def line_tokenize(self, non_literal, line_num):
"""
Tokenize a line, with string literals removed.
:param non_literal: text to be tokenize, no string literal
:param line_num: the line number
:return: None
"""
lst = normalize(non_literal)
for part in lst:
if part.isidentifier():
self.tokens.append(stl.IdToken(line_num, part))
elif is_float(part):
self.tokens.append(stl.NumToken(line_num, part))
elif is_integer(part):
self.tokens.append(stl.NumToken(line_num, part))
elif stl.is_in_all(part):
self.tokens.append(stl.IdToken(line_num, part))
elif part[:-1] in stl.OP_EQ:
self.tokens.append(stl.IdToken(line_num, part))
elif part == stl.EOL:
self.tokens.append(stl.IdToken(line_num, stl.EOL))
elif part in stl.OMITS:
pass
else:
raise stl.ParseException(
"Unknown symbol: '{}', at line {}".format(part, line_num))
def import_file(self, full_path, import_name):
"""
Imports an external sp file.
This method tokenize the imported file, and inserts all tokens except the EOF token of the imported
file into the current file.
:param full_path: the path of the file to be imported
:param import_name: the name to be used
"""
with open(full_path, "r") as file:
lexer = Tokenizer()
lexer.setup(self.spl_path,
full_path,
get_dir(full_path),
import_lang=False)
# lexer.script_dir = get_dir(full_path)
lexer.tokenize(file)
# print(lexer.tokens)
self.tokens.append(stl.IdToken(LINE_FILE, import_name))
self.tokens.append(stl.IdToken(LINE_FILE, full_path))
self.tokens.append(stl.IdToken(LINE_FILE, "{"))
self.tokens += lexer.tokens
self.tokens.pop() # remove the EOF token
self.tokens.append(stl.IdToken(LINE_FILE, "}"))
def restore_tokens(self, file: _io.BytesIO):
self.tokens.clear()
while True:
flag = int.from_bytes(file.read(1), "big")
if flag == 0:
self.tokens.append(stl.Token((stl.EOF, None)))
break
else:
line = int(stl.read_string(file))
file_name = stl.read_string(file)
lf = line, file_name
if flag == 1:
token: stl.NumToken = stl.NumToken(lf,
stl.read_string(file))
elif flag == 2:
token: stl.LiteralToken = stl.LiteralToken(
lf, stl.read_string(file))
elif flag == 3:
token: stl.IdToken = stl.IdToken(lf, stl.read_string(file))
elif flag == 4:
token: stl.DocToken = stl.DocToken(lf,
stl.read_string(file))
else:
raise stl.ParseException("Unknown flag: {}".format(flag))
self.tokens.append(token)
def tokenize(self, source):
"""
Tokenize the source spl source code into a list of tokens, stored in the memory of this Lexer.
:param source: the source code, whether an opened file or a list of lines.
:return: None
"""
self.tokens.clear()
if self.import_lang and self.file_name[-7:] != "lang.sp":
self.tokens += [
stl.IdToken(LINE_FILE, "import"),
stl.IdToken(LINE_FILE, "namespace"),
stl.LiteralToken(LINE_FILE, "lang")
]
self.find_import(0, 3)
if isinstance(source, list):
self.tokenize_text(source)
else:
self.tokenize_file(source)
def find_import(self, from_, to):
"""
Looks for import statement between the given slice of the tokens list.
:param from_: the beginning position of search
:param to: the end position of search
:return: None
"""
for i in range(from_, to, 1):
token = self.tokens[i]
if isinstance(token, stl.IdToken) and token.symbol == "import":
next_token: stl.Token = self.tokens[i + 1]
namespace_token = None
if isinstance(
next_token,
stl.IdToken) and next_token.symbol == "namespace":
namespace_token = next_token
self.tokens.pop(i + 1)
path_token: stl.LiteralToken = self.tokens[i + 1]
elif isinstance(next_token, stl.LiteralToken):
path_token: stl.LiteralToken = self.tokens[i + 1]
else:
raise stl.ParseException(
"Unexpected token in file '{}', at line {}".format(
next_token.file, next_token.line))
name = path_token.text
if name[-3:] == ".sp": # user lib
if len(self.script_dir) == 0:
file_name = name[:-3] + ".sp"
else:
file_name = self.script_dir + "{}{}".format(
"/", name[:-3]) + ".sp"
# file_name = "{}{}{}".format(self.script_dir, os.sep, name[:-3]).replace(".", "/") + ".sp"
if "/" in name:
import_name = name[name.rfind("/") + 1:-3]
else:
import_name = name[:-3]
else: # system lib
file_name = "{}{}lib{}{}.sp".format(
self.spl_path, os.sep, os.sep, name)
import_name = name
if len(self.tokens) > i + 2:
as_token: stl.IdToken = self.tokens[i + 2]
if as_token.symbol == "as":
if namespace_token is not None:
raise stl.ParseException(
"Unexpected combination 'import namespace ... as ...'"
)
name_token: stl.IdToken = self.tokens[i + 3]
import_name = name_token.symbol
self.tokens.pop(i + 1)
self.tokens.pop(i + 1)
self.tokens.pop(i + 1) # remove the import name token
self.import_file(file_name, import_name)
if namespace_token:
lf = namespace_token.line, namespace_token.file
self.tokens.append(namespace_token)
self.tokens.append(stl.IdToken(lf, import_name))
self.tokens.append(stl.IdToken(lf, stl.EOL))
break