initial commit

2 files changed, 147 insertions, 0 deletions

diff --git a/parsing.moon b/parsing.moon
new file mode 100644
index 0000000..f9efef4
--- /dev/null
+++ b/parsing.moon
@@ -0,0 +1,76 @@
+lpeg = require 'lpeg'
+
+class Atom
+  new: (@raw, @style='', @value) =>
+
+  stringify: =>
+    switch @style
+      when ''
+        @raw
+      when '"'
+        "\"#{@raw}\""
+      else
+        error!
+
+  @make_num: (match) -> Atom match, '', tonumber match
+  @make_sym: (match) -> Atom match, '', match
+  @make_str: (match) -> Atom match, '"', match
+
+class Xpr
+  new: (parts, @style='(') =>
+    @white = {}
+    @white[0] = parts[1]
+
+    for i = 2,#parts,2
+      @[i/2] = parts[i]
+      @white[i/2] = parts[i+1]
+
+  stringify: =>
+    buf = ''
+    buf ..= @white[0]
+    for i, frag in ipairs @
+      buf ..= frag\stringify!
+      buf ..= @white[i]
+
+    switch @style
+      when 'naked'
+        buf
+      else
+        '(' .. buf .. ')'
+
+  make_sexpr: (parts) -> Xpr parts, '('
+  make_nexpr: (parts) -> Xpr parts, 'naked'
+
+space = (lpeg.S ' \t\r\n') ^ 1 / 1
+mspace = (lpeg.S ' \t\r\n') ^ 0 / 1
+
+sym = ((lpeg.R 'az', 'AZ') + (lpeg.S '-_+*')) ^ 1 / Atom.make_sym
+str = '"' * (lpeg.C (1 - lpeg.P '"') ^ 0) * '"' / Atom.make_str
+num = (lpeg.R '09', 'AZ') ^ 1 / Atom.make_num
+atom = sym + num + str
+
+expr = (lpeg.V 'sexpr') + atom
+explist = lpeg.Ct mspace * (lpeg.V 'expr') * (space * atom) ^ 0 * mspace
+sexpr = (lpeg.P '(') * (lpeg.V 'explist') * (lpeg.P ')') / Xpr.make_sexpr
+
+nexpr = lpeg.P {
+  (lpeg.V 'explist') / Xpr.make_nexpr
+  :expr, :explist, :sexpr
+}
+
+sexpr = lpeg.P {
+  'sexpr'
+  :expr, :explist, :sexpr
+}
+
+program = nexpr
+
+{
+  :space
+  :atom
+  :expr
+  :explist
+  :sexpr
+  :nexpr
+  :program
+}
diff --git a/spec/peg_spec.moon b/spec/peg_spec.moon
new file mode 100644
index 0000000..1b25406
--- /dev/null
+++ b/spec/peg_spec.moon
@@ -0,0 +1,71 @@
+import space, atom, expr, explist, sexpr, nexpr, program from require 'parsing'
+
+describe 'atom parsing', ->
+  test 'symbols', ->
+    sym = atom\match 'some-toast help'
+    assert.is.equal 'some-toast', sym.raw
+    assert.is.equal 'some-toast', sym.value
+    assert.is.equal 'some-toast', sym\stringify!
+
+  test 'numbers', ->
+    num = atom\match '1234 nope'
+    assert.is.equal '1234', num.raw
+    assert.is.equal 1234, num.value
+    assert.is.equal '1234', num\stringify!
+
+  test 'strings', ->
+    str = atom\match '"help some stuff!" nope'
+    assert.is.equal 'help some stuff!', str.raw
+    assert.is.equal 'help some stuff!', str.value
+    assert.is.equal '"help some stuff!"', str\stringify!
+
+test 'whitespace parsing', ->
+  assert.is.equal '  ', space\match '  '
+  assert.is.equal '\n\t ', space\match '\n\t '
+
+describe 'nexpr parsing', ->
+  it 'handles leading whitespace', ->
+    node = nexpr\match ' 3\tok-yes'
+
+    assert.is.equal 2, #node
+    assert.is.equal 3, node[1].value
+    assert.is.equal 'ok-yes', node[2].value
+
+    assert.is.equal ' 3\tok-yes', node\stringify!
+
+  it 'handles trailing whitespace', ->
+    node = nexpr\match '3\tok-yes\n'
+
+    assert.is.equal 2, #node
+    assert.is.equal 3, node[1].value
+    assert.is.equal 'ok-yes', node[2].value
+
+    assert.is.equal '3\tok-yes\n', node\stringify!
+
+  it 'handles whitespace everywhere', ->
+    node = nexpr\match ' 3\tok-yes\n'
+
+    assert.is.equal 2, #node
+    assert.is.equal 3, node[1].value
+    assert.is.equal 'ok-yes', node[2].value
+
+    assert.is.equal ' 3\tok-yes\n', node\stringify!
+
+test 'sexpr parsing', ->
+  str = '( 3   ok-yes
+  "friend" )'
+  node = sexpr\match str
+
+  assert.is.equal '(', node.style
+  assert.is.equal 3, #node
+  assert.is.equal 3, node[1].value
+  assert.is.equal 'ok-yes', node[2].value
+  assert.is.equal 'friend', node[3].value
+
+  assert.is.equal str, node\stringify!
+
+test 'mixed parsing', ->
+  str = '( 3   ok-yes
+  "friend" )'
+  node = program\match str
+  assert.is.equal str, node\stringify!