head 1.3; access; symbols pkgsrc-2026Q3:1.2.0.4 pkgsrc-2026Q3-base:1.2 pkgsrc-2026Q2:1.2.0.2 pkgsrc-2026Q2-base:1.2 pkgsrc-2026Q1:1.1.0.26 pkgsrc-2026Q1-base:1.1 pkgsrc-2025Q4:1.1.0.24 pkgsrc-2025Q4-base:1.1 pkgsrc-2025Q3:1.1.0.22 pkgsrc-2025Q3-base:1.1 pkgsrc-2025Q2:1.1.0.20 pkgsrc-2025Q2-base:1.1 pkgsrc-2025Q1:1.1.0.18 pkgsrc-2025Q1-base:1.1 pkgsrc-2024Q4:1.1.0.16 pkgsrc-2024Q4-base:1.1 pkgsrc-2024Q3:1.1.0.14 pkgsrc-2024Q3-base:1.1 pkgsrc-2024Q2:1.1.0.12 pkgsrc-2024Q2-base:1.1 pkgsrc-2024Q1:1.1.0.10 pkgsrc-2024Q1-base:1.1 pkgsrc-2023Q4:1.1.0.8 pkgsrc-2023Q4-base:1.1 pkgsrc-2023Q3:1.1.0.6 pkgsrc-2023Q3-base:1.1 pkgsrc-2023Q2:1.1.0.4 pkgsrc-2023Q2-base:1.1 pkgsrc-2023Q1:1.1.0.2 pkgsrc-2023Q1-base:1.1; locks; strict; comment @# @; 1.3 date 2026.10.27.12.06.03; author wiz; state dead; branches; next 1.2; commitid PWCBa7bbjjcgW61H; 1.2 date 2026.05.07.16.34.35; author adam; state Exp; branches; next 1.1; commitid k78hT9E15ta7ZTEG; 1.1 date 2023.01.19.00.53.42; author wiz; state Exp; branches; next ; commitid jC5N3rCxkLnKt5aE; desc @@ 1.3 log @hunspell: update to 1.7.6. 2026-10-26: Hunspell 1.7.6 release: - add, add_with_flags, add_with_affix and remove apply the dictionary's ICONV rules to the word, as spell does, so a caller that converted the word itself must stop doing so (#616) - A personal dictionary exclusion such as *afterwards also rejects the word when the dictionary has it only as an affixed form (#1155) - stem and analyze give st: stem fields again with COMPLEXPREFIXES - analyze lists every analysis of a compound word, where a rejected split of the word could hide the splits after it - The text and XML parsers check the character after a character entity such as à (#686) - The interactive mode shows the first digit of a two digit suggestion number (#158) - The -a pipe interface checks a line of any length in one piece (#220) - The hunspell tool takes -- as the end of the options, the -u2 sed script escapes the word and the suggestion, and a zipped ODF file whose name starts with - can be checked - The hunspell tool no longer searches the OpenOffice.org 2.x and 3 folders or LibreOffice's user wordbook folder for dictionaries, and searches /Library/Spelling and ~/Library/Spelling only on macOS - On Windows, search the share\hunspell, share\myspell and share\myspell\dicts folders of the installation that holds hunspell.exe - New configure option --disable-tools to build and install only the library - A static-only build records that in the installed hunvisapi.h, so its users need not define HUNSPELL_STATIC - Install the Hungarian hunspell.5 in hu/man5 - Use Mozilla's own FileMgr and csutil code in MOZILLA_CLIENT builds - A CMake build of the library, as an alternative to configure (see README.md) - New configure option --enable-werror, off by default - Man page: the fallback dictionary is en_US, the SYNOPSIS lists -u, -U, -u2, -u3, -P and --trace (#1159), and -a reads a file with the pipe protocol too - Build fixes: MinGW, the BSDs, macOS, makealias without gawk or GNU sed, readline and curses detection, and the release tarball ships the CMake files and all the test data - oss-fuzz and Coverity fixes and hardening @ text @$NetBSD: patch-configure.ac,v 1.2 2026/05/07 16:34:35 adam Exp $ Check for symbol that is in libcurses, not in libterminfo. --- configure.ac.orig 2026-05-05 07:19:07.000000000 +0000 +++ configure.ac @@@@ -62,17 +62,17 @@@@ AC_ARG_WITH( [with_ui=no] ) AS_IF([test "x$with_ui" != xno], [ - AC_CHECK_LIB([ncursesw],[tparm],[ + AC_CHECK_LIB([ncursesw],[ungetch],[ CURSESLIB=-lncursesw AC_CHECK_LIB([tinfow],[tgetent], [ CURSESLIB="$CURSESLIB -ltinfow" ]) - ],[AC_CHECK_LIB([curses],[tparm],[ + ],[AC_CHECK_LIB([curses],[ungetch],[ CURSESLIB=-lcurses AC_CHECK_LIB([tinfo],[tgetent], [ CURSESLIB="$CURSESLIB -ltinfo" ]) - ],[AC_CHECK_LIB([ncurses],[tparm],[ + ],[AC_CHECK_LIB([ncurses],[ungetch],[ CURSESLIB=-lncurses AC_CHECK_LIB([tinfo],[tgetent], [ CURSESLIB="$CURSESLIB -ltinfo" @@@@ -106,7 +106,7 @@@@ AS_IF([test "x$with_readline" != xno], TERMLIB=-ltinfow ],[AC_CHECK_LIB([tinfo],[tgetent], [ TERMLIB=-ltinfo - ],[AC_CHECK_LIB([curses],[tparm],[ + ],[AC_CHECK_LIB([curses],[ungetch],[ TERMLIB=-lncurses ],[AC_CHECK_LIB([termcap],[tgetent],[ TERMLIB=-ltermcap @ 1.2 log @hunspell: updated to 1.7.3 Hunspell 1.7.3 release: - Fix stack-buffer-overflow in Hunzip::getline (reported and fixed by MarkLee131) - Fix stack overflow in compound_check on Hungarian dictionaries under certain conditions (reported by Anthropic via Ada Logics) - Fix UB when SFX condition starts with '^' - Bounds-check continuation bytes in u8_u16 - oss-fuzz timeout/OOM hardenings - Fix 715 CHECKCOMPOUNDCASE considers digits uppercase - Fix 748 hzip: cannot write file - Fix 1024 std::string bounds check - Fix 1044 tools/analyze crash - Fix 1076 flags 65520/65521 wrongly rejected - Fix 1058 don't suggest the input word as its own correction - Fix 1002 exact word marked as a near miss - Fix tdf125600 dotted-I regression - Partial Unicode table refresh for Mc combining marks - Add Hunspell_add_with_flags / Hunspell::add_with_flags - New SPELL_BEST_SUG flag, MAXBREAKDEPTH limit - Replace clock() with std::chrono for suggestion time limits - Improve exception safety - Document analyze/stem/generate requirements - Report iconv direction on private dic load failures - Show dic load errors unconditionally - Rename es_EU to eu - Build fixes: out-of-tree, Windows ARM64, MSVC hzip tmpfile --disable-shared with mingw32 ncurses with separate tinfo - New fuzzers: hzfuzzer, persdicfuzzer, parserfuzzer, affdicfuzzer - Coverity-flagged fixes - Merge in weblate translations @ text @d1 1 a1 1 $NetBSD: patch-configure.ac,v 1.1 2023/01/19 00:53:42 wiz Exp $ @ 1.1 log @hunspell: update to 1.7.2 * Crash fixes, code clean-up in ~200 commits * tdf#136306 don't accept/suggest typos as 3-or-more-word compound words * Prepare optional spelling mode of LibreOffice to not accept/suggest not dictionary-based words as compound words (#517) * Merge in weblate translations While here: Fix libcurses detection on NetBSD. However, switch to ncursesw to get Unicode support, hunspell doesn't recognize wide character support in NetBSD curses. @ text @d1 1 a1 1 $NetBSD$ d3 1 a3 1 Check for symbol that is in libcurses not, not in in libterminfo. d5 1 a5 1 --- configure.ac.orig 2022-12-29 20:10:49.000000000 +0000 d7 1 a7 1 @@@@ -52,11 +52,11 @@@@ AC_ARG_WITH( d14 3 d20 3 d26 5 a30 5 ])])]) if test "$CURSESLIB" != "" ; then @@@@ -82,7 +82,7 @@@@ READLINELIB="" AS_IF([test "x$with_readline" != xno], [AC_CHECK_LIB([tinfo],[tgetent], [ d32 2 a33 2 - ],[AC_CHECK_LIB([curses],[tparm],[ + ],[AC_CHECK_LIB([curses],[ungetch],[ @