diff options
Diffstat (limited to '')
| -rw-r--r-- | crates/yt_dlp/src/lib.rs | 722 |
1 files changed, 316 insertions, 406 deletions
diff --git a/crates/yt_dlp/src/lib.rs b/crates/yt_dlp/src/lib.rs index 970bfe2..6be5e87 100644 --- a/crates/yt_dlp/src/lib.rs +++ b/crates/yt_dlp/src/lib.rs @@ -1,6 +1,6 @@ // yt - A fully featured command line YouTube client // -// Copyright (C) 2024 Benedikt Peetz <benedikt.peetz@b-peetz.de> +// Copyright (C) 2025 Benedikt Peetz <benedikt.peetz@b-peetz.de> // SPDX-License-Identifier: GPL-3.0-or-later // // This file is part of Yt. @@ -8,461 +8,371 @@ // You should have received a copy of the License along with this program. // If not, see <https://www.gnu.org/licenses/gpl-3.0.txt>. -// The pyo3 `pyfunction` proc-macros call unsafe functions internally, which trigger this lint. -#![allow(unsafe_op_in_unsafe_fn)] -#![allow(clippy::missing_errors_doc)] +//! The `yt_dlp` interface is completely contained in the [`YoutubeDL`] structure. -use std::env; -use std::io::stdout; -use std::{fs::File, io::Write}; +use std::path::PathBuf; -use std::{path::PathBuf, sync::Once}; - -use crate::{duration::Duration, logging::setup_logging, wrapper::info_json::InfoJson}; - -use bytes::Bytes; -use log::{info, log_enabled, Level}; -use pyo3::types::{PyString, PyTuple, PyTupleMethods}; +use log::{debug, info}; use pyo3::{ - pyfunction, - types::{PyAnyMethods, PyDict, PyDictMethods, PyList, PyListMethods, PyModule}, - wrap_pyfunction, Bound, PyAny, PyResult, Python, + Bound, Py, PyAny, Python, intern, + types::{PyAnyMethods, PyDict, PyIterator, PyList}, }; -use serde::Serialize; -use serde_json::{Map, Value}; use url::Url; -pub mod duration; -pub mod logging; -pub mod wrapper; - -#[cfg(test)] -mod tests; - -/// Synchronisation helper, to ensure that we don't setup the logger multiple times -static SYNC_OBJ: Once = Once::new(); - -/// Add a logger to the yt-dlp options. -/// If you have an logger set (i.e. for rust), than this will log to rust -/// -/// # Panics -/// This should never panic. -pub fn add_logger_and_sig_handler<'a>( - opts: Bound<'a, PyDict>, - py: Python<'_>, -) -> PyResult<Bound<'a, PyDict>> { - setup_logging(py, "yt_dlp")?; - - let logging = PyModule::import(py, "logging")?; - let ytdl_logger = logging.call_method1("getLogger", ("yt_dlp",))?; - - // Ensure that all events are logged by setting the log level to NOTSET (we filter on rust's side) - // Also use this static, to ensure that we don't configure the logger every time - SYNC_OBJ.call_once(|| { - // Disable the SIGINT (Ctrl+C) handler, python installs. - // This allows the user to actually stop the application with Ctrl+C. - // This is here because it can only be run in the main thread and this was here already. - py.run( - c"\ -import signal -signal.signal(signal.SIGINT, signal.SIG_DFL)", - None, - None, - ) - .expect("This code should always work"); - - let config_opts = PyDict::new(py); - config_opts - .set_item("level", 0) - .expect("Setting this item should always work"); - - logging - .call_method("basicConfig", (), Some(&config_opts)) - .expect("This method exists"); - }); - - // This was taken from `ytcc`, I don't think it is still applicable - // ytdl_logger.setattr("propagate", false)?; - // let logging_null_handler = logging.call_method0("NullHandler")?; - // ytdl_logger.setattr("addHandler", logging_null_handler)?; - - opts.set_item("logger", ytdl_logger).expect("Should work"); - - Ok(opts) -} +use crate::{ + info_json::{InfoJson, json_dumps, json_loads}, + python_error::{IntoPythonError, PythonError}, +}; -#[pyfunction] -#[allow(clippy::too_many_lines)] -#[allow(clippy::missing_panics_doc)] -#[allow(clippy::items_after_statements)] -#[allow( - clippy::cast_possible_truncation, - clippy::cast_sign_loss, - clippy::cast_precision_loss -)] -pub fn progress_hook(py: Python<'_>, input: &Bound<'_, PyDict>) -> PyResult<()> { - // Only add the handler, if the log-level is higher than Debug (this avoids covering debug - // messages). - if log_enabled!(Level::Debug) { - return Ok(()); - } +pub mod info_json; +pub mod options; +pub mod post_processors; +pub mod progress_hook; +pub mod python_error; + +#[macro_export] +macro_rules! json_get { + ($value:expr, $name:literal, $into:ident) => {{ + match $value.get($name) { + Some(val) => $crate::json_cast!(@log_key $name, val, $into), + None => panic!( + concat!( + "Expected '", + $name, + "' to be a key for the '", + stringify!($value), + "' object: {:#?}" + ), + $value + ), + } + }}; +} - // ANSI ESCAPE CODES Wrappers {{{ - // see: https://en.wikipedia.org/wiki/ANSI_escape_code#Control_Sequence_Introducer_commands - const CSI: &str = "\x1b["; - fn clear_whole_line() { - print!("{CSI}2K"); - } - fn move_to_col(x: usize) { - print!("{CSI}{x}G"); - } - // }}} - - let input: Map<String, Value> = serde_json::from_str(&json_dumps( - py, - input - .downcast::<PyAny>() - .expect("Will always work") - .to_owned(), - )?) - .expect("Python should always produce valid json"); - - macro_rules! get { - (@interrogate $item:ident, $type_fun:ident, $get_fun:ident, $name:expr) => {{ - let a = $item.get($name).expect(concat!( - "The field '", - stringify!($name), - "' should exist." - )); - - if a.$type_fun() { - a.$get_fun().expect( - "The should have been checked in the if guard, so unpacking here is fine", - ) +#[macro_export] +macro_rules! json_try_get { + ($value:expr, $name:literal, $into:ident) => {{ + if let Some(val) = $value.get($name) { + if val.is_null() { + None } else { - panic!( - "Value {} => \n{}\n is not of type: {}", - $name, - a, - stringify!($type_fun) - ); + Some(json_cast!(@log_key $name, val, $into)) } - }}; + } else { + None + } + }}; +} - ($type_fun:ident, $get_fun:ident, $name1:expr, $name2:expr) => {{ - let a = get! {@interrogate input, is_object, as_object, $name1}; - let b = get! {@interrogate a, $type_fun, $get_fun, $name2}; - b - }}; +#[macro_export] +macro_rules! json_cast { + ($value:expr, $into:ident) => {{ + let value_name = stringify!($value); + json_cast!(@log_key value_name, $value, $into) + }}; + + (@log_key $name:expr, $value:expr, $into:ident) => {{ + match $value.$into() { + Some(result) => result, + None => panic!( + concat!( + "Expected to be able to cast '{}' value (which is '{:?}') ", + stringify!($into) + ), + $name, + $value + ), + } + }}; +} - ($type_fun:ident, $get_fun:ident, $name:expr) => {{ - get! {@interrogate input, $type_fun, $get_fun, $name} - }}; - } +macro_rules! py_kw_args { + ($py:expr => $($kw_arg_name:ident = $kw_arg_val:expr),*) => {{ + use $crate::python_error::IntoPythonError; - macro_rules! default_get { - (@interrogate $item:ident, $default:expr, $get_fun:ident, $name:expr) => {{ - let a = if let Some(field) = $item.get($name) { - field.$get_fun().unwrap_or($default) - } else { - $default - }; - a - }}; - - ($get_fun:ident, $default:expr, $name1:expr, $name2:expr) => {{ - let a = get! {@interrogate input, is_object, as_object, $name1}; - let b = default_get! {@interrogate a, $default, $get_fun, $name2}; - b - }}; - - ($get_fun:ident, $default:expr, $name:expr) => {{ - default_get! {@interrogate input, $default, $get_fun, $name} - }}; - } + let dict = PyDict::new($py); - macro_rules! c { - ($color:expr, $format:expr) => { - format!("\x1b[{}m{}\x1b[0m", $color, $format) - }; - } + $( + dict.set_item(stringify!($kw_arg_name), $kw_arg_val).wrap_exc($py)?; + )* - fn format_bytes(bytes: u64) -> String { - let bytes = Bytes::new(bytes); - bytes.to_string() + Some(dict) } + .as_ref()}; +} +pub(crate) use py_kw_args; - fn format_speed(speed: f64) -> String { - #[allow(clippy::cast_possible_truncation, clippy::cast_sign_loss)] - let bytes = Bytes::new(speed.floor() as u64); - format!("{bytes}/s") +/// The core of the `yt_dlp` interface. +#[derive(Debug)] +pub struct YoutubeDL { + inner: Py<PyAny>, + options: serde_json::Map<String, serde_json::Value>, +} + +impl YoutubeDL { + /// Fetch the underlying `yt_dlp` and `python` version. + /// + /// # Errors + /// If python attribute access fails. + pub fn version(&self) -> Result<(String, String), PythonError> { + Python::with_gil(|py| { + let yt_dlp = py + .import(intern!(py, "yt_dlp")) + .wrap_exc(py)? + .getattr(intern!(py, "version")) + .wrap_exc(py)? + .getattr(intern!(py, "__version__")) + .wrap_exc(py)? + .extract() + .wrap_exc(py)?; + + let python = py.version(); + + Ok((yt_dlp, python.to_owned())) + }) } - let get_title = |add_extension: bool| -> String { - match get! {is_string, as_str, "info_dict", "ext"} { - "vtt" => { - format!( - "Subtitles ({})", - default_get! {as_str, "<No Subtitle Language>", "info_dict", "name"} - ) - } - title_extension @ ("webm" | "mp4" | "m4a") => { - if add_extension { - format!( - "{} ({})", - default_get! { as_str, "<No title>", "info_dict", "title"}, - title_extension - ) - } else { - default_get! { as_str, "<No title>", "info_dict", "title"}.to_owned() - } - } - other => panic!("The extension '{other}' is not yet implemented"), - } - }; - - match get! {is_string, as_str, "status"} { - "downloading" => { - let elapsed = default_get! {as_f64, 0.0f64, "elapsed"}; - let eta = default_get! {as_f64, 0.0, "eta"}; - let speed = default_get! {as_f64, 0.0, "speed"}; - - let downloaded_bytes = get! {is_u64, as_u64, "downloaded_bytes"}; - let (total_bytes, bytes_is_estimate): (u64, &'static str) = { - let total_bytes = default_get!(as_u64, 0, "total_bytes"); - if total_bytes == 0 { - let maybe_estimate = default_get!(as_u64, 0, "total_bytes_estimate"); - - if maybe_estimate == 0 { - // The download speed should be in bytes per second and the eta in seconds. - // Thus multiplying them gets us the raw bytes (which were estimated by `yt_dlp`, from their `info.json`) - let bytes_still_needed = (speed * eta).ceil() as u64; - - (downloaded_bytes + bytes_still_needed, "~") - } else { - (maybe_estimate, "~") - } - } else { - (total_bytes, "") - } - }; - let percent: f64 = { - if total_bytes == 0 { - 100.0 - } else { - (downloaded_bytes as f64 / total_bytes as f64) * 100.0 - } + /// Download a given list of URLs. + /// Returns the paths they were downloaded to. + /// + /// # Errors + /// If one of the downloads error. + pub fn download(&self, urls: &[Url]) -> Result<Vec<PathBuf>, extract_info::Error> { + let mut out_paths = Vec::with_capacity(urls.len()); + + for url in urls { + info!("Started downloading url: '{url}'"); + let info_json = self.extract_info(url, true, true)?; + + // Try to work around yt-dlp type weirdness + let result_string = if let Some(filename) = json_try_get!(info_json, "filename", as_str) + { + PathBuf::from(filename) + } else { + PathBuf::from(json_get!( + json_cast!( + json_get!(info_json, "requested_downloads", as_array)[0], + as_object + ), + "filename", + as_str + )) }; - clear_whole_line(); - move_to_col(1); - - print!( - "'{}' [{}/{} at {}] -> [{} of {}{} {}] ", - c!("34;1", get_title(true)), - c!("33;1", Duration::from(Some(elapsed))), - c!("33;1", Duration::from(Some(eta))), - c!("32;1", format_speed(speed)), - c!("31;1", format_bytes(downloaded_bytes)), - c!("31;1", bytes_is_estimate), - c!("31;1", format_bytes(total_bytes)), - c!("36;1", format!("{:.02}%", percent)) - ); - stdout().flush()?; - } - "finished" => { - println!("-> Finished downloading."); + out_paths.push(result_string); + info!("Finished downloading url"); } - "error" => { - panic!("-> Error while downloading: {}", get_title(true)) - } - other => unreachable!("'{other}' should not be a valid state!"), - }; - Ok(()) -} - -pub fn add_hooks<'a>(opts: Bound<'a, PyDict>, py: Python<'_>) -> PyResult<Bound<'a, PyDict>> { - if let Some(hooks) = opts.get_item("progress_hooks")? { - let hooks = hooks.downcast::<PyList>()?; - hooks.append(wrap_pyfunction!(progress_hook, py)?)?; - - opts.set_item("progress_hooks", hooks)?; - } else { - // No hooks are set yet - let hooks_list = PyList::new(py, &[wrap_pyfunction!(progress_hook, py)?])?; - - opts.set_item("progress_hooks", hooks_list)?; + Ok(out_paths) } - Ok(opts) -} - -/// `extract_info(self, url, download=True, ie_key=None, extra_info=None, process=True, force_generic_extractor=False)` -/// -/// Extract and return the information dictionary of the URL -/// -/// Arguments: -/// @param url URL to extract -/// -/// Keyword arguments: -/// @param download Whether to download videos -/// @param process Whether to resolve all unresolved references (URLs, playlist items). -/// Must be True for download to work -/// @param `ie_key` Use only the extractor with this key -/// -/// @param `extra_info` Dictionary containing the extra values to add to the info (For internal use only) -/// @`force_generic_extractor` Force using the generic extractor (Deprecated; use `ie_key`='Generic') -#[allow(clippy::unused_async)] -#[allow(clippy::missing_panics_doc)] -pub async fn extract_info( - yt_dlp_opts: &Map<String, Value>, - url: &Url, - download: bool, - process: bool, -) -> PyResult<InfoJson> { - Python::with_gil(|py| { - let opts = json_map_to_py_dict(yt_dlp_opts, py)?; - - let instance = get_yt_dlp(py, opts)?; - let args = (url.as_str(),); - - let kwargs = PyDict::new(py); - kwargs.set_item("download", download)?; - kwargs.set_item("process", process)?; - - let result = instance.call_method("extract_info", args, Some(&kwargs))?; - - // Remove the `<generator at 0xsome_hex>`, by setting it to null - if !process { - result.set_item("entries", ())?; - } + /// `extract_info(self, url, download=True, ie_key=None, extra_info=None, process=True, force_generic_extractor=False)` + /// + /// Extract and return the information dictionary of the URL + /// + /// Arguments: + /// - `url` URL to extract + /// + /// Keyword arguments: + /// :`download` Whether to download videos + /// :`process` Whether to resolve all unresolved references (URLs, playlist items). + /// Must be True for download to work + /// + /// # Panics + /// If expectations about python fail to hold. + /// + /// # Errors + /// If python operations fail. + pub fn extract_info( + &self, + url: &Url, + download: bool, + process: bool, + ) -> Result<InfoJson, extract_info::Error> { + Python::with_gil(|py| { + let inner = self + .inner + .bind(py) + .getattr(intern!(py, "extract_info")) + .wrap_exc(py)?; + + let result = inner + .call( + (url.to_string(),), + py_kw_args!(py => download = download, process = process), + ) + .wrap_exc(py)? + .downcast_into::<PyDict>() + .expect("This is a dict"); + + // Resolve the generator object + if let Ok(generator) = result.get_item(intern!(py, "entries")) { + if generator.is_instance_of::<PyList>() { + // already resolved. Do nothing + } else if let Ok(generator) = generator.downcast::<PyIterator>() { + // A python generator object. + let max_backlog = json_try_get!(self.options, "playlistend", as_u64) + .map_or(10000, |playlistend| { + usize::try_from(playlistend).expect("Should work") + }); + + let mut out = vec![]; + for output in generator { + out.push(output.wrap_exc(py)?); + + if out.len() == max_backlog { + break; + } + } - let result_str = json_dumps(py, result)?; + result.set_item(intern!(py, "entries"), out).wrap_exc(py)?; + } else { + // Probably some sort of paged list (`OnDemand` or otherwise) + let max_backlog = json_try_get!(self.options, "playlistend", as_u64) + .map_or(10000, |playlistend| { + usize::try_from(playlistend).expect("Should work") + }); - if let Ok(confirm) = env::var("YT_STORE_INFO_JSON") { - if confirm == "yes" { - let mut file = File::create("output.info.json")?; - write!(file, "{result_str}").unwrap(); - } - } + let next = generator.getattr(intern!(py, "getslice")).wrap_exc(py)?; - Ok(serde_json::from_str(&result_str) - .expect("Python should be able to produce correct json")) - }) -} + let output = next + .call((), py_kw_args!(py => start = 0, end = max_backlog)) + .wrap_exc(py)?; -/// # Panics -/// Only if python fails to return a valid URL. -pub fn unsmuggle_url(smug_url: &Url) -> PyResult<Url> { - Python::with_gil(|py| { - let utils = get_yt_dlp_utils(py)?; - let url = utils - .call_method1("unsmuggle_url", (smug_url.as_str(),))? - .downcast::<PyTuple>()? - .get_item(0)?; - - let url: Url = url - .downcast::<PyString>()? - .to_string() - .parse() - .expect("Python should be able to return a valid url"); - - Ok(url) - }) -} + result + .set_item(intern!(py, "entries"), output) + .wrap_exc(py)?; + } + } -/// Download a given list of URLs. -/// Returns the paths they were downloaded to. -/// -/// # Panics -/// Only if `yt_dlp` changes their `info_json` schema. -pub async fn download( - urls: &[Url], - download_options: &Map<String, Value>, -) -> PyResult<Vec<PathBuf>> { - let mut out_paths = Vec::with_capacity(urls.len()); - - for url in urls { - info!("Started downloading url: '{}'", url); - let info_json = extract_info(download_options, url, true, true).await?; - - // Try to work around yt-dlp type weirdness - let result_string = if let Some(filename) = info_json.filename { - filename - } else { - info_json.requested_downloads.expect("This must exist")[0] - .filename - .clone() - }; + let result = self.prepare_info_json(&result, py)?; - out_paths.push(result_string); - info!("Finished downloading url: '{}'", url); + Ok(result) + }) } - Ok(out_paths) -} - -fn json_map_to_py_dict<'a>( - map: &Map<String, Value>, - py: Python<'a>, -) -> PyResult<Bound<'a, PyDict>> { - let json_string = serde_json::to_string(&map).expect("This must always work"); + /// Take the (potentially modified) result of the information extractor (i.e., + /// [`Self::extract_info`] with `process` and `download` set to false) + /// and resolve all unresolved references (URLs, + /// playlist items). + /// + /// It will also download the videos if 'download' is true. + /// Returns the resolved `ie_result`. + /// + /// # Panics + /// If expectations about python fail to hold. + /// + /// # Errors + /// If python operations fail. + pub fn process_ie_result( + &self, + ie_result: InfoJson, + download: bool, + ) -> Result<InfoJson, process_ie_result::Error> { + Python::with_gil(|py| { + let inner = self + .inner + .bind(py) + .getattr(intern!(py, "process_ie_result")) + .wrap_exc(py)?; + + let result = inner + .call( + (json_loads(ie_result, py),), + py_kw_args!(py => download = download), + ) + .wrap_exc(py)? + .downcast_into::<PyDict>() + .expect("This is a dict"); - let python_dict = json_loads(py, json_string)?; + let result = self.prepare_info_json(&result, py)?; - Ok(python_dict) -} + Ok(result) + }) + } -fn json_dumps(py: Python<'_>, input: Bound<'_, PyAny>) -> PyResult<String> { - // json.dumps(yt_dlp.sanitize_info(input)) + /// Close this [`YoutubeDL`] instance, and stop all currently running downloads. + /// + /// # Errors + /// If python operations fail. + pub fn close(&self) -> Result<(), close::Error> { + Python::with_gil(|py| { + debug!("Closing YoutubeDL."); - let yt_dlp = get_yt_dlp(py, PyDict::new(py))?; - let sanitized_result = yt_dlp.call_method1("sanitize_info", (input,))?; + let inner = self + .inner + .bind(py) + .getattr(intern!(py, "close")) + .wrap_exc(py)?; - let json = PyModule::import(py, "json")?; - let dumps = json.getattr("dumps")?; + inner.call0().wrap_exc(py)?; - let output = dumps.call1((sanitized_result,))?; + Ok(()) + }) + } - let output_str = output.extract::<String>()?; + fn prepare_info_json<'py>( + &self, + info: &Bound<'py, PyDict>, + py: Python<'py>, + ) -> Result<InfoJson, prepare::Error> { + let sanitize = self + .inner + .bind(py) + .getattr(intern!(py, "sanitize_info")) + .wrap_exc(py)?; - Ok(output_str) -} + let value = sanitize.call((info,), None).wrap_exc(py)?; -fn json_loads_str<T: Serialize>(py: Python<'_>, input: T) -> PyResult<Bound<'_, PyDict>> { - let string = serde_json::to_string(&input).expect("Correct json must be pased"); + let result = value.downcast::<PyDict>().expect("This should stay a dict"); - json_loads(py, string) + Ok(json_dumps(result)) + } } -fn json_loads(py: Python<'_>, input: String) -> PyResult<Bound<'_, PyDict>> { - // json.loads(input) - - let json = PyModule::import(py, "json")?; - let dumps = json.getattr("loads")?; +#[allow(missing_docs)] +pub mod close { + use crate::python_error::PythonError; - let output = dumps.call1((input,))?; - - Ok(output - .downcast::<PyDict>() - .expect("This should always be a PyDict") - .clone()) + #[derive(Debug, thiserror::Error)] + pub enum Error { + #[error(transparent)] + Python(#[from] PythonError), + } } +#[allow(missing_docs)] +pub mod process_ie_result { + use crate::{prepare, python_error::PythonError}; -fn get_yt_dlp_utils(py: Python<'_>) -> PyResult<Bound<'_, PyAny>> { - let yt_dlp = PyModule::import(py, "yt_dlp")?; - let utils = yt_dlp.getattr("utils")?; + #[derive(Debug, thiserror::Error)] + pub enum Error { + #[error(transparent)] + Python(#[from] PythonError), - Ok(utils) + #[error("Failed to prepare the info json")] + InfoJsonPrepare(#[from] prepare::Error), + } } -fn get_yt_dlp<'a>(py: Python<'a>, opts: Bound<'a, PyDict>) -> PyResult<Bound<'a, PyAny>> { - // Unconditionally set a logger - let opts = add_logger_and_sig_handler(opts, py)?; - let opts = add_hooks(opts, py)?; +#[allow(missing_docs)] +pub mod extract_info { + use crate::{prepare, python_error::PythonError}; - let yt_dlp = PyModule::import(py, "yt_dlp")?; - let youtube_dl = yt_dlp.call_method1("YoutubeDL", (opts,))?; + #[derive(Debug, thiserror::Error)] + pub enum Error { + #[error(transparent)] + Python(#[from] PythonError), - Ok(youtube_dl) + #[error("Failed to prepare the info json")] + InfoJsonPrepare(#[from] prepare::Error), + } +} +#[allow(missing_docs)] +pub mod prepare { + use crate::python_error::PythonError; + + #[derive(Debug, thiserror::Error)] + pub enum Error { + #[error(transparent)] + Python(#[from] PythonError), + } } |