0.1 - original version, known as "rtf2htm" 0.2 - first Linux version 0.3 - addition of hash storage for all strings 0.4 - support for a few HTML special exprs 0.5 - tables finally work 0.6 - addition of usage() 0.7 - added special chars, \tab, info group, 0.8 - font table support 0.9.0 - images are stored to files 0.9.1 - pict files extensions a bit more accurate 0.9.2 - font size changes (\fs) more accurate. 0.9.3 - fixed bug regarding where goes 0.9.4 - added -nopict option 0.9.5 - fixed table data printing. 0.9.6 - if >1 \cell in a group, font attrs affect all. 0.9.7 - Chunk class renamed Word 0.10.0 - input file type is checked, -echo renamed -dump 0.11.0 - addition of ANSI->HTML conversion, plus limited Mac->HTML 0.11.1 - unknown chars are printed as 0.12.0 - separated parser from word module 0.12.1 - added -version option 0.12.2 - backslash-newline is now \par 0.12.3 - added default font families (\fswiss etc) 0.12.4 - added blurb about program in the output HTML 0.12.5 - added recognition of files with ".RTF" ending, not just ".rtf" 0.12.6 - added charset tables for CP437 and CP850 0.12.7 - fixed bug preventing \'## sequences from being translated 0.12.8 - removed size limit for words, verified DJGPP compilation works. 0.12.9 - compilation bugfix, fixed author/date information 0.12.10 - added --simple switch for generating HTML without SPAN/DIV tags 0.13.0 - paragraph alignment added 0.13.1 - fix for words getting lost after \par 0.13.2 - fix for color table not being recognized 0.14.0 - separation of word and convert modules 0.14.1 - fixed color table indexing, added support for font background color. 0.14.2 - updated parsing mechanism 0.14.3 - fix for DIV nesting; fix for NS's hiding any table within a span. 0.14.4 - fix for color table parsing; added better underline/strikethru support. 0.15.0 - hashing of RTF keywords in convert.c; moved much code out of convert.c 0.15.1 - removed null entries from special.c to reduce program size 0.15.2 - removed no-op entries from hash in convert.c to reduce program size 0.15.3 - first 16-bit MSDOS executable 0.15.4 - fixed paragraph alignment not getting cleared by \pard 0.15.5 - updated code to produce as few paragraph alignment tags as possible 0.16.0 - added input buffering to improve speed 0.16.1 - bugfix for printing of debugging strings stored in command hash 0.17.0 - hyperlink support added 0.17.1 - added support for several font attributes e.g. smallcaps, caps, \expand 0.17.2 - added Amiga(TM) compilation support 0.17.3 - fixed problem in interpretation of RTF command words 0.17.4 - changed attr.c to use AttrStack (stack of stacks) paradigm ----program renamed UnRTF---- 0.17.5 - began implementation of output personalities; wrote HTML personality. 0.17.6 - added text and vt personalities 0.17.7 - began addition of PostScript(TM) and LaTeX and WPML output 0.17.8 - first fully commented version, removed "last change" lines 0.17.9 - fixed translation between character sets 0.17.10 - PS now supports underline, strikethrough, shadowed text 0.17.11 - PS now uses ISOLatin1Encoding 0.17.12 - PS now supports these fonts: Times,Courier,Symbol,Helvetica ----program is now GNU UnRTF---- 0.18.0 - updates to comments, documentation to reflect inclusion in GNU suite. 0.18.1 - updated manual 0.19.1 - new maintainers: minor fixes to prevent segmentation violations; further special character code; minor cleanups 0.19.2 - updated some files; 0.19.3 - fix bug #225592: null pointer param in convert.c 0.19.4 - added unicode support 0.19.5 - removed defective PS support and non-free text files more unicode support improved symbol font support - no longer puts entities in latex output Bug#266020 concerning double slashes fixed Bug#269054 concerning Doctype fixed Bug#287038 security breach fixed (thanks to Joey Hess ) 0.19.6 - fix some latex problems 0.19.7 - updated FSF address 0.19.8 - minor fixes 0.19.9 - included verbose mode 0.20.0 - new build system by Claudio Fontana fixed creation date spelling and added support for accented characters in titles from Laurent Monin 0.20.1 - minor fix in convert.c for Solaris System 0.20.2 - minor fix to unterminated comment output, added attr_get_param() 0.20.3 - Greek font support added, fix to UTF character processing 0.20.4 - nopict mode bug fixes 0.20.5 - added --noremap option, some html entity corrections, updated to GPL v3 0.21.3 - - added '--quiet' flag which suppresses leading comments in output - default directory for config files is now /usr/local/share/unrtf and this can be changed when configure is run at compilation time - subdirs in config directory eliminated making it easier for a user to supply their own in a single directory specified with -P - code for two hex characters introduced in 0.21.2 completely replaced by more generalised string handling. - major improvements to codepage handling - major improvements in unicode handling - latex config file improved. Some Greek characters output properly. - troff_mm config file improved such that this option is now useful 0.21.2 - added feof() calls to fgets() and fgetchar() instances so end-of file is correctly determined under AIX - supplemented code to handle special characters specified with two hex characters (e.g. \'81\'83) as output by some RTF writers. This code is experimental, and uses the SHIFT_JIS codepage for conversion. - definitions in html.config changed to ensure values of markup variables are enclosed in double quotes, e.g. size="3" instead of size=3. This is a first step towards supporting production of XHTML. 0.21.1 - fixed strings in unicode_to_string() (unicode.c) to be null terminated instead of newline terminated. This prevents spurious newline output after characters in Symbol font. It also overcomes a fault whereby characters in the symbol font were being output as unicode instead of the desired entities defined in the output specification. - suppressed output of as this is wrong if characters are changed to unicode or entities - changed path set with -P to be directory containing the output charmaps specifications, and changed the default path to /usr/local/lib/unrtf for the same reason. This makes it possible to set a different default path, useful for example if one cannot install files in /usr/local/lib. - dropped "outputs" and "charmaps" sub-directories in the configuration directory. Output configuration files are now named "html.conf", "text.conf" etc. and the symbol font charmap file is "SYMBOL.charmap". - charmap specifications now sought in the search path - compensate for RTF with unicodes specified as negative numbers, possibly due to overflow on a 16bit machine? - changed processing of --html (etc.) args so that "-t html" and "--html" are equivalent, i.e. both subject to search path set with "-P" 0.21.3: - major changes to code page support and UTF handling - improved LaTeX output - /usr/local/lib no longer hard wired - set by configure 0.21.4: - ignore embedded picture which cannot be rendered - allow unnested bold/italic to be interpreted as expected. Such input may be outside the RTF spec, but can be created by TextEdit. The change has been limited to the bold command (cmd_b in convert.c) on an experimental basis. If no bugs appear it could be more generally applied. 0.21.5: - improved code for the unnested case and extended to more RTF commands 0.21.6: - improved man page - improved USAGE string - fix to attr.c for clang compilation 0.21.7: - improved code for creation of image files when RTF files containing images processed - prevent segmentation violations with RTF input containing corrupt \info content 0.21.8: - Changed bug reporting in unrtf.1 - Test file for newly supported EMF image type. - Fixed all -Wall compilation warning - Bogus test in op_translate_buffer() caused error messages and data truncation - cmd_maybe_ignore(): test next command presence in the cmd hashes instead of listing locally the commands that should be processed - Fix unrtf looping when closing any underlining except the basic (ul) one - Open Picture files in wb mode instead of w for Windows portability. Fixes bug #21274 - html output doctype value needs to be quoted - Skip binary data after \bin command - Simplified shared string storage in hash.c and word.c: suppressed the "index" value and replaced it by string address. Simpler and faster. Other changes are making shared string refs const - Better processing of \pict including addition of EMF type. - Fix access overflow occurring while translating unicode data from garbled or truncated file - Fix buffer access overflow caused by truncated file - Need to check word_string() output not NULL in cmd_maybe_ignore() - String equality test while searching the command table for command having parameter would access byte out of allocated memory and produce potentially random result. Fixes AddressSanitizer error found by Hanno Boeck - String equality test while searching the command table for command having parameter would access byte out of allocated memory and produce potentially random result. Fixes AddressSanitizer error found by Hanno Boeck - Substitute appropriate prefix-depending paths in the man page prior to installing - Limit depth of group embedding to guard against malicious files with thousands of consecutive { - check index before access into month_strings array - attr_get_param(): silence warning message in frequent and normal case (Fabian Keil) - attrstack_drop(): Properly drop the last stack element (Fabian Keil) + get rid of unused variable in attrstack_push() - hash_get_string(): Reject invalid indexes (author: Fabian Keil) 0.21.9: - Support EndNote reference citations in convert.c - Improved handling of some tables - Fixed --text processing broken in 0.21.8 - outputs/text.conf 0.21.10: - added gentoo automake-fix and iconv-detection patches