unicoverage.c File Reference

unicoverage - Show the coverage of Unicode plane scripts for a GNU Unifont hex glyph file More...

#include <stdio.h>
#include <stdlib.h>
#include <string.h>

Include dependency graph for unicoverage.c:

Go to the source code of this file.

Macros
#define	MAXBUF   256
	Maximum input line length - 1.

Functions
int	main (int argc, char *argv[])
	The main function.
int	nextrange (FILE *coveragefp, unsigned *cstart, unsigned *cend, char *coverstring)
	Get next Unicode range.
void	print_subtotal (FILE *outfp, int print_n, int nglyphs, unsigned cstart, unsigned cend, char *coverstring)
	Print the subtotal for one Unicode script range.

Detailed Description

unicoverage - Show the coverage of Unicode plane scripts for a GNU Unifont hex glyph file

Author

Paul Hardy, unifoundry <at> unifoundry.com, 6 January 2008

Copyright

Copyright (C) 2008, 2013 Paul Hardy

Synopsis: unicoverage [-ifont_file.hex] [-ocoverage_file.txt]

This program requires the file "coverage.dat" to be present in the directory from which it is run.

Definition in file unicoverage.c.

Macro Definition Documentation

MAXBUF

#define MAXBUF   256

Maximum input line length - 1.

Definition at line 63 of file unicoverage.c.

Function Documentation

main()

int main	(	int	argc,
		char *	argv[]
	)

The main function.

Parameters

[in]	argc	The count of command line arguments.
[in]	argv	Pointer to array of command line arguments.

Returns

This program exits with status 0.

Definition at line 74 of file unicoverage.c.

{
 
   int      print_n=0;        /* print # of glyphs, not percentage   */
   unsigned i;                /* loop variable                       */
   unsigned slen;             /* string length of coverage file line */
   char     inbuf[256];       /* input buffer                        */
   unsigned thischar;         /* the current character               */
 
   char *infile="", *outfile="";  /* names of input and output files        */
   FILE *infp, *outfp;        /* file pointers of input and output files    */
   FILE *coveragefp;          /* file pointer to coverage.dat file          */
   unsigned cstart, cend;     /* current coverage start and end code points */
   char coverstring[MAXBUF];  /* description of current coverage range      */
   int nglyphs;               /* number of glyphs in this section           */
 
   /* to get next range & name of Unicode glyphs */
   int nextrange (FILE *coveragefp, unsigned *cstart, unsigned *cend,
                  char *coverstring);
 
   void print_subtotal (FILE *outfp, int print_n, int nglyphs,
                        unsigned cstart, unsigned cend, char *coverstring);
 
   if ((coveragefp = fopen ("coverage.dat", "r")) == NULL) {
      fprintf (stderr, "\nError: data file \"coverage.dat\" not found.\n\n");
      exit (0);
   }
 
   if (argc > 1) {
      for (i = 1; i < argc; i++) {
         if (argv[i][0] == '-') {  /* this is an option argument */
            switch (argv[i][1]) {
               case 'i':  /* name of input file */
                  infile = &argv[i][2];
                  break;
               case 'n':  /* print number of glyphs instead of percentage */
                  print_n = 1;
               case 'o':  /* name of output file */
                  outfile = &argv[i][2];
                  break;
               default:   /* if unrecognized option, print list and exit */
                  fprintf (stderr, "\nSyntax:\n\n");
                  fprintf (stderr, "   %s -p<Unicode_Page> ", argv[0]);
                  fprintf (stderr, "-i<Input_File> -o<Output_File> -w\n\n");
                  exit (1);
            }
         }
      }
   }
   /*
      Make sure we can open any I/O files that were specified before
      doing anything else.
   */
   if (strlen (infile) > 0) {
      if ((infp = fopen (infile, "r")) == NULL) {
         fprintf (stderr, "Error: can't open %s for input.\n", infile);
         exit (1);
      }
   }
   else {
      infp = stdin;
   }
   if (strlen (outfile) > 0) {
      if ((outfp = fopen (outfile, "w")) == NULL) {
         fprintf (stderr, "Error: can't open %s for output.\n", outfile);
         exit (1);
      }
   }
   else {
      outfp = stdout;
   }
 
   /*
      Print header row.
   */
   if (print_n) {
      fprintf (outfp, "# Glyphs      Range        Script\n");
      fprintf (outfp, "--------      -----        ------\n");
   }
   else {
      fprintf (outfp, "Covered      Range        Script\n");
      fprintf (outfp, "-------      -----        ------\n\n");
   }
 
   slen = nextrange (coveragefp, &cstart, &cend, coverstring);
   nglyphs = 0;
 
   /*
      Read in the glyphs in the file
   */
   while (slen != 0 && fgets (inbuf, MAXBUF-1, infp) != NULL) {
      sscanf (inbuf, "%x", &thischar);
 
      /* Read a character beyond end of current script. */
      while (cend < thischar && slen != 0) {
         print_subtotal (outfp, print_n, nglyphs, cstart, cend, coverstring);
 
         /* start new range total */
         slen = nextrange (coveragefp, &cstart, &cend, coverstring);
         nglyphs = 0;
      }
      nglyphs++;
   }
 
   print_subtotal (outfp, print_n, nglyphs, cstart, cend, coverstring);
 
   exit (0);
}

Here is the call graph for this function:

nextrange()

int nextrange	(	FILE *	coveragefp,
		unsigned *	cstart,
		unsigned *	cend,
		char *	coverstring
	)

Get next Unicode range.

This function reads the next Unicode script range to count its glyph coverage.

Parameters

[in]	coveragefp	File pointer to Unicode script range data file.
[in]	cstart	Starting code point in current Unicode script range.
[in]	cend	Ending code point in current Unicode script range.
[out]	coverstring	String containing <cstart>-<cend> substring.

Returns

Length of the last string read, or 0 for end of file.

Definition at line 196 of file unicoverage.c.

{
   int i;
   static char inbuf[MAXBUF];
   int retval;         /* the return value */
 
   retval = 0;
 
   do {
      if (fgets (inbuf, MAXBUF-1, coveragefp) != NULL) {
         retval = strlen (inbuf);
         if ((inbuf[0] >= '0' && inbuf[0] <= '9') ||
             (inbuf[0] >= 'A' && inbuf[0] <= 'F') ||
             (inbuf[0] >= 'a' && inbuf[0] <= 'f')) {
            sscanf (inbuf, "%x-%x", cstart, cend);
            i = 0;
            while (inbuf[i] != ' ') i++;  /* find first blank */
            while (inbuf[i] == ' ') i++;  /* find next non-blank */
            strncpy (coverstring, &inbuf[i], MAXBUF);
         }
         else retval = 0;
      }
      else retval = 0;
   } while (retval == 0 && !feof (coveragefp));
 
   return (retval);
}

Here is the caller graph for this function:

print_subtotal()

void print_subtotal	(	FILE *	outfp,
		int	print_n,
		int	nglyphs,
		unsigned	cstart,
		unsigned	cend,
		char *	coverstring
	)

Print the subtotal for one Unicode script range.

Parameters

[in]	outfp	Pointer to output file.
[in]	print_n	1 = print number of glyphs, 0 = print percentage.
[in]	nglyphs	Number of glyphs in current range.
[in]	cstart	Starting code point for current range.
[in]	cend	Ending code point for current range.
[in]	coverstring	Character string of "<cstart>-<cend>".

Definition at line 237 of file unicoverage.c.

                                                                        {
 
   /* print old range total */
   if (print_n) {  /* Print number of glyphs, not percentage */
      fprintf (outfp, " %6d ", nglyphs);
   }
   else {
      fprintf (outfp, " %5.1f%%", 100.0*nglyphs/(1+cend-cstart));
   }
 
   if (cend < 0x10000)
      fprintf (outfp, "  U+%04X..U+%04X   %s",
               cstart, cend, coverstring);
   else
      fprintf (outfp, " U+%05X..U+%05X  %s",
               cstart, cend, coverstring);
 
   return;
}

Here is the caller graph for this function: