external/liborcus/0001-add-xml-path.patch

   1 diff --git a/include/orcus/xml_structure_tree.hpp b/include/orcus/xml_structure_tree.hpp
   2 index 58cabfd116fa24e35ff27cf8d7512b6e73df33f4..c88808d24bd74c175fa4017328d3e54b4c588c5e 100644
   3 --- a/include/orcus/xml_structure_tree.hpp
   4 +++ b/include/orcus/xml_structure_tree.hpp
   5 @@ -127,6 +127,20 @@ public:
   6          size_t get_xmlns_index(xmlns_id_t ns) const;
   7
   8          std::string get_xmlns_short_name(xmlns_id_t ns) const;
   9 +
  10 +        /**
  11 +         * Get a XPath like ID for the element inside of the XML tree.
  12 +         *
  13 +         */
  14 +        std::string get_path() const;
  15 +
  16 +        /**
  17 +         * Select an element by a path expression. The path expression may be
  18 +         * generated by <code>xml_structure_tree::walker::get_path</code>.
  19 +         *
  20 +         * @param path a simple XPath like expression
  21 +         */
  22 +        element select_by_path(const std::string& path);
  23      };
  24
  25      xml_structure_tree(xmlns_context& xmlns_cxt);
  26 diff --git a/src/liborcus/xml_structure_tree.cpp b/src/liborcus/xml_structure_tree.cpp
  27 index 2778bc05f32841a9441bf471913872e119256895..6622bc57cd2595f12bba80d4bbdb5c24cd6e7bc6 100644
  28 --- a/src/liborcus/xml_structure_tree.cpp
  29 +++ b/src/liborcus/xml_structure_tree.cpp
  30 @@ -12,6 +12,7 @@
  31  #include "orcus/exception.hpp"
  32
  33  #include "orcus/string_pool.hpp"
  34 +#include "string_helper.hpp"
  35
  36  #include <iostream>
  37  #include <sstream>
  38 @@ -275,6 +276,15 @@ struct xml_structure_tree_impl
  39      {
  40          delete mp_root;
  41      }
  42 +
  43 +    std::string get_element_str(const xml_structure_tree::entity_name& name) const
  44 +    {
  45 +        ostringstream ss;
  46 +        if (m_xmlns_cxt.get_index(name.ns) != index_not_found)
  47 +            ss << m_xmlns_cxt.get_short_name(name.ns) << ":";
  48 +        ss << name.name;
  49 +        return ss.str();
  50 +    }
  51  };
  52
  53  struct xml_structure_tree::walker_impl
  54 @@ -423,6 +433,66 @@ string xml_structure_tree::walker::get_xmlns_short_name(xmlns_id_t ns) const
  55      return mp_impl->m_parent_impl.m_xmlns_cxt.get_short_name(ns);
  56  }
  57
  58 +string xml_structure_tree::walker::get_path() const
  59 +{
  60 +    ostringstream ss;
  61 +    for (auto& element : mp_impl->m_scopes)
  62 +    {
  63 +        ss << "/" << mp_impl->m_parent_impl.get_element_str(element.name);
  64 +    }
  65 +
  66 +    return ss.str();
  67 +}
  68 +
  69 +xml_structure_tree::element xml_structure_tree::walker::select_by_path(const std::string& path)
  70 +{
  71 +    pstring p(path);
  72 +    std::vector<pstring> parts = string_helper::split_string(p, '/');
  73 +    if (parts.empty())
  74 +        throw general_error("invalid format for path");
  75 +
  76 +    // string_helper::split_string will create an empty first element due to leading '/'
  77 +    if (parts[0] != "")
  78 +    {
  79 +        throw general_error("invalid format for path");
  80 +    }
  81 +    else
  82 +    {
  83 +        parts.erase(parts.begin());
  84 +    }
  85 +
  86 +    if (parts.empty())
  87 +        throw general_error("invalid format for path");
  88 +
  89 +    element_ref root_ref(mp_impl->mp_root->name, &mp_impl->mp_root->prop);
  90 +    if (pstring(mp_impl->m_parent_impl.get_element_str(root_ref.name)) != parts[0])
  91 +        throw general_error("path does not match any element");
  92 +
  93 +    std::vector<element_ref> scopes;
  94 +    scopes.push_back(root_ref);
  95 +
  96 +    for (size_t i = 1; i < parts.size(); ++i)
  97 +    {
  98 +        const elem_prop& prop = *scopes.back().prop;
  99 +        bool found = false;
 100 +        for (auto& child : prop.child_elements)
 101 +        {
 102 +            if (pstring(mp_impl->m_parent_impl.get_element_str(child.first)) == parts[i])
 103 +            {
 104 +                scopes.emplace_back(child.first, child.second);
 105 +                found = true;
 106 +                break;
 107 +            }
 108 +        }
 109 +        if (!found)
 110 +            throw general_error("path does not match any element");
 111 +    }
 112 +
 113 +    std::swap(mp_impl->m_scopes, scopes);
 114 +    const element_ref& ref = mp_impl->m_scopes.back();
 115 +    return element(ref.name, ref.prop->repeat);
 116 +}
 117 +
 118  xml_structure_tree::xml_structure_tree(xmlns_context& xmlns_cxt) :
 119      mp_impl(new xml_structure_tree_impl(xmlns_cxt)) {}