#!/usr/bin/env python3
"""Reproduces the 154-sonnet letter-skip (ELS) null-test in cipher-arguments.html
on shakespeare-enquiry.org. Reads sonnets-1041.txt (Project Gutenberg eBook #1041,
154 sonnets, poem text only, editorial numbering ignored), reduces it to A-Z, and
runs a single-pass equidistant-letter search (both directions, no wrap-around)."""
import re, os
from collections import Counter

TEXT = os.path.join(os.path.dirname(__file__), "sonnets-1041.txt")
L = re.sub(r'[^A-Za-z]', '', open(TEXT, encoding='utf-8').read()).upper()
N = len(L)

def min_skip(target):
    m = len(target); rt = target[::-1]
    for s in range(1, (N-1)//(m-1) + 1):
        for r in range(s):
            line = L[r::s]
            if target in line or rt in line:
                return s
    return None

def present(word):  # appears as an ELS at some skip <=120, either direction
    for s in range(1, 121):
        for r in range(s):
            line = L[r::s]
            if word in line or word[::-1] in line:
                return True
    return False

if __name__ == "__main__":
    c = Counter(L)
    print(f"letters: {N}   X={c['X']}  A={c['A']}  S={c['S']}")
    for t in ["VERE","HENRY","DEVERE","BASSANO","EMILIA","LANIER","BACON",
              "MARLOWE","NIXON","OBAMA","ELVIS","OXFORD","WRIOTHESLEY"]:
        print(f"  {t:12} {min_skip(t)}")
    S = set()
    for s in range(1, 121):
        for r in range(s):
            line = L[r::s]
            for i in range(len(line)-3):
                w = line[i:i+4]; S.add(w); S.add(w[::-1])
    print(f"distinct 4-letter strings (skip<=120, both dir): {len(S)} of 456976 ({100*len(S)/456976:.1f}%)")
    print("examples present:", {w: present(w) for w in ["VERE","BASS","FAPP","MECK"]})