Missed cnl header parser added

include/cnl_header_reader.hpp

@@ -0,0 +1,177 @@
+//! \brief CNL format header reader
+//! ATTENTION: this file is shared for both gecmi and onmi
+//!
+//! \license Apache License, Version 2.0: http://www.apache.org/licenses/LICENSE-2.0.html
+//! > 	Simple explanation: https://tldrlegal.com/license/apache-license-2.0-(apache-2.0)
+//!
+//! Copyright (c)
+//! \authr Artem Lutov
+//! \email luart@ya.ru
+//! \date 2017-02-10
+
+#ifndef CNL_HEADER_READER
+#define CNL_HEADER_READER
+
+#include <sstream>  // istringstream
+#include <string>
+#include <cstring>  // strtok
+#include <cmath>  // sqrt
+
+#ifdef __unix__
+#include <sys/stat.h>
+#endif // __unix__
+
+
+using std::string;
+using std::istream;
+using std::istringstream;
+
+//! \brief  Parse the header of CNL file and validate the results
+//! \post clsnum <= ndsnum if ndsnum > 0. 0 means not specified
+//!
+//! \param fsm istream&  - the reading file
+//! \param line string&  - processing line (string, header) being read from the file
+//! \param[out] clsnum size_t&  - resulting number of clusters if specified, 0 in case of parsing errors
+//! \param[out] ndsnum size_t&  - resulting number of nodes if specified, 0 in case of parsing errors
+//! \return void
+void parseHeader(istream& fsm, string& line, size_t& clsnum, size_t& ndsnum) {
+	// Process the header, which is a special initial comment
+	// The target header is:  # Clusters: <cls_num>[,] Nodes: <cls_num>
+	const string  clsmark = "clusters";
+	const string  ndsmark = "nodes";
+	while(getline(fsm, line)) {
+		// Skip empty lines
+		if(line.empty())
+			continue;
+		// Consider only subsequent comments
+		if(line[0] != '#')
+			break;
+
+		// 1. Replace the staring comment mark '#' with space to allow "#clusters:"
+		// 2. Replace ':' with space to allow "Clusters:<clsnum>"
+		for(size_t pos = 0; pos != string::npos; pos = line.find(':', pos + 1))
+			line[pos] = ' ';
+
+		istringstream fields(line);
+		string field;
+
+		// Read clusters specification
+		fields >> field;
+		if(field.length() != clsmark.length())
+			continue;
+		// Convert to lower case
+		for(size_t i = 0; i < field.length(); ++i)
+			field[i] = tolower(field[i]);
+		if(field != clsmark)
+			continue;
+		fields >> clsnum;
+
+		// Read nodes specification
+		fields >> field;
+		// Allow optional ','
+		if(!field.empty() && field[0] == ',')
+			fields >> field;
+		if(field.length() == ndsmark.length()) {
+			for(size_t i = 0; i < field.length(); ++i)
+				field[i] = tolower(field[i]);
+			if(field == ndsmark)
+				fields >> ndsnum;
+		}
+		// Validate and correct the number of clusters if required
+		// Note: it's better to reallocate a container a few times than too much overconsume the memory
+		if(ndsnum && clsnum > ndsnum) {
+			fprintf(stderr, "WARNING parseHeader(), clsnum (%lu) should not increase ndsnum (%lu)"
+				", fixed\n", clsnum, ndsnum);
+			clsnum = ndsnum;
+			//assert(0 && "parseHeader(), clsnum typically should be less than ndsnum");
+		}
+		// Get following line for the unified subsequent processing
+		getline(fsm, line);
+		break;
+	}
+}
+
+//! \brief Estimate zeroized values
+//!
+//! \param ndsnum size_t&  - the estimate number of nodes if 0, otherwise omit it
+//! \param clsnum size_t&  - the estimate number of clusters if 0, otherwise omit it
+//! \param cmsbytes size_t  - the number of bytes in the file
+//! \param membership=1.f float  - average membership of nodes in the clusters, > 0,
+//! 	=> 1 in case all nodes belong to any cluster and overlaps are present (> 1)
+//! \return bool  - the estimation is made
+bool estimateSizes(size_t& ndsnum, size_t& clsnum, size_t cmsbytes, float membership=1.f)
+{
+	if(membership <= 0) {
+		fprintf(stderr, "WARNING estimateSizes(), invalid membership = %G specified"
+			", reseted to 1\n", membership);
+		membership = 1;
+//		throw invalid_argument("estimateSizes(), membership = " + to_string(membership)
+//			+ " should be positive");
+	}
+
+	if(clsnum && !ndsnum) {
+		// Typically the number of nodes is at least square of the number of clusters
+		ndsnum = clsnum * clsnum / membership;
+		return true;
+	}
+
+	bool  estimated = false;
+	if(!ndsnum) {
+		// Estimate the number of nodes from the file size
+		if(!cmsbytes)
+			return false;
+
+		size_t  magn = 10;  // Decimal ids magnitude
+		unsigned  img = 1;  // Index of the magnitude (10^1)
+		size_t  reminder = cmsbytes % magn;  // Reminder in bytes
+		ndsnum = reminder / ++img;  //  img digits + 1 delimiter for each element
+		while(cmsbytes >= magn) {
+			magn *= 10;
+			ndsnum += (cmsbytes - reminder) % magn / ++img;
+			reminder = cmsbytes % magn;
+		}
+		ndsnum /= membership;
+		estimated = true;
+	}
+
+	// Usually the number of clusters does not increase square root of the number of nodes
+	if(!clsnum) {
+		clsnum = sqrt(ndsnum * membership) / membership + 1;  // Note: +1 to consider rounding down
+		estimated = true;
+	}
+
+	return estimated;
+}
+
+//! \brief Identify size of the input in bytes
+//!
+//! \param input istream&  - input stream
+//! \param fname=nullptr const char*  - file name of the corresponding input stream
+//! \return size_t  - resulting size of the input or 0 on errors
+size_t inputSize(istream& input, const char* fname=nullptr)
+{
+	size_t  inpsize = 0;
+#ifdef __unix__
+	if(fname) {
+		struct stat  filest;
+		if(!stat(fname, &filest))
+			inpsize = filest.st_size;
+	}
+	//fprintf(stderr, "# %s: %lu bytes\n", fname, inpsize);
+#endif // __unix
+	// Get length of the file
+	if(!inpsize) {
+		input.seekg(0, input.end);
+		inpsize = input.tellg();  // The number of bytes in the input communities
+		if(inpsize == size_t(-1)) {
+			//assert(!input && "inpsize = -1 should be only for the failed file operation");
+			fputs("WARNING inputSize(), file size evaluation failed\n", stderr);
+			inpsize = 0;
+		}
+		input.seekg(0, input.beg);
+	}
+
+	return inpsize;
+}
+
+#endif // CNL_HEADER_READER

onmi.cpp

@@ -16,10 +16,8 @@
  *     You should have received a copy of the GNU General Public License
  *     along with this program.  If not, see <http://www.gnu.org/licenses/>.
  */
-#include <getopt.h>
 #include <cmath>
 #include <iostream>
-#include <sstream>
 #include <fstream>
 #include <exception>
 #include <limits>
@@ -32,9 +30,7 @@
 #include <unordered_map>
 #include <unordered_set>
 
-#ifdef __unix__
-#include <sys/stat.h>
-#endif // __unix__
+#include <getopt.h>
 
 #include "cmdline.h"
 #include "cnl_header_reader.hpp"