clear
set mem 500m

/*
File:	cepr_basic_read_all.do
Date:	Nov 21, 2006
Desc:	Retrieves raw CPS Basic data and adds data labels
*/

	/*Windows vs. GNU/Linux*/
global gnulin = 1 /*Set gnulin=0 if you run Windows; 1 if GNU/Linux*/

if $gnulin==1 {
	/*directories*/
global locbdo "/ceprdata/CPS_Basic/CEPR/DoFiles"
global locbas "/ceprdata/CPS_Basic/NBER"
	/*executables*/
global gzip "/usr/bin/gzip"
global copy "/bin/cp"
global erase "/bin/rm"
}

if $gnulin==0 {
	/*directories*/
global locbdo "F:\DATA\CPS_Basic\CEPR\DoFiles"
global locbas "F:\DATA\CPS_Basic\NBER"
	/*executables*/
global gzip "F:\Software\Gzip\gzip.exe"
global copy "copy"
global erase "erase"
}

/* 2003 */
capture program drop b2003
program define b2003
local i=1
while `i'<=12 	{
cd $locbas
cd 2003
!$gzip -d cpsb_2003_`i'.txt.gz
!$copy cpsb_2003_`i'.txt cpsb2003.txt
cd $locbdo
do cps_basic_read_2003.do
gen int year=2003
compress
cd $locbas
cd 2003
save cps_basic_raw_2003_`i'.dta, replace
!$gzip -vN cpsb_2003_`i'.txt /* recompress */
!$erase cpsb2003.txt
local i=`i'+1
		} /* loop through months */
end

	/* 	from May 2004, the CPS changed coding of sub-state
		geographic areas; the change has no effect on the
		current version of the CEPR CPS ORG and BASIC CPS
		extracts, which do not provide consistent versions
		of sub-state geographic areas; in all other respects,
		the raw CPS data were unchanged, allowing continued
		use of Jan 2003 coding for data from May 2004
	*/

/* 2004 */
capture program drop b2004
program define b2004
local i=1
while `i'<=12 {
cd $locbas
cd 2004
!$gzip -d cpsb_2004_`i'.txt.gz
/* use Jan 2003 coding, even after May 2004: */
/* 2004 uses Jan 2003 coding */
if $gnulin==1 {
	!$copy cpsb_2004_`i'.txt "$locbas/2003/cpsb2003.txt"
}
if $gnulin==0 {
	!$copy cpsb_2004_`i'.txt "$locbas\2003\cpsb2003.txt"
}
cd $locbdo
do cps_basic_read_2003.do /* uses Jan 2003 coding */
gen int year=2004
compress
cd $locbas
cd 2004
save cps_basic_raw_2004_`i'.dta, replace
!$gzip -vN cpsb_2004_`i'.txt /* recompress */
cd $locbas
cd 2003
!$erase cpsb2003.txt  /* 2004 uses Jan 2003 coding */
local i=`i'+1
	} /* loop through months */
end

	/* 	in August 2005, BLS/Census introduced a new
		data dictionary 
	*/

/* 2005 */
capture program drop b2005
program define b2005
local i=1
while `i'<=12 {
cd $locbas
cd 2005
	if `i'<=7 	{ /* Jan-Jul 2005 */
!$gzip -d cpsb_2005_`i'.txt.gz
/* use Jan 2003 coding, even after May 2004: */
/* 2005 uses Jan 2003 coding */
if $gnulin==1 {
!$copy cpsb_2005_`i'.txt "$locbas/2003/cpsb2003.txt"
}
if $gnulin==0 {
!$copy cpsb_2005_`i'.txt "$locbas\2003\cpsb2003.txt"
}
cd $locbdo
do cps_basic_read_2003.do /* uses Jan 2003 coding */
gen int year=2005
cd $locbas
cd 2005
save cps_basic_raw_2005_`i'.dta, replace
!$gzip -vN cpsb_2005_`i'.txt /* recompress */
cd $locbas
cd 2003
!$erase cpsb2003.txt
			} /* Jan-Jul 2005 */
	if 8<=`i' & `i'<=12 { /* Aug-Sep 2005 */
cd $locbas
cd 2005
!$gzip -d cpsb_2005_`i'.txt.gz
!$copy cpsb_2005_`i'.txt cpsb2005.txt
cd $locbdo
do cps_basic_read_2005.do /* from Aug 2005 */
gen int year=2005
compress
cd $locbas
cd 2005
save cps_basic_raw_2005_`i'.dta, replace
!$gzip -vN cpsb_2005_`i'.txt /* recompress */
!$erase cpsb2005.txt
			} /* Aug-Sep 2005 */
local i=`i'+1
	} /* loop through months */
 
end

/* 2006 */
capture program drop b2006
program define b2006
local i=1
while `i'<=12 {
	/* Jan-Dec 2006 */
	if 1<=`i' & `i'<=12 { /* Jan-Dec 2006 */
		cd $locbas
		cd 2006
		!$gzip -d cpsb_2006_`i'.txt.gz
		/* use Aug 2005 coding for Jan-Dec 2006*/
		if $gnulin==1 {
			!$copy cpsb_2006_`i'.txt "$locbas/2005/cpsb2005.txt"
		}
		if $gnulin==0 {
			!$copy cpsb_2006_`i'.txt "$locbas\2005\cpsb2005.txt"
		}
		cd $locbdo
		do cps_basic_read_2005.do /* from Aug 2005 */
		gen int year=2006
		compress
		cd $locbas
		cd 2006
		save cps_basic_raw_2006_`i'.dta, replace
		!$gzip -vN cpsb_2006_`i'.txt /* recompress */
		cd $locbas
		cd 2005
		!$erase cpsb2005.txt
	} /* Jan-Dec 2006 */
local i=`i'+1
} /* loop through months */
 
end