external/hunspell/0001-Keep-only-REP-ph-or-2-word-dictionary-phrase-suggest.patch

   1 From b88f9ea57bdb9b219f3c1d2c67f4f882f1f23194 Mon Sep 17 00:00:00 2001
   2 From: =?UTF-8?q?L=C3=A1szl=C3=B3=20N=C3=A9meth?= <nemeth@numbertext.org>
   3 Date: Sun, 14 May 2023 22:15:15 +0200
   4 Subject: [PATCH] Keep only REP, ph: or 2-word dictionary phrase suggestions
   5
   6 These are the best suggestions, no need to search other
   7 ones to avoid annoying redundant and long list.
   8
   9 For example to suggest only "a lot" to the bad form "alot",
  10 add the 2-word phrase "a lot" to the dic file.
  11
  12 Or for a very typical spelling mistake, enough to specify the
  13 bad form with a ph: in the dictionary file to remove the other
  14 suggestions.
  15
  16 Note: partial revert of commit de9fe28008eb0761c33bd83847f282602c599fda
  17 "fix up some warnings seen with -Wall -Wextra".
  18 ---
  19  src/hunspell/atypes.hxx     |  1 +
  20  src/hunspell/suggestmgr.cxx | 31 ++++++++++++++++++++++++++-----
  21  src/hunspell/suggestmgr.hxx |  2 +-
  22  tests/ph.sug                |  4 ++--
  23  tests/rep.sug               |  2 +-
  24  5 files changed, 31 insertions(+), 9 deletions(-)
  25
  26 diff --git a/src/hunspell/atypes.hxx b/src/hunspell/atypes.hxx
  27 index 7e5a5c0..6e3ed1b 100644
  28 --- a/src/hunspell/atypes.hxx
  29 +++ b/src/hunspell/atypes.hxx
  30 @@ -82,6 +82,7 @@ static inline void HUNSPELL_WARNING(FILE*, const char*, ...) {}
  31  #define SPELL_ORIGCAP (1 << 5)
  32  #define SPELL_WARN (1 << 6)
  33  #define SPELL_COMPOUND_2 (1 << 7)  // permit only 2 dictionary words in the compound
  34 +#define SPELL_BEST_SUG (1 << 8)    // limit suggestions for the best ones, i.e. ph:
  35
  36  #define MINCPDLEN 3
  37  #define MAXCOMPOUND 10
  38 diff --git a/src/hunspell/suggestmgr.cxx b/src/hunspell/suggestmgr.cxx
  39 index 19a24f8..ba688aa 100644
  40 --- a/src/hunspell/suggestmgr.cxx
  41 +++ b/src/hunspell/suggestmgr.cxx
  42 @@ -242,8 +242,11 @@ bool SuggestMgr::suggest(std::vector<std::string>& slst,
  43      if ((slst.size() < maxSug) && (!cpdsuggest || (slst.size() < oldSug + maxcpdsugs))) {
  44        size_t i = slst.size();
  45        replchars(slst, word, cpdsuggest, info);
  46 -      if (slst.size() > i)
  47 +      if (slst.size() > i) {
  48          good_suggestion = true;
  49 +        if (info & SPELL_BEST_SUG)
  50 +          return true;
  51 +      }
  52      }
  53      if (clock() > timelimit + TIMELIMIT_SUGGESTION)
  54        return good_suggestion;
  55 @@ -365,7 +368,10 @@ bool SuggestMgr::suggest(std::vector<std::string>& slst,
  56      // we always suggest them, in despite of nosplitsugs, and
  57      // drop compound word and other suggestions)
  58      if (!cpdsuggest || (!nosplitsugs && slst.size() < oldSug + maxcpdsugs)) {
  59 -      good_suggestion = twowords(slst, word, cpdsuggest, good_suggestion, info);
  60 +      good_suggestion = twowords(slst, word, cpdsuggest, good_suggestion, info);
  61 +
  62 +      if (info & SPELL_BEST_SUG)
  63 +        return true;
  64      }
  65      if (clock() > timelimit + TIMELIMIT_SUGGESTION)
  66        return good_suggestion;
  67 @@ -506,15 +512,23 @@ int SuggestMgr::replchars(std::vector<std::string>& wlst,
  68        candidate.assign(word, 0, r);
  69        candidate.append(entry.outstrings[type]);
  70        candidate.append(word, r + entry.pattern.size(), std::string::npos);
  71 +      size_t sp = candidate.find(' ');
  72 +      size_t oldns = wlst.size();
  73        testsug(wlst, candidate, cpdsuggest, NULL, NULL, info);
  74 +      if (oldns < wlst.size()) {
  75 +        int patlen = entry.pattern.size();
  76 +        int replen = entry.outstrings[type].size();
  77 +        // REP suggestions are the best, don't search other type of suggestions
  78 +        info |= SPELL_BEST_SUG;
  79 +      }
  80 +
  81        // check REP suggestions with space
  82 -      size_t sp = candidate.find(' ');
  83        if (sp != std::string::npos) {
  84          size_t prev = 0;
  85          while (sp != std::string::npos) {
  86            std::string prev_chunk = candidate.substr(prev, sp - prev);
  87            if (checkword(prev_chunk, 0, NULL, NULL)) {
  88 -            size_t oldns = wlst.size();
  89 +            oldns = wlst.size();
  90              std::string post_chunk = candidate.substr(sp + 1);
  91              testsug(wlst, post_chunk, cpdsuggest, NULL, NULL, info);
  92              if (oldns < wlst.size()) {
  93 @@ -854,11 +868,15 @@ bool SuggestMgr::twowords(std::vector<std::string>& wlst,
  94      // alot -> a lot, alto, slot...
  95      *p = ' ';
  96      if (!cpdsuggest && checkword(candidate, cpdsuggest, NULL, NULL)) {
  97 +      // best solution
  98 +      info |= SPELL_BEST_SUG;
  99 +
 100        // remove not word pair suggestions
 101        if (!good) {
 102          good = true;
 103          wlst.clear();
 104        }
 105 +
 106        wlst.insert(wlst.begin(), candidate);
 107      }
 108
 109 @@ -867,6 +885,9 @@ bool SuggestMgr::twowords(std::vector<std::string>& wlst,
 110        *p = '-';
 111
 112        if (!cpdsuggest && checkword(candidate, cpdsuggest, NULL, NULL)) {
 113 +        // best solution
 114 +        info |= SPELL_BEST_SUG;
 115 +
 116          // remove not word pair suggestions
 117          if (!good) {
 118            good = true;
 119 diff --git a/tests/ph.sug b/tests/ph.sug
 120 index 8daee56..ccd936e 100644
 121 --- a/tests/ph.sug
 122 +++ b/tests/ph.sug
 123 @@ -1,11 +1,11 @@
 124  a lot
 125 -in spite, inspire
 126 +in spite
 127  what
 128  what
 129  Wednesday
 130  Wednesday
 131  Wednesday
 132  Wednesday
 133 -which, witch, winch, wish
 134 +which, witch
 135  Oh, my gosh!
 136  OH, MY GOSH!
 137 diff --git a/tests/rep.sug b/tests/rep.sug
 138 index b48a5b8..424731c 100644
 139 --- a/tests/rep.sug
 140 +++ b/tests/rep.sug
 141 @@ -5,4 +5,4 @@ a lot, lot
 142  un alunno
 143  bar
 144  vinte e un
 145 -auto's, auto
 146 +auto's
 147 --
 148 2.25.1
 149