pyrom/app/lib/babycode_parser.py

# originally written in lua by kaesa

import re

PAT_EMOTE = r"[^\s:]"
PAT_BBCODE_TAG = r"\w"
PAT_BBCODE_ATTR = r"[^\s\]]"
PAT_LINK = r"https?:\/\/[\w\-_.?:\/=&~@#%]+[\w\-\/]"

class Parser:
    def __init__(self, src_str):
        self.valid_bbcode_tags = []
        self.valid_emotes = []
        self.bbcode_tags_only_text_children = [],
        self.source = src_str
        self.position = 0
        self.position_stack = []


    def advance(self, count = 1):
        self.position += count


    def is_end_of_source(self, offset = 0):
        return self.position + offset >= len(self.source)


    def save_position(self):
        self.position_stack.append(self.position)


    def restore_position(self):
        self.position = self.position_stack.pop()


    def forget_position(self):
        self.position_stack.pop()


    def peek_char(self, offset = 0):
        if self.is_end_of_source(offset):
            return ""
        return self.source[self.position + offset]


    def get_char(self):
        char = self.peek_char()
        self.advance()
        return char


    def check_char(self, wanted):
        char = self.peek_char()

        if char == wanted:
            self.advance()
            return True

        return False


    def check_str(self, wanted):
        self.save_position()

        # for each char in wanted
        for i in range(len(wanted)):
            if not self.check_char(wanted[i]):
                self.restore_position()
                return False

        self.forget_position()
        return True


    def match_pattern(self, pattern):
        buf = ""
        while not self.is_end_of_source():
            ch = self.peek_char()

            if not re.match(pattern, ch):
                break

            self.advance()
            buf = buf + ch

        return buf


    def parse_emote(self):
        self.save_position()

        if not self.check_char(":"):
            self.restore_position()
            return None

        name = self.match_pattern(PAT_EMOTE)

        if not self.check_char(":"):
            self.restore_position()
            return None

        if not name in self.valid_emotes:
            self.restore_position()
            return None

        self.forget_position()
        return {
            "type": "emote",
            "name": name
        }


    def parse_bbcode_open(self):
        self.save_position()

        if not self.check_char("["):
            self.restore_position()
            return None, None

        name = self.match_pattern(PAT_BBCODE_TAG)

        if name == "":
            self.restore_position()
            return None, None

        attr = None

        if self.check_char("="):
            attr = self.match_pattern(PAT_BBCODE_ATTR)

        if not self.check_char("]"):
            self.restore_position()
            return None, None

        if not name in self.valid_bbcode_tags:
            self.restore_position()
            return None, None

        self.forget_position()
        return name, attr


    def parse_bbcode(self):
        self.save_position()

        name, attr = self.parse_bbcode_open()

        if name is None:
            self.restore_position()
            return None

        children = []

        while not self.is_end_of_source():
            if self.check_str(f"[/{name}]"):
                break

            if name in self.bbcode_tags_only_text_children:
                ch = self.get_char()

                if len(children) == 0:
                    children.append(ch)
                else:
                    children[0] = children[0] + ch
            else:
                element = self.parse_element(children)

                if element is None:
                    self.restore_position()
                    return None

                children.append(element)

        self.forget_position()
        return {
            "type": "bbcode",
            "name": name,
            "attr": attr,
            "children": children,
        }


    def parse_rule(self):
        if not self.check_str("---"):
            return None

        return {
            "type": "rule"
        }


    def parse_link(self):
        self.save_position()

        # extract printable chars (extreme hack edition)
        word = self.match_pattern(r'[ -~]')

        if not re.match(PAT_LINK, word):
            self.restore_position()
            return None

        self.forget_position()
        return {
            "type": "link",
            "url": word
        }


    def parse_element(self, siblings):
        if self.is_end_of_source():
            return None

        element = self.parse_emote() \
            or self.parse_bbcode() \
            or self.parse_rule() \
            or self.parse_link()

        if element is None:
            if len(siblings) > 0:
                last = siblings[-1]

                if isinstance(last, str):
                    siblings.pop()
                    return last + self.get_char()

            return self.get_char()

        return element


    def parse(self):
        elements = []

        while True:
            element = self.parse_element(elements)
            if element is None:
                break

            elements.append(element)

        return elements