Ejemplo n.º 1
0
def reading_article_xml(file_xml_path, move_success=True):

    article = files.read_file(file_xml_path)
    obj_xml = etree.fromstring(article)
    medias = xml.find_medias(obj_xml)

    if medias:
        logger.info("%s possui midias", file_xml_path)

    if move_success:
        files.move_xml_conversion2success(
            file_xml_path.replace(config.get("CONVERSION_PATH"), ""))
Ejemplo n.º 2
0
def conversion_article_xml(file_xml_path):
    article = files.read_file(file_xml_path)

    obj_xml = etree.fromstring(article)
    obj_html_body = xml.parser_body_xml(obj_xml)

    # sobrecreve o html escapado anterior pelo novo xml tratado
    remove = obj_xml.find("body/p")
    remove.getparent().replace(remove, obj_html_body)

    new_file_xml_path = os.path.join(
        config.get("CONVERSION_PATH"), os.path.split(file_xml_path)[1]
    )
    files.write_file(new_file_xml_path, etree.tostring(obj_xml).decode("utf-8"))
def manage_error_file(errors, err_file, converted_file):
    if os.path.isfile(err_file):
        try:
            os.unlink(err_file)
        except:
            pass

    if errors:
        msg = []
        for err, data in errors.items():
            msg.append(err)
            msg.extend([
                "{}:{}".format(ln, text)
                for ln, text in zip(data["lineno"], data["message"])
            ])

        files.write_file(
            err_file,
            "%s %s\n%s" %
            (files.read_file(converted_file), "=" * 80, "\n".join(msg)),
        )
Ejemplo n.º 4
0
def get_document_renditions(
    folder: str, file_prefix: str, storage: object
) -> List[dict]:
    """Faz o envio das manifestações do pacote com base no arquivo
    manifest.json e retorna o resultado do envio destas manifestações
    para o object store."""

    mimetypes = MimeTypes()
    _renditions = []

    try:
        _manifest_json = json.loads(
            files.read_file(os.path.join(folder, "manifest.json"))
        )
    except Exception as exc:
        logger.error("Could not read manifest: %s", str(exc))
    else:
        _manifest = {lang: urlparse(url).path for lang, url in _manifest_json.items()}
        logger.debug("Renditions lang and legacy url: %s", _manifest)

        for lang, legacy_url in _manifest.items():
            rendition = os.path.basename(legacy_url)
            _mimetype = mimetypes.guess_type(rendition)[0]
            _rendition_path = os.path.join(folder, rendition)
            _rendition = {
                "filename": rendition,
                "url": storage.register(
                    _rendition_path, file_prefix, _manifest.get(lang)
                ),
                "size_bytes": os.path.getsize(_rendition_path),
                "mimetype": _mimetype,
                "lang": lang,
            }

            _renditions.append(_rendition)

    return _renditions
Ejemplo n.º 5
0
 def test_read_file(self):
     data = files.read_file(
         os.path.join(SAMPLES_PATH, "S0036-36341997000100001.xml"))
     self.assertIn("0036-3634", data)
Ejemplo n.º 6
0
def read_json_file(file_path: str) -> List[dict]:
    """Ler um arquivo JSON e retorna o resultado
    em formato de estruturas Python"""

    return json.loads(files.read_file(file_path))