_bold )
-{
-  const char *last_italic = finish_italic && ( finish_bold   || finish_a    ) ? strrstr( xml_text->c_str(), "<em>" ) : NULL;
-  const char *last_bold   = finish_bold   && ( finish_italic || finish_a    ) ? strrstr( xml_text->c_str(), "<strong>" ) : NULL;
-  const char *last_a      = finish_a      && ( finish_italic || finish_bold ) ? strrstr( xml_text->c_str(), "<a " ) : NULL;
-  if( finish_a && ( finish_italic || finish_bold ) && last_a > ( last_italic > last_bold ? last_italic : last_bold ) ) {
-    xml_text->append("</a>");
-    finish_a = false;
-  }
-  if( finish_italic && finish_bold && last_italic > last_bold ){
-    xml_text->append("</em>");
-    finish_italic = false;
-  }
-  if( finish_bold )
-    xml_text->append("</strong>");
-  if( finish_italic )
-    xml_text->append("</em>");
-  if( finish_a )
-    xml_text->append("</a>");
-}
-
-void XMLPage::coalesce() {
-    XMLString *str1, *str2, *str3;
-    XMLFont *hfont1, *hfont2;
-    double space, hor_space, vert_space, vert_overlap, size, x_limit;
-    bool add_space, found;
-    int n, i;
-    double cur_x, cur_y;
-
-    str1 = this->yx_strings;
-
-    if( !str1 ) return;
-
-    //----- discard duplicated text (fake boldface, drop shadows)
-  
-  	while (str1)
-	{
-		size = str1->y_max - str1->y_min;
-		x_limit = str1->x_min + size * 0.2;
-		found = false;
-		for (str2 = str1, str3 = str1->yx_next;
-			str3 && str3->x_min < x_limit;
-			str2 = str3, str3 = str2->yx_next)
-		{
-			if (str3->length() == str1->length() &&
-				!memcmp(str3->text, str1->text, str1->length() * sizeof(Unicode)) &&
-				fabs(str3->y_min - str1->y_min) < size * 0.2 &&
-				fabs(str3->y_max - str1->y_max) < size * 0.2 &&
-				fabs(str3->x_max - str1->x_max) < size * 0.2)
-			{
-				found = true;
-				//printf("found duplicate!\n");
-				break;
-			}
-		}
-		if (found)
-		{
-			str2->xy_next = str3->xy_next;
-			str2->yx_next = str3->yx_next;
-			delete str3;
-		}
-		else
-		{
-			str1 = str1->yx_next;
-		}
-  }		
-  
-  str1 = yx_strings;
-  
-  hfont1 = this->fonts->at(str1->font_idx);
-  if( hfont1->is_bold() )
-    str1->xml_text->insert(0, "<strong>");
-  if( hfont1->is_italic() )
-    str1->xml_text->insert(0, "<em>");
-  if (str1->get_link())
-      str1->xml_text->insert(0, str1->get_link()->get_link_start());
-  cur_x = str1->x_min; cur_y = str1->y_min;
-
-  while (str1 && (str2 = str1->yx_next)) {
-    hfont2 = this->fonts->at(str2->font_idx);
-    space = str1->y_max - str1->y_min;
-    hor_space = str2->x_min - str1->x_max;
-    vert_space = str2->y_min - str1->y_max;
-
-    vert_overlap = 0;
-    if (str2->y_min >= str1->y_min && str2->y_min <= str1->y_max)
-    {
-        vert_overlap = str1->y_max - str2->y_min;
-    } else if (str2->y_max >= str1->y_min && str2->y_max <= str1->y_max)
-    {
-        vert_overlap = str2->y_max - str1->y_min;
-    }     
-    if (
-	(
-	 (
-	   (str2->y_min < str1->y_max) 
-	   &&
-	  (hor_space > -0.5 * space && hor_space < space)
-	 ) 
-	) &&
-	(hfont1->eq_upto_inline(*hfont2)) && 
-	str1->dir == str2->dir // text direction the same
-       ) 
-    {
-      n = str1->length() + str2->length();
-      if ((add_space = hor_space > 0.1 * space)) {
-        ++n;
-      }
-  
-      str1->text->reserve((n + 15) & ~15);
-      str1->x_right->reserve((n + 15) & ~15);
-      if (add_space) {
-		  str1->text->push_back(0x20);
-		  str1->xml_text->push_back(' ');
-		  str1->x_right->push_back(str2->x_min);
-      }
-      
-      for (i = 0; i < str2->length(); i++) {
-    	str1->text->push_back(str2->text->at(i));
-	    str1->x_right->push_back(str2->x_right->at(i));
-      }
-
-      /* fix <i>, <b> if str1 and str2 differ and handle switch of links */
-      XMLLink *hlink1 = str1->get_link();
-      XMLLink *hlink2 = str2->get_link();
-      bool switch_links = !hlink1 || !hlink2 || !((*hlink1) == (*hlink2));
-      bool finish_a = switch_links && hlink1 != NULL;
-      bool finish_italic = hfont1->is_italic() && ( !hfont2->is_italic() || finish_a );
-      bool finish_bold   = hfont1->is_bold()   && 
-          ( !hfont2->is_bold()   || finish_a || finish_italic );
-      close_tags( str1->xml_text, finish_a, finish_italic, finish_bold );
-      if( switch_links && hlink2 != NULL ) {
-        string ls = hlink2->get_link_start();
-        str1->xml_text->append(ls);
-      }
-      if( ( !hfont1->is_italic() || finish_italic ) && hfont2->is_italic() )
-	    str1->xml_text->append("<em>");
-      if( ( !hfont1->is_bold() || finish_bold ) && hfont2->is_bold() )
-	    str1->xml_text->append("<strong>");
-
-
-      str1->xml_text->append(*str2->xml_text);
-      // str1 now contains href for link of str2 (if it is defined)
-      str1->link = str2->link; 
-      hfont1 = hfont2;
-      if (str2->x_max > str1->x_max) {
-    	str1->x_max = str2->x_max;
-      }
-      if (str2->y_max > str1->y_max) {
-    	str1->y_max = str2->y_max;
-      }
-      str1->yx_next = str2->yx_next;
-      delete str2;
-    } else { // keep strings separate
-      bool finish_a = str1->get_link() != NULL;
-      bool finish_bold   = hfont1->is_bold();
-      bool finish_italic = hfont1->is_italic();
-      close_tags( str1->xml_text, finish_a, finish_italic, finish_bold );
-     
-      str1->x_min = cur_x; str1->y_min = cur_y; 
-      str1 = str2;
-      cur_x = str1->x_min; cur_y = str1->y_min;
-      hfont1 = hfont2;
-      if ( hfont1->is_bold() )
-    	str1->xml_text->insert(0, "<strong>");
-      if( hfont1->is_italic() )
-    	str1->xml_text->insert(0, "<em>");
-      if( str1->get_link() != NULL ) {
-	    str1->xml_text->insert(0, str1->get_link()->get_link_start());
-      }
-    }
-  }
-  str1->x_min = cur_x; str1->y_min = cur_y;
-
-  bool finish_bold   = hfont1->is_bold();
-  bool finish_italic = hfont1->is_italic();
-  bool finish_a = str1->get_link() != NULL;
-  close_tags( str1->xml_text, finish_a, finish_italic, finish_bold );
-
-}
-
-
-//------------------------------------------------------------------------
-// XMLOutputDev
-//------------------------------------------------------------------------
-
-XMLOutputDev::XMLOutputDev(PDFDoc *doc) : 
-    current_page(NULL), output(new ofstream("index.xml", ios::trunc)),
-    fonts(new Fonts()), catalog(NULL), images(new XMLImages()), doc(doc)
-{
-    if (!(*this->output)) {
-        throw ReflowException(strerror(errno));
-    }
-    (*this->output) << "<pdfreflow>" << endl;
-    (*this->output) << "\t<pages>" << endl;
-    if (!(*this->output)) throw ReflowException(strerror(errno));
-}
-
-XMLOutputDev::~XMLOutputDev() {
-    (*this->output) << "\t</pages>" << endl;
-    if (!(*this->output)) throw ReflowException(strerror(errno));
-    (*this->output) << "\t<fonts>" << endl;
-    if (!(*this->output)) throw ReflowException(strerror(errno));
-    for (Fonts::const_iterator it = this->fonts->begin(); it < this->fonts->end(); it++) {
-        (*this->output) << "\t\t" << (*it)->str(it - this->fonts->begin()) << endl;
-        if (!(*this->output)) throw ReflowException(strerror(errno));
-    }
-    (*this->output) << "\t</fonts>" << endl;
-    if (!(*this->output)) throw ReflowException(strerror(errno));
-    (*this->output) << "</pdfreflow>" << endl;
-    if (!(*this->output)) throw ReflowException(strerror(errno));
-    this->output->close();
-    delete this->output;
-    delete this->fonts;
-    delete this->images;
-}
-
-static string get_link_dest(LinkAction *link, PDFDoc *doc) {
-  unsigned int page = 1;
-  ostringstream oss;
-
-  switch(link->getKind()) 
-  {
-      case actionGoTo:
-	  { 
-        LinkGoTo *ha = (LinkGoTo *)link;
-        LinkDest *dest = NULL;
-        if (ha->getDest() != NULL)
-            dest = ha->getDest()->copy();
-        else if (ha->getNamedDest() != NULL) {
-            dest = doc->findDest(ha->getNamedDest());
-        }
-            
-        if (dest) { 
-            if (dest->isPageRef()) {
-                Ref pageref = dest->getPageRef();
-                page = doc->findPage(pageref.num, pageref.gen);
-            }
-            else {
-                page = dest->getPageNum();
-            }
-
-            oss << "#" << page
-                << setiosflags(ios::fixed) << setprecision(2)
-                << ":l=" << dest->getLeft() 
-                << "t=" << dest->getTop();
-                //<< "r=" << dest->getRight()
-                //<< "b=" << dest->getBottom();
-            delete dest;
-        }
-        break;
-	  }
-
-      case actionGoToR:
-	  {
-        LinkGoToR *ha = (LinkGoToR *) link;
-        LinkDest *dest = NULL;
-        bool has_file = false;
-        if (ha->getFileName()) {
-            oss << ha->getFileName()->getCString();
-            has_file = true;
-        }
-        if (ha->getDest() != NULL) dest=ha->getDest()->copy();
-
-        if (dest && has_file) {
-            if (!(dest->isPageRef()))  page = dest->getPageNum();
-            delete dest;
-            oss << '#' << page;
-        }
-        break;
-      }
-      case actionURI:
-        { 
-        LinkURI *ha=(LinkURI *) link;
-        oss << ha->getURI()->getCString();
-        break;
-      }
-      case actionLaunch:
-      {
-        LinkLaunch *ha = (LinkLaunch *) link;
-        oss << ha->getFileName()->getCString();
-        break;
-	  }
-      case actionNamed: break;
-      case actionMovie: break;
-      case actionRendition: break;
-      case actionSound: break;
-      case actionJavaScript: break;
-      case actionUnknown: break;
-      default: break;
-  }
-  return oss.str();
-}
-
-void XMLOutputDev::process_link(AnnotLink* link){
-
-  double _x1, _y1, _x2, _y2;
-  int x1, y1, x2, y2;
-  
-  link->getRect(&_x1, &_y1, &_x2, &_y2);
-  cvtUserToDev(_x1, _y1, &x1, &y1);
-  
-  cvtUserToDev(_x2, _y2, &x2, &y2); 
-
-  LinkAction *a = link->getAction();
-  if (!a) return;
-  string dest = get_link_dest(a, this->doc);
-  if (dest.length() > 0) {
-      XMLLink *t = new XMLLink((double)x1, (double)y2, (double)x2, (double)y1,
-              dest.c_str());
-      this->current_page->add_link(t); 
-  }
-}
-
-
-void XMLOutputDev::endPage() {
-#ifdef POPPLER_PRE_20
-    Links *slinks = catalog->getPage(current_page->number())->getLinks(catalog);
-#else
-    Links *slinks = catalog->getPage(current_page->number())->getLinks();
-#endif
-
-    for (int i = 0; i < slinks->getNumLinks(); i++)
-    {
-        this->process_link(slinks->getLink(i));
-    }
-    delete slinks;
-    
-    this->current_page->end();
-    vector<string*> images = this->images->str();
-    for (vector<string*>::iterator it = images.begin(); it < images.end(); it++) {
-        (*this->output) << "\t\t\t" << *(*it) << endl;
-        if (!(*this->output)) throw ReflowException(strerror(errno));
-        delete *it;
-    }
-    this->images->clear();
-    delete this->current_page;
-    this->current_page = NULL;
-}
-
-
-void XMLOutputDev::drawImageMask(GfxState *state, Object *ref, Stream *str,
-				  int width, int height, GBool invert,
-				  GBool interpolate, GBool inlineImg) {
-    OutputDev::drawImageMask(state, ref, str, width, height,
-            invert, interpolate, inlineImg);
-    //this->images->add_mask();
-    cerr << "mask requested" << endl;
-}
-
-void XMLOutputDev::drawImage(GfxState *state, Object *ref, Stream *str,
-			      int width, int height, GfxImageColorMap *colorMap,
-			      GBool interpolate, int *maskColors, GBool inlineImg) {
-    this->images->add(state, ref, str,
-            static_cast<unsigned int>(width), static_cast<unsigned int>(height),
-            colorMap, interpolate, maskColors, inlineImg);
-}
-
-Reflow::Reflow(char *pdfdata, size_t sz) :
-    pdfdata(pdfdata), current_font_size(-1), doc(NULL), obj()
-{
-    int err;
-    this->obj.initNull();
-    if (globalParams == NULL) {
-        globalParams = new GlobalParams();
-        if (!globalParams)
-            throw ReflowException("Failed to allocate Globalparams");
-    }
-    MemStream *str = new MemStream(pdfdata, 0, sz, &this->obj);
-    this->doc = new PDFDoc(str, NULL, NULL);
-
-    if (!this->doc->isOk()) {
-        err = this->doc->getErrorCode();
-        ostringstream stm;
-        if (err == errEncrypted) 
-            stm << "PDF is password protected.";
-        else {
-            stm << "Failed to open PDF file";
-            stm << " with error code: " << err;
-        }
-        delete this->doc;
-        this->doc = NULL;
-        throw ReflowException(stm.str().c_str());
-    }
-
-}
-
-int
-Reflow::render(int first_page, int last_page) {
-
-    if (!this->doc->okToCopy()) 
-        cout << "Warning, this document has the copy protection flag set, ignoring." << endl;
-
-    globalParams->setTextEncoding(encoding);
-
-    int doc_pages = doc->getNumPages();
-    if (last_page < 1 || last_page > doc_pages) last_page = doc_pages;
-    if (first_page < 1) first_page = 1;
-    if (first_page > last_page) first_page = last_page;
-
-    XMLOutputDev *xml_out = new XMLOutputDev(this->doc);
-    doc->displayPages(xml_out, first_page, last_page,
-              96, //hDPI
-              96, //vDPI
-              0, //rotate
-		      true, //UseMediaBox
-              true, //Crop
-              false //Printing
-    );
-    
-    if (last_page - first_page == doc_pages - 1)
-        this->dump_outline();
-
-    delete xml_out;
-
-    return doc_pages;
-}
-
-void Reflow::dump_outline() {
-	Outline *outline = this->doc->getOutline();
-    if (!outline) return;
-    GooList *items = outline->getItems();
-    if ( !items || items->getLength() < 1 )
-        return;
-
-    ostringstream *output = new ostringstream();
-    (*output) << "<outline>" << endl;
-    this->outline_level(output, items);
-    (*output) << "</outline>" << endl;
-    ofstream of("outline.xml", ios::trunc);
-    of << output->str();
-    if (!of) throw ReflowException("Error writing outline file");
-    of.close();
-    delete output;
-}
-
-static inline void outline_tabs(ostringstream *o, int level) {
-    for (int i = 0; i < level; i++)
-        (*o) << "\t";
-}
-
-void Reflow::outline_level(ostringstream *oss, GooList *items, int level)
-{
-    int num_of_items = items->getLength();
-    if (num_of_items > 0) {
-        outline_tabs(oss, level);
-        (*oss) << "<links level=\"" << level << "\">" << endl;
-
-        for (int i = 0; i < num_of_items; i++) {
-            OutlineItem* item = (OutlineItem *)items->get(i);
-            Unicode *u = item->getTitle();
-            string title = encode_unicode_chars(u, item->getTitleLength());
-            if (title.size() < 1) continue;
-            outline_tabs(oss, level+1);
-            (*oss) << "<link open=\"" << (item->isOpen()?"yes":"no") << "\"";
-            LinkAction *a = item->getAction();
-            if (a != NULL)
-                (*oss) << " dest=\"" << get_link_dest(a, this->doc) << "\"";
-            (*oss) << ">" << title << "</link>" << endl;
-            item->open();
-            GooList *children = item->getKids();
-            if (children)
-                outline_level(oss, children, level+1);
-        }
-    }
-}
-
-Reflow::~Reflow() {
-    delete this->doc;
-}
-
-map<string, string> Reflow::get_info() {
-    Object info;
-    map<string, string> ans;
-    string val;
-    globalParams->setTextEncoding(encoding);
-
-    this->doc->getDocInfo(&info);
-    if (info.isDict()) {
-        for(size_t i = 0; i < num_info_keys; i++) {
-            val = this->decode_info_string(info.getDict(), info_keys[i]);
-            if (val.size() > 0) {
-                ans[string(info_keys[i])] = string(val);
-            }
-        }
-    }
-    return ans;
-}
-
-string Reflow::decode_info_string(Dict *info, const char *key) const {
-    Object obj;
-    GooString *s1;
-    bool is_unicode;
-    Unicode u;
-    char buf[8];
-    int i, n;
-    ostringstream oss;
-    char *tmp = new char[strlen(key)+1];
-    strncpy(tmp, key, strlen(key)+1);
-    UnicodeMap *umap;
-    if (!(umap = globalParams->getTextEncoding())) {
-        throw ReflowException("Failed to allocate unicode map.");
-    }
-
-
-    if (info->lookup(tmp, &obj)->isString()) {
-        s1 = obj.getString();
-        if ((s1->getChar(0) & 0xff) == 0xfe &&
-        (s1->getChar(1) & 0xff) == 0xff) {
-            is_unicode = true;
-            i = 2;
-        } else {
-            is_unicode = false;
-            i = 0;
-        }
-        while (i < obj.getString()->getLength()) {
-            if (is_unicode) {
-                u = ((s1->getChar(i) & 0xff) << 8) |
-                (s1->getChar(i+1) & 0xff);
-                i += 2;
-            } else {
-                u = pdfDocEncoding[s1->getChar(i) & 0xff];
-                ++i;
-            }
-            n = umap->mapUnicode(u, buf, sizeof(buf));
-            buf[n] = 0;
-            oss << buf;
-        }
-    }
-    obj.free();
-    delete[] tmp;
-    return oss.str();
-}
-
-vector<char>* Reflow::render_first_page(bool use_crop_box, double x_res,
-        double y_res) {
-    if (this->numpages() < 1) throw ReflowException("Document has no pages.");
-    globalParams->setTextEncoding(encoding);
-    globalParams->setEnableFreeType(yes);
-    globalParams->setAntialias(yes);
-    globalParams->setVectorAntialias(yes);
-    
-    SplashColor paper_color;
-    paper_color[0] = 255;
-    paper_color[1] = 255;
-    paper_color[2] = 255;
-    SplashOutputDev *out = new SplashOutputDev(splashModeRGB8, 4, false, paper_color, true, true);
-    out->setVectorAntialias(true);
-    if (!out) {
-        throw ReflowException("Failed to allocate SplashOutputDev");
-    }
-    try {
-#ifdef POPPLER_PRE_20
-        out->startDoc(doc->getXRef());
-#else
-        out->startDoc(doc);
-#endif
-        out->startPage(1, NULL);
-
-        double pg_w, pg_h;
-        int pg = 1;
-
-        if (use_crop_box) {
-            pg_w = this->doc->getPageCropWidth(pg);
-            pg_h = this->doc->getPageCropHeight(pg);
-        } else {
-            pg_w = this->doc->getPageMediaWidth(pg);
-            pg_h = this->doc->getPageMediaHeight(pg);
-        }
-
-        pg_w *= x_res/72.;
-        pg_h *= y_res/72.;
-
-        int x=0, y=0;
-        this->doc->displayPageSlice(out, pg, x_res, y_res, 0,
-                !use_crop_box, false, false, x, y, pg_w, pg_h);
-    } catch(...) { delete out; throw; }
-
-    SplashBitmap *bmp = out->takeBitmap();
-    out->endPage();
-    delete out; out = NULL;
-    PNGMemWriter writer;
-    vector<char> *buf = new vector<char>();
-    try {
-        writer.init(buf, bmp->getWidth(), bmp->getHeight()); 
-        writer.write_splash_bitmap(bmp);
-        writer.close();
-    } catch(...) { delete buf; delete bmp; throw; }
-    delete bmp;
-    return buf;
-}
-
-class MemOutStream : public OutStream {
-    private:
-        ostringstream out;
-
-    public:
-        MemOutStream() :OutStream() {}
-        ~MemOutStream() {}
-        void close() {}
-        int getPos() { return out.tellp(); }
-        void put(char c) { out.put(c); }
-        void printf (const char *format, ...) {
-            vector<char> buf;
-            size_t written = strlen(format)*5;
-            va_list ap;
-            do {
-                buf.reserve(written + 20);
-                va_start(ap, format);
-                written = vsnprintf(&buf[0], buf.capacity(), format, ap);
-                va_end(ap);
-            } while (written >= buf.capacity());
-            out.write(&buf[0], written);            
-        }
-};
-
-string Reflow::set_info(map<char *, char *> sinfo) {
-    XRef *xref = this->doc->getXRef();
-    if (!xref) throw ReflowException("No XRef table");
-    Object *trailer_dict = xref->getTrailerDict();
-    if (!trailer_dict || !trailer_dict->isDict()) throw ReflowException("No trailer dictionary");
-    Object tmp;
-    char INFO[5] = "Info";
-    Object *info = trailer_dict->dictLookup(INFO, &tmp);
-    if (!info) {
-        info = new Object();
-        info->initDict(xref);
-    }
-    if (!info->isDict()) throw ReflowException("Invalid info object");
-
-    for (map<char *, char *>::iterator it = sinfo.begin(); it != sinfo.end(); it++) {
-        Object *tmp = new Object();
-        tmp->initString(new GooString((*it).second));
-        info->dictSet((*it).first, tmp);
-    }
-
-    trailer_dict->dictSet(INFO, info);
-    char out[20] = "/t/out.pdf";
-    this->doc->saveAs(new GooString(out), writeForceRewrite);
-    string ans;
-    return ans;
-}
-

src/calibre/ebooks/pdf/reflow.h

@@ -1,253 +0,0 @@
-/**
- * Copyright 2009 Kovid Goyal <kovid@kovidgoyal.net>
- * License: GNU GPL v2+
- * Based on pdftohtml from the poppler project.
- */
-
-#pragma once
-#define UNICODE
-
-#ifdef _WIN32 
-#include <poppler/Object.h>
-#elif defined(_OSX)
-#include <poppler/Object.h>
-#else
-#include <Object.h>
-#endif
-
-#include <PDFDoc.h>
-#include <GlobalParams.h>
-#include <GfxState.h>
-#include <GfxFont.h>
-#include <OutputDev.h>
-#include <Link.h>
-#include <UnicodeMap.h>
-#include <cmath>
-#include <exception>
-#include <string>
-#include <sstream>
-#include <vector>
-#include <iostream>
-#include <algorithm>
-#include <fstream>
-#include <iomanip>
-#include <map>
-#include <errno.h>
-#include "fonts.h"
-#include "links.h"
-#include "images.h"
-
-using namespace std;
-
-namespace calibre_reflow {
-
-#ifdef POPPLER_OLD_LINK_TYPE
-#define AnnotLink Link
-#endif
-
-enum UnicodeTextDirection {
-  text_dir_unknown,
-  text_dir_left_right,
-  text_dir_right_left,
-  text_dir_top_bottom
-};
-
-class Reflow {
-
-    private:
-        char *pdfdata;
-        double current_font_size;
-        PDFDoc *doc;
-        Object obj;
-
-        string decode_info_string(Dict *info, const char *key) const;
-        void outline_level(ostringstream *oss, GooList *items,
-                            int level=1);
-
-    public:
-        Reflow (char *xpdfdata, size_t sz);
-        ~Reflow();
-        
-        /* Convert the PDF to XML. All files are output to the current directory */
-        int render(int first_page, int last_page);
-
-        /* Get the PDF Info Dictionary */
-        map<string, string> get_info();
-
-        /* True if the PDF is encrypted */
-        bool is_locked() const { return !this->doc || this->doc->isEncrypted(); }
-
-        /* Return the first page of the PDF, rendered as a PNG image */
-        vector<char>* render_first_page(bool use_crop_box=true, double x_res=150.0,
-            double y_res = 150.0);
-
-        /* Dump the PDF outline as the file outline.xml in the current directory */
-        void dump_outline();
-
-        /* Set the info dictionary. Currently broken. */
-        string set_info(map<char *, char *> info);
-
-        /* Number of pages in the document */
-        int numpages() { return this->doc->getNumPages(); }
-};
-
-class XMLString {
-    private:
-        vector<Unicode> *text;  // the text
-        vector<double> *x_right; // right-hand x coord of each char
-        XMLString *yx_next;	    	// next string in y-major order
-        XMLString *xy_next;		    // next string in x-major order
-        Fonts *fonts;
-        Fonts::size_type font_idx;
-        string *xml_text;
-        XMLLink *link;
-
-        double x_min, x_max;		// bounding box x coordinates
-        double y_min, y_max;		// bounding box y coordinates
-        int col;		        	// starting column
-        UnicodeTextDirection dir;	// direction (left to right/right to left)
-
-        friend class XMLPage;
-
-    public:
-        XMLString(GfxState *state, GooString *s, double current_font_size, Fonts *fonts);
-        ~XMLString();
-
-        bool character_does_not_belong_to_string(GfxState *state, double x1) {
-            return this->length() > 0 && 
-                fabs(x1 - x_right->at(this->length()-1)) > 0.1 * (y_max - y_min);
-        }
-        
-        void add_char(GfxState *state, double x, double y,
-            	       double dx, double dy, Unicode u);
-
-        void end_string();
-        inline int length() const { return this->text->size(); }
-        inline double height() const { return y_max - y_min; }
-        void encode();
-        XMLLink* get_link() { return this->link; }
-        string str() const;
-};
-
-class XMLPage {
-    private:
-        XMLString *current_string;
-        unsigned int num;
-        ofstream *output;
-        double current_font_size;
-        XMLString *yx_strings;	// strings in y-major order
-        XMLString *xy_strings;	// strings in x-major order
-        XMLString *yx_cur1, *yx_cur2;	// cursors for yxStrings list
-        Fonts *fonts;
-        XMLLinks *links;
-        void coalesce();
-
-    public:
-        XMLPage(unsigned int num, GfxState *state, ofstream *output, Fonts* fonts);
-        ~XMLPage();
-
-        void update_font(GfxState *state);
-
-        void begin_string(GfxState *state, GooString *s) {
-            this->current_string = new XMLString(state, s,
-                    this->current_font_size, this->fonts);
-        }
-        
-        void draw_char(GfxState *state, double x, double y,
-                double dx, double dy,
-                double originX, double originY,
-                CharCode code, int nBytes, Unicode *u, int uLen);
-
-        void end_string();
-
-        void end();
-
-        void add_link(XMLLink *t) { this->links->push_back(t); }
-
-        unsigned int number() const { return this->num; }
-};
-
-class XMLOutputDev : public OutputDev {
-  public:
-    XMLOutputDev(PDFDoc *doc);
-    virtual ~XMLOutputDev();
-    //---- get info about output device
-
-    // Does this device use upside-down coordinates?
-    // (Upside-down means (0,0) is the top left corner of the page.)
-    virtual GBool upsideDown() { return gTrue; }
-
-    // Does this device use drawChar() or drawString()?
-    virtual GBool useDrawChar() { return gTrue; }
-
-    // Does this device use beginType3Char/endType3Char?  Otherwise,
-    // text in Type 3 fonts will be drawn with drawChar/drawString.
-    virtual GBool interpretType3Chars() { return gFalse; }
-
-    // Does this device need non-text content?
-    virtual GBool needNonText() { return gTrue; }
-
-    //----- initialization and control
-
-    virtual GBool checkPageSlice(Page *page, double hDPI, double vDPI,
-                                int rotate, GBool useMediaBox, GBool crop,
-                                int sliceX, int sliceY, int sliceW, int sliceH,
-                                GBool printing, Catalog * catalogA,
-                                GBool (* abortCheckCbk)(void *data) = NULL,
-                                void * abortCheckCbkData = NULL)
-    {
-    this->catalog = catalogA;
-    return gTrue;
-    }
-
-
-    // Start a page.
-    virtual void startPage(int page_num, GfxState *state) {
-        this->current_page = new XMLPage(page_num, state, this->output, this->fonts);
-    }
-
-
-    // End a page.
-    virtual void endPage();
-
-    //----- update text state
-    virtual void updateFont(GfxState *state) {current_page->update_font(state);}
-
-    //----- text drawing
-    virtual void beginString(GfxState *state, GooString *s) {
-        this->current_page->begin_string(state, s);
-    }
-    virtual void endString(GfxState *state) {
-        this->current_page->end_string();
-    }
-    virtual void drawChar(GfxState *state, double x, double y,
-                double dx, double dy,
-                double originX, double originY,
-                CharCode code, int nBytes, Unicode *u, int uLen) {
-        this->current_page->draw_char(state, x, y, dx, dy, originX,
-                originY, code, nBytes, u, uLen);
-    }
-    
-    virtual void drawImageMask(GfxState *state, Object *ref, 
-                    Stream *str,
-                    int width, int height, GBool invert,
-                    GBool interpolate, GBool inlineImg);
-    virtual void drawImage(GfxState *state, Object *ref, Stream *str,
-                int width, int height, GfxImageColorMap *colorMap,
-                GBool interpolate, int *maskColors, GBool inlineImg);
-
-    //new feature    
-    virtual int DevType() {return 1234;}
-
-  private:
-    XMLPage *current_page;
-    ofstream *output;                   // xml file
-    Fonts *fonts;
-    Catalog *catalog;
-    XMLImages *images;
-    PDFDoc *doc;
-
-    void process_link(AnnotLink* link);
-
-};
-}