From ed5fbd5e72c3a20de4bb321a3923fc7e7e87a596 Mon Sep 17 00:00:00 2001
From: Marshall Lochbaum <mwlochbaum@gmail.com>
Date: Sat, 5 Sep 2020 22:33:04 -0400
Subject: Merge JS and dbqn versions into UTF-16-independent (!) compiler

---
 c.bqn | 188 ++++++++----------------------------------------------------------
 1 file changed, 21 insertions(+), 167 deletions(-)

(limited to 'c.bqn')

diff --git a/c.bqn b/c.bqn
index 5867d6bd..ae4f3cc5 100755
--- a/c.bqn
+++ b/c.bqn
@@ -1,67 +1,33 @@
 #! ./dzref
 
-nl←@+10
-charSet←∾charGroups←⟨
-  chF←"+-×÷⋆√⌊⌈|¬∧∨<>≠=≤≥≡≢⊣⊢⥊∾≍↑↓↕⌽⍉/⍋⍒⊏⊑⊐⊒∊⍷⊔!" # Function
-  "˙˜˘¨⌜⁼´˝`"           # Modifier
-  "∘○⊸⟜⌾⊘◶⎉⚇⍟"          # Composition
-  nl∾"⋄,"               # Separator
-  "←↩→"                 # Gets
-  "(){}⟨⟩"              # Bracket
-  "‿"                   # Ligature
-  "·"                   # nOthing
-  ¯1⊏˘∘‿2⥊"𝕊𝕏𝕎𝔽𝔾𝕤𝕩𝕨𝕗𝕘"  # Input (𝕣 pending; ℝ not allowed)
-  "@"                   # nUll character
-  '0'+↕10               # Digit
-  "¯.π∞"                # Numeric
-  "_"∾˜⥊"aA"+⌜↕na←26    # Alphabetic
-  1↑"𝕨"                 # Prefix for input (hack around UTF-16)
-  " "∾@+9               # Whitespace
-# #'" eliminated during tokenization
-⟩
-ErrUnknownChars←{
-  "Unclosed quote"!¬∨´⟨''','"'⟩∊𝕩
-  ⟨"Unknown character","s"/˜1<≠𝕩,": ",𝕩⟩∾⊸!0
-}
-CharCode←charSet{
-  Chk ← ⊢⊣ErrUnknownChars∘(≠/⊣)⍟≢⟜(⊏⟜𝕗)
-  g←⍋𝕗 ⋄ ⊢ Chk g⊏˜1-˜1⌈(g⊏𝕗)⍋⊢
-}
+i←"⟨charGroups,Tokenize,ReadNum,Compile⟩"
+⟨charGroups,Tokenize,ReadNum,Compile⟩ ← ⟨•path∾"src/c.bqn",i⟩ •EX •path∾"dzref"
+
+chF←"+-×÷⋆√⌊⌈|¬∧∨<>≠=≤≥≡≢⊣⊢⥊∾≍↑↓↕⌽⍉/⍋⍒⊏⊑⊐⊒∊⍷⊔!"
 bF‿bM‿bC‿bS‿bG‿bB‿bL‿bO‿bI‿bU‿bD‿bN‿bA‿bP‿bW←≍¨˜⟜(+`≠↑0∾⊢)≠¨charGroups
-bA_e←4+⊑bA
 M←1⊸⊑(0⊸≤∧>)-⟜⊑
 vi←+´≠¨9↑charGroups
-charRole←((vi-1⊑bI)↑/0∾≠¨3↑charGroups)∾(5/⌽↕2)∾0
-spc←⥊3‿5‿6-⌜3‿0
 
-Tokenize←{
-  r←𝕩='#'⋄s←/(≠↑2⊸↓)⊸∧𝕩='''⋄d←/dm←𝕩='"'
-  g←⍋q←∾⟨  s⋄¯1↓d⋄/r⟩ ⋄q↩g⊏q
-  e← g⊏∾⟨2+s⋄ 1↓d⋄(⊢-¯1↓0∾⊢)∘⊏⟜(0∾+`r)⊸//(𝕩=nl)∾1⟩
-  Se←{(⊏˜𝕨)Se 1¨⌾((𝕩/𝕨)⊸⊏)𝕩}⍟{0=⊑⌽𝕩}
-  st←¯1↓Se⟜(1↑˜≠)∾⟜≠q⍋e⋄b←st/q∾˘e
-  ToI←¯1↓·/⁼(≠𝕩)∾˜⥊
-  f←¬≠`bm←ToI b⋄f∧↩¬qe←f∧bm∧(1↓0∾˜⊢)⊸∧dm
-  cb←(¬(st/q)⊏r)/b
-  lu←⍷lit←𝕩⊔˜1-˜(+`qe-˜ToI⊑˘cb)×≠`ToI cb
-  cl←f/ToI⊑∘⌽˘cb
 
-  c←cl-˜CharCode(⊑charSet)¨⌾(cl⊸/)f/𝕩
-  w←(≠↑0∾⊢)⊸<l←c M (⊑bD)≍+´1⊑¨bD‿bN‿bA
-  u←c=⊑bU⋄l+↩u⋄w+↩u
-  us←c=¯1++´bA
-  id←(1-˜(us<l)×+`w)⊔na(⊢-⊣×+⟜(⊑bA)⊸≤)c
-  ti←na⌊∘÷˜(⊑bA)-˜w/c
-  iu←(≠↑¯1∾⌈`)⊸<is←⊐id⋄ni←≠in←0>iu/ti
-  ti↩(us/˜(1↓0∾˜⊢)⊸<l)(⊢+∧⟜(2⊸=))0⌈ti
-
-  c↩(w∨¬l∨c M (⊑bP)≍+´1⊑¨bP‿bW)/(vi+is⊏⍋⍋in)⌾(w⊸/)(vi+ni+lu⊐lit)⌾(cl⊸/)c
-  c/˜↩¬(≠↑1∾(c∊2‿4+⊑bB)∨⊢)⊸∧c M bS⋄c/˜↩¬(1↓1∾˜c∊3‿5+⊑bB)∧c M bS
-  ic←(0⊸≤∧<⟜ni)c-vi⋄t←ti⌾(ic⊸/)(vi⌊c)⊏charRole
-  c+↩5×c M 5≍˜⊑bI
-  ⟨c,t,+´¬in,((in//iu)⊏id)∾lu⟩
+# Targetting dzaima/BQN bytecode:
+prims←⟨
+  +,-,×,÷,⋆,√,⌊,⌈,|,¬,∧,∨,<,>,≠,=,≤,≥,≡,≢,⊣,⊢,⥊,∾,≍,↑,↓,↕,⌽,⍉,/,⍋,⍒,⊏,⊑,⊐,⊒,∊,⍷,⊔,!
+  ˙,˜,˘,¨,⌜,⁼,´,˝,`
+  ∘,○,⊸,⟜,⌾,⊘,◶,⎉,⚇,⍟
+⟩
+special←<˘∘‿2⥊"𝕤𝕩𝕨𝕣𝕗𝕘"
+DRun←{
+  ⟨bc,o,blk⟩ ← prims Compile 𝕩
+  blk ↩ {
+    ⟨t,i,l,n⟩←𝕩
+    s←(3×i)↓(t⊑3‿5‿6)↑special
+    ⟨t⊑"fmd",i,l,s∾⥊¨'a'+↕n-≠s⟩
+  }¨blk
+  •COMP ⟨bc,o,⟨⟩,blk⟩
 }
 
+
+# Targetting WebAssembly (very incomplete)
 WParse←{
   a←𝕩M(2≍˜⊑bG)⋄at←1⌽a⋄𝕩/˜↩¬a⋄a/˜↩¬at
   l←≠𝕩⋄sep←𝕩M bS⋄𝕩↩(bF⊑⊸+⊑chF⊐<'⊣')¨⌾(sep⊸/)𝕩⋄sep∨↩𝕩=2+⊑bB
@@ -78,17 +44,6 @@ Enc2←{2|⌊∘÷⟜2⍟(↕𝕨)𝕩}
 
 Hex←16 Base⟜⌽ (∾"0A"+⟜↕¨10‿6)⊸⊐
 
-ReadNum←{
-  n‿d‿p‿i←4↕⊸+⊑bN  # ¯.π∞
-  Nat←+⟜(10⊸×)´∘⌽ -⟜(⊑bD)
-  Int←(n=⊑)◶⟨Nat,-·Nat 1⊸↓⟩
-  Dec←⊐⟜(<d)⊸(Nat∘↑ + ·(0<≠)◶⟨0,Nat÷10⋆≠⟩+⟜1⊸↓)
-  Pos←(⊑p‿i⊐⊏)◶⟨π,∞,Dec⟩
-  e←⊑𝕩⊐bA_e
-  m←(n=⊑)◶⟨Pos,-·Pos 1⊸↓⟩e↑𝕩
-  𝕩×⟜(10⋆·Int(e+1)↓⊣)˜⍟(e<≠𝕩)m
-}
-LitVal←(⊑⟨⊑bU,''','"'⟩⊐⊏)◶⟨@,1⊸⊑,1⊸↓,ReadNum⟩
 GenF64←{
   0:8⥊0;
   l←2(⌊⋆⁼)𝕩
@@ -148,104 +103,3 @@ WCompile←{
   f←⟨rcp,or,body⟩
   Gen ⟨(≠¨f)/(1‿2‿0∾¨1)⥊¨¨f64 ⋄ 2 ⋄ ∾f⟩
 }
-
-
-# Targetting dzaima/BQN bytecode:
-
-LEBv ← {
-  b←128
-  l←⌊b⋆⁼1⌈𝕩
-  o←⍋⍋((↕¯1⊑⊢)-/⟜(≠↑0∾⊢))⟜(+`)1+l
-  o⊏l{f←×𝕨⋄(𝕨-1)(b⊸(×⟜f+|)∾𝕊⟜(⌊÷⟜b)○(f⊸/))⍟(∨´f)𝕩}𝕩
-}
-
-ErrMismatchedBrackets←{
-  Lcs ← ¯1 ⊑ 0˘∘⊢ {𝕩⌈⌈`𝕨+0∾˜1↓𝕩}˝ =⌜
-  _mis_ ← {"Missing "∾𝕗∾" "∾charSet⊏˜𝕘+⊢/˜≠∘⊣=⊒}
-  Msg ← >○≠◶⟨"opening"_mis_ 0, "closing"_mis_ 1˜⟩
-  0 !˜ 𝕨 (Lcs<⌊○≠)◶⟨Msg,"Malformed bracket nesting"⟩ 𝕩
-}
-ErrUndeclared←{
-  0 !˜ "Undefined identifier"∾(1<≠𝕩)/"s"
-}
-Parse ← {nv‿r←𝕨
-  g←⍋pd←+`p←𝕩(Mׯ1⋆2|-⟜⊑)bB⋄br←p×𝕩M 2≍˜2+⊑bB
-  p(=⟜1⊸/ErrMismatchedBrackets⍟≢1-˜=⟜¯1⊸/)○(g⊸⊏)𝕩
-  "Swapped open and closed brackets" ! 0≤(⊑g)⊑pd
-  sl←1⊸⌽⊸∨𝕩 M bL⋄sr←¯1⊸⌽⌾(g⊸⊏)sl⋄sa←sl∨sr
-  g⊏˜↩⍋g⊏sl⋄r×↩¬sa⋄o←p>0
-  pt←sa<𝕩=1+⊑bB
-  rev←⍋+`1+¯1↓g((¯1∾⊣)(⊣⍋⊸⊏⊏˜-⊏˜⟜⍋)⟜⍋1(+`∘∾-∾˜)⊏)o∨sl∨𝕩 M bS
-
-  gf←⍋fd←+`rev⊏br
-  rev⊏˜↩gf⋄fd⊏˜↩gf
-  𝕩⊏˜↩rev⋄bv←rev⊏br⋄b←/bv<0⋄fi←+`c←bv>0
-
-  FC←-⟜(≠↑0∾⊢)(c∾1)/·+`∾⟜0
-  H←0<1↓FC∘=
-  ft←(0∾1⊸H+2×2⊸H⌈2×3⊸H)(0‿3‿4‿5+5+⊑bI)⍋𝕩
-  fsc←ft⊏spc
-  r↩((1↓ft)⊏(1+2⊸≤)⊸/↕4)⌾((b⊏rev)⊸⊏)r
-  r+↩pt(⊣∧¬⊸∨=○(⌈`(1+↕∘≠)⊸×)⊢)(¯1⊸⌽⌾(g⊸⊏)3=r)∨1≤r
-
-  id←/idm←(0⊸≤∧<⟜nv)xv←𝕩-vi⋄idx←id⊏𝕩⋄sp←/𝕩 M bI
-  ad←(⌈`↕∘≠⊸×𝕩 M bG)⊏𝕩=⊑bG
-
-  g↩⍋+`rev⊏p-br⋄gr←g⊏rev⋄sll←1+2÷˜0(<-○/>)gr⊏sr-sl⋄l←/g⊏𝕩=5+⊑bB
-  g⊏˜↩gs←⍋gr⊏sl⋄gr↩g⊏rev⋄𝕩⊏˜↩g⋄r⊏˜↩gr⋄o⊏˜↩gr⋄si←/gr⊏sr>sl
-  gi←⍋g⋄l⊏↩⍋gs
-  s←𝕩 M bS⋄ps←s∨o∨gr⊏sl⋄a←𝕩 M 2≍˜⊑bG
-  r-↩ps∨a⋄op←r≥2⋄fe←(r≥1)∨1⌽r=3
-  tr←(⌈`↕∘≠⊸ׯ1⌽ps)⊏fe
-  ma←tr<(𝕩=1+⊑bG)∧1⌽fe
-  ro←op∨a<1⌽r=3⋄os←⌽↕∘≠⊸(⊣-⌈`∘×)⌽¬ro∨ma⋄at←(⊢+1+⊏⟜os)/a
-  aa←g⊏asn←¯1⌽+`gi⊏-⟜(≠↑0∾⊢)¯1↓/⁼(≠𝕩)∾˜at-1⋄r↩¯1¨⌾(aa⊸/)r
-  tf←(a≤○(⌈`(1+↕∘≠)⊸×)ps)∧(⊢∧2(|<≤)ps(⊢-⌈`∘×)+`)¬ro
-  opa←op>1↓0∾˜ps∨a
-  oa←⌽/opa⋄fa←/(1⌽aa)<(tf∨¬tr)∧(ro∧1⌽opa)<(r=1)∨op<¯1⌽opa
-  dy←fa⊏1↓0∾˜(𝕩≠⊑bO)∧(tr∧r≥0)∨ro<r=0
-  n←𝕩≥vi+nv⋄cn←/n∨𝕩≤¯1++´bC⋄u←∧⍷ob←cn⊏𝕩
-  lo←(o/𝕩)=4+⊑bB⋄ll←1+lo/1(↓--⊸↓)(o∾1)/+`(s∾0)-(1∾o)∧ps∾1
-  dr←/s>o+`⊸⊏0∾lo⋄rt←/𝕩=2+⊑bB
-
-  d←id⊏dec←idm∧ad∧asn⋄lc←FC dec
-  ixf←(⊏⟜(¯1∾c/gf)∾d⊸/⊏((≠𝕩)∾b⊏gf)˜)id⊏fi
-  ig←(⍋⊏⟜(ixx←∾⟜(d⊸/)idx))⊸⊏⍋ixf
-  "Multiple definitions" ! 1∧´ixx∨○((≠↑0∾⊢)⊸≠((≠id)⊸≤⊸/ig)⊏⊢)ixf
-  ig↩<⟜(≠id)⊸/(⍋+`∘⊏⟜(d∾-/˜d))⊸⊏ig
-  d⊏˜↩ig⋄ErrUndeclared∘/˜⍟(0∨´⊢)⟜(d<(≠↑0∾⊢)⊸≠)ig⊏idx
-  ui←(⌈`↕∘≠⊸×d)⊸⊏⌾(ig⊸⊏)id
-  ii←(ui⊏(fi⊏fsc)-(⌈`c⊸×)⊸¬+`dec)∾((sp⊏fi)⊏3×2|ft)++⟜(0⊸≤)2+sp⊏xv
-  idor←∾3⥊<is←id∾sp
-  idbc←⟨21+is⊏asn,(id-○(⊏⟜fd)ui)∾0¨sp,ii⟩
-
-  b⊏↩gi
-  or←⍋idor∾g⊏˜∾⟨cn,cn,b,b,rt,dr,l,l,si,si,at,oa+1⌈oa⊏os,(dy×⊏⟜os)⊸+fa+dy⟩
-  bc0←∾idbc∾⟨0¨cn,u⊐ob,15¨b,1+↕≠b⟩
-  bc←or⊏bc0∾∾⟨25¨rt,14¨dr,3+l⊏aa,ll,3+si⊏aa,sll,(11-⊑bG)+a/ma+𝕩,5+oa⊏r,(⊢+9×10⊸=)16+dy-7×fa⊏tr⟩
-  fs←(ft⊏⥊3≍⌜⟜⌽○↕2)∾¨((+`0∾1+⌊128⋆⁼1⌈bc)⊏˜/1∾or M bc0≍○≠rt)≍○<¨lc+fsc
-  ⟨LEBv bc∾25,u,fs⟩
-}
-
-Compile←{
-  ⟨t,r,nVar,lits⟩←Tokenize𝕩
-  ⟨bc,u,blk⟩←nVar‿r Parse t
-  o←(u-(vi+nVar-≠𝕨)×u≥vi)⊏𝕨∾LitVal¨lits
-  ⟨bc,o,blk⟩
-}
-
-prims←⟨
-  +,-,×,÷,⋆,√,⌊,⌈,|,¬,∧,∨,<,>,≠,=,≤,≥,≡,≢,⊣,⊢,⥊,∾,≍,↑,↓,↕,⌽,⍉,/,⍋,⍒,⊏,⊑,⊐,⊒,∊,⍷,⊔,!
-  ˙,˜,˘,¨,⌜,⁼,´,˝,`
-  ∘,○,⊸,⟜,⌾,⊘,◶,⎉,⚇,⍟
-⟩
-special←<˘∘‿2⥊"𝕤𝕩𝕨𝕣𝕗𝕘"
-DRun←{
-  ⟨bc,o,blk⟩ ← prims Compile 𝕩
-  blk ↩ {
-    ⟨t,i,l,n⟩←𝕩
-    s←(3×i)↓(t⊑3‿5‿6)↑special
-    ⟨t⊑"fmd",i,l,s∾⥊¨'a'+↕n-≠s⟩
-  }¨blk
-  •COMP ⟨bc,o,⟨⟩,blk⟩
-}
-- 
cgit v1.2.3