path: root/domllib.py

#!/usr/bin/env python
#
# Copyright 2026 David Vazgenovich Shakaryan

import re
import copy


WS = set(' \t')
WSNL = {*WS, '\n'}
ESCAPES = {
    '\n': '',
    '\\': '\\',
    '"': '"',
    'n': '\n',
    't': '\t',
}

RE_IDENT = re.compile(r'[A-Za-z0-9_]+')
RE_NUM = re.compile(
    r'[+-]?(?:'
        r'0x[0-9a-fA-F]+(?:_[0-9a-fA-F]+)*|'
        r'0o[0-7]+(?:_[0-7]+)*|'
        r'0b[01]+(?:_[01]+)*|'
        r'(?:0|[1-9][0-9]*(?:_[0-9]+)*)'
            r'(\.[0-9]+(?:_[0-9]+)*(?:[eE][+-]?[0-9]+(?:_[0-9]+)*)?|'
                r'[eE][+-]?[0-9]+(?:_[0-9]+)*)?|'
        r'(inf|nan)'
    r')')


def _raise_unexpected_eof(ctx):
    raise SyntaxError(f'Unexpected end of input while parsing {ctx}') from None


def _raise_expected_char(char, pos):
    if type(char) is tuple:
        s = ' or '.join(repr(c) for c in char)
    else:
        s = repr(char)
    raise SyntaxError(f'Expected {s} at position {pos}') from None


def _raise_unexpected_char(char, pos, ctx=None):
    ctx = f' {ctx}' if ctx else ''
    raise SyntaxError(
        f'Unexpected {repr(char)}{ctx} at position {pos}') from None


def _raise_invalid_escape(seq, pos):
    seq = ''.join(
        '\\' if (r := repr(c)[1:-1]) == '\\\\'
            else r if len(r) == 1
            else f'<{r}>'
        for c in seq)
    raise SyntaxError(f'Invalid escape sequence {seq} at pos {pos}') from None


def skip_chars(src, pos, chars, skip_comments=False):
    try:
        while True:
            char = src[pos]
            if skip_comments and char == '#':
                while src[pos] != '\n':
                    pos += 1
                continue
            if char == '\\':
                try:
                    if src[pos+1] != '\n':
                        _raise_expected_char('\n', pos + 1)
                except IndexError:
                    _raise_unexpected_eof('line continuation')
                pos += 2
                continue
            if char not in chars:
                break
            pos += 1
    except IndexError:
        pass
    return pos


def skip(src, pos, depth):
    if depth:
        return skip_chars(src, pos, WSNL, skip_comments=True)
    return skip_chars(src, pos, WS)


def ensure_path(d, path):
    curr = d
    for n in path:
        if type(curr) is not dict:
            raise SyntaxError(
                f'Cannot resolve {repr(n)} in {type(curr).__name__}')
        curr = curr.setdefault(n, {})
    return curr


def dict_unpack(d, val):
    if type(val) is not dict:
        raise SyntaxError(f'Cannot dict-unpack {type(val).__name__}')
    for k, v in val.items():
        d[k] = v


def dict_assign(d, path, val):
    tgt = ensure_path(d, path[:-1])
    if type(tgt) is not dict:
        raise SyntaxError(
            f"Cannot assign key '{path[-1]}' in {type(tgt).__name__}")
    tgt[path[-1]] = val


def dict_delete(d, path):
    try:
        for n in path[:-1]:
            d = d[n]
        del d[path[-1]]
    except KeyError, TypeError:
        return


def parse_dict_unpack(src, pos, depth, root, sect):
    try:
        if src[pos+1] == '*':
            pos += 2
        else:
            _raise_expected_char('*', pos + 1)
    except IndexError:
        _raise_unexpected_eof('dict unpack')
    pos = skip(src, pos, depth)
    return parse_value(src, pos, depth, root, sect)


def parse_sequence(src, pos, depth, delim, term, on_item):
    pos += 1
    depth += 1

    while True:
        pos = skip(src, pos, depth)

        try:
            char = src[pos]
        except IndexError:
            _raise_unexpected_eof('sequence')

        if char == term:
            return pos + 1

        pos = on_item(src, pos, depth)
        pos = skip(src, pos, depth)

        try:
            char = src[pos]
            if char == delim:
                pos += 1
            elif char != term:
                _raise_expected_char((delim, term), pos)
        except IndexError:
            _raise_unexpected_eof('sequence')


def parse_inline_list(src, pos, depth, root, sect):
    res = []

    def on_item(src, pos, depth):
        unpack = False
        if src[pos] == '*':
            unpack = True
            pos = skip(src, pos + 1, depth)

        pos, val = parse_value(src, pos, depth, root, sect)
        if unpack:
            if type(val) is not list:
                raise SyntaxError(f'Cannot list-unpack {type(val).__name__}')
            res.extend(val)
        else:
            res.append(val)
        return pos

    pos = parse_sequence(src, pos, depth, ',', ']', on_item)
    return pos, res


def parse_inline_dict(src, pos, depth, root, sect):
    res = {}

    def on_item(src, pos, depth):
        char = src[pos]
        if char == '*':
            pos, val = parse_dict_unpack(src, pos, depth, root, sect)
            dict_unpack(res, val)
        else:
            pos, key, val = parse_assign(src, pos, depth, root, sect)
            dict_assign(res, key, val)
        return pos

    pos = parse_sequence(src, pos, depth, ',', '}', on_item)
    return pos, res


def parse_str(src, pos, multiline=False):
    try:
        if multiline:
            pos += 3
            if src[pos] == '\n':
                pos += 1
        else:
            pos += 1

        start = pos
        res = ''

        while True:
            curr = src[pos]
            if curr == '"' and (not multiline or src.startswith('"""', pos)):
                res += src[start:pos]
                break

            if curr == '\\':
                res += src[start:pos]
                nxt = src[pos+1]

                if nxt == 'u':
                    if src[pos+2] != '{':
                        _raise_expected_char('{', pos + 2)
                    end = src.find('}', pos + 3)
                    if end == -1:
                        raise IndexError
                    code = src[pos+3:end]
                    try:
                        code_int = int(code, 16)
                        if not (0 <= code_int <= 0xd7ff or
                                0xe000 <= code_int <= 0x10ffff):
                            _raise_invalid_escape(f'\\u{{{code}}}', pos)
                        res += chr(code_int)
                    except ValueError:
                        _raise_invalid_escape(f'\\u{{{code}}}', pos)
                    pos = end + 1

                else:
                    try:
                        res += ESCAPES[nxt]
                    except KeyError:
                        _raise_invalid_escape(f'\\{nxt}', pos)
                    pos += 2

                start = pos
                continue

            code = ord(curr)
            if ((code < 32 and (not multiline or code != 10 and code != 9)) or
                    code == 127):
                _raise_unexpected_char(curr, pos, ctx='in string')
            pos += 1
    except IndexError:
        _raise_unexpected_eof('string')
    return pos + (3 if multiline else 1), res


def parse_raw_str(src, pos, multiline=False):
    try:
        if multiline:
            pos += 3
            if src[pos] == '\n':
                pos += 1
        else:
            pos += 1

        start = pos

        while (src[pos] != "'" or
                (multiline and not src.startswith("'''", pos))):
            code = ord(src[pos])
            if ((code < 32 and (not multiline or code != 10 and code != 9)) or
                    code == 127):
                _raise_unexpected_char(src[pos], pos, ctx='in raw string')
            pos += 1
    except IndexError:
        _raise_unexpected_eof('raw string')
    return pos + (3 if multiline else 1), src[start:pos]


def parse_literal(src, pos, depth, root, sect):
    char = src[pos] # guaranteed to exist by parse_value()

    if char == "'":
        return parse_raw_str(src, pos, multiline=src.startswith("'''", pos))
    if char == '"':
        return parse_str(src, pos, multiline=src.startswith('"""', pos))

    if char == 't':
        if src.startswith('true', pos):
            return pos + 4, True
    if char == 'f':
        if src.startswith('false', pos):
            return pos + 5, False
    if char == 'n':
        if src.startswith('null', pos):
            return pos + 4, None

    if char == '{':
        return parse_inline_dict(src, pos, depth, root, sect)
    if char == '[':
        return parse_inline_list(src, pos, depth, root, sect)

    if (m := RE_NUM.match(src, pos)):
        return m.end(), float(m[0]) if m.lastindex else int(m[0], 0)

    _raise_unexpected_char(char, pos)


def parse_value(src, pos, depth, root, sect):
    try:
        char = src[pos]
    except IndexError:
        _raise_unexpected_eof('value')

    ref = root if char == '$' else sect if char == '@' else None

    if ref is None:
        pos, val = parse_literal(src, pos, depth, root, sect)

        if type(val) is str:
            while True:
                pos = skip(src, pos, depth)
                try:
                    char = src[pos]
                except IndexError:
                    break
                if char == "'":
                    pos, v = parse_raw_str(
                        src, pos, multiline=src.startswith("'''", pos))
                    val += v
                elif char == '"':
                    pos, v = parse_str(
                        src, pos, multiline=src.startswith('"""', pos))
                    val += v
                else:
                    break
    else:
        pos = skip(src, pos + 1, depth)
        pos, key = parse_key_segment(src, pos)
        try:
            val = ref[key]
        except KeyError:
            if ref is root:
                raise SyntaxError(
                    f'Invalid global reference {repr(key)}') from None
            else:
                raise SyntaxError(
                    f'Invalid local reference {repr(key)}') from None

    pos, val = parse_selectors(src, pos, depth, root, sect, val, copy_=ref)

    return pos, val


def parse_selectors(src, pos, depth, root, sect, val, copy_=False):
    while True:
        pos = skip(src, pos, depth)
        try:
            char = src[pos]
        except IndexError:
            break

        if char == '.':
            pos = skip(src, pos + 1, depth)
            pos, key = parse_key_segment(src, pos)

        elif char == '[':
            depth += 1
            pos = skip(src, pos + 1, depth)
            try:
                nxt = src[pos]
            except IndexError:
                _raise_unexpected_eof('selector')
            continue_ = False

            if nxt == '^':
                pos = skip(src, pos + 1, depth)
                def on_item(src, pos, depth):
                    nonlocal val, copy_
                    pos, key = parse_key(src, pos, depth)
                    if copy_:
                        val = copy.deepcopy(val)
                        copy_ = False
                    dict_delete(val, key)
                    return pos
                if src[pos] == '{':
                    pos = parse_sequence(src, pos, depth, ',', '}', on_item)
                else:
                    pos = on_item(src, pos, depth)
                continue_ = True

            elif nxt == '&':
                pos = skip(src, pos + 1, depth)
                nval = {}
                def on_item(src, pos, depth):
                    pos, key = parse_key(src, pos, depth)
                    v = val
                    for n in key:
                        try:
                            v = v[n]
                        except KeyError, TypeError:
                            return pos
                    dict_assign(nval, key, v)
                    return pos
                if src[pos] == '{':
                    pos = parse_sequence(src, pos, depth, ',', '}', on_item)
                else:
                    pos = on_item(src, pos, depth)
                val = nval
                continue_ = True

            else:
                if nxt == ':':
                    key = None
                else:
                    pos, key = parse_value(src, pos, depth, root, sect)
                pos = skip(src, pos, depth)
                try:
                    nxt = src[pos]
                except IndexError:
                    _raise_unexpected_eof('selector')
                if nxt == ':':
                    pos = skip(src, pos + 1, depth)
                    try:
                        nxt = src[pos]
                    except IndexError:
                        _raise_unexpected_eof('selector')
                    if nxt == ']':
                        stop = None
                    else:
                        pos, stop = parse_value(src, pos, depth, root, sect)
                    key = slice(key, stop)

            pos = skip(src, pos, depth)
            try:
                if src[pos] != ']':
                    _raise_expected_char(']', pos)
            except IndexError:
                _raise_unexpected_eof('selector')
            pos += 1

            depth -= 1
            if continue_:
                continue

        else:
            break

        try:
            val = val[key]
        except KeyError, IndexError, TypeError:
            raise SyntaxError(
                f'Cannot resolve {repr(key)} in {type(val).__name__}'
            ) from None

    return pos, copy.deepcopy(val) if copy_ else val


def parse_key_segment(src, pos):
    try:
        char = src[pos]
    except IndexError:
        _raise_unexpected_eof('key segment')

    if char == "'":
        return parse_raw_str(src, pos)
    if char == '"':
        return parse_str(src, pos)

    if (m := RE_IDENT.match(src, pos)):
        return m.end(), m[0]
    else:
        raise SyntaxError(f'Invalid key identifier at position {pos}')


def parse_key(src, pos, depth):
    pos, seg = parse_key_segment(src, pos)
    key = (seg,)
    while True:
        pos = skip(src, pos, depth)
        try:
            char = src[pos]
        except IndexError:
            return pos, key
        if char != '.':
            return pos, key
        pos = skip(src, pos + 1, depth)
        pos, seg = parse_key_segment(src, pos)
        key += (seg,)


def parse_assign(src, pos, depth, root, sect):
    pos, key = parse_key(src, pos, depth)

    pos = skip(src, pos, depth)
    try:
        char = src[pos]
    except IndexError:
        char = None
    if char != '=':
        _raise_expected_char('=', pos)
    pos += 1

    pos = skip(src, pos, depth)
    pos, val = parse_value(src, pos, depth, root, sect)

    return pos, key, val


def parse_dict_header(src, pos, depth, root):
    depth += 1
    pos = skip(src, pos + 1, depth)
    try:
        if src[pos] == ']':
            return pos + 1, root
    except IndexError:
        _raise_unexpected_eof('header')

    pos, key = parse_key(src, pos, depth)
    try:
        if src[pos] != ']':
            _raise_expected_char(']', pos)
    except IndexError:
        _raise_unexpected_eof('header')

    sect = ensure_path(root, key)
    if type(sect) is not dict:
        raise SyntaxError(f'Cannot enter {type(sect).__name__} context: {key}')
    return pos + 1, sect


def parse_list_header(src, pos, depth, root):
    depth += 1
    pos = skip(src, pos + 2, depth)
    pos, key = parse_key(src, pos, depth)
    if not src.startswith(']]', pos):
        _raise_expected_char(']]', pos)

    if len(key) > 1:
        d = ensure_path(root, key[:-1])
        if type(d) is not dict:
            raise SyntaxError(
                f'Cannot resolve {repr(key[-1])} in {type(d).__name__}')
    else:
        d = root

    l = d.setdefault(key[-1], [])
    if type(l) is not list:
        raise SyntaxError(f'Cannot append to {type(l).__name__}: {key}')
    sect = {}
    l.append(sect)
    return pos + 2, sect


def loads(src):
    root = {}
    sect = root
    pos = 0

    while True:
        pos = skip_chars(src, pos, WSNL, skip_comments=True)

        try:
            char = src[pos]
        except IndexError:
            break

        if char == '[':
            try:
                char = src[pos+1]
            except IndexError:
                _raise_unexpected_eof('header')

            if char == '[':
                pos, sect = parse_list_header(src, pos, 0, root)
            else:
                pos, sect = parse_dict_header(src, pos, 0, root)
        elif char == '*':
            pos, val = parse_dict_unpack(src, pos, 0, root, sect)
            dict_unpack(sect, val)
        else:
            pos, key, val = parse_assign(src, pos, 0, root, sect)
            dict_assign(sect, key, val)

        pos = skip_chars(src, pos, WS, skip_comments=True)
        try:
            if src[pos] != '\n':
                _raise_expected_char('\n', pos)
        except IndexError:
            pass

    return root


def load(f):
    b = f.read()
    try:
        src = b.decode()
    except AttributeError:
        raise TypeError('File must be opened in binary mode')
    return loads(src)