path: root/host/utils/uhd_images_downloader.py.in

#!@RUNTIME_PYTHON_EXECUTABLE@
#
# Copyright 2018 Ettus Research, a National Instruments Company
# Copyright 2020 Ettus Research, a National Instruments Brand
#
# SPDX-License-Identifier: GPL-3.0-or-later
#
"""
Download image files required for USRPs.

Usage: The `uhd_images_downloader` should work, "out of the box", with no command line arguments.
Assuming your computer has an internet connection to [files.ettus.com], simply run the utility
every time you update UHD, and the installed images for your devices should always be up to date.


Images will be downloaded on a per-target basis. That is, there are image packages for a desired
device and configuration. Users can specify which image packages they would plan to use. To see a
list of available targets, run `uhd_images_downloader --list-targets`. The left column of the
printout will be a list of available image archives. From there, you can construct a regular
expression which matches to the targets you wish to download. For example, in order to download all
image packages related to the X300 product line, users may run
`uhd_images_downloader --types x3.*`.


The `uhd_images_downloader` uses a manifest to look-up the URLs of image packages to download.
Downloaded images are recorded in an inventory file that lives in the images install location.
This allows the downloader to skip images that were previously downloaded, and haven't changed
since.

Manifests are built into the downloader, but can also be accessed at uhd/images/manifest.txt.

Inventory files are JSON files called `inventory.json`, by default. It is possible to specify the
inventory file in command line arguments, but we don't recommend using this functionality unless
you're really sure you need it.
"""
import argparse
import hashlib
import json
import math
import os
import re
import shutil
import sys
import tempfile
import zipfile
import platform
from urllib.parse import urljoin  # Python 3

# For all the non-core-library imports, we will be extra paranoid and be very
# nice with error messages so that everyone understands what's up.
try:
    import requests
except ImportError:
    sys.stdout.write(
        "[ERROR] Missing module 'requests'! Please install it, e.g., by "
        "running 'pip install requests' or any other tool that can install "
        "Python modules.\n")
    if platform.system() == 'Windows':
        input('Hit Enter to continue.')
    exit(0)

# pylint: disable=bad-whitespace
_USERNAME_VARIABLE        = "UHD_IMAGES_USER"
_PASSWORD_VARIABLE        = "UHD_IMAGES_PASSWORD"
_DEFAULT_TARGET_REGEX     = "(fpga|fw|windrv)_default"
_BASE_DIR_STRUCTURE_PARTS = ["share", "uhd", "images"]
_DEFAULT_INSTALL_PATH     = os.path.join("@CMAKE_INSTALL_PREFIX@", *_BASE_DIR_STRUCTURE_PARTS)
_DEFAULT_BASE_URL         = "https://files.ettus.com/binaries/cache/"
_INVENTORY_FILENAME       = "inventory.json"
_CONTACT                  = "support@ettus.com"
_DEFAULT_BUFFER_SIZE      = 8192
_DEFAULT_DOWNLOAD_LIMIT   = 100 * 1024 * 1024 # Bytes
_ARCHIVE_DEFAULT_TYPE     = "zip"
_UHD_VERSION              = "@UHD_VERSION@"
_LOG_LEVELS = {
    "TRACE": 1,
    "DEBUG": 2,
    "INFO":  3,
    "WARN":  4,
    "ERROR": 5,
}
_LOG_LEVEL = _LOG_LEVELS["INFO"]
_YES = False
_PROXIES = {}
_BASE_URL_ENV_VAR = "UHD_IMAGES_URL"
# pylint: enable=bad-whitespace
# Note: _MANIFEST_CONTENTS are placed at the bottom of this file for aesthetic reasons

def log(level, message):
    """Logging function"""
    message_log_level = _LOG_LEVELS.get(level, 0)
    if message_log_level >= _LOG_LEVEL:
        sys.stderr.write(
            "[{level}] {message}\n".format(level=level, message=message))


def parse_args():
    """
    Setup argument parser and parse. Also does some sanity checks and sets some
    global variables we want to use.
    """
    parser = argparse.ArgumentParser(formatter_class=argparse.ArgumentDefaultsHelpFormatter,
                                     description=__doc__)
    parser.add_argument('-t', '--types', action='append',
                        help="RegEx to select image sets from the manifest file.")
    parser.add_argument('-i', '--install-location', type=str, default=None,
                        help="Set custom install location for images")
    parser.add_argument('-m', '--manifest-location', type=str, default=None,
                        help="Set custom location for the manifest file")
    parser.add_argument('-I', '--inventory-location', type=str, default=None,
                        help="Set custom location for the inventory file")
    parser.add_argument('-l', '--list-targets', action="store_true", default=False,
                        help="Print targets in the manifest file to stdout, and exit.\n"
                        "To get relative paths only, specify an empty base URL (-b '').")
    parser.add_argument('--url-only', action="store_true", default=False,
                        help="With -l, only print the URLs, nothing else.")
    parser.add_argument("--buffer-size", type=int, default=_DEFAULT_BUFFER_SIZE,
                        help="Set download buffer size")
    parser.add_argument("--download-limit", type=int, default=_DEFAULT_DOWNLOAD_LIMIT,
                        help="Set threshold for download limits. Any download "
                             "larger than this will require approval, either "
                             "interactively, or by providing --yes.")
    parser.add_argument("--http-proxy", type=str,
                        help="Specify HTTP(S) proxy in the format "
                        "http[s]://user:pass@1.2.3.4:port\n"
                        "If this this option is not given, the environment "
                        "variables HTTP_PROXY/HTTPS_PROXY can also be used to specify a proxy.")
    parser.add_argument("-b", "--base-url", type=str,
                        help="Set base URL for images download location. "
                        "Defaults to ${} if set, or {} otherwise.".format(
                            _BASE_URL_ENV_VAR, _DEFAULT_BASE_URL))
    parser.add_argument("-k", "--keep", action="store_true", default=False,
                        help="Keep the downloaded images archives in the image directory")
    parser.add_argument("-T", "--test", action="store_true", default=False,
                        help="Verify the downloaded archives before extracting them")
    parser.add_argument("-y", "--yes", action="store_true", default=False,
                        help="Answer all questions with 'yes' (for scripting purposes).")
    parser.add_argument("-n", "--dry-run", action="store_true", default=False,
                        help="Print selected target without actually downloading them.")
    parser.add_argument("--refetch", action="store_true", default=False,
                        help="Ignore the inventory file and download all images.")
    parser.add_argument('-V', '--version', action='version', version=_UHD_VERSION)
    parser.add_argument('-q', '--quiet', action='count', default=0,
                        help="Decrease verbosity level")
    parser.add_argument('-v', '--verbose', action='count', default=0,
                        help="Increase verbosity level")
    args = parser.parse_args()
    # Set the verbosity
    global _LOG_LEVEL
    log("TRACE", "Default log level: {}".format(_LOG_LEVEL))
    _LOG_LEVEL = _LOG_LEVEL - args.verbose + args.quiet
    # Some sanitation that's easier to handle outside of the argparse framework:
    if not args.base_url:
        if os.environ.get(_BASE_URL_ENV_VAR):
            args.base_url = os.environ.get(_BASE_URL_ENV_VAR)
        else:
            args.base_url = _DEFAULT_BASE_URL
    log("INFO", "Using base URL: {}".format(args.base_url))
    if not args.base_url.endswith('/') and args.base_url != "":
        args.base_url += '/'
    if args.yes:
        global _YES
        _YES = True
    if args.http_proxy:
        global _PROXIES
        _PROXIES['http'] = args.http_proxy
        _PROXIES['https'] = args.http_proxy
    return args


def get_images_dir(args):
    """
    Figure out where to store the images.
    """
    if args.install_location:
        return args.install_location
    if os.environ.get("UHD_IMAGES_DIR"):
        log("DEBUG",
            "UHD_IMAGES_DIR environment variable is set, using to set "
            "install location.")
        return os.environ.get("UHD_IMAGES_DIR")
    return _DEFAULT_INSTALL_PATH


def ask_permission(question):
    """
    Ask the question, and have the user type y or n on the keyboard. If the
    global variable _YES is true, this always returns True without asking the
    question. Otherwise, return True if the answer is 'yes', 'y', or something
    similar.
    """
    if _YES:
        log("DEBUG", "Assuming the answer is 'yes' for this question: " +
            question)
        return True
    answer = input(question + " [y/N]")
    if answer and answer[0].lower() == 'y':
        return True
    return False


class TemporaryDirectory:
    """Class to create a temporary directory"""
    def __enter__(self):
        try:
            self.name = tempfile.mkdtemp()
            return self.name
        except Exception as ex:
            log("ERROR", "Failed to create a temporary directory (%s)" % ex)
            raise ex

    # Can return 'True' to suppress incoming exception
    def __exit__(self, exc_type, exc_value, traceback):
        try:
            shutil.rmtree(self.name)
            log("TRACE", "Temp directory deleted.")
        except Exception as ex:
            log("ERROR", "Could not delete temporary directory: %s (%s)" % (self.name, ex))
        return exc_type is None


def get_manifest_raw(args):
    """
    Return the raw content of the manifest (i.e. the text file). It
    needs to be parsed to be of any practical use.
    """
    # If we're given a path to a manifest file, use it
    if args.manifest_location:
        manifest_fn = args.manifest_location
        if os.path.exists(manifest_fn):
            log("INFO", "Using manifest file at location: {}".format(manifest_fn))
            with open(manifest_fn, 'r') as manifest_file:
                manifest_raw = manifest_file.read()
        else:
            log("WARN", "Invalid manifest file specified: {}, using default"
                .format(manifest_fn))
            manifest_raw = _MANIFEST_CONTENTS       
    # Otherwise, use the CMake Magic manifest
    else:
        manifest_raw = _MANIFEST_CONTENTS
    log("TRACE", "Raw manifest contents: {}".format(manifest_raw))
    return manifest_raw


def parse_manifest(manifest_contents):
    """Parse the manifest file, returns a dictionary of potential targets"""
    manifest = {}
    for line in manifest_contents.split('\n'):
        line_unpacked = line.split()
        try:
            # Check that the line isn't empty or a comment
            if not line_unpacked or line.strip().startswith('#'):
                continue

            target, repo_hash, url, sha256_hash = line_unpacked
            manifest[target] = {"repo_hash": repo_hash,
                                "url": url,
                                "sha256_hash": sha256_hash,
                               }
        except ValueError:
            log("WARN", "Warning: Invalid line in manifest file:\n"
                "         {}".format(line))
            continue
    return manifest


def parse_inventory(inventory_fn):
    """Parse the inventory file, returns a dictionary of installed files"""
    try:
        if not os.path.exists(inventory_fn):
            log("INFO", "No inventory file found at {}. Creating an empty one.".format(inventory_fn))
            return {}
        with open(inventory_fn, 'r') as inventory_file:
            # TODO: verify the contents??
            return json.load(inventory_file)
    except Exception as ex:
        log("WARN", "Error parsing the inventory file. Assuming an empty inventory: {}".format(ex))
        return {}


def write_inventory(inventory, inventory_fn):
    """Writes the inventory to file"""
    try:
        with open(inventory_fn, 'w') as inventory_file:
            json.dump(inventory, inventory_file)
            return True
    except Exception as ex:
        log("ERROR", "Error writing the inventory file. Contents may be incomplete or corrupted.\n"
                     "Error message: {}".format(ex))
        return False


def lookup_urls(regex_l, manifest, inventory, refetch=False):
    """Takes a list of RegExs to match within the manifest, returns a list of tuples with
    (hash, URL) that match the targets and are not in the inventory"""
    selected_targets = []
    # Store whether or not we've found a target in the manifest that matches the requested type
    found_one = False
    for target in manifest.keys():
        # Iterate through the possible targets in the manifest.
        # If any of them match any of the RegExs supplied, add the URL to the
        # return list
        if all(map((lambda regex: re.findall(regex, target)), regex_l)):
            found_one = True
            log("TRACE", "Selected target: {}".format(target))
            target_info = manifest.get(target)
            target_url = target_info.get("url")
            target_hash = target_info.get("repo_hash")
            target_sha256 = target_info.get("sha256_hash")
            filename = os.path.basename(target_url)
            # Check if the same filename and hash appear in the inventory
            if not refetch and inventory.get(target, {}).get("repo_hash", "") == target_hash:
                # We already have this file, we don't need to download it again
                log("INFO", "Target {} is up to date.".format(target))
            else:
                # We don't have that exact file, add it to the list
                selected_targets.append({"target": target,
                                         "repo_hash": target_hash,
                                         "filename": filename,
                                         "url": target_url,
                                         "sha256_hash": target_sha256})
    if not found_one:
        log("INFO", "No targets matching '{}'".format(regex_l))
    return selected_targets


def print_target_list(manifest, args):
    """
    Print a list of targets.
    """
    char_offset = max(len(x) for x in manifest.keys())
    if not args.url_only:
        # Print a couple helpful lines,
        # then print each (Target, URL) pair in the manifest
        log("INFO", "Potential targets in manifest file:\n"
                    "{} : {}".format(
                        "# TARGET".ljust(char_offset),
                        "URL" if args.base_url else "RELATIVE_URL"))
        for key, value in sorted(manifest.items()):
            print("{target} : {base}{relpath}".format(
                target=key.ljust(char_offset),
                base=args.base_url,
                relpath=value["url"]))
    else:
        for manifest_item in manifest.items():
            print(args.base_url+manifest_item[1]["url"])


def parse_auth(env):
    """
    Parse a (potentially None) environment variable string as a username:password
    for consumption by requests.
    """
    username = env.get(_USERNAME_VARIABLE, None)
    password = env.get(_PASSWORD_VARIABLE, None)
    if not username and not password:
        return None

    if not username:
        raise RuntimeError(
            "Password variable {} was specified, so username variable {} must be set as well"
            .format(_PASSWORD_VARIABLE, _USERNAME_VARIABLE)
        )

    if not password:
        raise RuntimeError(
            "Username variable {} was specified, so password variable {} must as well"
            .format(_USERNAME_VARIABLE, _PASSWORD_VARIABLE)
        )

    return (username, password)


def download(
        images_url,
        filename,
        buffer_size=_DEFAULT_BUFFER_SIZE,
        print_progress=False,
        download_limit=None
    ):
    """ Run the download, show progress """
    download_limit = download_limit or _DEFAULT_DOWNLOAD_LIMIT
    log("TRACE", "Downloading {} to {}".format(images_url, filename))
    auth = parse_auth(os.environ)
    try:
        resp = requests.get(images_url, stream=True, proxies=_PROXIES,
                            headers={'User-Agent': 'UHD Images Downloader'},
                            auth=auth)
    except TypeError:
        # requests library versions pre-4c3b9df6091b65d8c72763222bd5fdefb7231149
        # (Dec.'12) workaround
        resp = requests.get(images_url, prefetch=False, proxies=_PROXIES,
                            headers={'User-Agent': 'UHD Images Downloader'},
                            allow_redirects=True, auth=auth)
    if resp.status_code != 200:
        raise RuntimeError("URL does not exist: {}".format(images_url))
    filesize = float(resp.headers.get('content-length', -1))
    if filesize > download_limit:
        if not ask_permission(
                "The file size for this target ({:.1f} MiB) exceeds the "
                "download limit ({:.1f} MiB). Continue downloading?".format(
                    filesize/1024**2, download_limit/1024**2)):
            return 0, 0, ""
    if filesize == -1:
        if not ask_permission(
                "The file size for this target could not be determined. "
                "Continue downloading?"):
            return 0, 0, ""
    filesize_dl = 0
    base_filename = os.path.basename(filename)
    if print_progress and not sys.stdout.isatty():
        print_progress = False
        log("INFO", "Downloading {}, total size: {} kB".format(
            base_filename, filesize/1000 if filesize > 0 else "-unknown-"))
    with open(filename, "wb") as temp_file:
        sha256_sum = hashlib.sha256()
        for buff in resp.iter_content(chunk_size=buffer_size):
            if buff:
                temp_file.write(buff)
                filesize_dl += len(buff)
                sha256_sum.update(buff)
            if print_progress:
                status = r"%05d kB " % int(math.ceil(filesize_dl / 1000.))
                if filesize > 0:
                    status += r"/ %05d kB (%03d%%) " % (
                        int(math.ceil(filesize / 1000.)),
                        int(math.ceil(filesize_dl * 100.) / filesize),
                    )
                status += base_filename
                if os.name == "nt":
                    status += chr(8) * (len(status) + 1)
                else:
                    sys.stdout.write("\x1b[2K\r")  # Clear previous line
                sys.stdout.write(status)
                sys.stdout.flush()
    if print_progress:
        print('')
    if filesize <= 0:
        filesize = filesize_dl
    return filesize, filesize_dl, sha256_sum.hexdigest()


def delete_from_inv(target_info, inventory, images_dir):
    """
    Uses the inventory to delete the contents of the archive file specified by in `target_info`
    """
    target = inventory.get(target_info.get("target"), {})
    target_name = target.get("target")
    log("TRACE", "Removing contents of {} from inventory ({})".format(
        target, target.get("contents", [])))
    dirs_to_delete = []
    # Delete all of the files
    for image_fn in target.get("contents", []):
        image_path = os.path.join(images_dir, image_fn)
        if os.path.isfile(image_path):
            os.remove(image_path)
            log("TRACE", "Deleted {} from inventory".format(image_path))
        elif os.path.isdir(image_path):
            dirs_to_delete.append(image_fn)
        else: # File doesn't exist
            log("WARN", "File {} in inventory does not exist".format(image_path))
    # Then delete all of the (empty) directories
    for dir_path in dirs_to_delete:
        try:
            if os.path.isdir(dir_path):
                os.removedirs(dir_path)
        except os.error as ex:
            log("ERROR", "Failed to delete dir: {}".format(ex))
    inventory.pop(target_name, None)
    return True


def extract(archive_path, images_dir, test_zip=False):
    """
    Extract the contents of `archive_path` into `images_dir`

    Returns a list of files that were extracted.

    If test_zip is set, it will verify the zip file before extracting.
    """
    log("TRACE", "Attempting to extracted files from {}".format(archive_path))
    with zipfile.ZipFile(archive_path) as images_zip:
        # Check that the Zip file is valid, in which case `testzip()` returns
        # None.  If it's bad, that function will return a list of bad files
        try:
            if test_zip and images_zip.testzip():
                log("ERROR", "Could not extract the following invalid Zip file:"
                             " {}".format(archive_path))
                return []
        except OSError:
            log("ERROR", "Could not extract the following invalid Zip file:"
                         " {}".format(archive_path))
            return []
        images_zip.extractall(images_dir)
        archive_namelist = images_zip.namelist()
        log("TRACE", "Extracted files: {}".format(archive_namelist))
        return archive_namelist


def update_target(target_info, temp_dir, images_dir, inventory, args):
    """
    Handle the updating of a single target.
    """
    target_name = target_info.get("target")
    target_sha256 = target_info.get("sha256_hash")
    filename = target_info.get("filename")
    temp_path = os.path.join(temp_dir, filename)
    # Add a trailing slash to make sure that urljoin handles things properly
    full_url = urljoin(args.base_url+'/', target_info.get("url"))
    _, downloaded_size, downloaded_sha256 = download(
        images_url=full_url,
        filename=temp_path,
        buffer_size=args.buffer_size,
        print_progress=(_LOG_LEVEL <= _LOG_LEVELS.get("INFO", 3))
    )
    if downloaded_size == 0:
        log("INFO", "Skipping target: {}".format(target_name))
        return
    log("TRACE", "{} successfully downloaded ({} Bytes)"
        .format(temp_path, downloaded_size))
    # If the SHA256 in the manifest has the value '0', this is a special case
    # and we just skip the verification step
    if target_sha256 == '0':
        log("DEBUG", "Skipping SHA256 check for {}.".format(full_url))
    # If the check fails, print an error and don't unzip the file
    elif downloaded_sha256 != target_sha256:
        log("ERROR", "Downloaded SHA256 does not match manifest for {}!"
            .format(full_url))
        return
        # Note: this skips the --keep option, so we'll never keep image packages
        #       that fail the SHA256 checksum
    ## Now copy the contents to the final destination (the images directory)
    delete_from_inv(target_info, inventory, images_dir)
    if os.path.splitext(temp_path)[1].lower() == '.zip':
        archive_namelist = extract(
            temp_path,
            images_dir,
            args.test)
        if args.keep:
            # If the user wants to keep the downloaded archive,
            # save it to the images directory and add it to the inventory
            shutil.copy(temp_path, images_dir)
            archive_namelist.append(filename)
    else:
        archive_namelist = []
        shutil.copy(temp_path, images_dir)
    ## Update inventory
    inventory[target_name] = {"repo_hash": target_info.get("repo_hash"),
                              "contents": archive_namelist,
                              "filename": filename}


def main():
    """
    Main function; does whatever the user requested (download or list files).

    Returns True on successful execution.
    """
    args = parse_args()
    images_dir = get_images_dir(args)
    log("INFO", "Images destination: {}".format(os.path.abspath(images_dir)))
    try:
        manifest = parse_manifest(get_manifest_raw(args))
        if args.list_targets:
            print_target_list(
                manifest,
                args
            )
            return True
        log("TRACE", "Manifest:\n{}".format(
            "\n".join("{}".format(item) for item in manifest.items())
        ))

        # Strip down path until we find first parent directory of images_dir
        # that exists and then check if we have proper write permissions
        images_dir_parent = images_dir
        while not os.path.exists(os.path.abspath(images_dir_parent)):
            path_pair = os.path.split(images_dir_parent)
            if images_dir_parent == path_pair[0]:
                # Prevent infinite loop
                break
            images_dir_parent = path_pair[0]
        if not os.access(os.path.abspath(images_dir_parent), os.W_OK):
            log("ERROR", "Invalid permissions to write images destination")
            return False

        # Read the inventory into a dictionary we can perform lookups on
        default_inventory_fn = os.path.join(images_dir, _INVENTORY_FILENAME)
        if args.inventory_location:
            if os.path.isfile(args.inventory_location):
                inventory_fn = args.inventory_location
            else:
                log("WARN", "Invalid inentory file specified: {}, using default: {}"
                    .format(args.inventory_location, default_inventory_fn))
                inventory_fn = default_inventory_fn
        else:
            inventory_fn = default_inventory_fn
        inventory = parse_inventory(inventory_fn=inventory_fn)
        log("TRACE", "Inventory: {}\n{}".format(
            os.path.abspath(inventory_fn),
            "\n".join("{}".format(item) for item in inventory.items())
        ))

        # Determine the URLs to download based on the input regular expressions
        if not args.types:
            types_regex_l = [_DEFAULT_TARGET_REGEX]
        else:
            types_regex_l = args.types

        log("TRACE", "RegExs for target selection: {}".format(types_regex_l))
        targets_info = lookup_urls(types_regex_l, manifest, inventory, args.refetch)
        # Exit early if we don't have anything to download
        if targets_info:
            target_urls = [info.get("url") for info in targets_info]
            log("DEBUG", "URLs to download:\n{}".format(
                "\n".join("{}".format(item) for item in target_urls)
            ))
        else:
            return True

        ## Now download all the images archives into a temp directory
        if args.dry_run:
            for target_info in targets_info:
                log("INFO", "[Dry Run] Fetch target: {}".format(
                    target_info.get("filename")))
            return True
        with TemporaryDirectory() as temp_dir:
            for target_info in targets_info:
                update_target(
                    target_info,
                    temp_dir,
                    images_dir,
                    inventory,
                    args
                )
        ## Update inventory with all the new content
        write_inventory(inventory, inventory_fn)

    except Exception as ex:
        log("ERROR", "Downloader raised an unhandled exception: {ex}\n"
            "You can run this again with the '--verbose' flag to see more information\n"
            "If the problem persists, please email the output to: {contact}"
            .format(contact=_CONTACT, ex=ex))
        # Again, we wait on Windows systems because if this is executed in a
        # window, and immediately fails, the user doesn't have a way to see the
        # error message, and if they're not very savvy, they won't know how to
        # execute this in a shell.
        if not _YES and platform.system() == 'Windows':
            input('Hit Enter to continue.')
        return False
    log("INFO", "Images download complete.")
    return True

# Placing this near the end of the file so we don't clutter the top
_MANIFEST_CONTENTS = """@CMAKE_MANIFEST_CONTENTS@"""
if __name__ == "__main__":
    sys.exit(not main())