"""A picture's content (see reading.py): the content the most scanners return, raw bytes and the text a scanner makes of them counting as one content, and the content as the page may show it, with digests that pair pictures of one content whichever scanners read them. """ from __future__ import annotations import hashlib def _texts(p: bytes) -> set[bytes]: """In UTF-8, the texts a scanner may make of raw bytes that aren't UTF-8: ISO-8859-1 (QR byte mode's default) and, where they decode, Shift JIS (QR kanji mode).""" try: p.decode("utf-8") return set() except UnicodeDecodeError: pass out = {p.decode("latin-1").encode("utf-8")} try: out.add(p.decode("shift_jis").encode("utf-8")) except UnicodeDecodeError: pass return out def _same(p: bytes, q: bytes) -> bool: """One content: the same bytes, or raw bytes and a text made of them.""" return p == q or p in _texts(q) or q in _texts(p) def _is_text(p: bytes) -> bool: try: p.decode("utf-8") return True except UnicodeDecodeError: return False def _consensus(contents: dict[tuple[str, int], list[bytes]]) -> bytes | None: forms = {p for ps in contents.values() for p in ps} if not forms: return None def votes(c: bytes) -> tuple[int, int]: hits = [(s, k) for (s, k), ps in contents.items() if any(_same(p, c) for p in ps)] return len({s for s, _ in hits}), len(hits) return max(forms, key=lambda c: (*votes(c), _is_text(c), -len(c), c)) def _raws(p: bytes) -> list[bytes]: """Every set of raw bytes a content stands for: p, when p is raw bytes already (not UTF-8); else, in this order, the ISO-8859-1 and the Shift JIS bytes of which p is a text a scanner made, or p itself when it is neither. A text with only ISO-8859-1 characters ("25°C") can be made of either, so it stands for both.""" try: text = p.decode("utf-8") except UnicodeDecodeError: return [p] # raw bytes already out = [] for encoding in ("latin-1", "shift_jis"): try: raw = text.encode(encoding) except UnicodeEncodeError: continue if raw != p and p in _texts(raw): out.append(raw) return out or [p] def _raw(p: bytes) -> bytes: """The raw bytes content.sha256 is taken from: the first of _raws, so one content (_same) has one digest whichever scanners read it, but for a text that stands for two byte strings, which content.sha256s covers.""" return _raws(p)[0] def describe(content: bytes | None) -> dict | None: """The content as the page may show it: printable text (UTF-8, or else ISO-8859-1), otherwise only its length; its SHA-256 as one content's raw bytes (_raw); and "sha256s", the sorted SHA-256s of every set of raw bytes it stands for (_raws), which two pictures of one content always share. Only the visitor's browser receives them, to pair pictures.""" if content is None: return None digest = hashlib.sha256(_raw(content)).hexdigest() digests = sorted({hashlib.sha256(r).hexdigest() for r in _raws(content)}) # pairing: one content shares one try: text = content.decode("utf-8") except UnicodeDecodeError: text = content.decode("latin-1") # QR byte mode's default; binary content shows control characters if all(c.isprintable() or c in "\r\n\t" for c in text): return {"text": text[:300], "cut": len(text) > 300, "bytes": len(content), "sha256": digest, "sha256s": digests} return {"text": None, "cut": False, "bytes": len(content), "sha256": digest, "sha256s": digests}