def get_keywords_from_text(text_lines, taxonomy=None, output_mode="text", output_limit=CFG_BIBCLASSIFY_DEFAULT_OUTPUT_NUMBER, spires=False, match_mode="full", no_cache=False, with_author_keywords=False, rebuild_cache=False, only_core_tags=False): """Returns a formatted string containing the keywords for a single document.""" global _SKWS global _CKWS if not _SKWS: if taxonomy is not None: _SKWS, _CKWS = get_regular_expressions(taxonomy, rebuild=rebuild_cache, no_cache=no_cache) else: write_message("ERROR: Please specify an ontology in order to " "extract keywords.", stream=sys.stderr, verbose=1) text_lines = cut_references(text_lines) fulltext = normalize_fulltext("\n".join(text_lines)) author_keywords = None if with_author_keywords: author_keywords = get_author_keywords(_SKWS, _CKWS, fulltext) if match_mode == "partial": fulltext = _get_partial_text(fulltext) single_keywords = get_single_keywords(_SKWS, fulltext) composite_keywords = get_composite_keywords(_CKWS, fulltext, single_keywords) return _get_keywords_output(single_keywords, composite_keywords, taxonomy, author_keywords, output_mode, output_limit, spires, only_core_tags)
def get_keywords_from_local_file(local_file, taxonomy, rebuild_cache=False, match_mode="full", no_cache=False, with_author_keywords=False): text_lines = text_lines_from_local_file(local_file) global _SKWS global _CKWS if not _SKWS: if taxonomy is not None: _SKWS, _CKWS = get_regular_expressions(taxonomy, rebuild=rebuild_cache, no_cache=no_cache) else: write_message("ERROR: Please specify an ontology in order to " "extract keywords.", stream=sys.stderr, verbose=1) text_lines = cut_references(text_lines) fulltext = normalize_fulltext("\n".join(text_lines)) author_keywords = None if with_author_keywords: author_keywords = get_author_keywords(_SKWS, _CKWS, fulltext) if match_mode == "partial": fulltext = _get_partial_text(fulltext) single_keywords = get_single_keywords(_SKWS, fulltext) composite_keywords = get_composite_keywords(_CKWS, fulltext, single_keywords) return (single_keywords, composite_keywords)
def extract_author_keywords(skw_db, ckw_db, fulltext): """Finds out human defined keyowrds in a text string. Searches for the string "Keywords:" and its declinations and matches the following words. @var skw_db: list single kw object @var ckw_db: list of composite kw objects @var fulltext: utf-8 string @return: dictionary of matches in a formt { <keyword object>, [matched skw or ckw object, ....] } or empty {} """ akw = {} K = reader.KeywordToken for k, v in keyworder.get_author_keywords(skw_db, ckw_db, fulltext).items(): akw[K(k, type='author-kw')] = v return akw