# The lexer extract NIT tokens from an input stream.
# It is better user with the Parser
class Lexer
+ super TablesCapable
# Last peeked token
var _token: nullable Token
_stream = stream
_stream_pos = -1
_stream_buf = new Buffer
- build_goto_table
- build_accept_table
end
# Give the next token (but do not consume it)
var accept_pos = -1
var accept_line = -1
- var goto_table = _goto_table[_state]
- var accept = _accept_table[_state]
var text = _text
text.clear
- while true do
+ loop
var c = get_char
if c != -1 then
text.add(c.ascii)
- var first_loop = true # aka until
- while dfa_state < -1 or first_loop do
+ loop
var old_state = dfa_state
if dfa_state < -1 then
old_state = -2 - dfa_state
dfa_state = -1
- var tmp0 = goto_table[old_state]
var low = 0
- var high = tmp0.length - 1
+ var high = lexer_goto(old_state, 0) - 1
if high >= 0 then
- var tmp1 = tmp0.intern_items
while low <= high do
var middle = (low + high) / 2
- var tmp2 = tmp1[middle].intern_items
+ var offset = middle * 3 + 1 # +1 because length is at 0
- if c < tmp2[0] then
+ if c < lexer_goto(old_state, offset) then
high = middle - 1
- else if c > tmp2[1] then
+ else if c > lexer_goto(old_state, offset+1) then
low = middle + 1
else
- dfa_state = tmp2[2]
- low = high + 1 # aka break
+ dfa_state = lexer_goto(old_state, offset+2)
+ break
end
end
end
- first_loop = false # aka until
+ if dfa_state > -2 then break
end
_cr = cr
end
if dfa_state >= 0 then
- if accept[dfa_state] != -1 then
+ var tok = lexer_accept(dfa_state)
+ if tok != -1 then
accept_state = dfa_state
- accept_token = accept[dfa_state]
+ accept_token = tok
accept_length = text.length
accept_pos = _pos
accept_line = _line
end
else
if accept_state != -1 then
+ var location = new Location(_filename, start_line + 1, accept_line + 1, start_pos + 1, accept_pos)
+ _pos = accept_pos
+ _line = accept_line
+ push_back(accept_length)
$ foreach {//token}
if accept_token == ${position()-1} then
- var location = new Location(_filename, start_line + 1, accept_line + 1, start_pos + 1, accept_pos)
-$ if {not(@text)}
-$ if {@parser_index}
- var token_text = text.substring(0, accept_length)
- var token = new @ename.init_tk(token_text, location)
-$ end
-$ else
- var token = new @ename.init_tk(location)
-$ end
- push_back(accept_length)
- _pos = accept_pos
- _line = accept_line
$ if {count(transition[@from!=@to])!=0}
var state_id = _state
$ foreach transition in {transition[@from!=@to]}
$ end
$ end if
$ if {@parser_index}
- return token
+$ if {not(@text)}
+ var token_text = text.substring(0, accept_length)
+ return new @ename.init_tk(token_text, location)
+$ else
+ return new @ename.init_tk(location)
+$ end
$ else
return null
$ end
end
$ end foreach
else
- var location = new Location(_filename, start_line + 1, accept_line + 1, start_pos + 1, accept_pos)
+ var location = new Location(_filename, start_line + 1, start_line + 1, start_pos + 1, start_pos + 1)
if text.length > 0 then
- var token = new PError.init_error("Unknown token: {text}", location)
+ var token = new PError.init_error("Syntax error: unknown token {text}.", location)
return token
else
var token = new EOF(location)
end
end
end
+ if false then break # FIXME remove once unreach loop exits are in c_src
end
- return null
+ return null # FIXME remove once unreach loop exits are in c_src
end
# Read the next character.
i = i - 1
end
end
+end
- var _goto_table: Array[Array[Array[Array[Int]]]]
- private fun build_goto_table
- do
- _goto_table = once [
+$ end template
+
+
+
+$ template make_lexer_table()
$ foreach {lexer_data/goto_table/state}
- [
$ foreach {row}
$ if {count(goto)!=0}
- [
+static const int lexer_goto_row${position()}[] = {
+ ${count(goto)},
$ foreach {goto}
- [@low, @high, @state][-sep ','-]
+ @low, @high, @state[-sep ','-]
$ end foreach
- ][-sep ','-]
+};
+$ end
+$ end foreach
+static const int lexer_goto_row_null[] = {0};
+const int* const lexer_goto_table[] = {
+$ foreach {row}
+$ if {count(goto)!=0}
+ lexer_goto_row${position()}[-sep ','-]
$ else
- nil_array[-sep ','-]
+ lexer_goto_row_null[-sep ','-]
$ end
$ end foreach
- ][-sep ','-]
+};
$ end foreach
- ]
- end
- private fun nil_array: Array[Array[Int]]
- do
- return once new Array[Array[Int]]
- end
-
- var _accept_table: Array[Array[Int]]
- private fun build_accept_table do
- _accept_table = once [
$ foreach {lexer_data/accept_table/state}
- [
- [-foreach {i}-]${.}[-sep ','-][-end foreach-]
-
- ][-sep ','-]
+const int lexer_accept_table[] = {
+ [-foreach {i}-]${.}[-sep ','-][-end foreach-]
+};
$ end foreach
- ]
- end
-end
$ end template
nitlanguage / nit.git / blobdiff