diff --git a/modules/io/src/mol/CMakeLists.txt b/modules/io/src/mol/CMakeLists.txt
index 1201eb58fcff984f0bf0b5887a29f011ed6fd7ac..8551f74234504a85005e20e4822e565383f594a2 100644
--- a/modules/io/src/mol/CMakeLists.txt
+++ b/modules/io/src/mol/CMakeLists.txt
@@ -12,6 +12,7 @@ load_entity.cc
surface_io_msms_handler.cc
load_surface.cc
chemdict_parser.cc
+mmcif_parser.cc
io_profile.cc
dcd_io.cc
star_parser.cc
@@ -20,6 +21,7 @@ PARENT_SCOPE
set(OST_IO_MOL_HEADERS
chemdict_parser.hh
+mmcif_parser.hh
star_parser.hh
io_profile.hh
dcd_io.hh
diff --git a/modules/io/src/mol/star_parser.cc b/modules/io/src/mol/star_parser.cc
index 049b026a2f7826f9a4c49b4224942314bfc82045..d21d00278bacc559ad382fbd2246f37bff67c65e 100644
--- a/modules/io/src/mol/star_parser.cc
+++ b/modules/io/src/mol/star_parser.cc
@@ -27,10 +27,11 @@
namespace ost { namespace io {
-StarParser::StarParser(std::istream& stream):
- stream_(stream), line_num_(0), has_current_line_(false), current_line_()
+StarParser::StarParser(std::istream& stream, bool items_as_row):
+ stream_(stream), line_num_(0), has_current_line_(false), current_line_(),
+ items_row_header_(), items_row_columns_(), items_row_values_()
{
-
+ items_as_row_ = items_as_row;
}
bool StarParser::SplitLine(const StringRef& line,
@@ -200,6 +201,47 @@ void StarParser::ParseLoop()
}
}
+void StarParser::ParseLastDataItemRow()
+{
+ if (items_row_header_.GetCategory().size() > 0) {
+ if (this->OnBeginLoop(items_row_header_)) {
+ this->OnDataRow(items_row_header_, items_row_columns_);
+ this->OnEndLoop();
+ }
+ items_row_values_.clear();
+ items_row_columns_.clear();
+ items_row_header_.Clear();
+ }
+}
+
+void StarParser::ParseDataItemOrRow(StarDataItem& item)
+{
+ if (items_as_row_) {
+ // header
+ if (StringRef(items_row_header_.GetCategory().c_str(),
+ items_row_header_.GetCategory().size())!=item.GetCategory()) {
+ this->ParseLastDataItemRow();
+ // set category for new section
+ items_row_header_.SetCategory(item.GetCategory());
+ }
+
+ // row
+ items_row_header_.Add(item.GetName());
+ items_row_values_.push_back(item.GetValue().str());
+ items_row_columns_.push_back(StringRef(items_row_values_.back().data(),
+ items_row_values_.back().length()).trim());
+ } else {
+ this->OnDataItem(item);
+ }
+}
+
+void StarParser::ParseEndDataItemRow()
+{
+ if (items_as_row_) {
+ this->ParseLastDataItemRow();
+ }
+}
+
void StarParser::ParseDataItem()
{
StringRef line;
@@ -242,7 +284,7 @@ void StarParser::ParseDataItem()
StringRef value_ref=StringRef(value.data(),
value.length()).trim();
StarDataItem data_item(cat, name, value_ref);
- this->OnDataItem(data_item);
+ this->ParseDataItemOrRow(data_item);
} else {
if (nv.size()!=2) {
std::cout << "ERROR:" << line_num_ << ":" << line << std::endl;
@@ -253,10 +295,9 @@ void StarParser::ParseDataItem()
StringRef cat=nv[0].substr(1, i-nv[0].begin()-1);
StringRef name=nv[0].substr(i-nv[0].begin()+1);
StarDataItem data_item(cat, name, nv[1]);
- this->OnDataItem(data_item);
+ this->ParseDataItemOrRow(data_item);
this->ConsumeLine();
}
-
}
void StarParser::ParseData()
@@ -283,6 +324,7 @@ void StarParser::ParseData()
break;
case 'd':
if (tline.length()>=5 && StringRef("data_", 5)==tline.substr(0, 5)) {
+ this->ParseEndDataItemRow();
this->OnEndData();
return;
}
@@ -297,6 +339,7 @@ void StarParser::ParseData()
break;
case 'l':
if (tline==StringRef("loop_", 5)) {
+ this->ParseEndDataItemRow();
this->ParseLoop();
break;
}
@@ -310,6 +353,7 @@ void StarParser::ParseData()
return;
}
}
+ this->ParseEndDataItemRow();
this->OnEndData();
}
diff --git a/modules/io/src/mol/star_parser.hh b/modules/io/src/mol/star_parser.hh
index 272a11898b1a9dc8197b0ab596a3d9bcb7068362..47aa98b0eb3fff87af2844a611e338b400b4570f 100644
--- a/modules/io/src/mol/star_parser.hh
+++ b/modules/io/src/mol/star_parser.hh
@@ -71,6 +71,12 @@ public:
{
return index_map_.size();
}
+ void Clear()
+ {
+ category_.clear();
+ index_map_.clear();
+ }
+
const String& GetCategory() const { return category_; }
private:
String category_;
@@ -95,8 +101,14 @@ private:
/// a list of column names and values.
class DLLEXPORT_OST_IO StarParser {
public:
- StarParser(std::istream& istream);
-
+ /// \brief create a StarParser
+ ///
+ /// \param stream input stream
+ /// \param item_as_row if true, data-items are first gathered (identifier as
+ /// header, values as row) and then parsed like a loop
+ /// (OnBeginLoop(), OnDataRow(), OnEndLoop())
+ explicit StarParser(std::istream& stream, bool items_as_row=false);
+
virtual ~StarParser() { }
// callback interface
public:
@@ -108,7 +120,7 @@ public:
/// \brief invoked when leaving a loop
///
/// OnEndLoop() is only invoked, when OnBeginLoop() returned true.
- virtual void OnEndLoop() { }
+ virtual void OnEndLoop() { }
/// \brief invoked when a data row in a loop is encountered.
/// \param header describes the row format
/// \param columns contains the data columns
@@ -135,6 +147,14 @@ public:
std::vector<StringRef>& parts, bool clear=true);
private:
void ParseLoop();
+ /// \brief Calls the loop parsing functions on the last data item fetched to
+ /// be read as loop
+ void ParseLastDataItemRow();
+ /// \brief Calls functions for parsing a data item, either as singleton
+ /// (default) or loop.
+ void ParseDataItemOrRow(StarDataItem& item);
+ /// \brief If enabled, calls ParseLastDataItemRow()
+ void ParseEndDataItemRow();
private:
/// \brief read next line, replacing the current line
bool NextLine(StringRef& str)
@@ -171,6 +191,10 @@ private:
int line_num_;
bool has_current_line_;
String current_line_;
+ bool items_as_row_;
+ StarLoopDesc items_row_header_;
+ std::vector<StringRef> items_row_columns_;
+ std::vector<String> items_row_values_;
};
}}
diff --git a/modules/io/tests/CMakeLists.txt b/modules/io/tests/CMakeLists.txt
index abbd589147f64663b2ab685d7deab0ef8ffa8679..dc2c8bfdfc87a8a9780da0749d4648552b3678cb 100644
--- a/modules/io/tests/CMakeLists.txt
+++ b/modules/io/tests/CMakeLists.txt
@@ -9,6 +9,7 @@ set(OST_IO_UNIT_TESTS
test_iomanager.cc
tests.cc
test_star_parser.cc
+ test_mmcif_parser.cc
)
ost_unittest(MODULE io
diff --git a/modules/io/tests/test_star_parser.cc b/modules/io/tests/test_star_parser.cc
index 0b31e3dd829ddcdd3298731ecaa6b4b1215b4a2f..fc58a63b62265264e3c06d27abd15524cfae3f1e 100644
--- a/modules/io/tests/test_star_parser.cc
+++ b/modules/io/tests/test_star_parser.cc
@@ -20,6 +20,7 @@
Author: Marco Biasini
*/
#include <fstream>
+#include <math.h>
#include <ost/mol/mol.hh>
#include <ost/io/io_exception.hh>
#include <ost/io/mol/star_parser.hh>
@@ -52,6 +53,16 @@ public:
s4=item.GetValue().str();
}
}
+ virtual bool OnBeginLoop(const StarLoopDesc& header)
+ {
+ return true;
+ }
+ virtual void OnDataRow(const StarLoopDesc& header,
+ const std::vector<StringRef>& columns)
+ {
+ BOOST_CHECK_EQUAL(header.GetCategory(), "loop");
+ BOOST_CHECK_EQUAL(columns.size(), size_t(4));
+ }
String s1;
String s2;
String s3;
@@ -66,7 +77,7 @@ public:
{
BOOST_CHECK_EQUAL(header.GetCategory(), "loop");
return true;
- }
+ }
virtual void OnDataRow(const StarLoopDesc& header,
const std::vector<StringRef>& columns)
{
@@ -103,6 +114,90 @@ public:
bool visit_two;
};
+class ItemsAsRowTestParser : public StarParser {
+public:
+ ItemsAsRowTestParser(std::istream& stream): StarParser(stream, true),
+ category("")
+ { }
+
+ virtual bool OnBeginLoop(const StarLoopDesc& header)
+ {
+ if ((header.GetCategory() == "data-item1") ||
+ (header.GetCategory() == "data-item2") ||
+ (header.GetCategory() == "data-item3") ||
+ (header.GetCategory() == "data-item4")) {
+ return true;
+ }
+ return false;
+ }
+
+ virtual void OnDataRow(const StarLoopDesc& header,
+ const std::vector<StringRef>& columns)
+ {
+ BOOST_CHECK_EQUAL(columns.size(), size_t(4));
+ category = header.GetCategory();
+ if (header.GetCategory() == "data-item1") {
+ s1 = columns[header.GetIndex("s1")].str();
+ s2 = columns[header.GetIndex("s2")].str();
+ s3 = columns[header.GetIndex("s3")].str();
+ s4 = columns[header.GetIndex("s4")].str();
+ return;
+ }
+ if (header.GetCategory() == "data-item2") {
+ i1 = columns[header.GetIndex("i1")].to_int().second;
+ i2 = columns[header.GetIndex("i2")].to_int().second;
+ i3 = columns[header.GetIndex("i3")].to_int().second;
+ i4 = columns[header.GetIndex("i4")].to_int().second;
+ return;
+ }
+ if (header.GetCategory() == "data-item3") {
+ f1 = columns[header.GetIndex("f1")].to_float().second;
+ f2 = columns[header.GetIndex("f2")].to_float().second;
+ f3 = columns[header.GetIndex("f3")].to_float().second;
+ f4 = columns[header.GetIndex("f4")].to_float().second;
+ return;
+ }
+ if (header.GetCategory() == "data-item4") {
+ s5 = columns[header.GetIndex("s1")].str();
+ s6 = columns[header.GetIndex("s2")].str();
+ s7 = columns[header.GetIndex("s3")].str();
+ s8 = columns[header.GetIndex("s4")].str();
+ return;
+ }
+ }
+
+ void OnEndLoop()
+ {
+ if (category == "data-item3") {
+ f1 = ceil(f1*2);
+ f2 = ceil(f2*2);
+ f3 = ceil(f3*2);
+ f4 = ceil(f4*2);
+ }
+ }
+
+ String category;
+
+ String s1;
+ String s2;
+ String s3;
+ String s4;
+ String s5;
+ String s6;
+ String s7;
+ String s8;
+
+ int i1;
+ int i2;
+ int i3;
+ int i4;
+
+ float f1;
+ float f2;
+ float f3;
+ float f4;
+};
+
BOOST_AUTO_TEST_SUITE( io );
BOOST_AUTO_TEST_CASE(star_split1)
@@ -182,6 +277,30 @@ BOOST_AUTO_TEST_CASE(star_loop)
BOOST_CHECK_EQUAL(star_p.lines[4][2], "15");
}
+BOOST_AUTO_TEST_CASE(star_items_as_row)
+{
+ std::ifstream s("testfiles/items-as-row.cif");
+ ItemsAsRowTestParser star_p(s);
+ star_p.Parse();
+ BOOST_CHECK_EQUAL(star_p.s1, "a");
+ BOOST_CHECK_EQUAL(star_p.s2, "a b c");
+ BOOST_CHECK_EQUAL(star_p.s3, "a\nb\nc");
+ BOOST_CHECK_EQUAL(star_p.s4, "a'b");
+ BOOST_CHECK_EQUAL(star_p.i1, 1);
+ BOOST_CHECK_EQUAL(star_p.i2, 2);
+ BOOST_CHECK_EQUAL(star_p.i3, 3);
+ BOOST_CHECK_EQUAL(star_p.i4, 4);
+ BOOST_CHECK_EQUAL(ceil(star_p.f1), ceil(2.22));
+ BOOST_CHECK_EQUAL(ceil(star_p.f2), ceil(2.44));
+ BOOST_CHECK_EQUAL(ceil(star_p.f3), ceil(2.66));
+ BOOST_CHECK_EQUAL(ceil(star_p.f4), ceil(2.88));
+
+ BOOST_CHECK_EQUAL(star_p.s5, "1.11");
+ BOOST_CHECK_EQUAL(star_p.s6, "1.22");
+ BOOST_CHECK_EQUAL(star_p.s7, "1.33");
+ BOOST_CHECK_EQUAL(star_p.s8, "1.44");
+}
+
BOOST_AUTO_TEST_CASE(star_missing_data)
{
std::ifstream s("testfiles/missing_data.cif");
diff --git a/modules/io/tests/testfiles/items-as-row.cif b/modules/io/tests/testfiles/items-as-row.cif
new file mode 100644
index 0000000000000000000000000000000000000000..db06b3565b3fbd4a5d4aadf225c2ade79008c6ff
--- /dev/null
+++ b/modules/io/tests/testfiles/items-as-row.cif
@@ -0,0 +1,52 @@
+data_items-as-row
+
+_data-item1.s1 a
+_data-item1.s2 'a b c'
+_data-item1.s3
+;
+a
+b
+c
+;
+_data-item1.s4 'a'b'
+_data-item2.i1 1
+_data-item2.i2 '2'
+_data-item2.i3
+;
+3
+;
+_data-item2.i4
+4
+
+data_more-items-as-row
+
+_data-item3.f1 1.11
+_data-item3.f2 1.22
+_data-item3.f3
+;
+1.33
+;
+_data-item3.f4 1.44
+
+loop_
+_loop.s1
+_loop.s2
+_loop.s3
+_loop.s4
+1 2 3 4
+
+loop_
+_loop.s1
+_loop.s2
+_loop.s3
+_loop.s4
+1 2 3 6
+4 5 6 8
+
+_data-item4.s1 1.11
+_data-item4.s2 1.22
+_data-item4.s3
+;
+1.33
+;
+_data-item4.s4 1.44
diff --git a/modules/io/tests/testfiles/missing_data.cif b/modules/io/tests/testfiles/missing_data.cif
new file mode 100644
index 0000000000000000000000000000000000000000..a239bd0ef9f8dd9fd79b8280b02323f18a5c0be7
--- /dev/null
+++ b/modules/io/tests/testfiles/missing_data.cif
@@ -0,0 +1,4 @@
+this file has no 'data_' tag.
+
+data_
+