[Top] -> [Works] -> [Unix] -> [catdoc]

catdoc & xls2csv

Overview

catdoc is program which reads one or more Microsoft word files and outputs text, contained insinde them to standard output. Therefore it does same work for .doc files, as unix cat command for plain ASCII files.

It is now accompanied by xls2csv - program which converts Excel spreadsheet into comma-separated value file

Optionaly, catdoc is able to translate some non-ASCII chars into correspoindig TeX escape sequences and convert charsets from Windows ANSI codepage to local codepage of target machine. (Becouse catdoc is russian program, by default it converts cp1251 to koi8-r, when running under UNIX and to cp866 when running under DOS.

Catdoc has rudimentary table handling. In TeX mode it inserts & when encounters field delimiter and \\ when encounters end of table row. No table headers are produced although.

Catdoc doesn't even try to preserver MS-Word character formatting. It's goal is to extract plain text and allow you to read it and, probably, reformat with TeX, according to TeXnical rules, most Word users haven't even heard about.

xls2csv does roughly same for Excel files. It extracts data and leaves out any formatting info and formulas. Concept is that you want to see data, not the way it was created.

Supported platforms

Unix. Catdoc was initially developed for Linux and Sparc Solaris. It also runs on variety of other Unices. For instance it is included in FreeBSD ports collection.
MS-DOS. Catdoc also runs on MS-DOS, even on XT machines. MS-DOS is only platform for which compiled executables are provided. These executables are 16-bit real mode. I think that protected mode version of xls2csv might be useful, but don't have time to support it.

There is no support for catdoc under Windows

Not because I hate windows. Just becouse I don't use it. Note that DOS catdoc is not intended to be used under windows. For example, it doesn't support long file names.

License

catdoc and xls2csv are distributed under GNU Public License. Commercial licensing of some parts of catdoc is also available upon request

Current status

Current development version of catdoc is 0.93.3. It finally is able to autodetect unicode/non-unicode Word files and also recognizes (and hopefully parses) MS-Write files and rtf. It also eliminates garbage which troubled prevoius version of catdoc. Note that footnotes and fastsaves still not handled.

Previous version is 0.92 This is first version where RTF support apperared. It seems that 0.93 series are just as stable as 0.92 and do their job better.

Old stable version of catdoc is 0.90.3 It supports UNICODE and uses runtime-configurable charset tables. Autodetection of Word files is also changed so catdoc can be used blindly on any file with .doc extension, producing reasonable results in most cases. These improvements have drawback that now it is not enough to copy executable to install catdoc properly.

"Competing" products

There are several other free programs, which read MS-Word files

Microsoft own freeware viewer. It probably is best of all listed in handling of Word format. (And only one, which can handle Word 97) Unfortunately, requires Windows to run, although I have heard that it runs successifully under Wine
Laola, the perl library to read OLE files
word2x much more elabodate than catdoc C++ program. It sometimes reads files, which catdoc couldn't and vice versa.
wvware Elaborated system which parses Word files and converts them into something more interesting, i.e. HTML or LaTeX.
xlHtml Converter of Excel and PowerPoint files into readable and processable form (html, csv, xml)
Word2Html (link appears to be dead). Very simple program initially developed for OS/2. Doesn't handle tables and charsets, but finds start and end of textstream correctly, based on code, borrowed from LAOLA.
Filters project Project to create GPLed filters for various proprietary formats. Now has C library cole, which does same job as LAOLA. Aimed to completely convert all information from Word file (including formatting) to XML.
w2tPerl program which clames to convert word to LaTeX. Based on LAOLA.
antiword
Spreadsheet::ParseExcel Perl module to read Excel files.
doc2xml Python script which reads post-97 Word files. Handles fast-save

Revision history

0.93.3 Nov 15 2003: It was planned as feature release. It has support for Excel Date formatting, output of blank cells and help window in wordview. Unfortunately, during its development important bug was found in ole parser code. So I have to publish this release real soon after previous
0.93.2 Nov 14 2003: Improved performance of OLE parser, fixed problems with unicode chars 0xFF00-0xFFFF in catdoc, rewrite wordview for unicode-aware version of Tcl, with support of displaying text in language different from current locale. Reworked autoconf configuration.
0.93.1 Sep 24 2003: Fixed numerous bugs in newer OLE and RTF code, including problem with incorrectly interpreting last (incomplete) 256-byte block of text as Unicode. Restored support for pre-OLE Word versions, which was accidently lost in 0.93
0.93 July 29 2003: Added proper handling of OLE structure (by Alex Ott).
0.92 June 16 2003: Added RTF parser at last (contributed by Alex Ott). MS-DOS executable for xls2csv is included. Some code clean up and splitting.
0.91.6 May 24 2003: Added autodetect of output charset from current locale. Fixed handling of RK and MULRK records in xls2csv. No more missing numbers. Fixed long-standing bug with loosing of first 8 symbols when recoding text file. I finally began to provide MS-DOS executables for 0.91.x series
0.91.5 January 30 2002: I finally got to catdoc again. UTF-8 output is added. Just specify utf-8 as output charset.
0.91.4 December 30 1999: Fixed important bug in xls2csv - improper recognition of numeric cells (as opposed to formula). Fixed segfault when catdoc is used to recode plain text files.
0.91.3 December 14 1999: Mainly xls2csv fixes - xls2csv now recognizes some options (man page is in sync), added endianess check to configure, so xls2csv compilies correctly out of the box on big-endian machines
0.91.2 October 19 1999: This is first verison which includes xls2csv program. Also, some long-standed bugs are fixed and newly-introduced bug when catdoc hangs on broken files. Although these files are not read properly without -b switch. New charset koi8-u is added to distribution. If you want to use it in the stable version, just download it from here and put in the catdoc library directory.
New switch -l is added. It causes catdoc to list available charsets in current charset path.
0.91.1 October 15 1999: As it was expected it was wrong decision to believe information about extended charset from word document header. Now we analyze encoding for each 256-byte page separately (becouse it is possible that first ones would be 8-bit and other 16-bit). When processing non-word files (i.e. plain text) encodings are converted and -u is taken into account, so catdoc can be used as generic character converter, which supports utf8 and utf16 (both byteorders) as input.
0.91.0 October 12 1999: Implemented new format analyzis. Now most versions of word format as well as MS-Write and rtf are detected. Boundaries of main text stream are also detected, so no more garbage is produced at the end of file
0.90.3 August 11 1999: Fixed small OS-specific bugs - broken isspace in Turbo C under DOS and %x was replaced %i for compatibility with SunOS 4.
0.90.2 May 24 1999: Artem Chuprina pointed out to segfault error when non-existent charset is specified in command line. It turned out to be silly bug in check_charset function with oneline fix. You can get one-line patch.
0.90.1 Nov 26 1998: Duncan Simpson pointed out to numerous places in catdoc source where paranoid sysadmin could suspect buffer overflow. They was investigated and either rewritten or commented why they are safe. Also fixed minor bug in Makefile (make args are now propagated to subdirs) and wordview (saving files now works).
0.90 Oct 29 1998: Fixed bug with redeclaring source_csname and target_csname in main. Fixed bug in configure when dealing with wish 8.0.3. Catdoc considered stable enough to be released.
0.90b5 Oct 14 1998: Fixed handling of 0x1F char (soft hyphen in Word 6.0), now it is translated to 0x00AD (unicode soft hyphen) Fixed permissions for manual page Added --with-install-root configure arg to simplify building of binary packages.
0.90b4 September 17 1998: Added proper configuration of library dir in wordview. Added --disable-charset-check config option Added 0x2026 symbol in ascii.rpl Added more Windows codepages in distribution
0.90b3 September 11 1998: added -x option to simplify debugging of substitution maps
0.90b2 September 10 1998: Added replacement sequences for some special charachers which present in cp1251 and cp1252. Fixed some filename-handling problems in wordview
0.90b1 September 8 1998: Added cyrillic transliteration into ascii.replchars fixed some bugs in configuration. Added us-ascii output charset.
0.90a3 September 7 1998: Fixed small bug in table handling, which caused catdoc to output extra column delimiter just before row delimiter. Added autoconf configuration. install is back, although not for charsets
0.90a2 August 18 1998: version 0.90 was tested on BSDI and Solaris platform. Makefile was rewritten to avoid use of highly incompatible /usr/{ucb,bin}/install
0.90a1 August 13 1998: Catdoc undergone major rewrite. Now it has proper charset handling, including UNICODE and runtime configurability.
0.35 - June 5 1998: Fixed bug with -s switch which prevents catdoc from returning non-zero code when invoked on UNIX text file
0.34 - Apr 28 1998: Files now opened in binary mode thus allowing catdoc to work on DOS and simular systems. All specs arrays now have terminating NULL
0.33 - October 1997: Fixed missing terminating NUL in specs array, which caused random seqfaults on Linux and many other systems, becouse specs is searched by strchr fynction
0.32 - August 1997: First mayor public release, uploaded to CTAN. Tk interface appeared, manual page was written. Unfortunately, this release was buggy.

[Top] -> [Works] -> [Unix] -> [catdoc]