Database conversion

A range of UNIX scripts was used to convert the input database formats from the individual projects into the normalised database form, and to join the attributes into one database using the orthography as a join attribute.

Perhaps the most interesting script is the one designed to convert the C programme of B1 into database format, which is shown below.

Code:

#!/bin/sh
# makeb1
# D. Gibbon 6.12.1994

grep -v "^\ *$" b1.wl |

gawk 'NR > 30 {print $0}' |

grep -v "\/\*" |

grep -v "[\{\}]" |
grep -v "short" |
grep -v "long" |
grep -v "wnr" |
grep -v "wanr" |

sed -e "s/insert_sim_value(simval=crea_simval(),/value/g" |
sed -e "s/);//g" |
sed -e "s/crea_sem(wnr);//g" |
sed -e "s/insert_ent_//g" |
sed -e "s/(ent, simval//g" |
sed -e "s/wortart(ent,/wortart/g" |

sed -e "s/(void)//g" |
sed -e "s/create_lexikon_/create_lexikon /g" |

gawk '
$1 == "create_lexikon" {word = $2}
$1 == "value" {value = $2}
$1 == "wortart" {print word, "wortart_" $2}
$1 != "create_lexikon" && $1 != "value" &&
      $1 != "wortart" {print word, $1 "_" value}
' |

sort -u > b1.att.val

gawk '{print $1}' b1.att.val | sort -u > b1.orth.att
gawk '{print $2}' b1.att.val > b1.orth.val

# check for duplicate keywords and create disjunctions

gawk ' { if ($1 == previous_1 && $2 != previous_2 )
	 	{ printf(";%s",$2) }
	else	{ previous_1 = $1
		  previous_2 = $2
		  printf("\n%s", $2) }
	} ' b1.att.val | grep -v "^$" > b1.dis

paste b1.orth.att b1.dis | sed -e "s/	/ /g" |
grep -v "^\ *$" |
sort -u |

gawk '{print $1, "B1", $2}' > b1.lex

echo Finished