diff options
Diffstat (limited to 'gather.py')
| -rwxr-xr-x | gather.py | 179 |
1 files changed, 179 insertions, 0 deletions
diff --git a/gather.py b/gather.py new file mode 100755 index 0000000..1948d8e --- /dev/null +++ b/gather.py @@ -0,0 +1,179 @@ +#!/usr/bin/python3 +from sys import argv +import logging +from time import localtime, mktime, time +import requests +from base64 import b64decode +try: + from sqlalchemy import Table, MetaData, Integer, BigInteger, String, Column, Table, ForeignKey, create_engine, select + from sqlalchemy.orm import declarative_base, relationship, Session +except ModuleNotFoundError: + raise ModuleNotFoundError("emerge dev-python/sqlalchemy or pip install SQLAlchemy") +try: + from bs4 import BeautifulSoup +except ModuleNotFoundError: + raise ModuleNotFoundError("emerge dev-python/beautifulsoup4 or pip install beautifulsoup4") + +if len(argv) != 1+1: + raise ValueError("1st argument specifies the db URL in sqlalchemy format. example for sqlite: sqlite:///db") + +engine = create_engine(argv[1], echo=True, future=True) + +Base = declarative_base() + +class Book(Base): + __tablename__ = "books" + isbn = Column(BigInteger, primary_key=True, nullable=False, doc="book isbn. found in URL http://www/isbn/987 and in acsm: resource, dc:identifier (sometimes not), thumbnailURL") + title = Column(String, nullable=True, doc="title of the book, dcc:title in acsm") + creator = Column(String, nullable=True, doc="author of the book, dc:creator in acsm") + publisher = Column(String, nullable=True, doc="publisher of the book, dc:publisher in acsm") + identifier = Column(String, nullable=True, doc="if dc:identifier can't be derived from isbn, it's stored here. if dc:identifier element is missing, a literal string noidentifier is stored.") + thumbnail_extension = Column(String, nullable=True, doc="thumbnails come in predictable URLs, derived from ISBN, apart from the extension. I've observed both jpg and png") + format = Column(String, nullable=True, doc="format of the file. I've seen application/pdf and application/epub+zip") + language = Column(String, nullable=True, doc="language of the book. I've seen sl.") + borrows = relationship("Borrow", back_populates="book"); + def __repr__(self): + return f"Book(isbn={self.isbn!r}, title={self.title!r}, creator={self.creator!r}, publisher={self.publisher!r})" + +class Borrow(Base): + __tablename__ = "borrows" + id = Column(Integer, primary_key=True, nullable=False, doc="id in transaction element of acsm or in filename of acsm on http") + isbn = Column(ForeignKey("books.isbn"), nullable=False, doc="foreign key that leads to a book") + purchase = Column(String, nullable=True, doc="acsm purchase element: iso8601 of purchase of book, including timezone") + expiration = Column(String, nullable=True, doc="acsm expiration element: iso8601 of expiration of acsm, including timezone") + obtained = Column(BigInteger, nullable=False, doc="UNIX timestamp when this borrow was obtained as acsm from http") + book = relationship("Book", back_populates="borrows") + def __repr__(self): + return f"Borrow(id={self.id!r}, isbn={self.isbn!r}, purchase={self.purchase!r}, expiration={self.expiration!r}, obtained=mktime({localtime(self.obtained)!r}), book={self.book!r})" + +logging.basicConfig(level=logging.NOTSET) +logger = logging.getLogger(argv[0]) +logger.debug("welcome to %s", argv[0]) + +Base.metadata.create_all(engine) + +starting_acsm_id = 177238 + +logger.debug(f"created metadata.") +force_acsm_id = 0 + +valid_acsms = 0 +only_isbn_acsms = 0 +failed_acsms = 0 + +try: + with Session(engine) as session: + while True: + if force_acsm_id != 0: + acsm_id = force_acsm_id + force_acsm_id = 0 + else: + borrow = session.scalars(select(Borrow).order_by(Borrow.id.desc())).first() + acsm_id = starting_acsm_id + if borrow is None: + logger.info(f"oooh, it looks like this is a fresh start, db contains no borrows. I'll start with hardcoded acsm id {starting_acsm_id}") + else: + logger.info(f"continuing from latest {borrow}") + acsm_id = borrow.id+1 + r = requests.get(f"https://www.biblos.si/izposoja/prenesi/{acsm_id}.acsm") + if r.status_code != 200: + logger.info(f"we are done for now, as server responded with {r.status_code} for queried acsm id {acsm_id}. latest borrow is {borrow}") + break + elif r.text.startswith('<error xmlns="http://ns.adobe.com/adept" data="E_URLLINK_NO_SUCH_RESOURCE resid urn:uuid:00000000-1002-0000-0009-78'): + isbn = int(r.text.split()[4].split("-").pop())+int(9e12) + borrow = Borrow(id=acsm_id, isbn=isbn, obtained=int(time())) + logger.warning(f"received 'no such resource' from server and stored a quite empty {borrow}") + session.add(borrow) + session.commit() + only_isbn_acsms += 1 + elif r.text.startswith('<error xmlns="http://ns.adobe.com/adept" data="E_URLLINK_PARAMETER_SYNTAX rights lrt http://cs.alliance.inkbook.eu:443/fulfillment/URLLink.acsm"/>'): + logger.warning(f"received urllink parameter syntax error with no usable data for acsm {acsm_id}, so I did not store anything") + force_acsm_id = acsm_id+1 + failed_acsms += 1 + else: + acsm = BeautifulSoup(r.text, "xml") + ft = acsm.fulfillmentToken + expected = f"ACS-BIBL-L-{acsm_id}" + if ft.transaction.string != expected: + raise ValueError(f"expected {expected} in transaction.string, but instead received {ft.transaction.string} in acsm {acsm_id}") + isbn = int(ft.resourceItemInfo.resource.string.split("-").pop())+int(9e12) + identifier_is_isbn = True + identifier_to_isbn = 0 + identifier = "noidentifier" + try: + identifier = ft.resourceItemInfo.metadata.identifier.string + identifier_to_isbn = int(identifier.split(":").pop().replace("-", "")) + except (ValueError, AttributeError): + identifier_is_isbn = False + if identifier_to_isbn == 0: + identifier_is_isbn = False + expected = ft.resourceItemInfo.resource.string + if ft.licenseToken.resource.string != expected: + raise ValueError(f"expected {expected} in ft.resourceItemInfo.licenseToken.resource.string but instead received {ft.resourceItemInfo.licenseToken.resource.string} in acsm {acsm_id}") + uuid = expected.split(":").pop() + expected = f"https://cs.alliance.inkbook.eu/books/{uuid}." + if ft.resourceItemInfo.metadata.thumbnailURL.string.startswith(expected) != True: + raise ValueError(f"expected {expected} in ft.resourceItemInfo.metadata.thumbnailURL.string but instead received {ft.resourceItemInfo.metadata.thumbnailURL.string} in acsm {acsm_id}") + thumbnail_extension = ft.resourceItemInfo.metadata.thumbnailURL.string.split(".").pop() + duration = int(ft.resourceItemInfo.licenseToken.permissions.display.duration.string) + if duration != int(ft.resourceItemInfo.licenseToken.permissions.play.duration.string): + raise ValueError(f"expected {duration} in fr.int(resourceItemInfo.licenseToken.permissions.play.duration.string) but instead received {int(resourceItemInfo.licenseToken.permissions.play.duration.string)} in acsm {acsm_id}") + hmac = b64decode(ft.hmac.string, validate=True) + title = ft.resourceItemInfo.metadata.find(name="dc:title").string + creator = ft.resourceItemInfo.metadata.creator.string + publisher = ft.resourceItemInfo.metadata.publisher.string + language = ft.resourceItemInfo.metadata.language.string + format = ft.resourceItemInfo.metadata.format.string + purchase = ft.purchase.string + expiration = ft.expiration.string + if identifier_is_isbn: + identifier = None + book = session.get(Book, isbn) + if book == None: + book = Book(identifier=identifier, isbn=isbn, title=title, creator=creator, publisher=publisher, thumbnail_extension=thumbnail_extension, language=language, format=format) + else: + book.identifier = identifier + book.isbn = isbn + book.title = title + book.creator = creator + book.publisher = publisher + book.thumbnail_extension = thumbnail_extension + book.language = language + book.format = format + borrow = Borrow(id=acsm_id, isbn=isbn, purchase=purchase, expiration=expiration, obtained=int(time()), book=book) + logger.info(f"found a new {borrow!r}") + session.add(borrow) + session.commit() + valid_acsms += 1 +except KeyboardInterrupt: + logger.warning(f"Keyboard interrupt. Exiting. I hope this terminated cleanly. Last requested acsm was discarded.") + +logger.info(f"In this session, {valid_acsms} valid acsms were stored, {only_isbn_acsms} acsms had only isbn and no other data available and {failed_acsms} acsms failed to be received. Last valid requested acsm was {acsm_id}. Thank you for cooperation.") + +""" +metadata = MetaData() +books = Table( + "books", + metadata, + Column("title", String, nullable=False, doc="title of the book, dcc:title in acsm"), + Column("creator", String, nullable=False, doc="author of the book, dc:creator in acsm"), + Column("publisher", String, nullable=False, doc="publisher of the book, dc:publisher in acsm") +) +borrows = Table( + "borrows", + metadata, + Column("id", Integer, primary_key=True, nullable=False, doc="id in transaction element of acsm or in filename of acsm on http"), + Column("isbn", ForeignKey(books.c.isbn), nullable=False, doc="foreign key that leads to a book"), + Column("purchase", String, nullable=False, doc="acsm purchase element: iso8601 of purchase of book, including timezone"), + Column("expiration", String, nullable=False, doc="acsm expiration element: iso8601 of expiration of acsm, including timezone"), + Column("duration", String, nullable=False, doc="acsm duration element, specifying borrow time of the book") +) +metadata.create_all(engine) +""" + +""" +with sqlalchemy.orm.Session(engine) as session: + result = session.execute(sqlalchemy.text("CREATE TABLE IF NOT EXISTS borrows (id INT PRIMARY KEY, isbn INT NOT NULL, purchase TEXT NOT NULL, expiration TEXT NOT NULL, title TEXT NOT NULL, creator TEXT NOT NULL, publisher TEXT NOT NULL, duration INT NOT NULL, hmac BLOB NOT NULL) STRICT")) + session.commit() +# cur.execute("CREATE TABLE IF NOT EXISTS borrows(id INT PRIMARY KEY, isbn INT NOT NULL, purchase TEXT NOT NULL, expiration TEXT NOT NULL, title TEXT NOT NULL, creator TEXT NOT NULL, publisher TEXT NOT NULL, duration INT NOT NULL, hmac BLOB NOT NULL) STRICT") +""" |