CHAPTER SIXTEEN: PATTERN MATCHING (Part 9)

The Art of ASSEMBLY LANGUAGE PROGRAMMING

Chapter Sixteen (Part 8)	Table of Content	Chapter Sixteen (Part 10)

16.8.2 Processing Dates

Another useful program that converts English text to numeric form is a date processor. A date processor takes strings like "Jan 23 1997" and converts it to three integer values representing the month day and year. Of course while we're at it it's easy enough to modify the grammar for date strings to allow the input string to take any of the following common date formats:

In each of these cases the date processing routines should store one into the variable month 23 into the variable day and 1997 into the year variable (we will assume all years are in the range 1900-1999 if the string supplies only two digits for the year). Of course we could also allow dates like "January twenty-third nineteen hundred and ninety seven" by using an number processing parser similar to the one presented in the previous section. However that is an exercise left to the reader.

Date		EngMon Integer Integer \| Integer EngMon Integer \| Integer / Integer / Integer \| Integer - Integer - Integer
EngMon		JAN \| JANUARY \| FEB \| FEBRUARY \| \| DEC \| DECEMBER
Integer		digit Integer \| digit
digit		0 \| 1 \| 2 \| 3 \| 4 \| 5 \| 6 \| 7 \| 8 \| 9

We will use some semantic rules to place some restrictions on these strings. For example the grammar above allows integers of any size; however months must fall in the range 1-12 and days must fall in the range 1-28 1-29 1-30 or 1-31 depending on the year and month. Years must fall in the range 0-99 or 1900-1999.

; datepat.asm ; ; This program converts dates of various formats to a three integer ; component value- month day and year. .xlist .286 include stdlib.a includelib stdlib.lib matchfuncs .list .lall dseg segment para public 'data' ; The following three variables hold the result of the conversion. month word 0 day word 0 year word 0 ; StrPtr is a double word value that points at the string under test. ; The output routines use this variable. It is declared as two word ; values so it is easier to store es:di into it. strptr word 0 0 ; Value is a generic variable the ConvertInt routine uses value word 0 ; Number of valid days in each month (Feb is handled specially) DaysInMonth byte 31 28 31 30 31 30 31 31 30 31 30 31 ; Some sample strings to test the date conversion routines. Str0 byte "Feb 4 1956" 0 Str1 byte "July 20 1960" 0 Str2 byte "Jul 8 1964" 0 Str3 byte "1/1/97" 0 Str4 byte "1-1-1997" 0 Str5 byte "12-25-74" 0 Str6 byte "3/28/1981" 0 Str7 byte "January 1 1999" 0 Str8 byte "Feb 29 1996" 0 Str9 byte "30 June 1990" 0 Str10 byte "August 7 1945" 0 Str11 byte "30 September 1992" 0 Str12 byte "Feb 29 1990" 0 Str13 byte "29 Feb 1992" 0 ; The following grammar is what we use to process the dates ; ; Date -> EngMon Integer Integer ; | Integer EngMon Integer ; | Integer "/" Integer "/" Integer ; | Integer "-" Integer "-" Integer ; ; EngMon-> Jan | January | Feb | February | ... | Dec | December ; Integer-> digit integer | digit ; digit -> 0 | 1 | ... | 9 ; ; Some semantic rules this code has to check: ; ; If the year is in the range 0-99 this code has to add 1900 to it. ; If the year is not in the range 0-99 or 1900-1999 then return an error. ; The month must be in the range 1-12 else return an error. ; The day must be between one and 28 29 30 or 31. The exact maximum ; day depends on the month. separators pattern {spancset delimiters} ; DatePat processes dates of the form "MonInEnglish Day Year" DatePat pattern {sl_match2 EngMon DatePat2 DayYear} DayYear pattern {sl_match2 DayInteger 0 YearPat} YearPat pattern {sl_match2 YearInteger} ; DatePat2 processes dates of the form "Day MonInEng Year" DatePat2 pattern {sl_match2 DayInteger DatePat3 MonthYear} MonthYear pattern {sl_match2 EngMon 0 YearPat} ; DatePat3 processes dates of the form "mm-dd-yy" DatePat3 pattern {sl_match2 MonInteger DatePat4 DatePat3a} DatePat3a pattern {sl_match2 separators DatePat3b DatePat3b} DatePat3b pattern {matchchar '-' 0 DatePat3c} DatePat3c pattern {sl_match2 DayInteger 0 DatePat3d} DatePat3d pattern {sl_match2 separators DatePat3e DatePat3e} DatePat3e pattern {matchchar '-' 0 DatePat3f} DatePat3f pattern {sl_match2 YearInteger} ; DatePat4 processes dates of the form "mm/dd/yy" DatePat4 pattern {sl_match2 MonInteger 0 DatePat4a} DatePat4a pattern {sl_match2 separators DatePat4b DatePat4b} DatePat4b pattern {matchchar '/' 0 DatePat4c} DatePat4c pattern {sl_match2 DayInteger 0 DatePat4d} DatePat4d pattern {sl_match2 separators DatePat4e DatePat4e} DatePat4e pattern {matchchar '/' 0 DatePat4f} DatePat4f pattern {sl_match2 YearInteger} ; DayInteger matches an decimal string converts it to an integer and ; stores the result away in the Day variable. DayInteger pattern {sl_match2 Integer 0 SetDayPat} SetDayPat pattern {SetDay} ; MonInteger matches an decimal string converts it to an integer and ; stores the result away in the Month variable. MonInteger pattern {sl_match2 Integer 0 SetMonPat} SetMonPat pattern {SetMon} ; YearInteger matches an decimal string converts it to an integer and ; stores the result away in the Year variable. YearInteger pattern {sl_match2 Integer 0 SetYearPat} SetYearPat pattern {SetYear} ; Integer skips any leading delimiter characters and then matches a ; decimal string. The Integer0 pattern matches exactly the decimal ; characters; the code does a patgrab on Integer0 when converting ; this string to an integer. Integer pattern {sl_match2 separators 0 Integer0} Integer0 pattern {sl_match2 number 0 Convert2Int} number pattern {anycset digits 0 number2} number2 pattern {spancset digits} Convert2Int pattern {ConvertInt} ; A macro to make it easy to declare each of the 24 English month ; patterns (24 because we allow the full month name and an ; abbreviation). MoPat macro name next str str2 value local SetMo string full short string2 doMon name pattern {sl_match2 short next} short pattern {matchistr string2 full SetMo} full pattern {matchistr string 0 SetMo} string byte str byte 0 string2 byte str2 byte 0 SetMo pattern {MonthVal value} endm ; EngMon is a chain of patterns that match one of the strings ; JAN JANUARY FEB FEBRUARY etc. The last parameter to the ; MoPat macro is the month number. EngMon pattern {sl_match2 separators jan jan} MoPat jan feb "JAN" "JANUARY" 1 MoPat feb mar "FEB" "FEBRUARY" 2 MoPat mar apr "MAR" "MARCH" 3 MoPat apr may "APR" "APRIL" 4 MoPat may jun "MAY" "MAY" 5 MoPat jun jul "JUN" "JUNE" 6 MoPat jul aug "JUL" "JULY" 7 MoPat aug sep "AUG" "AUGUST" 8 MoPat sep oct "SEP" "SEPTEMBER" 9 MoPat oct nov "OCT" "OCTOBER" 10 MoPat nov decem "NOV" "NOVEMBER" 11 MoPat decem 0 "DEC" "DECEMBER" 12 ; We use the "digits" and "delimiters" sets from the standard library. include stdsets.a dseg ends cseg segment para public 'code' assume cs:cseg ds:dseg ; ConvertInt- Matches a sequence of digits and converts them to an integer. ConvertInt proc far push ds push es push di mov ax dseg mov ds ax lesi Integer0 ;Integer0 contains the decimal patgrab ; string we matched grab that atou ; string and convert it to an mov Value ax ; integer and save the result. free ;Free mem allocated by patgrab. pop di mov ax di ;Required by sl_match. pop es pop ds stc ;Always succeed. ret ConvertInt endp ; SetDay SetMon and SetYear simply copy value to the appropriate ; variable. SetDay proc far push ds mov ax dseg mov ds ax mov ax value mov day ax mov ax di pop ds stc ret SetDay endp SetMon proc far push ds mov ax dseg mov ds ax mov ax value mov Month ax mov ax di pop ds stc ret SetMon endp SetYear proc far push ds mov ax dseg mov ds ax mov ax value mov Year ax mov ax di pop ds stc ret SetYear endp ; MonthVal is a pattern used by the English month patterns. ; This pattern function simply copies the matchparm field to ; the month variable (the matchparm field is passed in si). MonthVal proc far push ds mov ax dseg mov ds ax mov Month si mov ax di pop ds stc ret MonthVal endp ; ChkDate- Checks a date to see if it is valid. Returns with the ; carry flag set if it is clear if not. ChkDate proc far push ds push ax push bx mov ax dseg mov ds ax ; If the year is in the range 0-99 add 1900 to it. ; Then check to see if it's in the range 1900-1999. cmp Year 100 ja Notb100 add Year 1900 Notb100: cmp Year 2000 jae BadDate cmp Year 1900 jb BadDate ; Okay make sure the month is in the range 1-12 cmp Month 12 ja BadDate cmp Month 1 jb BadDate ; See if the number of days is correct for all months except Feb: mov bx Month mov ax Day ;Make sure Day <> 0. test ax ax je BadDate cmp ah 0 ;Make sure Day < 256. jne BadDate cmp bx 2 ;Handle Feb elsewhere. je DoFeb cmp al DaysInMonth[bx-1] ;Check against max val. ja BadDate jmp GoodDate ; Kludge to handle leap years. Note that 1900 is *not* a leap year. DoFeb: cmp ax 29 ;Only applies if day is jb GoodDate ; equal to 29. ja BadDate ;Error if Day > 29. mov bx Year ;1900 is not a leap year cmp bx 1900 ; so handle that here. je BadDate and bx 11b ;Else Year mod 4 is a jne BadDate ; leap year. GoodDate: pop bx pop ax pop ds stc ret BadDate: pop bx pop ax pop ds clc ret ChkDate endp ; ConvertDate- ES:DI contains a pointer to a string containing a valid ; date. This routine converts that date to the three ; integer values found in the Month Day and Year ; variables. Then it prints them to verify the pattern ; matching routine. ConvertDate proc near ldxi DatePat xor cx cx match jnc NoMatch mov strptr di ;Save string pointer for mov strptr+2 es ; use by printf call ChkDate ;Validate the date. jnc NoMatch printf byte "%-20^s = Month: %2d Day: %2d Year: %4d\n" 0 dword strptr Month Day Year jmp Done NoMatch: printf byte "Illegal date ('%^s')" cr lf 0 dword strptr Done: ret ConvertDate endp Main proc mov ax dseg mov ds ax mov es ax meminit ;Init memory manager. ; Call ConvertDate to test several different date strings. lesi Str0 call ConvertDate lesi Str1 call ConvertDate lesi Str2 call ConvertDate lesi Str3 call ConvertDate lesi Str4 call ConvertDate lesi Str5 call ConvertDate lesi Str6 call ConvertDate lesi Str7 call ConvertDate lesi Str8 call ConvertDate lesi Str9 call ConvertDate lesi Str10 call ConvertDate lesi Str11 call ConvertDate lesi Str12 call ConvertDate lesi Str13 call ConvertDate Quit: ExitPgm Main endp cseg ends sseg segment para stack 'stack' stk db 1024 dup ("stack ") sseg ends zzzzzzseg segment para public 'zzzzzz' LastBytes db 16 dup (?) zzzzzzseg ends end Main

Feb 4 1956 = Month: 2 Day: 4 Year: 1956 July 20 1960 = Month: 7 Day: 20 Year: 1960 Jul 8 1964 = Month: 7 Day: 8 Year: 1964 1/1/97 = Month: 1 Day: 1 Year: 1997 1-1-1997 = Month: 1 Day: 1 Year: 1997 12-25-74 = Month: 12 Day: 25 Year: 1974 3/28/1981 = Month: 3 Day: 28 Year: 1981 January 1 1999 = Month: 1 Day: 1 Year: 1999 Feb 29 1996 = Month: 2 Day: 29 Year: 1996 30 June 1990 = Month: 6 Day: 30 Year: 1990 August 7 1945 = Month: 8 Day: 7 Year: 1945 30 September 1992 = Month: 9 Day: 30 Year: 1992 Illegal date ('Feb 29 1990') 29 Feb 1992 = Month: 2 Day: 29 Year: 1992