external/hunspell/hunspell-tdf95024-compound.patch

   1 From 42807f970ac2d65f0d13a7c57eb454b210e92240 Mon Sep 17 00:00:00 2001
   2 From: =?UTF-8?q?L=C3=A1szl=C3=B3=20N=C3=A9meth?=
   3  <laszlo.nemeth@collabora.com>
   4 Date: Mon, 12 Oct 2015 08:43:12 +0200
   5 Subject: [PATCH] fix compound handling for new Hungarian orthography
   6
   7 The frequent cases of this compound limitation are handled by
   8 the extended dictionary, but not these ones with both derivative
   9 and inflectional suffixes.
  10 ---
  11  src/hunspell/affixmgr.cxx | 19 +++++++++++++++----
  12  src/hunspell/affixmgr.hxx |  1 +
  13  2 files changed, 16 insertions(+), 4 deletions(-)
  14
  15 diff --git a/src/hunspell/affixmgr.cxx b/src/hunspell/affixmgr.cxx
  16 index 0992e6e..0950425 100644
  17 --- misc/hunspell-1.3.3/src/hunspell/affixmgr.cxx
  18 +++ misc/build/hunspell-1.3.3/src/hunspell/affixmgr.cxx
  19 @@ -139,8 +139,9 @@ AffixMgr::AffixMgr(const char * affpath, HashMgr** ptr, int * md, const char * k
  20    cpdvowels=NULL; // vowels (for calculating of Hungarian compounding limit, O(n) search! XXX)
  21    cpdvowels_utf16=NULL; // vowels for UTF-8 encoding (bsearch instead of O(n) search)
  22    cpdvowels_utf16_len=0; // vowels
  23 -  pfxappnd=NULL; // previous prefix for counting the syllables of prefix BUG
  24 -  sfxappnd=NULL; // previous suffix for counting a special syllables BUG
  25 +  pfxappnd=NULL; // previous prefix for counting syllables of the prefix BUG
  26 +  sfxappnd=NULL; // previous suffix for counting syllables of the suffix BUG
  27 +  sfxextra=0; // modifier for syllable count of sfxappnd BUG
  28    cpdsyllablenum=NULL; // syllable count incrementing flag
  29    checknum=0; // checking numbers, and word with numbers
  30    wordchars=NULL; // letters + spec. word characters
  31 @@ -1201,6 +1202,7 @@ struct hentry * AffixMgr::prefix_check(const char * word, int len, char in_compo
  32      pfx = NULL;
  33      pfxappnd = NULL;
  34      sfxappnd = NULL;
  35 +    sfxextra = 0;
  36
  37      // first handle the special case of 0 length prefixes
  38      PfxEntry * pe = pStart[0];
  39 @@ -1261,6 +1263,7 @@ struct hentry * AffixMgr::prefix_check_twosfx(const char * word, int len,
  40
  41      pfx = NULL;
  42      sfxappnd = NULL;
  43 +    sfxextra = 0;
  44
  45      // first handle the special case of 0 length prefixes
  46      PfxEntry * pe = pStart[0];
  47 @@ -1302,6 +1305,7 @@ char * AffixMgr::prefix_check_morph(const char * word, int len, char in_compound
  48
  49      pfx = NULL;
  50      sfxappnd = NULL;
  51 +    sfxextra = 0;
  52
  53      // first handle the special case of 0 length prefixes
  54      PfxEntry * pe = pStart[0];
  55 @@ -1353,6 +1357,7 @@ char * AffixMgr::prefix_check_twosfx_morph(const char * word, int len,
  56
  57      pfx = NULL;
  58      sfxappnd = NULL;
  59 +    sfxextra = 0;
  60
  61      // first handle the special case of 0 length prefixes
  62      PfxEntry * pe = pStart[0];
  63 @@ -1993,7 +1998,7 @@ struct hentry * AffixMgr::compound_check(const char * word, int len,
  64                  // XXX only second suffix (inflections, not derivations)
  65                  if (sfxappnd) {
  66                      char * tmp = myrevstrdup(sfxappnd);
  67 -                    numsyllable -= get_syllable(tmp, strlen(tmp));
  68 +                    numsyllable -= get_syllable(tmp, strlen(tmp)) + sfxextra;
  69                      free(tmp);
  70                  }
  71
  72 @@ -2512,7 +2517,7 @@ int AffixMgr::compound_check_morph(const char * word, int len,
  73                  // XXX only second suffix (inflections, not derivations)
  74                  if (sfxappnd) {
  75                      char * tmp = myrevstrdup(sfxappnd);
  76 -                    numsyllable -= get_syllable(tmp, strlen(tmp));
  77 +                    numsyllable -= get_syllable(tmp, strlen(tmp)) + sfxextra;
  78                      free(tmp);
  79                  }
  80
  81 @@ -2696,6 +2701,12 @@ struct hentry * AffixMgr::suffix_check (const char * word, int len,
  82                      sfx=sptr; // BUG: sfx not stateless
  83                      sfxflag = sptr->getFlag(); // BUG: sfxflag not stateless
  84                      if (!sptr->getCont()) sfxappnd=sptr->getKey(); // BUG: sfxappnd not stateless
  85 +// LANG_hu section: spec. Hungarian rule
  86 +                    else if (langnum == LANG_hu && sptr->getKeyLen() && sptr->getKey()[0] == 'i' &&
  87 +                        sptr->getKey()[1] != 'y' && sptr->getKey()[1] != 't') {
  88 +                            sfxextra = 1;
  89 +                    }
  90 +// END of LANG_hu section
  91                      return rv;
  92                  }
  93               }
  94 diff --git a/src/hunspell/affixmgr.hxx b/src/hunspell/affixmgr.hxx
  95 index 2679b7f..8839814 100644
  96 --- misc/hunspell-1.3.3/src/hunspell/affixmgr.hxx
  97 +++ misc/build/hunspell-1.3.3/src/hunspell/affixmgr.hxx
  98 @@ -155,6 +155,7 @@ class LIBHUNSPELL_DLL_EXPORTED AffixMgr
  99    char *              cpdsyllablenum;
 100    const char *        pfxappnd; // BUG: not stateless
 101    const char *        sfxappnd; // BUG: not stateless
 102 +  int                 sfxextra; // BUG: not stateless
 103    FLAG                sfxflag;  // BUG: not stateless
 104    char *              derived;  // BUG: not stateless
 105    SfxEntry *          sfx;      // BUG: not stateless
 106 --
 107 1.9.1
 108