call-to-text/old/parsers/novofon.py

import re
from urllib.parse import unquote, urlparse
from typing import Optional

from .base import CallMetadata, BaseEmailParser


NOVOFON_URL_PATTERN = re.compile(
    r"(https://(?:my|app)\.novofon\.ru/system/media/talk/\d+/[a-f0-9]+[^\s<>\"']*)"
)


class NovofonParser(BaseEmailParser):
    def extract_audio_url(self, body_text: str, body_html: str) -> Optional[str]:
        for source in (body_text, body_html):
            if not source:
                continue
            m = NOVOFON_URL_PATTERN.search(source)
            if m:
                return m.group(1).rstrip("/") + "/"
        return None

    def parse_url(self, url: str) -> Optional[CallMetadata]:
        meta = CallMetadata(audio_url=url)

        path = urlparse(url).path
        parts = path.rstrip("/").split("/")
        if len(parts) >= 3:
            meta.call_id = parts[-2]

        filename = unquote(parts[-1]) if parts else ""
        if filename:
            self._parse_filename(filename, meta)

        return meta

    def parse_email(self, subject: str, body_text: str, body_html: str) -> Optional[CallMetadata]:
        audio_url = self.extract_audio_url(body_text, body_html)
        if not audio_url:
            return None

        meta = self.parse_url(audio_url)

        # Try to parse metadata from email body
        self._parse_email_body(body_text, meta)
        self._parse_email_body(body_html, meta)
        self._parse_subject(subject, meta)

        return meta

    def _parse_subject(self, subject: str, meta: CallMetadata):
        if "исходящ" in subject.lower():
            meta.direction = "outbound"
        elif "входящ" in subject.lower():
            meta.direction = "inbound"

    def _parse_email_body(self, body: str, meta: CallMetadata):
        if not body:
            return

        from_number = None
        to_number = None
        employee_name = None
        duration = None
        duration_unit = None

        m = re.search(r"От:\s*(\+?\d[\d\s\-]*)", body)
        if m:
            from_number = self._clean_number(m.group(1))

        m = re.search(r"(?:На номер|на номер|на номере)\s*(\+?\d[\d\s\-]*)", body)
        if m:
            to_number = self._clean_number(m.group(1))

        m = re.search(r"Сотрудник\s*[-–—]\s*(.+?)(?:\s*(?:Продолжительность|Длительность)\s*|$)", body)
        if m:
            employee_name = m.group(1).strip()

        m = re.search(r"(?:Продолжительность|Длительность)\s*звонка\s*[-–—]\s*(\d+)\s*(сек|с|мин|м)", body)
        if m:
            duration = int(m.group(1))
            duration_unit = m.group(2)
            if duration_unit in ("мин", "м"):
                duration *= 60
            meta.duration_sec = duration

        if not meta.caller_number:
            is_outbound = "исходящ" in body.lower()
            if is_outbound:
                meta.caller_number = from_number
                meta.callee_number = to_number
                meta.direction = "outbound"
            else:
                meta.caller_number = from_number
                meta.callee_number = to_number
                meta.direction = "inbound"

        if not meta.employee and employee_name:
            meta.employee = employee_name

    def parse_attachment(self, subject: str, filename: str) -> Optional[CallMetadata]:
        meta = CallMetadata(audio_filename=filename)

        m = re.search(r"(\d+)-(\d+)\.\d+-(\d+)", filename)
        if m:
            meta.call_id = m.group(2)
            meta.callee_number = f"{m.group(1)}-{m.group(3)}"

        m = re.search(r"(\d{4}-\d{2}-\d{2})-(\d{6})", filename)
        if m:
            meta.datetime = f"{m.group(1)}T{m.group(2)[:2]}:{m.group(2)[2:4]}:{m.group(2)[4:]}Z"

        self._parse_subject(subject, meta)

        if not meta.direction:
            meta.direction = "inbound"

        return meta

    def _parse_filename(self, filename: str, meta: CallMetadata):
        m = re.search(r"От:\s*(\+?\d[\d\s\-]*)", filename)
        if m:
            meta.caller_number = self._clean_number(m.group(1))

        m = re.search(r"На номер\s*(\+?\d[\d\s\-]*)", filename)
        if m:
            meta.callee_number = self._clean_number(m.group(1))

        m = re.search(r"Сотрудник\s*-\s*(.+?)(?:Длительность|$)", filename)
        if m:
            meta.employee = m.group(1).strip()

        m = re.search(r"Длительность звонка\s*-\s*(\d+)\s*сек", filename)
        if m:
            meta.duration_sec = int(m.group(1))

        m = re.search(r"(\d{4}-\d{2}-\d{2})_(\d{2}-\d{2}-\d{2})", filename)
        if m:
            meta.datetime = f"{m.group(1)}T{m.group(2).replace('-', ':')}Z"

        if meta.caller_number and meta.callee_number:
            meta.direction = "inbound"
        else:
            m = re.search(r"from_(\d+)_to_(\d+)", filename)
            if m:
                meta.caller_number = self._clean_number(m.group(1))
                meta.callee_number = self._clean_number(m.group(2))
                meta.direction = "inbound"

    @staticmethod
    def _clean_number(raw: str) -> str:
        return "+" + re.sub(r"\D", "", raw)


PARSER_REGISTRY = {
    "novofon": NovofonParser,
}


def get_parser(name: str = "novofon") -> BaseEmailParser:
    cls = PARSER_REGISTRY.get(name)
    if not cls:
        raise ValueError(f"Unknown parser: {name}, available: {list(PARSER_REGISTRY.keys())}")
    return cls()