先数一数句子

（每道题开头都有同一段：T() 是上一站 lexer 的替身——源码里 token 之间用空格分开，它切成 (kind, text, line, col)；show() 把树写成一行：二元运算全加括号，语句之间用 |。）

贯穿的源码（接 compiler_02 那段）：

let rate = 12 ;
let msg = "hi" ;
total = rate >= 3 ;

还没写 parser，先用最土的办法：数 token 数、数分号（每句一个分号）。

KEYWORDS = ("let", "if", "else", "while")
CMP = (">=", "<=", "==", "!=", "<", ">")

def T(src):
    toks = []
    for ln, line in enumerate(src.split("\n"), 1):
        pos = 0
        for w in line.split():
            pos = line.index(w, pos)
            if w in KEYWORDS:
                k = "KEYWORD"
            elif w[0] == '"':
                k = "STRING"
            elif w.isdigit():
                k = "NUMBER"
            elif w[0].isalpha() or w[0] == "_":
                k = "IDENT"
            else:
                k = "OP"
            toks.append((k, w, ln, pos + 1))
            pos += len(w)
    toks.append(("EOF", "", ln, pos + 1))
    return toks

SRC = 'let rate = 12 ;\nlet msg = "hi" ;\ntotal = rate >= 3 ;'
toks = T(SRC)
n = len(toks) - 1
semis = sum(1 for t in toks if t[1] == ";")
print(str(n) + "/" + str(semis))