#include

#include

#include

#include

#include

#include

#include

#include "amap.h$"$

#include "aux.h$"$

$/*$ scanner.c $-$ process directories$/$files$,$ counting words in each

$*$ regular text file. Accumulate counts in the amap$\_$t$,$ then pass them

$*$ along to reducers via the reducer pipes.

$*$ Each reducer collects counts for a portion of the alphabet.

$*$ Which reducer pipe a $($word$,$count$)$ pair should be written to is

$*$ determined by the first letter of word, using the "whichpipe" global.

$*$ The definition of a "word" is a contiguous sequence of alphabetic

$*$ characters $-$ no numbers, whitespace, or punctuation.

$*/$

#define FTYPE$\_$REG $1$

#define FTYPE$\_$DIR $2$

#define FTYPE$\_$SKIP $3$

#define SAMPLESIZE $16$

#define NAMESIZE $1024$

$/*$ Map of characters to reducer pipe #s$.$ Init'ed in driver once for all. $*/$

extern int whichpipe$[$ALPHABETLEN$]$;

$/*$ what kind of file is this? Return value indicates type via constants above.

$*$ If it's a text file to be processed, the file descriptor is passed via

$*$ the $"$fd$"$ out parameter.

$*/$

static int check$\_$type$($char $*$name$,$ int $*$fd$)\{$

struct stat finfo;

int rv$,$ fildes;

char buf$[$SAMPLESIZE$]$;

rv $=$ stat$($name$,$&finfo$)$;

if $($rv $<0)\{$

return FTYPE$\_$SKIP;

$\}$

if $(($finfo$.$st$\_$mode & S$\_$IFMT$)==$ S$\_$IFREG$)\{$

fildes $=$ open$($name$,$O$\_$RDONLY$)$;

if $($fildes $<0)$

return FTYPE$\_$SKIP;

$/*$ Read a few characters to try to get type $*/$

if $(($rv$=$read$($fildes$,$buf,SAMPLESIZE$))<0)$

return FTYPE$\_$SKIP;

if $(($buf$[0]==0$x$7$f &&

buf$[1]==\text{'}$E$\text{'}$ &&

buf$[2]==\text{'}$L$\text{'}$ &&

buf$[3]==\text{'}$F$\text{'})//$ executable

$||$

$($buf$[0]==\text{'}$G$\text{'}$ &&

buf$[1]==\text{'}$I$\text{'}$ &&

buf$[2]==\text{'}$F$\text{'})//$ image

$||$

$($buf$[0]==0$x$89$ &&

buf$[1]==\text{'}$P$\text{'}$ &&

buf$[2]==\text{'}$N$\text{'}$ &&

buf$[3]==\text{'}$G$\text{'})//$ image

$||$

$($buf$[0]==\text{'}\%\text{'}$ &&

buf$[1]==\text{'}$P$\text{'}$ &&

buf$[2]==\text{'}$D$\text{'}$ &&

buf$[3]==\text{'}$F$\text{'}))\{//$ PDF

close$($fildes$)$;

return FTYPE$\_$SKIP;

$\}$

for $($int i$=0$; i