A very basic obicount

src/obicolumn.h

@@ -1,47 +0,0 @@
-/****************************************************************************
- * OBIColumn header file                                                    *
- ****************************************************************************/
-
-/**
- * @file obicolumn.h
- * @author Celine Mercier
- * @date 12 May 2015
- * @brief Header file for the shared elements of all the OBIColumn structures.
- */
-
-#ifndef OBICOLUMN_H_
-#define OBICOLUMN_H_
-
-#include <stdio.h>
-
-
-/**
- * @brief enum OBIDataType for the data type of the OBIColumns.
- */
-
-typedef enum OBIDataType {
-	OBI_VOID    = 0,     		/**< data type not specified */
-	OBI_INT32,					/**< type int32_t */
-	OBI_INT64,					/**< type int64_t */
-	OBI_UINT32,					/**< type uint32_t */
-	OBI_UNIT64,					/**< type uint64_t */
-	OBI_STRING					/**< type char* */
-} OBIDataType_t, *OBIDataType_p;
-
-
-/**
- * @brief OBIColumnHeader structure.
- */
-
-typedef struct OBIColumnHeader {
-	bool			endyan_byte_order;		/**< endyan byte order */
-	int				header_size_value;		/**< size of the header: a multiple of PAGESIZE */
-	int				line_count;				/**< number of lines of data */
-	OBIDataType_t	data_type;				/**< type of the data */
-	char*			creation_date;			/**< date of creation of the file */
-	int				version;				/**< version of the OBIColumn */
-	char*			comments;				/**< comments */>
-} OBIIntColumn_t, *OBIIntColumn_p;
-
-
-#endif /* OBICOLUMN_H_ */

src/obicount/Makefile

@@ -0,0 +1,18 @@
+CC = gcc
+CFLAGS = -c -Wall
+LDFLAGS = 
+SOURCES = obicount.c ../obidmscolumn.c
+OBJECTS = $(SOURCES:.c=.o)
+EXECUTABLE = obicount
+
+all: $(SOURCES) $(EXECUTABLE)
+    
+$(EXECUTABLE): $(OBJECTS) 
+	$(CC) $(LDFLAGS) $(OBJECTS) -o $@
+
+.c.o:
+	$(CC) $(CFLAGS) $< -o $@
+
+clean:
+	rm *o
+	rm $(EXECUTABLE)

src/obicount/obicount.c

@@ -0,0 +1,87 @@
+
+#include <stdio.h>
+#include <stdlib.h>
+#include <sys/mman.h>	/* mmap() is defined in this header */
+#include <stdint.h>
+#include <fcntl.h>
+#include <sys/stat.h>
+
+#include "../obitypes.h"
+#include "../obidmscolumn.h"
+
+
+/**
+ * @brief Computes the size to map.
+ *
+ * * @param OBIDMSColumn_file The file to map.
+ * @return The size to map.
+ */
+int get_size_to_map(int OBIDMSColumn_file)
+// compute size to map : file size minus size of the header
+{
+	int size;
+	struct stat s;
+
+	fstat(OBIDMSColumn_file, &s);
+	size = (s.st_size) - HEADER_SIZE;
+	return(size);
+}
+
+/**
+ * @brief Computes and prints the total number of sequences by summing their counts.
+ *
+ * * @param The count file.
+ */
+int main(int argc, char const *argv[])
+{
+	char* map;
+	int size;
+	int OBIDMSColumn_file;
+	int count;
+	char c;
+	char num_str[10] = "";
+	int num_int;
+	int i,j;
+
+	// initialize variables
+	OBIDMSColumn_file = open(argv[1], O_RDONLY);	//read only
+	count = 0;
+	j = 0;
+
+	// compute size to map
+	size = get_size_to_map(OBIDMSColumn_file);
+
+	// map the data
+	map = obi_map_read_only(OBIDMSColumn_file, HEADER_SIZE, size);
+
+	// sum the counts
+	for (i=0; i<size; i++)
+	{
+		c = map[i];
+
+		if (c != SEPARATOR)		// reading lines
+		{
+			num_str[j] = c;
+			j++;
+		}
+
+		else if (c == SEPARATOR)	// end of a line
+		{
+			num_int = atoi(num_str);	// turn number from character string to int
+			count = count + num_int;	// add the number to the sum
+			j = 0;
+			num_str[j] = '\0';
+		}
+	}
+
+	// print the final count of sequences
+	fprintf(stderr, "Sequence count = %d\n", count);
+
+	// unmap
+	obi_unmap(size);
+
+	// close file
+	close(OBIDMSColumn_file);
+
+	return(0);
+}

src/obidmscolumn.c

@@ -0,0 +1,46 @@
+/****************************************************************************
+ * OBIDMSColumn functions                                                   *
+ ****************************************************************************/
+
+/**
+ * @file obidmscolumn.h
+ * @author Celine Mercier
+ * @date 22 May 2015
+ * @brief Functions for the shared elements of all the OBIColumn structures.
+ */
+
+
+#include <stdio.h>
+#include <stdlib.h>
+#include <sys/mman.h>	/* mmap() is defined in this header */
+
+
+/**
+ * @brief Reads a memory block from a file.
+ *
+ * * @param file The file that should be mapped.
+ * * @param size The number of bytes that should be mapped.
+ * * @param start The position at which the mapping should start.
+ * @return A pointer on the first element of the block.
+ */
+char* obi_map_read_only(int file, int start, int size)
+{
+	char* map;
+	map = (char*) mmap(0, size, PROT_READ, MAP_PRIVATE, file, start);
+	if (map == MAP_FAILED)
+	{
+		fprintf(stderr, "\nmmap read-only error\n");
+		exit(1);
+	}
+	return(map);
+}
+
+/**
+ * @brief Unmaps a memory block.
+ *
+ * * @param size The size of the block that should be unmapped.
+ */
+void obi_unmap(int size)
+{
+	munmap(0, size);
+}

src/obidmscolumn.h

@@ -0,0 +1,63 @@
+/****************************************************************************
+ * OBIDMSColumn header file                                                    *
+ ****************************************************************************/
+
+/**
+ * @file obidmscolumn.h
+ * @author Celine Mercier
+ * @date 12 May 2015
+ * @brief Header file for the shared elements of all the OBIColumn structures.
+ */
+
+#ifndef OBIDMSCOLUMN_H_
+#define OBIDMSCOLUMN_H_
+
+#include <stdio.h>
+#include <sys/types.h>
+#include <unistd.h>
+#include <stdbool.h>
+
+#include "obitypes.h"
+
+/**
+ * @brief Value separator in OBIDMSColumn files.
+ */
+static const char SEPARATOR = '\n';
+
+
+/**
+ * @brief Header size in OBIDMSColumn files.
+ */
+static const int HEADER_SIZE = 4096;
+
+
+/**
+ * @brief Number of bytes to map
+ */
+static const int BUFFER_SIZE = 4;
+
+
+/**
+ * @brief OBIColumnHeader structure.
+ */
+
+typedef struct OBIDMSColumnHeader {
+	bool			endyan_byte_order;		/**< endyan byte order */
+	int				header_size_value;		/**< size of the header: a multiple of PAGESIZE */
+	bool			file_status;			/**< file status : 0 is Open and 1 is Closed */
+	pid_t			PID;					/**< PID of the process that created the file and is
+												 the only one allowed to modify it if it is open */
+	int				line_count;				/**< number of lines of data */
+	OBIType_t		data_type;				/**< type of the data */
+	char*			creation_date;			/**< date of creation of the file */
+	int				version;				/**< version of the OBIColumn */
+	char*			comments;				/**< comments */
+} OBIDMSColumnHeader_t, *OBIDMSColumnHeader_p;
+
+
+char* obi_map_read_only(int file, int start, int size);
+
+void obi_unmap(int size);
+
+
+#endif /* OBIDMSCOLUMN_H_ */

src/obiintcolumn.h

@@ -1,81 +0,0 @@
-/****************************************************************************
- * OBIIntColumn header file                                                 *
- ****************************************************************************/
- 
-/**
- * @file obiintcolumn.h
- * @author Celine Mercier
- * @date 11 May 2015
- * @brief Header file for the OBIIntColumn structure.
- */
-
-#ifndef OBIINTCOLUMN_H_
-#define OBIINTCOLUMN_H_
- 
-#include <stdio.h>
-#include <stdlib.h>
-
-#include "obicolumn.h"
-
-
-/**
- * @brief OBIIntColumn stucture.
- */
-
-typedef struct OBIIntcolumn {
-
-} OBIIntColumn_t, *OBIIntColumn_p;
-
-/**
- * @brief Reads a memory block from an OBIIntColumn file.
- *
- * @param OBIIntColumn The OBIColumn that should be read.
- * @return A pointer on the first element of the block.
- */
-
-int* readMemoryBlock(int* OBIIntcolumn, int startingPosition, int memoryBlockSize);
-
-
-/**
- * @brief Writes a memory block line by line at the end of a file.
- */
-
-void writeMemoryBlock(int* OBIIntcolumn, int memoryBlockSize, char* fileName);
-
-
-/**
- * @brief Returns one line of an OBIColumn.
- */
-
-int readLine(int line);
-
-
-/**
- * @brief Creates a new file for a new version of the OBIColumn.
- */
-
-void newColumn(OBIIntcolumn* column);
-
-
-/**
- * @brief malloc for OBIIntColumn.
- */
-
-int* OBIIntMalloc();
-
-
-/**
- * @brief realloc for OBIIntColumn.
- */
-
-int* OBIIntRealloc();
-
-
-/**
- * @brief free for OBIIntColumn.
- */
-
-void OBIIntFree();
-
-
-#endif /* OBIINTCOLUMN_H_ */

src/obitypes.h

@@ -0,0 +1,50 @@
+/****************************************************************************
+ * OBITypes header file                                                     *
+ ****************************************************************************/
+
+/**
+ * @file obitypes.h
+ * @author Celine Mercier
+ * @date 18 May 2015
+ * @brief Header file for the OBITypes.
+ */
+
+#ifndef OBITYPES_H_
+#define OBITYPES_H_
+
+#include <stdio.h>
+
+#define OBIInt_NA (INT32_MIN)
+#define OBIFloat_NA (NAN)
+#define OBIIdx_NA (SIZE_MAX)
+
+/**
+ * @brief enum for the boolean OBIType.
+ */
+
+typedef enum {
+    FALSE      = 0,
+    TRUE       = 1,
+    OBIBool_NA = 2
+} OBIBool_t, *OBIBool_p; 		/**< a boolean true/false value */
+
+
+/**
+ * @brief enum OBITypes for the data type of the OBIColumns.
+ */
+
+typedef enum OBIType {
+	OBI_VOID    = 0,     		/**< data type not specified */
+	OBIInt_t,					/**< a signed integer value (C type : int32_t) */
+	OBIFloat_t,					/**< a floating value (C type : double) */
+	OBIChar_t,					/**< a character (C type : char) */
+	OBIIdx_t					/**< an index in a data structure (C type : size_t) */
+} OBIType_t, *OBIType_p;
+
+typedef int32_t obiint_t;
+typedef double obifloat_t;
+typedef char obichar_t;
+typedef size_t obiidx_t;
+
+
+#endif /* OBITYPES_H_ */