initial datetime support

shogun-toolbox · May 9, 2019 · 6b855f6 · 6b855f6
1 parent 9255cd3
commit 6b855f6
Show file tree

Hide file tree

Showing 2 changed files with 49 additions and 21 deletions.
diff --git a/src/shogun/io/ARFFFile.cpp b/src/shogun/io/ARFFFile.cpp
@@ -7,7 +7,7 @@
 #include <shogun/io/ARFFFile.h>
 #include <shogun/mathematics/linalg/LinalgNamespace.h>
 
-#include <iostream>
+#include <ctime>
 
 using namespace shogun;
 using namespace shogun::arff_detail;
@@ -55,7 +55,8 @@ void ARFFDeserializer::read()
 			// store attribute name and type
 			std::string name;
 			std::string type;
-			auto inner_string = m_current_line.substr(strlen(m_attribute_string));
+			auto inner_string =
+			    m_current_line.substr(strlen(m_attribute_string));
 			left_trim(inner_string);
 			auto it = inner_string.begin();
 			while (it != inner_string.end())
@@ -71,14 +72,16 @@ void ARFFDeserializer::read()
 			}
 			if (it == inner_string.end())
 				SG_SERROR(
-				    "Could not split attibute name and type on line %d: \"%s\".\n",
+				    "Could not split attibute name and type on line %d: "
+				    "\"%s\".\n",
 				    m_line_number, m_current_line.c_str())
 			// check if it is nominal
 			if (type[0] == '{')
 			{
 				std::vector<std::string> attributes;
 				// split norminal values: "{A, B, C}" to vector{A, B, C}
-				split(type.substr(1, type.size() - 2), ", ", true,
+				split(
+				    type.substr(1, type.size() - 2), ", ", true,
 				    std::back_inserter(attributes));
 				m_nominal_attributes.emplace_back(
 				    std::make_pair(name, attributes));
@@ -91,23 +94,24 @@ void ARFFDeserializer::read()
 			{
 				std::vector<std::string> date_elements;
 				// split "date [[date-format]]" or "name date [[date-format]]"
-				split(type, " ", true,
-					  std::back_inserter(date_elements));
-				if (date_elements[0]=="date" && date_elements.size() < 3)
+				split(type, " ", true, std::back_inserter(date_elements));
+				if (date_elements[0] == "date" && date_elements.size() < 3)
 				{
 					// @attribute date [[date-format]]
 					if (type.size() == 1)
 						m_date_formats.emplace_back(m_default_date_format);
 					else
-						m_date_formats.push_back(javatime_to_cpptime(date_elements[1]));
+						m_date_formats.push_back(
+						    javatime_to_cpptime(date_elements[1]));
 				}
-				else if (date_elements[1]=="date" && date_elements.size() < 4)
+				else if (date_elements[1] == "date" && date_elements.size() < 4)
 				{
 					// @attribute name date [[date-format]]
 					if (date_elements.size() == 2)
 						m_date_formats.emplace_back(m_default_date_format);
 					else
-						m_date_formats.push_back(javatime_to_cpptime(date_elements[2]));
+						m_date_formats.push_back(
+						    javatime_to_cpptime(date_elements[2]));
 				}
 				else
 				{
@@ -172,12 +176,13 @@ void ARFFDeserializer::read()
 		{
 			return;
 		}
-		// assumes that until EOF we should expect tabular data to be parsed
+		// assumes that until EOF we should expect comma delimited values
 		else
 		{
 			std::vector<std::string> elems;
 			split(m_current_line, ",", true, std::back_inserter(elems));
 			auto nominal_pos = m_nominal_attributes.begin();
+			auto date_pos = m_date_formats.begin();
 			for (int i = 0; i < elems.size(); ++i)
 			{
 				Attribute type = m_attributes[i];
@@ -194,7 +199,8 @@ void ARFFDeserializer::read()
 					catch (const std::invalid_argument&)
 					{
 						SG_SERROR(
-						    "Failed to covert \"%s\" to numeric.\n", elems[i].c_str())
+						    "Failed to covert \"%s\" to numeric.\n",
+						    elems[i].c_str())
 					}
 				}
 				break;
@@ -219,7 +225,34 @@ void ARFFDeserializer::read()
 				}
 				break;
 				case (Attribute::Date):
-					SG_SERROR("Date parsing not implemented.\n")
+				{
+					tm t{};
+					if (date_pos == m_date_formats.end())
+						SG_SERROR(
+						    "Unexpected date value \"%s\" on line "
+						    "%d.\n",
+						    elems[i].c_str(), m_line_number);
+					if (strptime(elems[i].c_str(), (*date_pos).c_str(), &t))
+					{
+						auto value_timestamp = std::mktime(&t);
+						if (value_timestamp == -1)
+							SG_SERROR(
+							    "Error creating timestamp with \"%s\" with "
+							    "date format \"%s\" on line %d.\n",
+							    elems[i].c_str(), (*date_pos).c_str(),
+							    m_line_number)
+						else
+							m_data.emplace_back(value_timestamp);
+					}
+					else
+						SG_SERROR(
+						    "Error parsing date \"%s\" with date format \"%s\" "
+						    "on line %d.\n",
+						    elems[i].c_str(), (*date_pos).c_str(),
+						    m_line_number)
+					++date_pos;
+				}
+				break;
 				case (Attribute::String):
 					SG_SERROR("String parsing not implemented.\n")
 				}

diff --git a/src/shogun/io/ARFFFile.h b/src/shogun/io/ARFFFile.h
@@ -329,7 +329,7 @@ namespace shogun
 				    "have the right permissions to open it.\n",
 				    filename.c_str())
 			}
-			m_stream = static_cast<std::istream*>(file_stream);
+			m_stream = std::unique_ptr<std::istream>(static_cast<std::istream*>(file_stream));
 		}
 
 		/**
@@ -341,12 +341,7 @@ namespace shogun
 		 */
 		explicit ARFFDeserializer(std::istream stream)
 		{
-			m_stream = &stream;
-		}
-
-		~ARFFDeserializer()
-		{
-			delete m_stream;
+			m_stream = std::unique_ptr<std::istream>(&stream);
 		}
 
 		/**
@@ -449,7 +444,7 @@ namespace shogun
 		/** the string after m_relation_string*/
 		std::string m_relation;
 		/** the input stream */
-		std::istream* m_stream;
+		std::unique_ptr<std::istream> m_stream;
 		/** the string where comments are stored */
 		std::vector<std::string> m_comments;
 		/** the string representing the current line being parsed */