1. Developer's Guide
  2. Using the C/C++ API
  3. UNIX Implementation Details

5 UNIX Implementation Details

This chapter describes the implementation of the Web View Export SDK on the UNIX platform. The UNIX implementation of the Export product set is delivered as a set of shared libraries.

For a list of the currently supported platforms, see Outside In Technology and click links under Certified Platforms and Supported Formats from Get Started page.

This chapter includes the following sections:

5.1 Installation

To install the demo version of the SDK, copy the tgz file corresponding to your platform (available on the web site) to a local directory of your choice. Decompress the tgz file and then extract from the resulting tar file as follows:

gunzip tgzfile
tar xvf tarfile

The installation directory should contain the following directory structure.

Directory Description

/redist

Contains a working copy of the UNIX version of the technology.

/sdk/assets

Contains Javascript, CSS, and image files that are used by the Web View Export output.

/sdk/common

Contains the C include files needed to build or rebuild the technology.

/sdk/demo

Contains the compiled executables of the sample applications.

/sdk/resource

Contains localization resource files. For more details, see Changing Resources.

/sdk/samplecode

Contains a subdirectory holding the source code for a sample application. For more details, see Sample Applications.

/sdk/samplefiles

Contains sample files designed to exercise the technology.

5.1.1 NSF Support

Notes Storage Format (NSF) files are produced by the Lotus Notes Client or the Lotus Domino server. The NSF filter is the only Outside In filter that requires the native application to be present to filter the input documents. Due to integration with an outside application, NSF support will not work with redirected I/O nor will it work when an NSF file is embedded in another file. Lotus Domino version 8 must be installed on the same machine as OIT. The NSF filter is currently only supported on the Win32, Win x86-64, Linux x86-32, and Solaris Sparc 32 platforms. SCCOPT_LOTUSNOTESDIRECTORY is a Windows-only option and is ignored on Unix.

Additional steps must be taken to prepare the system. It is necessary to know the name of the directory in which Lotus Domino has been installed. On Linux, this default directory is /opt/ibm/lotus/notes/latest/linux. On Solaris, it is /opt/ibm/lotus/notes/latest/sunspa.

  • In the Lotus Domino directory, check for the existence of a file called "notes.ini". If the file "notes.ini" does not exist, create it in that directory and ensure that it contains the following single line:

    [Notes]

  • Add the Lotus Domino directory to the $LD_LIBRARY_PATH environment variable.

  • Set the environment variable $Notes_ExecDirectory to the Lotus Domino directory.

5.2 Libraries and Structure

On UNIX platforms the Oracle Outside In products are delivered with a set of shared libraries. All libraries should be installed to a single directory. Depending upon your application, you may also need to add that directory to the system's runtime search path. For more details, see Environment Variables.

The following is a brief description of the included libraries and support files. In instances where a file extension is listed as .*, the file extension varies for each UNIX platform (sl on HP-UX, so on Linux and Solaris).

5.2.1 API Libraries

These libraries implement the API. They should be linked with the developer's application.

Library Description HTML Export Image Export PDF Export Search Export XML Export Web View Export

libsc_da.*

Data Access module

X

X

X

X

X

X

libsc_ex.*

Export module

X

X

X

X

X

X

libsc_fi.*

File Identification module (identifies files based on their contents).

X

X

X

X

X

X

The File ID Specification may not be used directly by any application or workflow without it being separately licensed expressly for that purpose.

5.2.2 Support Libraries

The following libraries are used for support.

Library Description HTML Export Image Export PDF Export Search Export XML Export Web View Export

libccflex.*

A data model adapter that converts from stream model utilized by Outside In filters to the FlexionDoc Tree model used as a basis by XML Export.

X

libexpatw.*

A third-party XML parser.

X

liboc_emul.*

Output component emulation module

X

X

X

X

X

X

libos_gd.*

The internal rendering GDI implementation. This library is only supported on Linux (32- and 64-bit Intel), Solaris (32-bit SPARC), HP-UX (32-bit RISC), and AIX (32-bit PPC).

X

X

X

X

libos_pdf.*

PDF generation module

X

libos_xwin.*

The native GDI implementation

X

X

X

X

libsc_anno.*

The annotation module

X

X

X

X

libsc_ca.*

Content Access module (provides organized chunker data for the developer)

X

X

X

X

libsc_ch.*

Chunker (provides caching of and access to filter data for the export engines)

X

X

X

X

X

X

libsc_du.*

Display Utilities module (includes text formatting)

X

X

X

X

X

X

libsc_fmt.*

Formatting module (resolves numbers to formatted strings)

X

X

X

X

X

X

libsc_fut.*

Filter utility module

X

X

X

X

X

X

libsc_ind.*

Indexing engine. In Search Export, it handles common functionality.

X

X

X

X

X

libsc_lo.*

Localization library (all strings, menus, dialogs and dialog procedures reside here)

X

X

X

X

X

X

libsc_sd.*

Schema Definition Module Manager (brokers multiple Schema Definition Modules)

X

libsc_ut.*

Utility functions, including IO subsystem

X

X

X

X

X

X

libsc_xp.*

XPrinter bridge

X

X

X

X

libsdflex.*

Schema Definition module (handles conversion of XML string names and attribute values to compact binary representations and vice versa)

X

libwv_core.*

The Abstraction layer

X

X

X

X

X

X

libwv_gdlib.so

The GDI rendering module. This library is only supported on Linux (32- and 64-bit Intel), Solaris (32-bit SPARC), HP-UX (32-bit RISC), and AIX (32-bit PPC).

X

X

X

X

5.2.3 Engine Libraries

The following libraries are used for display purposes.

Library Description    HTML Export Image Export PDF Export Search Export XML Export Web View Export

libde_bmp.*

Raster rendering engine (TIFF, GIF, BMP, PNG, PCX…)

X

X

X

libde_vect.*

Vector/Presentation rendering engine (PowerPoint, Impress, Freelance)

X

X

X

X

X

libde_ss.*

Spreadsheet/Database (Excel, Calc, Lotus 123)

X

X

X

libde_tree*

Archive (ZIP, GZIP, TAR…)

X

X

libde_wp.*

Document (Word, Writer, WordPerfect)

X

X

X

X

5.2.4 Filter and Export Filter Libraries

The following libraries are used for filtering.

libex_gdsf must be linked with libsc_img.* at compile time. This forces the filter to be dependent on libsc_img.* at runtime, even though that module may not be used directly. If you want to reduce your application's physical footprint, you can experiment with unlinking libsc_img.*.

Library Description HTML Export Image Export PDF Export Search Export XML Export Web View Export

libvs_*.*

Filters for specific file types (there are more than 150 of these filters, covering more than 600 file formats)

X

X

X

X

X

X

libex_gdsf.*

Export filter for GIF, JPEG, and PNG graphics files

X

X

X

libex_h5.*

Export filter for HTML5 files

X

libsc_img.*

Image conversion module

X

X

X

X

libex_itext.*

Export filter for SearchText

X

libex_html.*

Export filter for HTML files

X

libex_img.*

Extended image conversion module

X

libex_xml.*

Export filter for XML files using the Flexiondoc schema

X

libex_page.*

Export filter for XML files using the PageML schema

X

libex_pagelayout.*

Page Layout module

X

libex_ixml.*

Export filters for XML files using the SearchML schema

X

libex_ihtml.*

Export filter for SearchHTML

X

5.2.5 Premier Graphics Filters

The following are graphics filters.

Library Description HTML Export Image Export PDF Export Search Export XML Export Web View Export

libi*.*

These files are the import filters for premier graphics formats.

X

X

X

X

X

X

libis_unx2.*

Interface to premier graphics filters

X

X

X

X

X

X

5.2.6 Additional Files

The following files are also used.

Library Description HTML Export Image Export PDF Export Search Export XML Export Web View Export

adinit.dat

Support file for the vsacad and vsacd2 filters

X

X

X

X

X

X

ccbf.so

Internal

X

cmmap000.bin

Tables for character mapping (all character sets)

X

X

X

X

X

X

cmmap000.sbc

Tables for character mapping (single-byte character sets). This file is located in the /common directory.

X

X

X

X

X

X

cmmap000.dbc

Identical to cmmap000.bin, but renamed for clarity (.dbc = double-byte character). This file is located in the common directory.

X

X

X

X

X

X

exbf.so

Internal

X

flexiondoc.dtd

The DTD version of the Flexiondoc schema

X

flexiondoc.xsd

The schema version of the Flexiondoc schema

X

libfreetype.so.6

TrueType font rendering module for the GD output solution. 32-bit Linux and Solaris Sparc only.

X

X

X

X

X

oitnsf.id

Support file for the vsnsf filter.

X

X

X

X

X

X

pageml.dtd

The Document Type Definition for the PageML schema

X

pageml.xsd

The Extensible Schema Definition for the PageML schema

X

searchml3.dtd

The Document Type Definitions for the SearchML schema

X

searchml3.xsd

The Extensible Schema Definitions for the SearchML schema

X

5.3 The Basics

Sample applications are provided with the SDK. These applications demonstrate most of the concepts described in this manual. For a complete description of the sample applications, see Sample Applications.

5.3.1 What You Need in Your Source Code

Any source code that uses this product should #include the file sccex.h and #define UNIX. For example, a 32-bit UNIX application might have a source file with the following lines: 

#define UNIX
#include <sccex.h>

and a 64-bit UNIX application might have a source file with the following lines:

#define UNIX
#define UNIX_64
#include <sccex.h>

5.3.2 Information Storage

This software is based on the Oracle Outside In Viewer Technology (or simply "Viewer Technology"). A file of default options is always created, and a list of available filters and a list of available display engines are also built by the technology, usually the first time the product runs (for UNIX implementations). You do not need to ship these lists with your application.

Lists are stored in the $HOME/.oit directory. If the $HOME environment variable is not set, the files are put in the same directory as the Oracle Outside In Technology. If a /.oit directory does not exist in the user's $HOME directory, the .oit directory is created automatically by the technology. The files are automatically regenerated if corrupted or deleted.

The files are:

  • *.d: Display engine list

  • *.opt: Persistent options

The filenames are intended to be unique enough to avoid conflict for any combination of machine name and install directory. This is intended to prevent problems with version conflicts when multiple versions of the Oracle Outside In Technology and/or other Oracle Outside In Technology-based products are installed on a single system. The filenames are built from an 11-character string derived from the directory the Oracle Outside In Technology resides in and the name of the machine it is being run on. The string is generated by code derived from the RSA Data Security, Inc. MD5 Message-Digest Algorithm.

The products still function if these files cannot be created for some reason. In that situation, however, significant performance degradation should be expected.

5.4 Character Sets

The strings passed in the UNIX API are ISO8859-1 by default.

To optimize performance on systems that do not require DBCS support, a second character mapping bin file, that does not contain any of the DBCS pages, is included. The second bin file gives additional performance benefits for English documents, but cannot handle DBCS documents. To use the new bin file, replace the cmmap000.bin with the new bin file, cmmap000.sbc. For clarity, a copy of the cmmap000.bin file (cmmap000.dbc) is also included. Both cmmap000.sbc and cmmap000.dbc are located in the /sdk/common directory of the technology.

5.5 Runtime Considerations

The following is information to consider during run-time.

5.5.1 OLE2 Objects

Some documents that the developer is attempting to convert may contain embedded OLE2 objects. There are platform-dependent limits on what the technology can do with OLE2 objects. However, Oracle Outside In attempts to take advantage of the fact that some documents accompany an OLE2 embedding with a graphic "snapshot," in the form of a Windows metafile.

On all platforms, when a metafile snapshot is available, the technology uses it to convert the object. When a metafile snapshot is not available on UNIX platforms, the technology is unable to convert the OLE2 object.

5.5.2 Signal Handling

These products trap and handle the following signals:

  • SIGABRT

  • SIGBUS

  • SIGFPE

  • SIGILL

  • SIGINT

  • SIGSEGV

  • SIGTERM

Developers who wish to override our default handling of these signals should set up their own signal handlers. This may be safely done after the developer's application has called DAInitEx().

Note:

The Java Native Interface (JNI) allows Java code to call and be called by native code (C/C++ in the case of OIT). You may run into problems if Java isn't allowed to handle signals and forward them to OIT. If OIT catches the signals and forwards them to Java, the JVMs will sometimes crash. OIT installs signal handlers when DAInitEx() is called, so if you call OIT after the JVM is created, you will need to use libjsig. Refer here for more information:

http://www.oracle.com/technetwork/java/javase/index-137495.html

5.5.3 Runtime Search Path and $ORIGIN

Libraries and sample applications are all built with the $ORIGIN variable as part of the binaries' runtime search path. This means that at runtime, OIT libraries will automatically look in the directory they were loaded from to find their dependent libraries. You don't necessarily need to include the technology directory in your LD_LIBRARY_PATH or SHLIB_PATH.

As an example, an application that resides in the same directory as the OIT libraries and includes $ORIGIN in its runtime search path will have its dependent OIT libraries found automatically. You will still need to include the technology directory in your linker's search path at link time using something like -L and possibly -rpath-link.

Another example is an application that loads OIT libraries from a known directory. The loading of the first OIT library will locate the dependent libraries.

Note:

This feature does not work on AIX and FreeBSD.

5.6 Environment Variables

Several environment variables may be used at run time. Following is a short summary of those variables and their usage.

Variable Description

$LD_LIBRARY_PATH (FreeBSD, HP-UX Itanium 64, Linux, Solaris)

$SHLIB_PATH (HP-UX PA-RISC 32)

$LIBPATH (AIX, iSeries)

These variables help your system's dynamic loader locate objects at runtime. If you have problems with libraries failing to load, try adding the path to the Oracle Outside In libraries to the appropriate environment variable. See your system's manual for the dynamic loader and its configuration for details.

Note that for products that have a 64-bit PA-RISC, 64-bit Solaris and Linux PPC/PPC64 distributable, they will also go under $LD_LIBRARY_PATH.

$HOME

Must be set to allow the system to write the option, filter and display engine lists. For details, see Information Storage.

5.7 Default Font Aliases

Outside In Technology (OIT) will use the fonts installed on the host system. If a file being converted with OIT uses fonts that are available on the host system, no substitution should occur and the original font from the input file will be used. If the original font used in the input file is not available on the host system, then OIT will first check to see if an alias has been set for the font using the SCCOPT_PRINTFONTALIAS option (see documentation for details on using this option). If there is an alias available, and the alias font is available on the host system, then OIT will use this font. If no alias is set or the alias font is not available on the host system, then a substitution will occur. The first attempt at a substitution will use the default font specified by the SCCOPT_DEFAULTPRINTFONT option. If this font has the glyphs to be rendered, it will be used ahead of all other potential substitutions. In some cases, the default font cannot be used because it does not contain the glyphs required to render the text from the input file.

For example, a default font of Arial may not contain glyphs required to render certain Asian languages. In these cases, OIT will pick another font that does have the glyphs, if one exists. The mechanism for picking that other font is not very predictable, and often leads to bad results (picking a serifed font for a non-serifed, variable width for a fixed width, etc.). Therefore, the best solution for users is to have as many fonts available to OIT as possible. This will avoid font substitutions and provide the most accurate rendering of the original file.

Note that font substitutions can lead to different wrapping, which can lead to different numbers of pages rendered by OIT versus the native application. This further underscores the importance of a host system with as many fonts as possible.

The technology includes the following default font alias map for UNIX platforms. The first value is the original font, and the second is the alias.

  • 61 = Liberation Sans

  • Andale Mono = Liberation Sans

  • Courier = Liberation Sans

  • Courier New = Liberation Sans

  • Lucida Console = Liberation Sans

  • MS Gothic = Liberation Sans

  • MS Mincho = Liberation Sans

  • OCR A Extended = Liberation Sans

  • OCR B = Liberation Sans

  • Agency FB = Liberation Sans

  • Arial = Liberation Sans

  • Arial Black = Liberation Sans

  • Arial Narrow = Liberation Sans

  • Arial Rounded MT = Liberation Sans

  • Arial Unicode MS = Liberation Sans

  • Berline Sans FB = Liberation Sans

  • Calibri = Liberation Sans

  • Frank Gothic Demi = Liberation Sans

  • Frank Gothic Medium Cond = Liberation Sans

  • Franklin Gothic Book = Liberation Sans

  • Futura = Liberation Sans

  • Geneva = Liberation Sans

  • Gill Sans = Liberation Sans

  • Gill Sans MT = Liberation Sans

  • Lucida Sans Regular = Liberation Sans

  • Lucida Sans Unicode = Liberation Sans

  • Modern No. 20 = Liberation Sans

  • Tahoma = Liberation Sans

  • Trebuchet MS = Liberation Sans

  • Tw Cen MT = Liberation Sans

  • Verdana = Liberation Sans

  • Albany = Liberation Sans

  • Franklin Gothic = Liberation Sans

  • Franklin Demi = Liberation Sans

  • Franklin Demi Cond = Liberation Sans

  • Franklin Gothic Heavy = Liberation Sans

  • Algerian = Liberation Serif

  • Baskerville = Liberation Serif

  • Bell MT = Liberation Serif

  • Bodoni MT = Liberation Serif

  • Bodoni MT Black = Liberation Serif

  • Book Antiqua = Liberation Serif

  • Bookman Old Style = Liberation Serif

  • Calisto MT = Liberation Serif

  • Cambria = Liberation Serif

  • Centaur = Liberation Serif

  • Century = Liberation Serif

  • Century Gothic = Liberation Serif

  • Century Schoolbook = Liberation Serif

  • Elephant = Liberation Serif

  • Footlight MT Light = Liberation Serif

  • Garamond = Liberation Serif

  • Georgia = Liberation Serif

  • Goudy Old Style = Liberation Serif

  • Lucida Bright = Liberation Serif

  • MS Serif = Liberation Serif

  • New York = Liberation Serif

  • Palatino = Liberation Serif

  • Perpetua = Liberation Serif

  • Times = Liberation Serif

  • times = Liberation Serif

  • Times New Roman = Liberation Serif

5.8 Changing Resources

All of the strings used in the UNIX versions of Oracle Outside In products are contained in the lodlgstr.h file. This file, located in the resource directory, can be modified for internationalization and other purposes. Everything necessary to rebuild the resource library to use the modified source file is included with the SDK.

In addition to lodlgstr.h, the scclo.o object file is provided. This is necessary for the linking phase of the build. A makefile has also been provided for building the library. The makefile allows building on all of the UNIX platforms supported by Oracle Outside In. It may be necessary to make minor modifications to the makefile so the system header files and libraries can be found for compiling and linking.

Standard INCLUDE and LIB make variables are defined for each platform in the makefile. Edit these variables to point to the header files and libraries on your particular system. Other make variables are:

  • TECHINCLUDE: May need to be edited to point to the location of the Oracle Outside In /common header files supplied with the SDK.

  • BUILDDIR: May need to be edited to point to the location of the makefile, lodlgstr.h, and scclo.o (which should all be in the same directory).

After these variables are set, change to the build directory and type make. The libsc_lo resource library is built and placed in the appropriate platform-specific directory. To use this library, copy it into the directory where the Oracle Outside In product is stored and the new, modified resource strings are used by the technology.

Menu constants are included in lomenu.h in the common directory.

5.9 Linux Compiling and Linking

This section discusses issues involving Linux compiling and linking.

5.9.1 Library Compatibility

This section discusses Linux compatibility issues when using libraries.

5.9.1.1 GLIBC and Compiler Versions

The following table indicates the compiler version used and the minimum required version of the GNU standard C library needed for Oracle Outside In operation.

Distribution Compiler Version GLIBC Version

x64 Linux

4.1.2

libc.so.6
5.9.1.2 Other Libraries

In addition to libc.so.6, Oracle Outside In is dependent upon the following libraries:

  • libstdc++.so.6

  • libgcc_so.1

The following table summarizes what is included with the RedHat and SUSE distributions supported by Oracle Outside In and what needs to be added/modified to make Oracle Outside In run on these systems.

5.9.1.2.1 Libraries on Linux Systems as Distributed (IA32)

SUSE 9.0

Included To be added

libc.so.6 version

/lib/libc.so.6 (GLIBC 2.3.4)

libstdc++

/usr/lib/libstdc++.so.5.0.6 + old libraries

libgcc_s.so.1

/lib/libgcc_s.so.1

Required to Use Oracle Outside In

  • Default system install has proper libstdc++.so.5

  • Default system install has libgcc_so.1

  • Update to >= libXm.so.3.0.2 (OpenMotif >=2.2.3)

  • Install X libraries

5.9.2 Compiling and Linking

The libsc_ex.so and libsc_da.so are the only libraries that must be linked with your applications. They can be loaded when your application starts by linking them directly at compile time or they can be loaded dynamically by your application using library load functions (for example, dlopen).

The following are example command lines used to compile the sample application exsimple from the /sdk/samplecode directory. This command line is only an example. The actual command line required on the developer's system may vary.

The example assumes that the include and library file search paths for the technology libraries are set correctly. If they are not set correctly, the search paths for the include and/or library files must be explicitly specified via the -I include file path and/or -L library file path options, respectively, so the compiler and linker can locate all required files.

5.9.2.1 Linux 64-bit 
gcc -w -o ../exsimple/unix/exsimple ../exsimple/unix/exsimple.c -I/usr/local/include -I../../common -L../../demo -L/usr/local/lib -lsc_ex -lsc_da -DUNIX_64 -Wl,-rpath,../../demo -Wl,-rpath,'${ORIGIN}'