xml-tree-builder.cpp

   1 #include "xml-tree-builder.hpp"
   2 #include <stdexcept>
   3 #include <cstdio>
   4 #include <utility>
   5
   6 using namespace SXSI;
   7
   8 xml_tree_builder::xml_tree_builder()
   9 {
  10   opened = false;
  11   par = 0;
  12   tags = 0;
  13   tag_ids = 0;
  14   text_positions = 0;
  15   disable_text_index = false;
  16   tc_builder = 0;
  17 }
  18
  19 xml_tree_builder::~xml_tree_builder()
  20 {
  21   if (opened) reset();
  22 }
  23
  24 void xml_tree_builder::reset()
  25 {
  26   delete par;
  27   delete tags;
  28   delete tag_ids;
  29   if (!disable_text_index){
  30     delete tc_builder;
  31     delete text_positions;
  32   };
  33 }
  34
  35
  36 int32_t xml_tree_builder::register_tag(std::string tag, int32_t id)
  37 {
  38   auto found = tag_ids->find(tag);
  39
  40   if (found == tag_ids->end()) {
  41     if (id != current_tag)
  42       throw std::runtime_error("xml-tree-builder: inconsistant tag id");
  43
  44     tag_ids->insert(std::make_pair(tag, id));
  45     current_tag++;
  46     return id;
  47   } else
  48     return found->second;
  49
  50 }
  51
  52 int32_t xml_tree_builder::register_tag(std::string tag)
  53 {
  54   return register_tag(tag, current_tag);
  55 }
  56
  57 void
  58 xml_tree_builder::open_document(bool disable_text_index,
  59                                 unsigned int sample_rate,
  60                                 TextCollectionBuilder::index_type_t idx_type)
  61 {
  62   if (opened) reset();
  63   opened = true;
  64   par = new bit_vector();
  65   tags = new std::vector<int32_t>();
  66   current_tag = 0;
  67   tag_ids = new std::unordered_map<std::string, int32_t>();
  68
  69   register_tag(xml_tree::DOCUMENT_OPEN_TAG, xml_tree::DOCUMENT_OPEN_TAG_ID);
  70   register_tag(xml_tree::DOCUMENT_OPEN_TAG, xml_tree::DOCUMENT_OPEN_TAG_ID);
  71
  72   register_tag(xml_tree::ATTRIBUTE_OPEN_TAG, xml_tree::ATTRIBUTE_OPEN_TAG_ID);
  73   register_tag(xml_tree::ATTRIBUTE_OPEN_TAG, xml_tree::ATTRIBUTE_OPEN_TAG_ID);
  74
  75   register_tag(xml_tree::PCDATA_OPEN_TAG, xml_tree::PCDATA_OPEN_TAG_ID);
  76   register_tag(xml_tree::PCDATA_OPEN_TAG, xml_tree::PCDATA_OPEN_TAG_ID);
  77
  78   register_tag(xml_tree::ATTRIBUTE_DATA_OPEN_TAG,
  79                xml_tree::ATTRIBUTE_DATA_OPEN_TAG_ID);
  80
  81   register_tag(xml_tree::ATTRIBUTE_DATA_OPEN_TAG,
  82                xml_tree::ATTRIBUTE_DATA_OPEN_TAG_ID);
  83
  84   register_tag(xml_tree::CLOSE_TAG,
  85                xml_tree::CLOSE_TAG_ID);
  86
  87   this->disable_text_index = disable_text_index;
  88   if (!disable_text_index){
  89     fprintf(stderr, "Sample rate is %u\n", sample_rate);
  90     tc_builder = TextCollectionBuilder::create(sample_rate, idx_type);
  91     text_positions = new bit_vector();
  92     text_index_type = idx_type;
  93   };
  94 }
  95
  96 void xml_tree_builder::open_tag(std::string tag)
  97 {
  98   int32_t id = register_tag(tag);
  99   tags->push_back(id);
 100   par->push_back(true);
 101   if (!disable_text_index) text_positions->set_le(text_positions->size(), false);
 102 }
 103
 104 void xml_tree_builder::close_tag(std::string)
 105 {
 106   xml_tree::tag_t t = xml_tree::CLOSE_TAG_ID;
 107   tags->push_back(t);
 108   par->push_back(false);
 109   if (!disable_text_index) text_positions->set_le(text_positions->size(), false);
 110 }
 111
 112 void xml_tree_builder::text(std::string s)
 113 {
 114   if (!disable_text_index){
 115     if (s.empty()) s = "\001";
 116     tc_builder->InsertText((const unsigned char *) s.c_str());
 117     text_positions->set_le(text_positions->size() - 1, true);
 118   }
 119 }
 120
 121 xml_tree *xml_tree_builder::close_document()
 122 {
 123   if (opened) {
 124     opened = false;
 125     auto tags_ = tags;
 126     auto tag_ids_ = tag_ids;
 127     auto par_ = par;
 128     auto tc_builder_ = tc_builder;
 129     auto text_positions_ = text_positions;
 130     tc_builder = 0;
 131     text_positions = 0;
 132     tags = 0;
 133     tag_ids = 0;
 134     par = 0;
 135     return new xml_tree(tags_, tag_ids_, par_,
 136                         disable_text_index,
 137                         tc_builder_,
 138                         text_index_type,
 139                         text_positions_);
 140   };
 141
 142   throw std::runtime_error("xml_tree_builder: inconsistent parser state");
 143 }
 144