libvspell/stridtree.cpp

   1 #include "stridtree.h"
   2
   3 using namespace std;
   4
   5 void BranchNode::get_leaves(std::vector<LeafNode*> &_nodes) const
   6 {
   7         const_np_range range;
   8         range = nodes.equal_range(sarch["<leaf>"]);
   9         node_map::const_iterator iter;
  10         for (iter = range.first;iter != range.second; ++iter)
  11                 if (iter->second->is_leaf())
  12                         _nodes.push_back((LeafNode*)iter->second.get());
  13 }
  14
  15 void BranchNode::get_branches(strid _id,std::vector<BranchNode*> &_nodes) const
  16 {
  17         const_np_range range;
  18         range = nodes.equal_range(_id);
  19         node_map::const_iterator iter;
  20         for (iter = range.first;iter != range.second; ++iter)
  21                 if (!iter->second->is_leaf())
  22                         _nodes.push_back((BranchNode*)iter->second.get());
  23 }
  24
  25 BranchNode* BranchNode::get_branch(strid _id) const
  26 {
  27         const_np_range range;
  28         range = nodes.equal_range(_id);
  29         node_map::const_iterator iter;
  30         for (iter = range.first;iter != range.second; ++iter)
  31                 if (!iter->second->is_leaf())
  32                         return (BranchNode*)iter->second.get();
  33         return NULL;
  34 }
  35
  36 void BranchNode::add(strid _id,NodeRef _branch)
  37 {
  38         nodes.insert(make_pair(_id,_branch));
  39 }
  40
  41 BranchNode* BranchNode::add_path(std::vector<strid> toks)
  42 {
  43         uint i,n = toks.size();
  44         BranchNode *me = this;
  45         for (i = 0;i < n;i ++) {
  46                 BranchNode *next = me->get_branch(toks[i]);
  47                 if (next == NULL) {
  48                         NodeRef branch(new BranchNode());
  49                         me->add(toks[i],branch);
  50                         next = (BranchNode*)branch.get();
  51                 }
  52                 me = next;
  53         }
  54         return me;
  55 }
  56
  57 bool WordArchive::load(const char* filename)
  58 {
  59         File ifs(filename,"rt");
  60
  61         if (ifs.error())
  62                 return false;
  63
  64         int nr_lines = 0;
  65         char *line;
  66         int start,len,tmp_char;
  67         string::size_type pos;
  68         char *str_pos;
  69         vector<string> toks;
  70         while ((line = ifs.getline())) {
  71                 start = 0;
  72                 len = strlen(line);
  73                 toks.clear();
  74                 while (start < len) {
  75                         str_pos = strchr(line+start,'#');
  76                         if (str_pos == NULL)
  77                                 pos = len;
  78                         else
  79                                 pos = str_pos - line;
  80                         tmp_char = line[pos];
  81                         line[pos] = 0;
  82                         toks.push_back(line+start);
  83                         line[pos] = tmp_char;
  84                         start = pos+1;
  85                 }
  86
  87                 if (!toks.size())
  88                         continue;                       // unrecoverable error
  89                 if (toks.size() < 2)
  90                         toks.push_back("N");    // assume N
  91                 if (toks.size() < 3)
  92                         toks.push_back("0");    // assume 0
  93
  94                 add_entry(toks);
  95
  96                 nr_lines ++;
  97         }
  98
  99         return true;
 100 }
 101
 102 void WordArchive::add_special_entry(strid tok)
 103 {
 104         NodeRef noderef(new LeafNode);
 105         LeafNode *leaf = (LeafNode*)noderef.get();
 106         leaf->set_id(tok);
 107         get_root()->add(sarch["<leaf>"],noderef);
 108 }
 109
 110 void WordArchive::add_entry(vector<string> toks)
 111 {
 112         unsigned start,len,pos;
 113         start = 0;
 114         len = toks[0].size();
 115         vector<VocabIndex> syllables;
 116         while (start < len) {
 117                 pos = toks[0].find(' ',start);
 118                 if (pos == string::npos)
 119                         pos = len;
 120                 string s = toks[0].substr(start,pos-start);
 121                 VocabIndex id = sarch[get_dic_syllable(s)];
 122                 syllables.push_back(id);
 123                 start = pos+1;
 124         }
 125
 126         vector<strid> path = syllables;
 127         BranchNode* branch = get_root()->add_path(path);
 128         NodeRef noderef(new LeafNode);
 129         LeafNode *leaf = (LeafNode*)noderef.get();
 130         branch->add("<leaf>",noderef);
 131
 132         // reconstruct word id
 133         string word;
 134         int i,nr_syllables = syllables.size();
 135         for (i = 0;i < nr_syllables;i ++) {
 136                 if (i)
 137                         word += "_";
 138                 word += sarch[syllables[i]];
 139         }
 140         leaf->set_id(sarch[word]);
 141
 142 }