external/hunspell/0001-Resolves-rhbz-2158548-allow-longer-words-for-hunspel.patch

   1 From e2fe9f86e1769b440972971240e9b8fb1cd53b97 Mon Sep 17 00:00:00 2001
   2 From: =?UTF-8?q?Caol=C3=A1n=20McNamara?= <caolanm@redhat.com>
   3 Date: Fri, 6 Jan 2023 16:20:45 +0000
   4 Subject: [PATCH] Resolves: rhbz#2158548 allow longer words for hunspell-ko
   5 MIME-Version: 1.0
   6 Content-Type: text/plain; charset=UTF-8
   7 Content-Transfer-Encoding: 8bit
   8
   9 https://github.com/hunspell/hunspell/issues/903
  10
  11 A problem since the sanity check added in:
  12
  13 commit 05e44e069e4cfaa9ce1264bf13f23fc9abd7ed05
  14 Author: Caolán McNamara <caolanm@redhat.com>
  15 Date:   Thu Sep 1 13:46:40 2022 +0100
  16
  17     Check word limit (#813)
  18
  19     * check against hentry blen max
  20 ---
  21  src/hunspell/hashmgr.cxx | 6 +++---
  22  src/hunspell/htypes.hxx  | 4 ++--
  23  tests/korean.dic         | 3 ++-
  24  3 files changed, 7 insertions(+), 6 deletions(-)
  25
  26 diff --git a/src/hunspell/hashmgr.cxx b/src/hunspell/hashmgr.cxx
  27 index 100916d..14201e9 100644
  28 --- a/src/hunspell/hashmgr.cxx
  29 +++ b/src/hunspell/hashmgr.cxx
  30 @@ -209,7 +209,7 @@ int HashMgr::add_word(const std::string& in_word,
  31    }
  32
  33    // limit of hp->blen
  34 -  if (word->size() > std::numeric_limits<unsigned char>::max()) {
  35 +  if (word->size() > std::numeric_limits<unsigned short>::max()) {
  36      HUNSPELL_WARNING(stderr, "error: word len %ld is over max limit\n", word->size());
  37      delete desc_copy;
  38      delete word_copy;
  39 @@ -235,8 +235,8 @@ int HashMgr::add_word(const std::string& in_word,
  40
  41    int i = hash(hpw, word->size());
  42
  43 -  hp->blen = (unsigned char)word->size();
  44 -  hp->clen = (unsigned char)wcl;
  45 +  hp->blen = (unsigned short)word->size();
  46 +  hp->clen = (unsigned short)wcl;
  47    hp->alen = (short)al;
  48    hp->astr = aff;
  49    hp->next = NULL;
  50 diff --git a/src/hunspell/htypes.hxx b/src/hunspell/htypes.hxx
  51 index 44366b1..2b896fb 100644
  52 --- a/src/hunspell/htypes.hxx
  53 +++ b/src/hunspell/htypes.hxx
  54 @@ -62,8 +62,8 @@
  55  #endif
  56
  57  struct hentry {
  58 -  unsigned char blen;    // word length in bytes
  59 -  unsigned char clen;    // word length in characters (different for UTF-8 enc.)
  60 +  unsigned short blen;   // word length in bytes
  61 +  unsigned short clen;   // word length in characters (different for UTF-8 enc.)
  62    short alen;            // length of affix flag vector
  63    unsigned short* astr;  // affix flag vector
  64    struct hentry* next;   // next word with same hash code
  65 diff --git a/tests/korean.dic b/tests/korean.dic
  66 index 95cb450..d76ea05 100644
  67 --- a/tests/korean.dic
  68 +++ b/tests/korean.dic
  69 @@ -1,3 +1,4 @@
  70 -2
  71 +3
  72  들어오세요
  73  안녕하세요
  74 +김수한무거북이와두루미삼천갑자동방삭치치카포사리사리세ᅡ워리워리세브리캉무드셀ᅡ구름위허ᅵ케ᅵᆫᅦ담벼락서생원에ᄀ양
  75 --
  76 2.38.1
  77