The upgrade from scopus (now pybliometrics) 0.x to 1.x saw many changes in pybliometrics' internal architecture, but also in four classes (see change log): ScopusAbstract(), ScopusAffiliation(), ScopusAuthor() and ScopusSearch().
To avoid too many issues resulting from missing backward-compatibility, new classes were introduced to gradually replace other ones: AbstractRetrieval() (replacing ScopusAbstract()), AuthorRetrieval() (replacing ScopusAuthor()) and ContentAffiliationRetrieval() (replacing ScopusAffiliation()). The corresponding old classes will remain until pybliometrics 2.x but their maintenance has been suspended immediately. Cached files that were downloaded with the old classes are not usable by the new classes.
ScopusSearch() had to be revamped completely; code that uses ScopusSearch() has to be updated, but not significantly.
The change to scopus 1.x was guided by five principles: 1. Use json rather than xml for the cached files to reduce overhead and lower maintenance efforts 2. Align class names, script names, attribution names and names of folders with the names the Scopus API uses 3. Use properties to return a high share of information provided by Scopus, and get functions to increase user experience 4. Allow users to set and change configuration via a configuration file 5. Return namedtuples when Scopus provides combined information to increase interoperability with other python modules
Class AbstractRetrieval() replaces ScopusAbstract(). This class has seen the most changes. The following attributes have been renamed but their return value stays the same (so that simply renaming it will suffice): citationLanguage becomes language, citationType becomes srctype, citingby_url becomes citingby_link, scopus_url becomes scopus_link. There are some attributes which are now properties: bibtex becomes get_bibtex(), html becomes get_html(), ris becomes get_ris() and latex becomes get_latex(). Properties affiliations (new: affiliation), subjectAreas (new: subject_areas), authkeywords and authors are entirely different now: They return namedtuples. Please see the examples for how to use them. Property nauthors has been removed; use len(AbstractRetrieval(<eid>).authors instead. Finally, method get_corresponding_author_info() has been removed, as Scopus does not provide this information any more.
Class AuthorRetrieval() replaces ScopusAuthor(). The following properties have been renamed but their value stays the same: author_id becomes identifier, coauthor_url becomes coauthor_link, firstname becomes given_name, hindex becomes h_index, lastname becomes surname, name becomes indexed_name, ncited_by becomes cited_by_count, ncoauthors becomes coauthor_count, ndocuments becomes document_count. Property current_affiliation has been renamed to affiliation_current but the return value is now the Scopus ID of the affiliation. Property publication_history has been renamed to journal_history and returns a list of namedtuples rather than a list of tuples. Property affiliation_history now returns a list of Scopus IDs instead of a list of ScopusAffiliation() objects. Property subject_areas now returns a list of namedtuples instead of a list of tuples.
Class ContentAffiliationRetrieval() replaces ScopusAffiliation. It will suffice to replace the class name in your scripts and rename the following attributes: nauthors becomes author_count, ndocuments becomes document_count, name becomes affiliation_name, org_url becomes org_URL, api_url becomes self_link, scopus_id becomes identifier.
Class ScopusSearch() remains but was revamped. The search results are now cached under a hex-ed filename to allow for complex queries. Files are now saved in a different folder (by default). results is now the main property, returning a list of namedtuples containing all useful information regarding the search results. For convenience, get_eids() returns just the list of EIDs of the articles, and property EIDS, which will be removed in a future release, returns just this list.