\ primes_bbs.fs -- WIP \ Sift files of 1,000,000 primes each from t5k.org into... \ ...those congruent-for-Blum Blum Shub and those not... \ ...creating a separate file for each C_**.hexp, N_**.hexp \ Also read a single prime from out of a *.hexp file. \ Version: 2026-07-04 Copyright Gan Uesli Starling \ License: None. Free for all uses. \ Load order matters S" ~/FYBB/edit_me.fs" INCLUDED \ You need to have edited this. S" defs.fs" include.file \ Some general user definitions S" kybd.fs" include.file \ So can query user S" dec_to_bin.fs" include.file S" math.fs" include.file S" bbs.fs" include.file : throw.file ( flg c-addr c -- ) ROT 0<> IF CR CR ." Oops! " TYPE CR ABORT ELSE 2DROP THEN ; CREATE path_dir 255 ALLOT ALIGN \ Directory CREATE path_p 255 ALLOT ALIGN CREATE path_c 255 ALLOT ALIGN CREATE path_n 255 ALLOT ALIGN CREATE hex_p 255 ALLOT ALIGN \ Prime read as HEX from *.hexp : path.erase ( addr -- addr ) 255 0 DO DUP I + 0 SWAP c! LOOP ; : path.load ( c-addr1 cnt1 c-addr2 -- ) \ String & Count path.erase 2DUP c! 1+ SWAP CMOVE ; S" /home/aplonis/Downloads" path_dir path.load 0 VALUE fd_p \ For primes from t5k.org 0 VALUE fd_c \ For congruent primes 0 VALUE fd_n \ For non-congruent primes 0 VALUE row_chars \ Width of output rows 0 VALUE c_tally \ Count of congruent primes 0 VALUE n_tally \ Count of non-congruent 0 VALUE out_# \ Which millionth prime set : open.file.p ( -- ) path_p COUNT R/O OPEN-FILE path_p COUNT S" @ open.file.p" cs.append throw.file TO fd_p ; : open.file.c ( -- ) path_c COUNT R/W OPEN-FILE path_c COUNT S" @ open.file.c" cs.append throw.file TO fd_c ; : open.file.n ( -- ) path_n COUNT R/w OPEN-FILE path_n COUNT S" @ open.file.n" cs.append throw.file TO fd_n ; : create.file.c ( -- ) path_c COUNT W/O CREATE-FILE path_c COUNT S" @ create.file.c" cs.append throw.file TO fd_c ; : create.file.n ( -- ) path_n COUNT W/O CREATE-FILE path_N COUNT S" @ create.file.n" cs.append throw.file TO fd_n ; : close.file.p ( -- ) fd_p CLOSE-FILE S" close.file.p" throw.file ; : close.file.c ( -- ) fd_c CLOSE-FILE S" close.file.c" throw.file ; : close.file.n ( -- ) fd_n CLOSE-FILE S" close.file.n" throw.file ; \ Runs slower than the mouse-macro colleting input files... \ ...so can clean up behind as it goes. Used only by cff.* words : delete.file.p ( 0|1 -- ) IF 1500 MS \ Wait for file to close path_p COUNT DELETE-FILE S" DELETE-FILE at cff.sift.files " throw.file THEN ; : file.paths.load ( -- ) path_dir COUNT S" /primes" cs.append out_# 0 <# #S #> cs.append S" .txt" cs.append path_p path.load path_dir COUNT S" /C_" cs.append out_# 0 <# # # #> cs.append S" .hexp" cs.append path_c path.load path_dir COUNT S" /N_" cs.append out_# 0 <# # # #> cs.append S" .hexp" cs.append path_n path.load ; \ NOTE 1 -- ALL WORDS NAMED t5k.* RELATE TO DOING THIS... \ 1. Sifting files named primes**.txt for congruent vs non-congruent primes \ 2. Writing out sifted primes into files named C_**.hexp and N_**.hexp \ NOTE 2 -- ALL WORDS NAMED cn.* ARE FOR... \ 1. Reading in a file produced by t5k words, either C_**.hexp or N_**.hexp \ 2. Obtaining from it a single prime, counted from either bottom or top \ Files from t5k.org are formated 125k rows of 8 cols : t5k.limit.row.col ( u1 u2 -- u3 u4 ) ABS 8 MOD SWAP \ Cols of primes in file ABS 125000 MOD SWAP \ Rows of primes in file ; \ Files from t5k.org have a header. Blank lines too. \ Read a full line of 8 primes onto PAD : t5k.get.line ( -- ) BEGIN PAD 81 fd_p READ-LINE ( u flg ior ) S" t5k.get.line READ-LINE" throw.file ( u flg ) DROP 81 <> ( flg ) \ 81 + CR + LF = 83 chars WHILE REPEAT ; \ Get Nth row of 8 primes from bottom of a t5k.org primes file. \ Calculate from EOF to avoid random-char-count header. : t5k.nth.line ( u -- u flg ) 125000 SWAP - 83 * S>D \ 1/8 M lines, 8 cols: chars = 83 (81 + CR + LF) fd_p FILE-SIZE \ S" t5k.nth.line FILE-SIZE" throw.file SWAP D- \ Wanted row end 83 S>D D- \ Wanted row start fd_p REPOSITION-FILE S" t5k.nth.line REPOSITION-FILE" throw.file t5k.get.line ( ) \ Line of primes on PAD ; \ Select single prime from row of eight in PAD : t5k.nth.column ( u -- u ) 10 * PAD + 10 string.to.integer ; : t5k.get.prime ( u1 u2 -- u ) \ Row & Column -- Prime t5k.limit.row.col ( u1' u2' ) SWAP t5k.nth.line ( u2' ) \ CR PAD 80 TYPE CR t5k.nth.column ( u ) \ DUP . CR ; : t5k.u.to.file ( u fd -- ) >R HEX 0 <# row_chars 0 DO # LOOP #> DECIMAL ( c-addr cnt ) R> WRITE-LINE S" t5k.u.to.file WRITE-LINE" throw.file ; \ Test for BBS congruency and choose output file. : sift.congruent ( u -- u fd ) rsa_p a.erase ( u ) \ Clear array DUP 1 rsa_p a.u! ( u ) \ Load one cell into array rsa_p bbs.congruent? ( u 0|1 ) \ Is it BBS congruent? Ref bbs.fs IF ( u ) c_tally 1+ TO c_tally ( u ) fd_c ( u fd ) \ Congruents into own file ELSE n_tally 1+ TO n_tally ( u ) fd_n ( u fd ) \ Why waste the dross? THEN ; \ Top and bottom rows of file prime50.txt \ 961748941 961748947 961748951 961748969 961748987 961748993 961749023 961749037 \ 982451501 982451549 982451567 982451579 982451581 982451609 982451629 982451653 : t5k.get.row ( u ) \ u = Row 8 0 DO \ I = Column DUP I t5k.get.prime ( row u ) \ Get prime from t5k.org file sift.congruent ( row u fd ) t5k.u.to.file ( row ) LOOP ( row ) DROP ( ) ; : t5k.get.rows ( highest lowest ) DO I t5k.get.row LOOP ; : t5k.first.row S" Width: " row_chars 0 <# #S #> cs.append S" Tally: " \ Ex-post-facto manual entry actual tally cs.append 2DUP fd_c WRITE-LINE S" t5k.first.row primes_bbs.fs" throw.file fd_n WRITE-LINE S" t5k.first.row primes_bbs.fs" throw.file ; \ How many HEX chars wide should output rows be? : t5k.col.bytewidth? ( -- ) 0 7 t5k.get.prime ( u ) \ Get last value from input HEX 0 <# #S #> DECIMAL ( c-array count ) \ Convert to HEX string NIP TO row_chars ( ) \ Store into value t5k.first.row ; \ Read all 1/8 million rows, obtaining 8 primes for each \ Test for Blum Blum Shug congruency, saving out to file : t5k.sift.primes open.file.p create.file.c create.file.n t5k.col.bytewidth? 125000 0 t5k.get.rows \ 1/8 million rows of 8 primes each fd_c FLUSH-FILE S" t5k.sift.primes FLUSH-FILE" throw.file fd_n FLUSH-FILE S" t5k.sift.primes FLUSH-FILE" throw.file close.file.n close.file.c close.file.p ; \ Sift N files of 1M primes from www.t5k.org : t5k.sift.files ( u u ) \ Smallest & highest files to sift SWAP DO 0 TO c_tally 0 TO n_tally I TO out_# file.paths.load CR CR ." Input file :" path_p COUNT TYPE CR t5k.sift.primes CR ." C = " c_tally . ." N = " n_tally . LOOP ; \ After having read header line into pad, parse out the Width... \ ...which could be multi-digit for very large primes. : cn.parse.width ( 22 -- c-addr c ) \ >Width: 8 Tally: 499905< in PAD PAD SWAP next.space 1+ ( pad 22 7 ) \ >Width: 8 Tally: 499905< 22 7 SWAP OVER - ( pad 7 15 ) \ >8 Tally: 499905< 15 -ROT + SWAP ( pad+7 15 ) \ >8 Tally: 499905< 2DUP next.space ( pad+7 15 pad+7 15 1 ) DUP 1+ >us NIP ( pad+7 15 pad+7 1 ) \ >8 Tally: 499905< >8< string.to.integer ( pad+7 15 8 ) \ >8 Tally: 499905< 8 TO row_chars ( pad+7 15 ) \ >8 Tally: 499905< SWAP us@ + ( 15 pad+9 ) SWAP us> - ( pad+9 13 ) \ >Tally: 499905< ; \ After having parsed width from pad, parse out left-over Tally... \ ...which might be any value. : cn.parse.tally ( c-addr u -- u ) \ >Tally: 499905< next.space 1+ ( pad+9 13 7 ) \ >Tally: 499905< 7 SWAP OVER - ( pad+9 7 6 ) \ >Tally: 499905< 7 6 -ROT + SWAP ( pad+16 6 ) \ >499905< 6 string.to.integer ( u ) \ 499905 ; : cn.parse.hdr ( fileid -- u ) \ Store row_chars, get tally of primes PAD 81 ROT READ-LINE S" cn.parse.hdr READ-LINE" throw.file ( u flg ) IF ( 22 ) \ Assuming PAD = >Width: 8 Tally: 499905< cn.parse.width ( 6 ) \ row_chars = 8 cn.parse.tally ( 499905 ) \ tally ELSE CR ." Oops! End of file at cn.parse.hdr " CR THEN DUP 0= IF CR ." Oops! Tally missing from header. " CR THEN ; \ Calculate file position from bottom-of-file : cn.up.fm.btm >R DUP FILE-SIZE S" cn.up.fm.btm FILE-SIZE" throw.file ( fileid u ud ) \ CR ." File-size = " 2DUP D. CR row_chars 1+ ( fileid file-size bytes/row ) \ inclusive of newline R> ( fileid file-size bytes/row prime-sought ) M* ( fileid file-size prime-offset ) \ offset from bottom-of-file D- ( fileid prime-location ) ; \ Calculate file position from top-of-file : cn.dn.fm.hdr >R DUP FILE-POSITION S" cn.dn.fm.top FILE-SIZE" throw.file ( fileid u ud ) \ CR ." File-size = " 2DUP D. CR row_chars 1+ ( fileid file-size bytes/row ) \ inclusive of newline R> ( fileid file-size bytes/row prime-sought ) M* ( fileid file-size prime-offset ) \ offset from bottom-of-file D+ ( fileid prime-location ) ; \ Read line into hex_p, returning count of chars read : cn.get.line ( fileid u -- u ) DUP 0> IF cn.up.fm.btm ELSE cn.dn.fm.hdr THEN 2 PICK REPOSITION-FILE S" cn.get.prime REPOSITION-FILE" throw.file hex_p 255 ROT READ-LINE S" cn.get.line READ-LINE" throw.file ( u flg ) IF ELSE CR ." Oops! End of file at cn.get.prime " CR THEN ; : cn.get.prime ( u fileid ) SWAP OVER ( fileid u fileid ) cn.parse.hdr ( fileid u u ) \ CR ." Tally = " DUP . CR MOD ( fileid u ) \ CR ." Nth = " DUP . CR cn.get.line ( u ) \ Line as HEX in hex_p ; : cn.get.c ( u -- u ) \ random -- congruent prime \ CR CR ." Source file (congruent): " path_c COUNT TYPE CR open.file.c fd_c cn.get.prime close.file.c ; : cn.get.n ( u -- u ) \ random -- non-congruent prime \ CR CR ." Source file (non-congruent): " path_n COUNT TYPE CR open.file.n fd_n cn.get.prime close.file.n ; : cn.get.p ( u -- u ) \ random -- a prime random 2 MOD 0= IF cn.get.c ELSE cn.get.n THEN ; \ Fetch the Nth-from-bottom prime from out of a sifted file. \ u1 = 10|11|12 ... 48|49|50 the ** in C_10.hexp or N_11.hexp files, etc. \ u2 = 543210 the Nth prime from bottom of said file \ 0|1|2 for whether a C_*.hexp, N_*.hexp. Zero for when don't care which : cn.prime ( u1 u2 0|1|2 -- u ) ROT TO out_# file.paths.load CASE 0 OF cn.get.p ENDOF \ Don't care whether congruent or not 1 OF cn.get.c ENDOF \ Get congruent 2 OF cn.get.n ENDOF \ Get non-congruent ENDCASE hex_p SWAP \ String & count for use by word a.load.hex in file math.fs ; \ Pillage primes in groups of 600 from website below... \ http://compoasso.free.fr/primelistweb/page/prime/liste_online_en.php \ Work around one aspect of file.paths.load from above ... \ ... preserving it as-is for use by earlier words above. : cff.open.file.p ( u -- ) DUP 0= IF DROP path_dir COUNT S" /export.txt" cs.append path_p path.load ELSE >R path_dir COUNT S" /export(" cs.append R> 0 <# #S #> cs.append S" ).txt" cs.append path_p path.load THEN open.file.p ; \ Files from compoasso.free.fr have no header or blank lines \ Read a full line of 10 primes onto PAD : cff.get.line ( -- c ) PAD 123 fd_p READ-LINE ( u flg ior ) S" cff.get.line READ-LINE" throw.file ( u flg ) DROP ; 0 VALUE cff_prime_last \ Parse out columns of numbers from line : cff.parse.numbers ( c i -- ) \ Count of chars read into PAD >R PAD SWAP 9 0 DO split.on.tab >us >us LOOP string.to.integer 9 0 DO us> us> string.to.integer LOOP R> 0= \ Index from loop of prior word IF DROP 9 \ First prime in file is dup of last prime in prior file ELSE 10 THEN 0 DO sift.congruent ( u -- u fd ) >R DUP TO cff_prime_last \ So can tell user final prime processed HEX 0 <# #S #> DECIMAL ( c-addr cnt ) \ 2DUP TYPE SPACE R> WRITE-LINE S" cff.parse.lines WRITE-LINE" throw.file LOOP ; \ Read in lines, parse for numbers, sift BBS congruent : cff.parse.lines ( -- ) 60 0 DO cff.get.line ( c ) \ Chars read DUP 0> ( c 0|1 ) \ Not an empty line PAD c@ ascii.decimal? ( c 0|1 0|1 ) \ Has numbers? Ref defs.fs AND ( c 0|1 ) \ Both at once? IF I cff.parse.numbers ( ) \ Get from PAD ELSE DROP THEN LOOP ; \ Close output files : cff.outputs.close fd_c FLUSH-FILE S" cff.sift.primes FLUSH-FILE" throw.file fd_n FLUSH-FILE S" cff.sift.primes FLUSH-FILE" throw.file close.file.n close.file.c ; \ Create output files : cff.outputs.create create.file.c create.file.n S" Width: Tally: " 2DUP fd_c WRITE-LINE S" cff.out.files? WRITE-LINE for fd_c" throw.file fd_n WRITE-LINE S" cff.out.files? WRITE-LINE for fd_n" throw.file ; : cff.outputs.fresh? 999999 \ A million less 1. Will actually be 1,000,330 n_tally c_tally + \ Total primes sifted < IF \ 1 million primes sifted? cff.outputs.close \ Close current ouput files 0 TO c_tally \ Zero both tallies 0 TO n_tally out_# 1+ TO out_# \ Increment total millions sifted file.paths.load cff.outputs.create \ Create new output files THEN ; \ Similar to t5k.sift.files except source files are... \ ...export(**).txt as downloaded from the compasso.free.fr webpage \ Append to files C_**.hexp and N_**.hexp where ** is numeric of \ top int on stack. : cff.sift.files ( 9999999 0 51 -- ) \ 168 for 2 when ready TO out_# 0 TO n_tally 0 TO c_tally file.paths.load cff.outputs.create \ Create new output files DO cff.outputs.fresh? \ New outputs at start and every 1 million sifted I cff.open.file.p \ Fresh input file, named like 'export(0).txt' CR ." Sifting 599 primes from " \ Assure user script has not stalled. path_p COUNT TYPE CR cff.parse.lines \ Sift primes in file close.file.p \ Close current input file 0 delete.file.p \ Clean up on the go? 0|1 is flag LOOP cff.outputs.close CR ." Final prime = " cff_prime_last . CR ; 120000 MS \ Delay 2 minutes 1670 2 * 0 51 cff.sift.files FALSE [IF] \ True for testing CR CR ." TESTING WORDS FROM primes_bbs.fs " CR FALSE [IF] CR ." This will take BLOODY AGES to finish!" CR ." For each of the 10th through 50th MILLION primes, two 4MB-plus files will be generated..." CR ." C_**.hexp containing primes congruent to MOD 3 and MOD 4, as needed by Blum Blum Shub" CR ." N_**.hexp segregating all the other primes which are not" 10 50 t5k.sift.files CR [THEN] CR ." Extracting single primes from out of *.hexp files." 46 -1 1 cn.prime CR TYPE \ 2nd prime after header in 46th million primes that are BBS congruent 47 0 2 cn.prime CR TYPE \ 1st prime after header in 47th million primes that aren't congruent 48 1 0 cn.prime CR TYPE \ Last prime of file in 48th million primes, congruent or not : test.cn.prime \ Over-shooting file capacity is no issue, as MOD will limit request 7 0 DO 50 40 random.range random 0 cn.prime CR TYPE 50 40 random.range random INVERT 0 cn.prime CR TYPE LOOP ; test.cn.prime CR CR ." ALL DONE! " CR [THEN]