From: ahmedsamyh <ahmedsamyh10@gmail.com>
Date: Sat, 16 Nov 2024 16:54:12 +0000 (+0500)
Subject: Can parse identifiers.
X-Git-Url: https://www.git.momoyon.org/?a=commitdiff_plain;h=45f58ef0edd4e8dd333244c177a7f12ba2727b18;p=lang.git

Can parse identifiers.

- Handle lines and bol
- [New test] 03-whitespaced-string.momo
- [New test] 04-identifier.momo
---

diff --git a/main.momo b/main.momo
index 794a03c..dad80ff 100644
--- a/main.momo
+++ b/main.momo
@@ -1,6 +1,2 @@
 
-
-
-
-
-"This is a long ass string"
+   foo : string = "Value of foo";
diff --git a/main.py b/main.py
index 0a40e28..0fee0a9 100644
--- a/main.py
+++ b/main.py
@@ -100,13 +100,31 @@ class Parser:
 
         return (string, string_loc)
 
+    def consume_identifier(self) -> (str, Loc):
+        # Identifiers can start with [a-z][A-Z]_ and contain [0-9] after the first char
+        assert self.current_char().isalpha() or self.current_char() == '_', "Called consume_identifier() at the wrong character!"
+        ident_loc: Loc = Loc(self.filename, self.line, self.row())
+        ident: str = self.consume_char()
+
+        c = self.consume_char()
+
+        while c.isalpha() or c == '_' or c.isdigit() or self.eof():
+            ident += c
+            c = self.consume_char()
+
+        return (ident, ident_loc)
+
     def left_trim(self):
         while self.current_char().isspace():
+            if self.current_char() == '\n':
+                self.line += 1
+                self.bol = self.cur + 1
             self.consume_char()
             # dlog(f"Skipping {self.current_char()}")
 
         # dlog(f"Char after left trim: '{self.current_char()}'")
 
+
     def next_token(self) -> Token | None:
         self.left_trim()
 
@@ -116,10 +134,14 @@ class Parser:
         if c == '"':
             value, loc = self.consume_string()
             t = Token(TokenType.STRING, value, loc)
+        elif c.isalpha() or c == '_':
+            ident, loc = self.consume_identifier()
+            t = Token(TokenType.IDENT, ident, loc)
         else:
             fatal(f"Unrecognized character '{c}'")
 
         return t
+
 def main():
     program: str = sys.argv.pop(0)
 
diff --git a/tests/03-whitespaced-string.momo b/tests/03-whitespaced-string.momo
new file mode 100644
index 0000000..c168e11
--- /dev/null
+++ b/tests/03-whitespaced-string.momo
@@ -0,0 +1,3 @@
+
+
+"This is a long ass string"
diff --git a/tests/03-whitespaced-string.momo.test b/tests/03-whitespaced-string.momo.test
new file mode 100644
index 0000000..0085efe
--- /dev/null
+++ b/tests/03-whitespaced-string.momo.test
@@ -0,0 +1 @@
+Token (2, 'This is a long ass string', ./tests/03-whitespaced-string.momo:3:0)
diff --git a/tests/04-identifier.momo b/tests/04-identifier.momo
new file mode 100644
index 0000000..dad80ff
--- /dev/null
+++ b/tests/04-identifier.momo
@@ -0,0 +1,2 @@
+
+   foo : string = "Value of foo";
diff --git a/tests/04-identifier.momo.test b/tests/04-identifier.momo.test
new file mode 100644
index 0000000..8eebbcc
--- /dev/null
+++ b/tests/04-identifier.momo.test
@@ -0,0 +1 @@
+Token (1, 'foo', ./tests/04-identifier.momo:2:3)