King_DuckZ/duckscraper
1
0
Fork 0
Code Issues Pull requests Releases Wiki Activity

Store the same tree for ApplyBlocks too.

Browse source
This commit is contained in:
King_DuckZ 2018-01-17 01:29:46 +00:00
parent 2fd4daf52c
commit 41bb315b02
2 changed files with 30 additions and 20 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

49
src/scraplang/apply.cpp
View file

@ -36,12 +36,6 @@ namespace duck { namespace sl {
namespace { namespace {
struct EntryNode; struct EntryNode;
struct ApplyEntry {
const SourceInfo* apply_to;
const std::vector<StructItem>* content;
const std::string* mustache_name;
};
struct MustacheEntry { struct MustacheEntry {
std::string text; std::string text;
mstch::map context; mstch::map context;
@ -49,7 +43,7 @@ namespace duck { namespace sl {
using EntryNodeList = std::vector<std::pair< using EntryNodeList = std::vector<std::pair<
const SourceInfo*, const SourceInfo*,
std::vector<EntryNode> EntryNode
>>; >>;
using MustacheEntryMap = std::map<std::string, MustacheEntry>; using MustacheEntryMap = std::map<std::string, MustacheEntry>;
@ -66,6 +60,21 @@ namespace duck { namespace sl {
std::vector<const XPathElement*> xpaths; std::vector<const XPathElement*> xpaths;
}; };
struct ApplyEntry {
ApplyEntry (const SourceInfo* parAppTo, std::string_view parMstchName) :
apply_to(parAppTo),
content(""),
mustache_name(parMstchName)
{
}
ApplyEntry (ApplyEntry&&) = default;
ApplyEntry& operator=(ApplyEntry&&) = default;
const SourceInfo* apply_to;
EntryNode content;
std::string_view mustache_name;
};
class StructItemExtractor : public boost::static_visitor<> { class StructItemExtractor : public boost::static_visitor<> {
public: public:
explicit StructItemExtractor (EntryNode& parRoot) : explicit StructItemExtractor (EntryNode& parRoot) :
@ -89,6 +98,16 @@ namespace duck { namespace sl {
EntryNode& m_root; EntryNode& m_root;
}; };
void store_entry_subtree (
const std::vector<StructItem>& parXPaths,
EntryNode& parCurrList
) {
for (auto& itm : parXPaths) {
StructItemExtractor extractor(parCurrList);
boost::apply_visitor(extractor, itm);
}
}
class DictBuilder : public boost::static_visitor<> { class DictBuilder : public boost::static_visitor<> {
public: public:
explicit DictBuilder (HtmlPoolBaseSP parHtmlPool) : explicit DictBuilder (HtmlPoolBaseSP parHtmlPool) :
@ -109,26 +128,18 @@ namespace duck { namespace sl {
#endif #endif
m_global_entries.emplace_back(std::make_pair( m_global_entries.emplace_back(std::make_pair(
&parVal.source, &parVal.source,
std::vector<EntryNode>() EntryNode("")
)); ));
auto& curr_list = m_global_entries.back().second;
for (auto& itm : parVal.xpaths) { store_entry_subtree(parVal.xpaths, m_global_entries.back().second);
curr_list.emplace_back("");
StructItemExtractor extractor(curr_list.back());
boost::apply_visitor(extractor, itm);
}
} }
void operator() (const ApplyBlock& parVal) { void operator() (const ApplyBlock& parVal) {
#if defined(APPLY_VERBOSE) #if defined(APPLY_VERBOSE)
std::cout << parVal << '\n'; std::cout << parVal << '\n';
#endif #endif
m_apply_entries.push_back(ApplyEntry { m_apply_entries.emplace_back(&parVal.source, parVal.mustache_model);
&parVal.source, store_entry_subtree(parVal.xpaths, m_apply_entries.back().content);
&parVal.xpaths,
&parVal.mustache_model
});
} }
void operator() (const MustacheBlock& parVal) { void operator() (const MustacheBlock& parVal) {

1
src/scraplang/parse.hpp
View file

@ -24,7 +24,6 @@
namespace duck { namespace sl { namespace duck { namespace sl {
std::vector<ScrapNode> parse ( const std::string& parData ); std::vector<ScrapNode> parse ( const std::string& parData );
//std::vector<element_def> get_xpath_definitions ( const ScrapNode& parAST );
}} //namespace duck::sl }} //namespace duck::sl
#endif #endif