Initial commit, WIP

1 files changed, 91 insertions, 0 deletions

diff --git a/service/apiImpl.cpp b/service/apiImpl.cpp
new file mode 100644
index 0000000..588d360
--- /dev/null
+++ b/service/apiImpl.cpp
@@ -0,0 +1,91 @@
+#include "apiImpl.h"
+
+#include <sql/getServices.sql.h>
+#include <buffer.h>
+#include <memory>
+
+#include <libxml/xpath.h>
+#include <libxml/xpathInternals.h>
+#include <libxml/HTMLparser.h>
+#include <libxml/HTMLtree.h>
+
+namespace MirrorSearch {
+	SearchImpl::SearchImpl(IceTray::DatabasePoolPtr db) :
+		IceTray::AbstractDatabaseClient(db),
+		log(LOGMANAGER()->getLogger<SearchImpl>())
+	{
+	}
+
+	SearchServices SearchImpl::getServices(const ::Ice::Current&)
+	{
+		return fetch<SearchServices>(sql::getServices);
+	}
+
+	typedef std::shared_ptr<xmlDoc> xmlDocSPtr;
+	typedef std::shared_ptr<xmlXPathContext> xmlXPathContextSPtr;
+	typedef std::shared_ptr<xmlXPathObject> xmlXPathObjectSPtr;
+
+	static auto getDoc(const ::std::string & url, int flags)
+	{
+		if (auto doc = xmlDocSPtr(htmlReadFile(url.c_str(), NULL, flags), xmlFreeDoc)) {
+			return doc;
+		}
+		throw XmlError("Failed to open " + url);
+	}
+
+	static auto getXPathCxt(const xmlDocSPtr & doc)
+	{
+		if (auto xpathCtx = xmlXPathContextSPtr(xmlXPathNewContext(doc.get()), xmlXPathFreeContext)) {
+			return xpathCtx;
+		}
+		throw XmlError("Failed to create xpath context");
+	}
+
+	static auto getXPathObj(const ::std::string & xpath, const xmlXPathContextSPtr & ctx, xmlXPathObjectType type)
+	{
+		if (auto xpathObj = xmlXPathObjectSPtr(xmlXPathEvalExpression(BAD_CAST xpath.c_str(), ctx.get()), xmlXPathFreeObject)) {
+			if (xpathObj->type != type) {
+				throw XmlError("Xpath evaluates to wrong type " + xpath);
+			}
+			return xpathObj;
+		}
+		throw XmlError("Failed to evaluate xpath " + xpath);
+	}
+
+	void SearchImpl::callService(const ::std::string & fn, const SearchServicePtr & s, SearchHits & sh) const
+	{
+		auto fmt = AdHoc::Buffer::getFormat(s->baseurl);
+		auto url = (*fmt % fn).str();
+		auto doc = getDoc(url,
+				HTML_PARSE_RECOVER | HTML_PARSE_NODEFDTD | HTML_PARSE_NOIMPLIED |
+				HTML_PARSE_NOWARNING | HTML_PARSE_NOERROR);
+		auto xpathCtx = getXPathCxt(doc);
+		auto xpathObj = getXPathObj(s->listxpath, xpathCtx, xmlXPathObjectType::XPATH_NODESET);
+		log->messagebf(LOG::INFO, "%d nodes matched %s", xpathObj->nodesetval->nodeNr, s->listxpath);
+		for (int row = 0; row < xpathObj->nodesetval->nodeNr; row += 1) {
+			xpathCtx->node = xpathObj->nodesetval->nodeTab[row];
+			auto xpathObjI = getXPathObj(s->urlxpath, xpathCtx, xmlXPathObjectType::XPATH_STRING);
+			if (xpathObjI->stringval && *xpathObjI->stringval) {
+				sh.push_back(new SearchHit(0, s->id, (const char *) xpathObjI->stringval));
+			}
+		}
+	}
+
+	SearchHits SearchImpl::getMatches(const ::std::string & fn, const ::Ice::Current & c)
+	{
+		SearchHits sh;
+		for (const auto & s : getServices(c)) {
+			callService(fn, s, sh);
+		}
+		return sh;
+	}
+
+	::IceUtil::Optional<::std::string> SearchImpl::feelingLucky(const ::std::string & fn, const ::Ice::Current & c)
+	{
+		const auto ms = getMatches(fn, c);
+		if (ms.empty())
+			return IceUtil::None;
+		return ms.front()->url;
+	}
+}
+