block_parser.py

import re
import string
from bisect import bisect_left
from typing import List, Match, Optional, Pattern, Set, Tuple

from .core import BlockState, Parser
from .helpers import (
    BLOCK_TAGS,
    HTML_ATTRIBUTES,
    HTML_TAGNAME,
    LINK_LABEL,
    PRE_TAGS,
    parse_link_href,
    parse_link_title,
    unescape_char,
)
from .list_parser import LIST_PATTERN, parse_list
from .util import escape_url, expand_leading_tab, expand_tab, unikey

DEFAULT_MAX_NESTED_LEVEL = 20

_INDENT_CODE_TRIM = re.compile(r"^ {1,4}", flags=re.M)
_ATX_HEADING_TRIM = re.compile(r"(\s+|^)#+\s*$")
_BLOCK_QUOTE_TRIM = re.compile(r"^ ?", flags=re.M)

_BLANK_TO_LINE = re.compile(r"[ \t]*\n")

_BLOCK_TAGS_PATTERN = "(" + "|".join(BLOCK_TAGS) + "|" + "|".join(PRE_TAGS) + ")"
_OPEN_TAG_END = re.compile(HTML_ATTRIBUTES + r"[ \t]*>[ \t]*(?:\n|$)")
_CLOSE_TAG_END = re.compile(r"[ \t]*>[ \t]*(?:\n|$)")
_BLOCK_QUOTE_LINE = re.compile(r"^ {0,3}>([^\n]*(?:\n|$))")


class BlockParser(Parser[BlockState]):
    state_cls = BlockState

    BLANK_LINE = re.compile(r"(^[ \t\v\f]*\n)+", re.M)

    RAW_HTML = (
        r"^ {0,3}("
        r"</?" + HTML_TAGNAME + r"|"
        r"<!--|"  # comment
        r"<\?|"  # script
        r"<![A-Z]|"
        r"<!\[CDATA\[)"
    )

    BLOCK_HTML = (
        r"^ {0,3}(?:"
        r"(?:</?" + _BLOCK_TAGS_PATTERN + r"(?:[ \t]+|\n|$))"
        r"|<!--"  # comment
        r"|<\?"  # script
        r"|<![A-Z]"
        r"|<!\[CDATA\[)"
    )

    SPECIFICATION = {
        "blank_line": r"(^[ \t\v\f]*\n)+",
        "atx_heading": r"^ {0,3}(?P<atx_1>#{1,6})(?!#+)(?P<atx_2>[ \t]*|[ \t]+.*?)$",
        "setex_heading": r"^ {0,3}(?P<setext_1>=|-){1,}[ \t]*$",
        "fenced_code": (
            r"^(?P<fenced_1> {0,3})(?P<fenced_2>`{3,}|~{3,})"
            r"[ \t]*(?P<fenced_3>.*?)$"
        ),
        "indent_code": (
            r"^(?: {4}| *\t)[^\n]+(?:\n+|$)"
            r"((?:(?: {4}| *\t)[^\n]+(?:\n+|$))|\s)*"
        ),
        "thematic_break": r"^ {0,3}((?:-[ \t]*){3,}|(?:_[ \t]*){3,}|(?:\*[ \t]*){3,})$",
        "ref_link": r"^ {0,3}\[(?P<reflink_1>" + LINK_LABEL + r")\]:",
        "block_quote": r"^ {0,3}>(?P<quote_1>.*?)$",
        "list": LIST_PATTERN,
        "block_html": BLOCK_HTML,
        "raw_html": RAW_HTML,
    }

    DEFAULT_RULES = (
        "fenced_code",
        "indent_code",
        "atx_heading",
        "setex_heading",
        "thematic_break",
        "block_quote",
        "list",
        "ref_link",
        "raw_html",
        "blank_line",
    )

    def __init__(
        self,
        block_quote_rules: Optional[List[str]] = None,
        list_rules: Optional[List[str]] = None,
        max_nested_level: int = DEFAULT_MAX_NESTED_LEVEL,
    ):
        super(BlockParser, self).__init__()

        if block_quote_rules is None:
            block_quote_rules = list(self.DEFAULT_RULES)

        if list_rules is None:
            list_rules = list(self.DEFAULT_RULES)

        self.block_quote_rules = block_quote_rules
        self.list_rules = list_rules
        self.max_nested_level = max_nested_level
        # register default parse methods
        self._methods = {name: getattr(self, "parse_" + name) for name in self.SPECIFICATION}

    def parse_blank_line(self, m: Match[str], state: BlockState) -> int:
        """Parse token for blank lines."""
        if state.cursor == 0:
            state.append_token({"type": "blank_line", "newlines": m.group(0).count("\n")})
        else:
            state.append_token({"type": "blank_line"})
        return m.end()

    def parse_thematic_break(self, m: Match[str], state: BlockState) -> int:
        """Parse token for thematic break, e.g. ``<hr>`` tag in HTML."""
        state.append_token({"type": "thematic_break"})
        # $ does not count '\n'
        return m.end() + 1

    def parse_indent_code(self, m: Match[str], state: BlockState) -> int:
        """Parse token for code block which is indented by 4 spaces."""
        # it is a part of the paragraph
        end_pos = state.append_paragraph()
        if end_pos:
            return end_pos

        code = m.group(0)
        end_pos = _trim_partial_next_line_indent(code, m.end())
        if end_pos != m.end():
            code = state.get_text(end_pos)
        code = expand_leading_tab(code)
        code = _INDENT_CODE_TRIM.sub("", code)
        code = code.strip("\n")
        state.append_token({"type": "block_code", "raw": code, "style": "indent"})
        return end_pos

    def parse_fenced_code(self, m: Match[str], state: BlockState) -> Optional[int]:
        """Parse token for fenced code block. A fenced code block is started with
        3 or more backtick(`) or tilde(~).

        An example of a fenced code block:

        .. code-block:: markdown

            ```python
            def markdown(text):
                return mistune.html(text)
            ```
        """
        spaces = m.group("fenced_1")
        marker = m.group("fenced_2")
        info = m.group("fenced_3")

        c = marker[0]
        if info and c == "`":
            # CommonMark Example 145
            # Info strings for backtick code blocks cannot contain backticks
            if info.find(c) != -1:
                return None

        _end = re.compile(r"^ {0,3}" + c + "{" + str(len(marker)) + r",}[ \t]*(?:\n|$)", re.M)
        cursor_start = m.end() + 1

        m2 = _end.search(state.src, cursor_start)
        if m2:
            code = state.src[cursor_start : m2.start()]
            end_pos = m2.end()
        else:
            code = state.src[cursor_start:]
            end_pos = state.cursor_max

        if spaces and code:
            _trim_pattern = re.compile("^ {0," + str(len(spaces)) + "}", re.M)
            code = _trim_pattern.sub("", code)

        token = {"type": "block_code", "raw": code, "style": "fenced", "marker": marker}
        if info:
            info = unescape_char(info)
            token["attrs"] = {"info": info.strip()}

        state.append_token(token)
        return end_pos

    def parse_atx_heading(self, m: Match[str], state: BlockState) -> int:
        """Parse token for ATX heading. An ATX heading is started with 1 to 6
        symbol of ``#``."""
        level = len(m.group("atx_1"))
        text = m.group("atx_2").strip(string.whitespace)
        # remove last #
        if text:
            text = _ATX_HEADING_TRIM.sub("", text)

        token = {"type": "heading", "text": text, "attrs": {"level": level}, "style": "atx"}
        state.append_token(token)
        return m.end() + 1

    def parse_setex_heading(self, m: Match[str], state: BlockState) -> Optional[int]:
        """Parse token for setex style heading. A setex heading syntax looks like:

        .. code-block:: markdown

            H1 title
            ========
        """
        if state.cursor in state.lazy_line_starts:
            return None

        last_token = state.last_token()
        if last_token and last_token["type"] == "paragraph":
            level = 1 if m.group("setext_1") == "=" else 2
            last_token["type"] = "heading"
            last_token["style"] = "setext"
            last_token["attrs"] = {"level": level}
            return m.end() + 1

        sc = self.compile_sc(["thematic_break", "list"])
        m2 = sc.match(state.src, state.cursor)
        if m2:
            return self.parse_method(m2, state)
        return None

    def parse_ref_link(self, m: Match[str], state: BlockState) -> Optional[int]:
        """Parse link references and save the link information into ``state.env``.

        Here is an example of a link reference:

        .. code-block:: markdown

            a [link][example]

            [example]: https://example.com "Optional title"

        This method will save the link reference into ``state.env`` as::

            state.env['ref_links']['example'] = {
                'url': 'https://example.com',
                'title': "Optional title",
            }
        """
        end_pos = state.append_paragraph()
        if end_pos:
            return end_pos

        label = m.group("reflink_1")
        key = unikey(label)
        if not key:
            return None

        href, href_pos = parse_link_href(state.src, m.end(), block=True)
        if href is None:
            return None

        assert href_pos is not None

        blank_pos = _find_next_blank_line(state, href_pos, self.BLANK_LINE)
        if blank_pos is None:
            max_pos = state.cursor_max
        else:
            max_pos = blank_pos

        title, title_pos = parse_link_title(state.src, href_pos, max_pos)
        if title_pos:
            m2 = _BLANK_TO_LINE.match(state.src, title_pos)
            if m2:
                title_pos = m2.end()
            else:
                title_pos = None
                title = None

        if title_pos is None:
            m3 = _BLANK_TO_LINE.match(state.src, href_pos)
            if m3:
                href_pos = m3.end()
            else:
                href_pos = None
                href = None

        end_pos = title_pos or href_pos
        if not end_pos:
            return None

        if key not in state.env["ref_links"]:
            assert href is not None
            href = unescape_char(href)
            data = {"url": escape_url(href), "label": label}
            if title:
                data["title"] = title
            state.env["ref_links"][key] = data
        return end_pos

    def extract_block_quote(self, m: Match[str], state: BlockState) -> Tuple[str, Optional[int], Set[int]]:
        """Extract text and cursor end position of a block quote."""

        text = _parse_block_quote_line(state.get_line(state.cursor))
        assert text is not None
        lazy_line_starts: Set[int] = set()

        sc = self.compile_sc(["blank_line", "indent_code", "fenced_code"])
        require_marker = bool(sc.match(text))

        state.cursor += len(state.get_line(state.cursor))

        end_pos: Optional[int] = None
        if require_marker:
            while state.cursor < state.cursor_max:
                quote = _parse_block_quote_line(state.get_line(state.cursor))
                if quote is None:
                    break
                text += quote
                state.cursor += len(state.get_line(state.cursor))
        else:
            prev_blank_line = False
            break_sc = self.compile_sc(
                [
                    "blank_line",
                    "thematic_break",
                    "fenced_code",
                    "list",
                    "block_html",
                ]
            )
            while state.cursor < state.cursor_max:
                quote = _parse_block_quote_line(state.get_line(state.cursor))
                if quote is not None:
                    text += quote
                    state.cursor += len(state.get_line(state.cursor))
                    if not quote.strip():
                        prev_blank_line = True
                    else:
                        prev_blank_line = False
                    continue

                if prev_blank_line:
                    # CommonMark Example 249
                    # because of laziness, a blank line is needed between
                    # a block quote and a following paragraph
                    break

                m4 = break_sc.match(state.src, state.cursor)
                if m4:
                    end_pos = self.parse_method(m4, state)
                    if end_pos:
                        break

                # lazy continuation line
                line = state.get_line(state.cursor)
                lazy_line_starts.add(len(text))
                text += expand_leading_tab(line, 3)
                state.cursor += len(line)

        # according to CommonMark Example 6, the second tab should be
        # treated as 4 spaces
        return expand_tab(text), end_pos, lazy_line_starts

    def parse_block_quote(self, m: Match[str], state: BlockState) -> int:
        """Parse token for block quote. Here is an example of the syntax:

        .. code-block:: markdown

            > a block quote starts
            > with right arrows
        """
        text, end_pos, lazy_line_starts = self.extract_block_quote(m, state)
        # scan children state
        child = state.child_state(text, lazy_line_starts=lazy_line_starts)
        if state.depth() >= self.max_nested_level - 1:
            # At the nesting limit, stop descending into any further container
            # blocks. Trimming only "block_quote" still allowed block quotes and
            # lists to recurse into each other without bound (RecursionError).
            rules = [rule for rule in self.block_quote_rules if rule not in ("block_quote", "list")]
        else:
            rules = self.block_quote_rules

        self.parse(child, rules)
        token = {"type": "block_quote", "children": child.tokens}
        if end_pos:
            state.prepend_token(token)
            return end_pos
        state.append_token(token)
        return state.cursor

    def parse_list(self, m: Match[str], state: BlockState) -> int:
        """Parse tokens for ordered and unordered list."""
        return parse_list(self, m, state)

    def parse_block_html(self, m: Match[str], state: BlockState) -> Optional[int]:
        return self.parse_raw_html(m, state)

    def parse_raw_html(self, m: Match[str], state: BlockState) -> Optional[int]:
        marker = m.group(0).strip()

        # rule 2
        if marker == "<!--":
            return _parse_html_to_end(state, "-->", m.end())

        # rule 3
        if marker == "<?":
            return _parse_html_to_end(state, "?>", m.end())

        # rule 5
        if marker == "<![CDATA[":
            return _parse_html_to_end(state, "]]>", m.end())

        # rule 4
        if marker.startswith("<!"):
            return _parse_html_to_end(state, ">", m.end())

        close_tag = None
        open_tag = None
        if marker.startswith("</"):
            close_tag = marker[2:].lower()
            # rule 6
            if close_tag in BLOCK_TAGS:
                return _parse_html_to_newline(state, self.BLANK_LINE)
        else:
            open_tag = marker[1:].lower()
            # rule 1
            if open_tag in PRE_TAGS:
                end_tag = "</" + open_tag + ">"
                return _parse_html_to_end(state, end_tag, m.end())
            # rule 6
            if open_tag in BLOCK_TAGS:
                return _parse_html_to_newline(state, self.BLANK_LINE)

        # Blocks of type 7 may not interrupt a paragraph.
        end_pos = state.append_paragraph()
        if end_pos:
            return end_pos

        # rule 7
        start_pos = m.end()
        end_pos = state.find_line_end()
        if (open_tag and _OPEN_TAG_END.match(state.src, start_pos, end_pos)) or (
            close_tag and _CLOSE_TAG_END.match(state.src, start_pos, end_pos)
        ):
            return _parse_html_to_newline(state, self.BLANK_LINE)

        return None

    def parse(self, state: BlockState, rules: Optional[List[str]] = None) -> None:
        if rules is None:
            rules = self.rules
        sc = self.compile_sc(rules)

        while state.cursor < state.cursor_max:
            if "indent_code" in rules:
                end_pos = self._parse_indent_code_fast(state)
                if end_pos is not None:
                    state.cursor = end_pos
                    continue

            m = sc.match(state.src, state.cursor)
            if not m and self._parse_plain_paragraph(state, sc):
                continue

            if not m:
                m = sc.search(state.src, state.cursor)
            if not m:
                break

            end_pos = m.start()
            if end_pos > state.cursor:
                text = state.get_text(end_pos)
                state.add_paragraph(text)
                state.cursor = end_pos

            end_pos2 = self.parse_method(m, state)
            if end_pos2:
                state.cursor = end_pos2
            else:
                end_pos3 = state.find_line_end()
                text = state.get_text(end_pos3)
                state.add_paragraph(text)
                state.cursor = end_pos3

        if state.cursor < state.cursor_max:
            text = state.src[state.cursor :]
            state.add_paragraph(text)
            state.cursor = state.cursor_max

    def _parse_indent_code_fast(self, state: BlockState) -> Optional[int]:
        line_end = state.find_line_end()
        line = state.get_text(line_end)
        if not _is_indent_code_line(line):
            return None

        if state.last_token() and state.last_token()["type"] == "paragraph":
            return state.append_paragraph()

        end_pos = _find_indent_code_end(state.src, line_end, state.cursor_max)
        code = state.get_text(end_pos)
        code = expand_leading_tab(code)
        code = _INDENT_CODE_TRIM.sub("", code)
        code = code.strip("\n")
        state.append_token({"type": "block_code", "raw": code, "style": "indent"})
        return end_pos

    def _parse_plain_paragraph(self, state: BlockState, sc: Pattern[str]) -> bool:
        if not _is_plain_paragraph_start(state.src, state.cursor):
            return False

        pos = state.cursor
        while pos < state.cursor_max:
            if pos > state.cursor and sc.match(state.src, pos):
                break

            line = state.get_line(pos)
            if not line.strip():
                break

            pos += len(line)

        if pos <= state.cursor:
            return False

        state.add_paragraph(state.get_text(pos))
        state.cursor = pos
        return True


def _parse_html_to_end(state: BlockState, end_marker: str, start_pos: int) -> int:
    marker_pos = state.src.find(end_marker, start_pos)
    if marker_pos == -1:
        text = state.src[state.cursor :]
        end_pos = state.cursor_max
    else:
        text = state.get_text(marker_pos)
        state.cursor = marker_pos
        end_pos = state.find_line_end()
        text += state.get_text(end_pos)

    state.append_token({"type": "block_html", "raw": text})
    return end_pos


def _parse_html_to_newline(state: BlockState, newline: Pattern[str]) -> int:
    m = newline.search(state.src, state.cursor)
    if m:
        end_pos = m.start()
        text = state.get_text(end_pos)
    else:
        text = state.src[state.cursor :]
        end_pos = state.cursor_max

    state.append_token({"type": "block_html", "raw": text})
    return end_pos


def _parse_block_quote_line(line: str) -> Optional[str]:
    m = _BLOCK_QUOTE_LINE.match(line)
    if not m:
        return None
    text = expand_leading_tab(m.group(1), 3)
    return _BLOCK_QUOTE_TRIM.sub("", text)


def _find_next_blank_line(state: BlockState, pos: int, pattern: Pattern[str]) -> Optional[int]:
    cache = state.env.get("__blank_line_starts__")
    if cache is None or cache[0] is not state.src:
        cache = (state.src, [m.start() for m in pattern.finditer(state.src)])
        state.env["__blank_line_starts__"] = cache

    positions = cache[1]
    index = bisect_left(positions, pos)
    if index < len(positions):
        return positions[index]
    return None


def _trim_partial_next_line_indent(text: str, end_pos: int) -> int:
    line_start = text.rfind("\n") + 1
    if line_start == 0:
        return end_pos

    suffix = text[line_start:]
    if suffix and suffix.strip(" \t") == "" and len(suffix.expandtabs(4)) < 4:
        return end_pos - len(suffix)
    return end_pos


def _is_indent_code_line(line: str) -> bool:
    if line.endswith("\n"):
        line = line[:-1]

    if len(line) > 4 and line.startswith("    "):
        return True

    index = 0
    while index < len(line) and line[index] == " ":
        index += 1
    return index + 1 < len(line) and line[index] == "\t"


def _find_indent_code_end(src: str, start: int, end: int) -> int:
    pos = start
    while pos < end:
        newline = src.find("\n", pos, end)
        line_end = end if newline < 0 else newline + 1
        line = src[pos:line_end]
        if _is_indent_code_line(line) or not line.strip():
            pos = line_end
            continue
        break
    return pos


def _is_plain_paragraph_start(src: str, pos: int) -> bool:
    if pos >= len(src):
        return False
    c = src[pos]
    return not c.isspace() and not c.isdigit() and c not in string.punctuation