external/hunspell/0001-Hunspell-patches-for-missing-OCONV-conversion.patch

   1 From e13ff056fd65990b88d29fb9eae304b411e58234 Mon Sep 17 00:00:00 2001
   2 From: Changwoo Ryu <cwryu@debian.org>
   3 Date: Wed, 8 Mar 2017 14:04:26 +0900
   4 Subject: [PATCH] Hunspell patches for missing OCONV conversion
   5
   6 4e2abfd Clean up PR #479
   7 cc2d71e Add oconv2 test to Makefile
   8 ca14fdb Avoid gotos across variable initialization
   9 7e5cb62 Use goto to reduce repetitive code
  10 f528192 Add missing OCONV conversion of root and morphemes output
  11 ---
  12  src/hunspell/hunspell.cxx | 59 +++++++++++++++++++++++++++++++++++++++--------
  13  tests/test.sh             | 23 +++++++++++++++---
  14  2 files changed, 70 insertions(+), 12 deletions(-)
  15
  16 diff --git a/src/hunspell/hunspell.cxx b/src/hunspell/hunspell.cxx
  17 index 1100a6f..87d1b4a 100644
  18 --- a/src/hunspell/hunspell.cxx
  19 +++ b/src/hunspell/hunspell.cxx
  20 @@ -98,10 +98,13 @@ public:
  21    std::vector<std::string> stem(const std::string& word);
  22    std::vector<std::string> stem(const std::vector<std::string>& morph);
  23    std::vector<std::string> analyze(const std::string& word);
  24 +  std::vector<std::string> analyze_internal(const std::string& word);
  25    int get_langnum() const;
  26    bool input_conv(const std::string& word, std::string& dest);
  27    bool spell(const std::string& word, int* info = NULL, std::string* root = NULL);
  28 +  bool spell_internal(const std::string& word, int* info = NULL, std::string* root = NULL);
  29    std::vector<std::string> suggest(const std::string& word);
  30 +  std::vector<std::string> suggest_internal(const std::string& word);
  31    const std::string& get_wordchars() const;
  32    const std::vector<w_char>& get_wordchars_utf16() const;
  33    const std::string& get_dict_encoding() const;
  34 @@ -415,6 +418,21 @@ bool Hunspell::spell(const std::string& word, int* info, std::string* root) {
  35  }
  36
  37  bool HunspellImpl::spell(const std::string& word, int* info, std::string* root) {
  38 +  bool r = spell_internal(word, info, root);
  39 +  if (r && root) {
  40 +    // output conversion
  41 +    RepList* rl = (pAMgr) ? pAMgr->get_oconvtable() : NULL;
  42 +    if (rl) {
  43 +      std::string wspace;
  44 +      if (rl->conv(*root, wspace)) {
  45 +        *root = wspace;
  46 +      }
  47 +    }
  48 +  }
  49 +  return r;
  50 +}
  51 +
  52 +bool HunspellImpl::spell_internal(const std::string& word, int* info, std::string* root) {
  53    struct hentry* rv = NULL;
  54
  55    int info2 = 0;
  56 @@ -834,6 +852,22 @@ std::vector<std::string> Hunspell::suggest(const std::string& word) {
  57
  58  std::vector<std::string> HunspellImpl::suggest(const std::string& word) {
  59    std::vector<std::string> slst;
  60 +  slst = suggest_internal(word);
  61 +  // output conversion
  62 +  RepList* rl = (pAMgr) ? pAMgr->get_oconvtable() : NULL;
  63 +  if (rl) {
  64 +    for (size_t i = 0; rl && i < slst.size(); ++i) {
  65 +      std::string wspace;
  66 +      if (rl->conv(slst[i], wspace)) {
  67 +        slst[i] = wspace;
  68 +      }
  69 +    }
  70 +  }
  71 +  return slst;
  72 +}
  73 +
  74 +std::vector<std::string> HunspellImpl::suggest_internal(const std::string& word) {
  75 +  std::vector<std::string> slst;
  76
  77    int onlycmpdsug = 0;
  78    if (!pSMgr || m_HMgrs.empty())
  79 @@ -1150,15 +1184,6 @@ std::vector<std::string> HunspellImpl::suggest(const std::string& word) {
  80    }
  81    slst.resize(l);
  82
  83 -  // output conversion
  84 -  rl = (pAMgr) ? pAMgr->get_oconvtable() : NULL;
  85 -  for (size_t j = 0; rl && j < slst.size(); ++j) {
  86 -    std::string wspace;
  87 -    if (rl->conv(slst[j], wspace)) {
  88 -      slst[j] = wspace;
  89 -    }
  90 -  }
  91 -
  92    return slst;
  93  }
  94
  95 @@ -1365,6 +1390,22 @@ std::vector<std::string> Hunspell::analyze(const std::string& word) {
  96
  97  std::vector<std::string> HunspellImpl::analyze(const std::string& word) {
  98    std::vector<std::string> slst;
  99 +  slst = analyze_internal(word);
 100 +  // output conversion
 101 +  RepList* rl = (pAMgr) ? pAMgr->get_oconvtable() : NULL;
 102 +  if (rl) {
 103 +    for (size_t i = 0; rl && i < slst.size(); ++i) {
 104 +      std::string wspace;
 105 +      if (rl->conv(slst[i], wspace)) {
 106 +        slst[i] = wspace;
 107 +      }
 108 +    }
 109 +  }
 110 +  return slst;
 111 +}
 112 +
 113 +std::vector<std::string> HunspellImpl::analyze_internal(const std::string& word) {
 114 +  std::vector<std::string> slst;
 115    if (!pSMgr || m_HMgrs.empty())
 116      return slst;
 117    if (utf8) {
 118 diff --git a/tests/test.sh b/tests/test.sh
 119 index 22e5087..9344f82 100755
 120 --- a/tests/test.sh
 121 +++ b/tests/test.sh
 122 @@ -112,7 +112,7 @@ if test -f $TESTDIR/$NAME.wrong; then
 123          echo "============================================="
 124          echo "Fail in $NAME.wrong. Bad words recognised as good:"
 125          tr -d '        ' <$TESTDIR/$NAME.wrong >$TEMPDIR/$NAME.wrong.detab
 126 -        diff $TEMPDIR/$NAME.wrong.detab $TEMPDIR/$NAME.wrong | grep '^<' | sed 's/^..//'
 127 +        diff -u $TEMPDIR/$NAME.wrong.detab $TEMPDIR/$NAME.wrong | grep '^<' | sed 's/^..//'
 128          rm -f $TEMPDIR/$NAME.wrong $TEMPDIR/$NAME.wrong.detab
 129          exit 1
 130      fi
 131 @@ -121,6 +121,23 @@ fi
 132
 133  check_valgrind_log "bad words"
 134
 135 +# Tests good words' root
 136 +if test -f $TESTDIR/$NAME.root; then
 137 +    # Extract the root words of the affixed words, after '+'
 138 +    hunspell $* -d $TESTDIR/$NAME <$TESTDIR/$NAME.good | grep -a '^+ ' | \
 139 +        sed 's/^+ //' >$TEMPDIR/$NAME.root
 140 +    if ! cmp $TEMPDIR/$NAME.root $TESTDIR/$NAME.root >/dev/null; then
 141 +        echo "============================================="
 142 +        echo "Fail in $NAME.root. Bad prefix or suffix?"
 143 +        diff -u $TESTDIR/$NAME.root $TEMPDIR/$NAME.root
 144 +        rm -f $TEMPDIR/$NAME.root
 145 +        exit 1
 146 +    fi
 147 +    rm -f $TEMPDIR/$NAME.root
 148 +fi
 149 +
 150 +check_valgrind_log "root"
 151 +
 152  # Tests morphological analysis
 153  if test -f $TESTDIR/$NAME.morph; then
 154      sed 's/    $//' $TESTDIR/$NAME.good >$TEMPDIR/$NAME.good
 155 @@ -129,7 +146,7 @@ if test -f $TESTDIR/$NAME.morph; then
 156      if ! cmp $TEMPDIR/$NAME.morph $TESTDIR/$NAME.morph >/dev/null; then
 157          echo "============================================="
 158          echo "Fail in $NAME.morph. Bad analysis?"
 159 -        diff $TESTDIR/$NAME.morph $TEMPDIR/$NAME.morph | grep '^<' | sed 's/^..//'
 160 +        diff -u $TESTDIR/$NAME.morph $TEMPDIR/$NAME.morph | grep '^<' | sed 's/^..//'
 161          rm -f $TEMPDIR/$NAME.morph
 162          exit 1
 163      fi
 164 @@ -145,7 +162,7 @@ if test -f $TESTDIR/$NAME.sug; then
 165      if ! cmp $TEMPDIR/$NAME.sug $TESTDIR/$NAME.sug >/dev/null; then
 166          echo "============================================="
 167          echo "Fail in $NAME.sug. Bad suggestion?"
 168 -        diff $TESTDIR/$NAME.sug $TEMPDIR/$NAME.sug
 169 +        diff -u $TESTDIR/$NAME.sug $TEMPDIR/$NAME.sug
 170          rm -f $TEMPDIR/$NAME.sug
 171          exit 1
 172      fi
 173 --
 174 2.7.4
 175