From 6bb849d6cfc00791acb72017ac0e7f7dea7bdc08 Mon Sep 17 00:00:00 2001
From: Marshall Lochbaum <mwlochbaum@gmail.com>
Date: Sat, 5 Sep 2020 22:17:48 -0400
Subject: Align self-hosted compiler with dzaima/BQN one, except for UTF-16
 hacks

---
 c.bqn       | 99 +++++++++++++++++++++++++++++++++++++++----------------------
 src/c.bqn   |  2 +-
 src/cjs.bqn |  4 +--
 test/t.js   |  2 +-
 4 files changed, 68 insertions(+), 39 deletions(-)

diff --git a/c.bqn b/c.bqn
index c7fa2d3d..5867d6bd 100755
--- a/c.bqn
+++ b/c.bqn
@@ -10,7 +10,7 @@ charSet←∾charGroups←⟨
   "(){}⟨⟩"              # Bracket
   "‿"                   # Ligature
   "·"                   # nOthing
-  ¯1⊏˘10‿2⥊"𝕊𝕏𝕎𝔽𝔾𝕤𝕩𝕨𝕗𝕘" # Input (𝕣 pending; ℝ not allowed)
+  ¯1⊏˘∘‿2⥊"𝕊𝕏𝕎𝔽𝔾𝕤𝕩𝕨𝕗𝕘"  # Input (𝕣 pending; ℝ not allowed)
   "@"                   # nUll character
   '0'+↕10               # Digit
   "¯.π∞"                # Numeric
@@ -19,13 +19,20 @@ charSet←∾charGroups←⟨
   " "∾@+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)⍋⊢
+}
 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
-special←<˘6‿2⥊"𝕤𝕩𝕨𝕣𝕗𝕘"
-spd←⥊⍉3‿0↓⌜↑⟜special⌜3‿5‿6
+spc←⥊3‿5‿6-⌜3‿0
 
 Tokenize←{
   r←𝕩='#'⋄s←/(≠↑2⊸↓)⊸∧𝕩='''⋄d←/dm←𝕩='"'
@@ -39,23 +46,23 @@ Tokenize←{
   lu←⍷lit←𝕩⊔˜1-˜(+`qe-˜ToI⊑˘cb)×≠`ToI cb
   cl←f/ToI⊑∘⌽˘cb
 
-  c←charSet⊐f/𝕩
+  c←cl-˜CharCode(⊑charSet)¨⌾(cl⊸/)f/𝕩
   w←(≠↑0∾⊢)⊸<l←c M (⊑bD)≍+´1⊑¨bD‿bN‿bA
   u←c=⊑bU⋄l+↩u⋄w+↩u
-  i←(1-˜l×+`w)⊔c
-  na←≠•a⋄us←¯1++´bA
-  in←na(⊢-⊣×+⟜(⊑bA)⊸≤)us⊸≠⊸/¨i
-  id←⍷(bA M˜w/c)⍒⊸⊏in⋄nv←+´bA M˜⊑¨id
+  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+id⊐in)⌾(w⊸/)(vi+(≠id)+lu⊐lit)⌾(cl⊸/)c
+  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
-  ti←(us=¯1⊸⊑¨i)(⊢+∧⟜(2⊸=))0⌈na⌊∘÷˜(⊑bA)-˜⊑¨i
-  t←ti⌾(((0⊸≤∧<⟜(≠id))c-vi)⊸/)(vi⌊c)⊏charRole
+  ic←(0⊸≤∧<⟜ni)c-vi⋄t←ti⌾(ic⊸/)(vi⌊c)⊏charRole
   c+↩5×c M 5≍˜⊑bI
-  ⟨c,t,nv,(nv↓id)∾lu⟩
+  ⟨c,t,+´¬in,((in//iu)⊏id)∾lu⟩
 }
 
-Parse←{
+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
   o←𝕩=⊑bB⋄c←𝕩=1+⊑bB⋄v←a-˜𝕩≥vi⋄f←¬o∨c∨v∨sep
@@ -69,11 +76,11 @@ Parse←{
 Base←{+⟜(𝕨⊸×)´𝕩}
 Enc2←{2|⌊∘÷⟜2⍟(↕𝕨)𝕩}
 
-Hex←16 Base⟜⌽ (•d∾"ABCDEF")⊸⊐
+Hex←16 Base⟜⌽ (∾"0A"+⟜↕¨10‿6)⊸⊐
 
 ReadNum←{
   n‿d‿p‿i←4↕⊸+⊑bN  # ¯.π∞
-  Nat←10 Base⟜⌽ -⟜(⊑bD)
+  Nat←+⟜(10⊸×)´∘⌽ -⟜(⊑bD)
   Int←(n=⊑)◶⟨Nat,-·Nat 1⊸↓⟩
   Dec←⊐⟜(<d)⊸(Nat∘↑ + ·(0<≠)◶⟨0,Nat÷10⋆≠⟩+⟜1⊸↓)
   Pos←(⊑p‿i⊐⊏)◶⟨π,∞,Dec⟩
@@ -81,7 +88,7 @@ ReadNum←{
   m←(n=⊑)◶⟨Pos,-·Pos 1⊸↓⟩e↑𝕩
   𝕩×⟜(10⋆·Int(e+1)↓⊣)˜⍟(e<≠𝕩)m
 }
-LitVal←(⊑((⊑bU)∾"'""")⊐⊏)◶⟨@,1⊸⊑,1⊸↓,ReadNum⟩
+LitVal←(⊑⟨⊑bU,''','"'⟩⊐⊏)◶⟨@,1⊸⊑,1⊸↓,ReadNum⟩
 GenF64←{
   0:8⥊0;
   l←2(⌊⋆⁼)𝕩
@@ -109,7 +116,7 @@ GenFn←{
   t⊏˜↩⍋+`-˝(2‿3+⊑bB)=⌜t
   nd←+´c←t=3+⊑bB
   t↩((vi+nVar+≠lits)+↕∘≠)⌾(c⊸/)t
-  ⟨a,na⟩←Parse t
+  ⟨a,na⟩←WParse t
   ops←⥊∾⟨
     ∾⟜(0⥊˜3∾˜vi-≠)fntab
     ⍉(Hex¨"20"‿"22"‿"21")∾⌜↕nVar
@@ -136,7 +143,7 @@ Gen←{
 
 rcp←1 GenFn "1÷x"
 or ←2 GenFn "(w+x)-w∧x"
-Compile←{
+WCompile←{
   body←GenFn 𝕩
   f←⟨rcp,or,body⟩
   Gen ⟨(≠¨f)/(1‿2‿0∾¨1)⥊¨¨f64 ⋄ 2 ⋄ ∾f⟩
@@ -148,12 +155,23 @@ Compile←{
 LEBv ← {
   b←128
   l←⌊b⋆⁼1⌈𝕩
-  o←⍋⍋⊒/1+l
+  o←⍋⍋((↕¯1⊑⊢)-/⟜(≠↑0∾⊢))⟜(+`)1+l
   o⊏l{f←×𝕨⋄(𝕨-1)(b⊸(×⟜f+|)∾𝕊⟜(⌊÷⟜b)○(f⊸/))⍟(∨´f)𝕩}𝕩
 }
 
-DParse ← {nv‿r𝕊𝕩:
-  g←⍋+`p←𝕩(Mׯ1⋆2|-⟜⊑)bB⋄br←p×𝕩M 2≍˜2+⊑bB
+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
@@ -166,6 +184,7 @@ DParse ← {nv‿r𝕊𝕩:
   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
 
@@ -190,10 +209,13 @@ DParse ← {nv‿r𝕊𝕩:
   dr←/s>o+`⊸⊏0∾lo⋄rt←/𝕩=2+⊑bB
 
   d←id⊏dec←idm∧ad∧asn⋄lc←FC dec
-  ig←⍋(⊏⟜(¯1∾c/gf)∾d⊸/⊏((≠𝕩)∾b⊏gf)˜)id⊏fi
-  ig↩<⟜(≠id)⊸/(⍋+`∘⊏⟜(d∾-/˜d))⊸⊏(⍋⊏⟜(∾⟜(d⊸/)idx))⊸⊏ig
-  ui←d(⌈`↕∘≠⊸×)⊸⊏⌾(ig⊸⊏)id
-  ii←(ui⊏(fi⊏ft⊏≠¨spd)-(⌈`c⊸×)⊸¬+`dec)∾((sp⊏fi)⊏3×2|ft)++⟜(0⊸≤)2+sp⊏xv
+  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⟩
 
@@ -201,8 +223,15 @@ DParse ← {nv‿r𝕊𝕩:
   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⊏⥊"fmd"≍⌜⌽↕2)∾¨((+`0∾1+⌊128⋆⁼1⌈bc)⊏˜/1∾or M bc0≍○≠rt)≍○<¨(ft⊏spd)∾⟜(⥊¨↑⟜•a)¨lc
-  ⟨LEBv bc,u,fs⟩
+  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←⟨
@@ -210,13 +239,13 @@ prims←⟨
   ˙,˜,˘,¨,⌜,⁼,´,˝,`
   ∘,○,⊸,⟜,⌾,⊘,◶,⎉,⚇,⍟
 ⟩
-DGenFn←{
-  p←𝕨⊣prims
-  ⟨t,r,nVar,lits⟩←Tokenize𝕩
-  ⟨bc,u,blk⟩←nVar‿r DParse t
-  o←(u-(vi+nVar-≠p)×u≥vi)⊏p∾LitVal¨lits
-  ⟨bc,o,⟨⟩,blk⟩
-}
+special←<˘∘‿2⥊"𝕤𝕩𝕨𝕣𝕗𝕘"
 DRun←{
-  •COMP DGenFn 𝕩
+  ⟨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⟩
 }
diff --git a/src/c.bqn b/src/c.bqn
index 1967591f..bd41f254 100644
--- a/src/c.bqn
+++ b/src/c.bqn
@@ -83,7 +83,7 @@ LEBv ← {
 
 ErrMismatchedBrackets←{
   Lcs ← ¯1 ⊑ 0˘∘⊢ {𝕩⌈⌈`𝕨+0∾˜1↓𝕩}˝ =⌜
-  _mis ← {"Missing "∾𝕗∾" "∾charSet⊏˜𝕘+⊢/˜≠∘⊣=⊒}
+  _mis_ ← {"Missing "∾𝕗∾" "∾charSet⊏˜𝕘+⊢/˜≠∘⊣=⊒}
   Msg ← >○≠◶⟨"opening"_mis_ 0, "closing"_mis_ 1˜⟩
   0 !˜ 𝕨 (Lcs<⌊○≠)◶⟨Msg,"Malformed bracket nesting"⟩ 𝕩
 }
diff --git a/src/cjs.bqn b/src/cjs.bqn
index e936b4fd..2b26f65d 100755
--- a/src/cjs.bqn
+++ b/src/cjs.bqn
@@ -1,6 +1,6 @@
 #! /usr/bin/env dbqn
 
-compile ← ⟨•path∾"../c.bqn"⋄"DGenFn"⟩ •EX •path∾"../dzref"
+compile ← ⟨•path∾"../c.bqn"⋄"Compile"⟩ •EX •path∾"../dzref"
 Comp ← ((<"runtime["∾⍕∾"]"˙)¨↕60)⊸Compile
 FP ← ∞⊸=◶⟨⍕,"Infinity"⟩
 Esc ← (⟨"\t","\n","\r","\'"⟩∾<∘⥊)⊑˜("'"∾˜@+9‿10‿13)⊑∘⊐⊢
@@ -9,7 +9,7 @@ Char ← "'"(∾∾⊣)Esc
 Fconst ← ≡◶⟨(⊑⍋)∘(∾⟜∞)◶⟨0⊸≤◶⟨"-"∾FP∘|,FP⟩,Char⟩⋄Str⋄⊑⟩
 L ← "["∾"]"∾˜1↓·∾","⊸∾⟜⍕¨
 J ← ∾∾⟜(@+10)¨
-Fout ← ⟨L∾⟜25,L Fconst¨,L⟨⊑"fmd"⊸⊐,⊢,⊢,≠⟩⊸(L{𝕎𝕩}¨)¨⟩ {𝕎𝕩}¨ 0‿1‿3 ⊏ ⊢
+Fout ← ⟨⊢,Fconst,L⟩ {L𝕎¨𝕩}¨ ⊢
 Frun ← ("provide.concat("∾∾⟜")")⌾(1⊸⊑) Fout
 Long ← ∾ (1‿2/⟨"  new Uint8Array("∾∾⟜")"⋄((@+10)∾" ,")⊸∾⟩) {𝕎𝕩}¨ ⊢
 • ← (⊑"r"‿"c"‿"fmt"⊐⊏)◶⟨
diff --git a/test/t.js b/test/t.js
index 05ee2004..c11bcc01 100755
--- a/test/t.js
+++ b/test/t.js
@@ -11,7 +11,7 @@ const t=load('cases.bqn').split('\n').filter(x=>x).map(x=>x.split(' % '))
 
 var compiler = execFile(
   __dirname+'/../cshim.bqn',
-  [ '{•←Compile𝕩}¨⟨'+test+'⟩' ],
+  [ '{•←WCompile𝕩}¨⟨'+test+'⟩' ],
   function (error, stdout, stderr) {
     const rslt=stdout.split('\n').filter(a=>a.length)
               .map(a=>runWasm(a.split("‿").map(n=>+n)))
-- 
cgit v1.2.3