browser-cli/browser_cli/sdk/dom.py

"""DOM, content-extraction, and page-info namespaces: ``b.dom.*``, ``b.extract.*``, ``b.page.*``."""
from __future__ import annotations

from browser_cli.sdk.base import Namespace, sdk_command

def _selector_args(self, selector):
  return {"selector": selector}

def _selector_value_args(self, selector, value):
  return {"selector": selector, "value": value}

def _extract_markdown(self, result, *args, **kwargs) -> str:
  from browser_cli.markdown import render_markdown

  return render_markdown(result)

class DomNS(Namespace):
  """Query and drive page elements in the active (or specified) tab."""

  @sdk_command("dom.query", _selector_args, default=[])
  def query(self, selector: str) -> list[dict]:
    """Return elements matching a CSS selector."""

  @sdk_command("dom.click", _selector_args, return_result=False)
  def click(self, selector: str) -> None:
    """Click the first element matching a CSS selector."""

  @sdk_command("dom.type", lambda self, selector, text: {"selector": selector, "text": text}, return_result=False)
  def type(self, selector: str, text: str) -> None:
    """Type text into the first matching element."""

  @sdk_command("dom.attr", lambda self, selector, attr: {"selector": selector, "attr": attr}, default=[])
  def attr(self, selector: str, attr: str) -> list[str]:
    """Return an attribute from all matching elements."""

  @sdk_command("dom.text", _selector_args, default=[])
  def text(self, selector: str) -> list[str]:
    """Return text from all matching elements."""

  @sdk_command("dom.exists", _selector_args, default=False)
  def exists(self, selector: str) -> bool:
    """Return whether a selector exists."""

  @sdk_command("dom.scroll", lambda self, selector=None, *, x=None, y=None: {"selector": selector, "x": x, "y": y}, return_result=False)
  def scroll(self, selector: str | None = None, *, x: int | None = None, y: int | None = None) -> None:
    """Scroll to a CSS selector or to pixel coordinates."""

  @sdk_command("dom.select", _selector_value_args, return_result=False)
  def select(self, selector: str, value: str) -> None:
    """Set the value of a <select> element."""

  @sdk_command("dom.eval", lambda self, code, tab_id=None: {"code": code, "tabId": tab_id})
  def eval(self, code: str, tab_id: int | None = None):
    """Evaluate JavaScript in the page's main world and return the result."""

  @sdk_command("dom.key", lambda self, key, selector=None: {"key": key, "selector": selector}, return_result=False)
  def key(self, key: str, selector: str | None = None) -> None:
    """Dispatch a keyboard event. key examples: 'Enter', 'Tab', 'Escape', 'ArrowDown'."""

  @sdk_command("dom.hover", _selector_args, return_result=False)
  def hover(self, selector: str) -> None:
    """Dispatch mouseover/mouseenter on an element."""

  @sdk_command("dom.check", _selector_args, return_result=False)
  def check(self, selector: str) -> None:
    """Check a checkbox."""

  @sdk_command("dom.uncheck", _selector_args, return_result=False)
  def uncheck(self, selector: str) -> None:
    """Uncheck a checkbox."""

  @sdk_command("dom.clear", _selector_args, return_result=False)
  def clear(self, selector: str) -> None:
    """Clear the value of an input element."""

  @sdk_command("dom.focus", _selector_args, return_result=False)
  def focus(self, selector: str) -> None:
    """Focus an element."""

  @sdk_command("dom.submit", _selector_args, return_result=False)
  def submit(self, selector: str) -> None:
    """Submit the form containing the matched element."""

  def poll(
    self,
    selector: str,
    pattern: str,
    *,
    attr: str | None = None,
    timeout: float = 30.0,
    interval: float = 0.5,
    tab_id: int | None = None,
  ) -> dict:
    """Poll selector's text/value until it matches regex pattern.

    Returns ``{"selector": ..., "value": ..., "pattern": ...}`` when matched.
    """
    return self.command("dom.poll", {
      "selector": selector,
      "pattern": pattern,
      "attr": attr,
      "timeout": int(timeout * 1000),
      "interval": int(interval * 1000),
      "tabId": tab_id,
    })

  def wait_for(
    self,
    selector: str,
    *,
    timeout: float = 10.0,
    visible: bool = False,
    hidden: bool = False,
    tab_id: int | None = None,
  ) -> dict:
    """Wait until a CSS selector appears (or disappears) in the DOM.

    Args:
      selector: CSS selector to watch.
      timeout: Max seconds to wait before raising ``RuntimeError``.
      visible: Wait until the element has non-zero dimensions.
      hidden: Wait until the element is absent or has ``offsetParent == null``.
      tab_id: Tab to watch. Defaults to the active tab.
    """
    return self.command("dom.wait_for", {
      "selector": selector,
      "timeout": int(timeout * 1000),
      "visible": visible,
      "hidden": hidden,
      "tabId": tab_id,
    })

class ExtractNS(Namespace):
  """Extract structured content from the active (or specified) tab."""

  @sdk_command("extract.links", lambda self, tab_id=None: {"tabId": tab_id}, default=[])
  def links(self, tab_id: int | None = None) -> list[dict]:
    """Return links from the active tab or *tab_id*."""

  @sdk_command("extract.images", lambda self, tab_id=None: {"tabId": tab_id}, default=[])
  def images(self, tab_id: int | None = None) -> list[dict]:
    """Return images from the active tab or *tab_id*."""

  @sdk_command("extract.text", lambda self, tab_id=None: {"tabId": tab_id}, default="")
  def text(self, tab_id: int | None = None) -> str:
    """Return plain text from the active tab or *tab_id*."""

  @sdk_command("extract.json", lambda self, selector, tab_id=None: {"selector": selector, "tabId": tab_id})
  def json(self, selector: str, tab_id: int | None = None):
    """Extract JSON-like structured data from a selector in the active tab or *tab_id*."""

  @sdk_command("extract.html", lambda self, tab_id=None: {"tabId": tab_id}, default="")
  def html(self, tab_id: int | None = None) -> str:
    """Return the full HTML source of the active tab or *tab_id*."""

  @sdk_command("extract.markdown", lambda self, selector=None, tab_id=None: {"selector": selector, "tabId": tab_id}, mapper=_extract_markdown)
  def markdown(self, selector: str | None = None, tab_id: int | None = None) -> str:
    """Extract the page's main content as clean Markdown from the active tab or *tab_id*.

    The extractor may return either Markdown or raw HTML; both are
    normalized to Markdown here so SDK and CLI callers get identical output.
    """

class PageNS(Namespace):
  """Inspect the active page or a specified tab."""

  def info(self, tab_id: int | None = None) -> dict:
    """Return metadata for the active page, or tab status for *tab_id*."""
    if tab_id is not None:
      return self.command("tabs.status", {"tabId": tab_id}) or {}
    return self.command("page.info", {}) or {}