Trailing-Edge
-
PDP-10 Archives
-
decuslib20-05
-
decus/20-0141/daverb.for
There are 2 other files named daverb.for in the archive. Click here to see a list.
SUBROUTINE DAVERB(LOWWRD,MAXWRD,IWORD ,LOWKNT,MAXKNT,
1 KNTLTR,IBUFFR,MAXBFR,LOWBFR,KIND ,MATCH ,LCNWRD,
2 LCNKNT,LCNBFR)
C RENBR(/IDENTIFY WORDS OR ABBREVIATIONS)
C
C DONALD BARTH, HARVARD BUSINESS SCHOOL
C
C DAVERB INTERPRETS AN ARRAY READ BY THE CALLING
C PROGRAM WITH A MULTIPLE OF AN A1 FORMAT AND
C IDENTIFIES THE WORDS AND WORD ABBREVIATIONS CONTAINED
C IN THIS ARRAY. THE WORDS ARE RECOGNIZED BY BEING
C MATCHED AGAINST A USER DEFINED DICTIONARY. IF THE
C ARRAY CONTAINS ABBREVIATIONS OF WORDS IN THE
C DICTIONARY, THEN DAVERB ALSO SPECIFIES WHETHER THESE
C ABBREVIATIONS ARE AMBIGUOUS.
C
C ARGUMENT LIST DEFINITIONS:
C
C LOWBFR IS USED FOR BOTH INPUT AND OUTPUT. KIND,
C MATCH, LCNWRD, LCNKNT AND LCNBFR ARE USED ONLY FOR
C OUTPUT. REMAINING ARGUMENTS ARE USED ONLY FOR INPUT.
C
C LOWWRD = SUBSCRIPT OF LOCATION IN IWORD ARRAY WHICH
C CONTAINS 1ST LETTER OF 1ST WORD. NOTE THAT
C IF KNTLTR(LOWKNT) IS NEGATIVE, THEN THE 1ST
C LETTER OF 1ST WORD WILL BE FOUND IN ARRAY
C LOCATION IWORD(LOWWRD-KNTLTR(LOWKNT)).
C MAXWRD = DIMENSION OF IWORD ARRAY.
C IWORD = DICTIONARY ARRAY CONTAINING CHARACTERS OF
C WORDS TO BE RECOGNIZED, 1 CHARACTER PER
C ARRAY LOCATION AS READ BY A1 FORMAT OR ELSE
C DEFINED BY 1H FIELD. SECTIONS OF A WORD CAN
C BE ABBREVIATED AND/OR SEPARATED BY SPACES OR
C TABS IF THE WORD IN IWORD CONTAINS A SINGLE
C SPACE BETWEEN EACH SUCH SECTION AND IF THE
C LENGTH STORED IN THE KNTLTR ARRAY IS 100
C MORE THAN THE ACTUAL LENGTH (INCLUDING THE
C SPACES). ALL LETTERS IN THE IWORD ARRAY
C MUST BE UPPER CASE.
C LOWKNT = SUBSCRIPT OF KNTLTR ARRAY LOCATION DEFINING
C LENGTH OF FIRST WORD WHICH CAN BE MATCHED IN
C THE IWORD ARRAY. THIS FIRST WORD WILL START
C AT IWORD(LOWWRD). IF NO WORDS ARE TO BE
C RECOGNIZED, THEN EITHER MAXKNT SHOULD BE
C LESS THAN LOWKNT, OR ELSE BOTH LOWKNT AND
C MAXKNT CAN POINT TO THE SAME ZERO ENTRY IN
C THE KNTLTR ARRAY.
C MAXKNT = SUBSCRIPT OF KNTLTR ARRAY LOCATION DEFINING
C LENGTH OF FINAL WORD WHICH CAN BE MATCHED IN
C THE IWORD ARRAY.
C KNTLTR = ARRAY CONTAINING THE NUMBERS OF CHARACTERS
C IN THE WORDS IN THE IWORD ARRAY. A ZERO OR
C NEGATIVE VALUE IN THE KNTLTR ARRAY OFFSETS
C THE NEXT POSSIBLE WORD WHICH CAN BE MATCHED
C IN THE IWORD ARRAY BY THE NUMBER OF LETTERS
C GIVEN BY THE ABSOLUTE VALUE OF THE NEGATIVE
C NUMBER IN THE KNTLTR ARRAY. DIMENSION OF
C KNTLTR MUST BE AT LEAST MAXKNT. FOR EXAMPLE
C TO RECOGNIZE THE WORDS
C
C YES, NO, MAYBE
C
C THE CONTENTS OF THE IWORD ARRAY WOULD BE
C
C 1HY,1HE,1HS,1HN,1HO,1HM,1HA,1HY,1HB,1HE
C
C AND CONTENTS OF THE KNTLTR ARRAY WOULD BE
C
C 3,2,5
C
C IF A WORD IN THE IWORD ARRAY CONTAINS
C EMBEDDED SPACES, THEN 100 MUST BE ADDED TO
C THE LENGTH STORED FOR THIS WORD IN THE
C KNTLTR ARRAY TO ALLOW THE PORTION OF THE
C WORD LEFT OF THE SPACE TO BE ABBREVIATED.
C VALUES 101 THROUGH 199 IN KNTLTR ARRAY THUS
C INDICATE WORDS CONTAINING SPACES WHICH HAVE
C LENGTHS OF 1 THROUGH 99 RESPECTIVELY. THE
C VALUE 100 IN THE KNTLTR ARRAY IS TREATED THE
C SAME AS A ZERO.
C IBUFFR = INPUT BUFFER ARRAY CONTAINING CHARACTERS
C TYPED BY USER, READ BY A MULTIPLE OF AN A1
C FORMAT, WHICH IS TO BE SEARCHED FOR WORDS.
C IBUFFR THEN CONTAINS 1 LETTER PER COMPUTER
C STORAGE LOCATION. LETTERS IN THE IBUFFR
C ARRAY CAN BE EITHER UPPER OR LOWER CASE.
C MAXBFR = MAXIMUM SUBSCRIPT OF IBUFFR ARRAY TO BE
C SEARCHED
C LOWBFR = SUBSCRIPT WITHIN THE IBUFFR ARRAY OF THE
C FIRST (LEFTMOST) CHARACTER WHICH CAN BE
C SCANNED FOR WORDS. LOWBFR WILL BE RETURNED
C POINTING TO THE NEXT CHARACTER BEYOND A
C MATCHED WORD IF A WORD IS FOUND. IF THERE
C IS NOTHING AT OR TO RIGHT OF LOWBFR, THEN
C LOWBFR WILL BE LEFT POINTING AT MAXBFR+1 AND
C KIND WILL BE RETURNED CONTAINING ONE.
C LOWBFR MUST BE SET BY CALLING PROGRAM BEFORE
C ANYTHING IS PROCESSED IN CURRENT CONTENTS
C OF THE IBUFFR ARRAY, BUT THEN SHOULD NOT BE
C MODIFIED BY CALLING PROGRAM UNTIL THE ENTIRE
C CONTENTS OF IBUFFR ARRAY HAS BEEN PROCESSED.
C KIND = RETURNED DESCRIBING THE KIND OF ITEM LOCATED
C IN THE IBUFFR ARRAY.
C = 1, NOTHING WAS FOUND AT OR TO THE RIGHT OF
C LOWBFR. THE CALLING PROGRAM SHOULD READ A
C NEW LINE INTO IBUFFR.
C = 2, ACCEPTABLE WORD OR ABBREVIATION THEREOF
C WAS NOT FOUND, BUT A PRINTING CHARACTER WAS
C FOUND WHICH DOES NOT BEGIN ANY WORD IN THE
C DICTIONARY. LOWBFR IS RETURNED POINTING TO
C THIS PRINTING CHARACTER.
C = RETURNED CONTAINING 3, 4 OR 5 IF A WORD IN
C THE DICTIONARY WAS MATCHED EVEN PARTIALLY.
C FOR EXAMPLE, IF DICTIONARY CONTAINED BOTH OF
C THE WORDS NO AND NONE, THEN
C A) INITIAL LETTER N IN THE BUFFER FOLLOWED
C BY SOME CHARACTER OTHER THAN THE LETTER O
C WOULD BE AMBIGUOUS ABBREVIATION AND THE
C POINTER NAMED MATCH WOULD BE RETURNED
C POINTING TO (CONTAINING SEQUENCE NUMBER
C WITHIN DICTIONARY OF) WHICHEVER WORD NO
C OR NONE APPEARED FIRST IN THE DICTIONARY.
C B) INITIAL LETTERS N AND O FOLLOWED BY SOME
C CHARACTER OTHER THAN THE LETTER N WOULD
C BE AN EXACT MATCH WITH THE WORD NO.
C C) INITIAL LETTERS N AND O AND N WOULD BE A
C PARTIAL BUT NONAMBIGUOUS MATCH WITH THE
C WORD NONE.
C LEADING SPACES AND/OR TABS ARE IGNORED. A
C STRING OF CHARACTERS CONTAINING EMBEDDED
C SPACES AND/OR TABS CAN MATCH A WORD IN THE
C DICTIONARY ONLY IF THE WORD IN DICTIONARY
C CONTAINS A SINGLE SPACE AT THE POSITION AT
C WHICH THE SPACES AND/OR TABS ARE ALLOWED
C (BUT NOT NECESSARY).
C = 3, A WORD IN THE IWORD ARRAY WAS MATCHED
C EXACTLY. MATCH IS RETURNED CONTAINING THE
C SEQUENCE NUMBER OF THE WORD MATCHED IN THE
C IWORD ARRAY.
C = 4, A NONAMBIGUOUS ABBREVIATION OF A WORD IN
C THE IWORD ARRAY WAS FOUND. MATCH IS
C RETURNED CONTAINING THE SEQUENCE NUMBER OF
C THE WORD IN THE IWORD ARRAY.
C = 5, AN AMBIGUOUS ABBREVIATION OF A WORD WAS
C FOUND. MATCH IS RETURNED CONTAINING THE
C SEQUENCE NUMBER OF THE FIRST WORD MATCHED IN
C THE IWORD ARRAY.
C MATCH = RETURNED CONTAINING THE SEQUENCE NUMBER OF A
C WORD MATCHED IN THE IWORD ARRAY IF KIND IS
C RETURNED CONTAINING 3, 4 OR 5. FOR EXAMPLE,
C IF THE SECOND WORD IS MATCHED, THEN MATCH
C WOULD BE RETURNED CONTAINING 2. THE
C SEQUENCE NUMBER OF THE WORD IN THE IWORD
C ARRAY DOES NOT INCLUDE THE LETTERS SKIPPED
C OVER BY THE VALUE OF LOWWRD, AND DOES NOT
C INCLUDE THE LETTERS SKIPPED OVER BY NEGATIVE
C VALUES ENCOUNTERED IN THE KNTLTR ARRAY.
C MATCH IS RETURNED CONTAINING KIND-2 IF KIND
C IS RETURNED .LE.2 INDICATING THAT NO WORD IN
C THE IWORD ARRAY COULD BE MATCHED EVEN
C PARTIALLY. THIS MEANS THAT IF THE CALLING
C PROGRAM TESTS FOR KIND=5 AFTER THE RETURN
C FROM DAVERB, AND IF KIND=4 IS TO BE TAKEN AS
C EQUIVALENT TO KIND=3, THEN CALLING PROGRAM
C CAN ADD 2 TO THE VALUE OF MATCH AND USE THIS
C SUM AS INDEX FOR A COMPUTED GO TO STATEMENT.
C LCNWRD = IF KIND IS RETURNED CONTAINING 3 OR GREATER,
C LOCWRD IS RETURNED WITH SUBSCRIPT OF IWORD
C LOCATION CONTAINING FIRST LETTER OF MATCHED
C WORD.
C LCNKNT = IF KIND IS RETURNED CONTAINING 3 OR GREATER,
C LCNKNT IS RETURNED WITH SUBSCRIPT OF KNTWRD
C LOCATION CONTAINING THE WORD LENGTH.
C LCNBFR = IF KIND IS RETURNED CONTAINING 3 OR GREATER,
C INDICATING THAT A WORD OR ITS ABBREVIATION
C WAS FOUND, THEN LCNBFR IS RETURNED
C CONTAINING THE SUBSCRIPT OF THE IBUFFR ARRAY
C LOCATION WHICH CONTAINS THE FIRST CHARACTER
C OF THE WORD OR ITS ABBREVIATION.
C
DIMENSION IBUFFR(MAXBFR),IWORD(MAXWRD),
1KNTLTR(MAXKNT),KONVRT(10),KAPITL(26),LOWER(26)
C
C CCCC AAA UU UU TTTTTTTT IIII OOOO NN NN
C CC AAAA UU UU TT II OO OO NNN NN
C CC AA AA UU UU TT II OO OO NNNN NN
C CC AA AA UU UU TT II OO OO NN NN NN
C CC AAAAAAA UU UU TT II OO OO NN NNNN
C CC AA AA UU UU TT II OO OO NN NNN
C CCCC AA AA UUU TT IIII OOOO NN NN
C
C TO CONVERT LOWER CASE LETTERS IN THE INPUT TEXT
C BUFFER INTO UPPER CASE LETTERS WHICH CAN BE MATCHED
C AGAINST THE DICTIONARY, THIS ROUTINE COMPARES THE
C CHARACTERS IN THE INPUT TEXT BUFFER AGAINST THE LOWER
C CASE LETTERS IN THE LOWER ARRAY. THE LETTERS IN THE
C LOWER ARRAY MUST BE ARRANGED IN INCREASING NUMERICAL
C ORDER. IF THE NUMERICAL ORDER IS NOT THE SAME AS THE
C ALPHABETICAL ORDER, THEN THE DATA STATEMENTS
C APPEARING BELOW MUST BE CHANGED OR ELSE SOME OR ALL
C LOWER CASE LETTERS IN THE INPUT TEXT BUFFER WILL NOT
C BE TREATED AS EQUIVALENT TO THE CORRESPONDING UPPER
C CASE LETTERS. ONCE THE LETTERS IN THE LOWER ARRAY
C ARE SORTED INTO INCREASING NUMERICAL ORDER, THE UPPER
C CASE LETTERS IN THE KAPITL ARRAY SHOULD BE REARRANGED
C SO THAT LOWER AND UPPER CASE VERSIONS OF EACH LETTER
C APPEAR IN LOCATIONS IN THE LOWER AND KAPITL ARRAYS
C HAVING THE SAME SUBSCRIPTS.
C
C IF THE COMPUTER UPON WHICH THIS ROUTINE IS USED DOES
C NOT SUPPORT LOWER CASE LETTERS, THEN BOTH THE LOWER
C AND KAPITL ARRAYS CAN CONTAIN THE LETTERS 1HA THROUGH
C 1HZ IN ALPHABETICAL ORDER (EVEN IF THIS IS NOT THE
C NUMERICALLY SORTED ORDER).
C
C KAPITL = UPPER CASE LETTERS A THROUGH Z SORTED ON
C LOWER ARRAY
DATA KAPITL/
11HA,1HB,1HC,1HD,1HE,1HF,1HG,1HH,1HI,1HJ,
21HK,1HL,1HM,1HN,1HO,1HP,1HQ,1HR,1HS,1HT,
31HU,1HV,1HW,1HX,1HY,1HZ/
C
C LOWER = LOWER CASE LETTERS A THROUGH Z SORTED INTO
C NUMERICALLY INCREASING ORDER
DATA LOWER/
11Ha,1Hb,1Hc,1Hd,1He,1Hf,1Hg,1Hh,1Hi,1Hj,
21Hk,1Hl,1Hm,1Hn,1Ho,1Hp,1Hq,1Hr,1Hs,1Ht,
31Hu,1Hv,1Hw,1Hx,1Hy,1Hz/
C
C IBLANK = THE BLANK OR SPACE CHARACTER
C ITAB = TABULATION CHARACTER, THIS CAN BE REPLACED
C BY SPACE IF TAB CHARACTER IS NOT AVAILABLE
DATA IBLANK,ITAB/1H ,1H /
C
C SEARCH FOR FIRST PRINTING CHARACTER
GO TO 2
1 LOWBFR=LOWBFR+1
2 IF(LOWBFR.GT.MAXBFR)GO TO 29
NOWLTR=IBUFFR(LOWBFR)
IF(NOWLTR.EQ.IBLANK)GO TO 1
IF(NOWLTR.EQ.ITAB)GO TO 1
C
C SET INITIAL CONSTANTS IF FIND PRINTING CHARACTER
LMTBFR=MAXBFR
LCNBFR=LOWBFR
IEND=LOWWRD
MSTSAM=1
KNTKNV=0
KNTWRD=LOWKNT-1
INDEX=0
3 IEXACT=1
4 KNTWRD=KNTWRD+1
IF(KNTWRD.GT.MAXKNT)GO TO 28
C
C GET NEXT WORD IN DICTIONARY
JEND=KNTLTR(KNTWRD)
KEND=JEND-100
IF(KEND.GE.0)JEND=KEND
IF(JEND.LE.0)GO TO 27
KEND=0
NXTCMP=IEND
IEND=IEND+JEND
JEXACT=-1
INDEX=INDEX+1
NXTBFR=LOWBFR
NOWSAM=1
C
C GET NEXT CHARACTERS TO BE COMPARED
5 IF(NXTBFR.GT.LMTBFR)GO TO 22
KOMPAR=IBUFFR(NXTBFR)
IF(KOMPAR.EQ.IBLANK)GO TO 15
IF(KOMPAR.EQ.ITAB)GO TO 15
IF(NOWSAM.LE.KNTKNV)GO TO 13
C
C DETERMINE UPPER CASE VERSION OF A LOWER CASE LETTER.
C THIS IS A TERNARY SEARCH TAKING ADVANTAGE OF THE SIZE
C OF ALPHABET BEING NEARLY 3**3. THE 3RD OF THE ARRAY
C CONTAINING THE DESIRED LETTER IS FIRST LOCATED, THEN
C THE 3RD OF THIS 3RD, AND FINALLY EACH OF THE
C REMAINING 3 LETTERS ARE TESTED INDIVIDUALLY. TO
C PREVENT TESTING AGAINST THE 27TH LETTER WHICH DOES
C NOT EXIST, UPPER 3RD IS TAKEN AS UPPER 9 SORTED
C LETTERS, RATHER THAN FROM 19TH THROUGH 27TH LETTERS,
C SO THAT LOWER(18) IS TESTED AGAINST IN UPPER 3RD EVEN
C THOUGH LETTER BEING MATCHED HAS ALREADY BEEN FOUND TO
C BE LARGER THAN THIS.
IF(KOMPAR.GT.LOWER(18))GO TO 7
IF(KOMPAR.GT.LOWER(9))GO TO 6
IF(KOMPAR.LT.LOWER(1))GO TO 11
KUT=3
GO TO 8
6 KUT=12
GO TO 8
7 IF(KOMPAR.GT.LOWER(26))GO TO 11
KUT=20
8 IF(KOMPAR.LE.LOWER(KUT))GO TO 9
KUT=KUT+3
IF(KOMPAR.GT.LOWER(KUT))KUT=KUT+3
9 IF(KOMPAR.EQ.LOWER(KUT))GO TO 10
KUT=KUT-1
IF(KOMPAR.EQ.LOWER(KUT))GO TO 10
KUT=KUT-1
IF(KOMPAR.NE.LOWER(KUT))GO TO 11
10 KOMPAR=KAPITL(KUT)
11 IF(KNTKNV.GE.10)GO TO 12
KNTKNV=KNTKNV+1
KONVRT(KNTKNV)=KOMPAR
C
C DETERMINE IF LETTER IN BUFFER MATCHES DICTIONARY.
12 IF(KOMPAR.EQ.IWORD(NXTCMP))GO TO 19
GO TO 14
13 IF(KONVRT(NOWSAM).EQ.IWORD(NXTCMP))GO TO 19
14 IF(KEND.LE.0)GO TO 23
GO TO 17
15 IF(KEND.LE.0)GO TO 23
16 NXTBFR=NXTBFR+1
IF(NXTBFR.GT.LMTBFR)GO TO 22
IF(IBUFFR(NXTBFR).EQ.IBLANK)GO TO 16
IF(IBUFFR(NXTBFR).EQ.ITAB)GO TO 16
17 KEND=0
18 IF(IWORD(NXTCMP).EQ.IBLANK)GO TO 20
JEXACT=0
NXTCMP=NXTCMP+1
IF(NXTCMP.LT.IEND)GO TO 18
GO TO 24
19 NOWSAM=NOWSAM+1
NEWBFR=NXTBFR
KEND=JEND
NXTBFR=NXTBFR+1
20 NXTCMP=NXTCMP+1
21 IF(NXTCMP.LT.IEND)GO TO 5
GO TO 24
C
C WORD CANNOT EXTEND FURTHER TO RIGHT
22 LMTBFR=NEWBFR
23 JEXACT=0
24 IF(NOWSAM.LT.MSTSAM)GO TO 4
IF(NOWSAM.GT.MSTSAM)GO TO 26
IF(IEXACT.GE.0)GO TO 25
IF(JEXACT.LT.0)GO TO 3
GO TO 4
25 IF(JEXACT.GE.0)GO TO 3
26 IEXACT=JEXACT
MSTSAM=NOWSAM
MATCH=INDEX
LSTBFR=NEWBFR+1
LCNKNT=KNTWRD
LCNWRD=IEND
GO TO 4
27 IEND=IEND-JEND
GO TO 4
C
C ENTIRE DICTIONARY HAS BEEN SEARCHED
28 IF(MSTSAM.LE.1)GO TO 30
LOWBFR=LSTBFR
KIND=4+IEXACT
JEND=KNTLTR(LCNKNT)
IF(JEND.GE.100)JEND=JEND-100
LCNWRD=LCNWRD-JEND
GO TO 31
C
C NO PRINTING CHARACTERS WERE FOUND TO BE IDENTIFIED
29 KIND=1
MATCH=-1
GO TO 31
C
C NOT EVEN A PARTIAL MATCH COULD BE MADE
30 KIND=2
MATCH=0
C
C RETURN TO CALLING PROGRAM
31 RETURN
C
C IEXACT = -1, EXACT MATCH FOUND BUT MUST CHECK THAT
C A LONGER MATCH CANNOT BE FOUND WITH ANOTHER
C WORD (FOR EXAMPLE, IF IWORD ARRAY CONTAINS
C BOTH OF THE WORDS NO AND NONE, THEN THE
C BUFFER CONTENTS "NON" WOULD MATCH WORD NO
C EXACTLY, BUT THE PARTIAL MATCH WITH WORD
C NONE WOULD BE BETTER)
C = 0, A PARTIAL MATCH HAS BEEN FOUND
C = 1, NO MATCH FOUND OR DUPLICATE PARTIAL
C MSTSAM = 1 + MAXIMUM NUMBER OF LETTERS MATCHED
C NOWSAM = 1 + NUMBER OF LETTERS MATCHING CURRENT WORD
C864241272470abcdefghijklmnopqrstuvwxyz
END