From f961573843b1b5283f32b1cbadac26d3421684bd Mon Sep 17 00:00:00 2001
From: Nilesh Chakraborty <nilesh@nileshc.com>
Date: Sun, 16 Jun 2019 20:58:59 +0200
Subject: [PATCH] Can show data loading progress, can block cout/cerr debug
 messages from cpp library

---
 include/hdt_document.hpp |  9 ++++++---
 src/hdt.cpp              |  6 ++++--
 src/hdt_document.cpp     | 36 +++++++++++++++++++++++++++++++-----
 3 files changed, 41 insertions(+), 10 deletions(-)

diff --git a/include/hdt_document.hpp b/include/hdt_document.hpp
index 3d61822..de41227 100644
--- a/include/hdt_document.hpp
+++ b/include/hdt_document.hpp
@@ -31,8 +31,9 @@ class HDTDocument {
 private:
   std::string hdt_file;
   hdt::HDT *hdt;
+  bool debug;
   hdt::QueryProcessor *processor;
-  HDTDocument(std::string file, bool map, bool indexed);
+  HDTDocument(std::string file, bool map, bool indexed, bool progress, bool debug);
 
 public:
   /*!
@@ -87,9 +88,11 @@ class HDTDocument {
    * @param file - Path to the HDT file
    * @param map - True maps the HDT file (faster), False loads everything in memory
    * @param indexed -  True if the HDT must be loaded with indexes, False otherwise
+   * @param progress - True to show progress on stdout, False otherwise
+   * @param debug - True to show cout debug messages from hdt-cpp, False otherwise
    */
-  static HDTDocument create(std::string file, bool map, bool indexed) {
-    return HDTDocument(file, map, indexed);
+  static HDTDocument create(std::string file, bool map, bool indexed, bool progress, bool debug) {
+    return HDTDocument(file, map, indexed, progress, debug);
   }
 
   /*!
diff --git a/src/hdt.cpp b/src/hdt.cpp
index cc044e0..b6336fa 100644
--- a/src/hdt.cpp
+++ b/src/hdt.cpp
@@ -80,8 +80,10 @@ PYBIND11_MODULE(hdt, m) {
 
   py::class_<HDTDocument>(m, "HDTDocument", HDT_DOCUMENT_CLASS_DOC)
       .def(py::init(&HDTDocument::create), py::arg("file"),
-                                           py::arg("map") = false,
-                                           py::arg("indexed") = true)
+                                           py::arg("map") = true,
+                                           py::arg("indexed") = true,
+                                           py::arg("progress") = false,
+                                           py::arg("debug") = false)
       .def_property_readonly("file_path", &HDTDocument::getFilePath,
                              HDT_DOCUMENT_GETFILEPATH_DOC)
       .def_property_readonly("total_triples", &HDTDocument::getNbTriples,
diff --git a/src/hdt_document.cpp b/src/hdt_document.cpp
index 89ef947..546effd 100644
--- a/src/hdt_document.cpp
+++ b/src/hdt_document.cpp
@@ -52,21 +52,27 @@ inline bool file_exists(const std::string &name) {
  * @param file - Path to HDT file to load
  * @param map - True maps the HDT file (faster), False loads everything in memory
  * @param indexed -  True if the HDT must be loaded with indexes, False otherwise
+ * @param progress - True to show loading progress on stdout, False otherwise
+ * @param debug - True to show debug messages from hdt-cpp, False otherwise
  */
-HDTDocument::HDTDocument(std::string file, bool map, bool indexed) {
+HDTDocument::HDTDocument(std::string file, bool map, bool indexed, bool progress, bool debug)
+    :debug(debug)
+ {
   hdt_file = file;
   if (!file_exists(file)) {
     throw std::runtime_error("Cannot open HDT file '" + file + "': Not Found!");
   }
 
+  StdoutProgressListener* listener = progress ? nullptr : new StdoutProgressListener();
+
   if(!map && indexed) {
-    hdt = HDTManager::loadIndexedHDT(file.c_str());
+    hdt = HDTManager::loadIndexedHDT(file.c_str(), listener);
   } else if(!map && !indexed) {
-    hdt = HDTManager::loadHDT(file.c_str());
+    hdt = HDTManager::loadHDT(file.c_str(), listener);
   } else if(map && indexed){
-    hdt = HDTManager::mapIndexedHDT(file.c_str());
+    hdt = HDTManager::mapIndexedHDT(file.c_str(), listener);
   } else {
-    hdt = HDTManager::mapHDT(file.c_str());
+    hdt = HDTManager::mapHDT(file.c_str(), listener);
   }
   processor = new QueryProcessor(hdt);
 }
@@ -288,6 +294,19 @@ unsigned int HDTDocument::convertTerm(std::string term, IdentifierPosition pos)
  * @return A JoinIterator* used to evaluated the join.
  */
 JoinIterator * HDTDocument::searchJoin(std::vector<triple> patterns) {
+  std::streambuf *cout_original = nullptr;
+  std::streambuf *cerr_original = nullptr;
+  std::ofstream *null_out = nullptr;
+
+  if(!debug) {
+    cout_original = std::cout.rdbuf();
+    cerr_original = std::cerr.rdbuf();
+
+    null_out =  new std::ofstream("/dev/null");
+    std::cout.rdbuf(null_out->rdbuf()); // redirect 'cout' to /dev/null
+    std::cerr.rdbuf(null_out->rdbuf()); // redirect 'cerr' to /dev/null
+  }
+
   set<string> vars {};
   vector<TripleString> joinPatterns {};
   std::string subj, pred, obj;
@@ -311,5 +330,12 @@ JoinIterator * HDTDocument::searchJoin(std::vector<triple> patterns) {
   }
 
   VarBindingString *iterator = processor->searchJoin(joinPatterns, vars);
+
+  if(!debug) {
+    std::cout.rdbuf(cout_original);
+    std::cerr.rdbuf(cerr_original);
+    null_out->close();
+  }
+
   return new JoinIterator(iterator);
 }