def convert_text_simple(text): '''This function convert a tex document in a single wikitext output collapsing the content at the -1 level''' config = {"input_path":filein, "output_path":fileout, "doc_title":"texla_min", "lang":"it", "keywords":json.loads( open('lang.txt').read())['it'], "collapse_content_level":-1, "base_path":'', "create_index":0, } logging.info('######## STARTING PARSING ########') p = Parser(config) tree = p.parse(text) n_blocks = tree.n_blocks() logging.info('PARSED %i Blocks', n_blocks) logging.info('######## STARTING RENDERING ########') #rendering rend = MediaWikiRenderer(config) rend.start_rendering(tree) #collpasing tree tree = rend.tree tree.collapse_content_level(config['collapse_content_level']) tree.collapse_urls() tree.fix_references() #getting text of root_page output_text = rend.tree.root_page.text return output_text logging.info('Finished')
def execute_texla_mediawiki(config): p = Parser(config) a = open(config['input_path'], 'r').read() try: tree = p.parse(a) except (PreparserError, ParserError) as err: err.print_error() err.print_complete_tree('debug/crash_tree') exit() f = open(config['output_path'] + '.tree', 'w') json_tree = tree.print_raw_tree() n_blocks = tree.get_number_blocks() logging.info('PARSED %i Blocks', n_blocks) f.write(json_tree) logging.info( '\033[0;34m############### STARTING RENDERING ###############\033[0m') #creating Reporter reporter = Reporter(tree) #rendering rend = MediaWikiRenderer(config, reporter) rend.start_rendering(tree) o = open(config['output_path'] + '.json', 'w') o.write(json.dumps(rend.tree.get_tree_json(), indent=3)) #print page tree before POST-PROCESSING logging.info('PageTree:\n' + rend.tree.get_tree_debug()) #collpasing logging.info( '\033[0;34m############### STARTING POST-PROCESSING ###############\033[0m' ) tree = rend.tree tree.collapse_tree(config['collapse_content_level'], config['collapse_pages_level']) #printing tree after POST-PROCESSING logging.info('PageTree:\n' + rend.tree.get_tree_debug()) oc = open(config['output_path'] + '-coll.json', 'w') oc.write(json.dumps(rend.tree.get_tree_json(), indent=3)) logging.info('\033[0;34m############### EXPORTING ###############\033[0m') if config['create_index']: tree.create_indexes(config["export_book_page"]) exporter.exportPages(tree.pages, config['output_path'] + '.mw', config['export_format']) if config['export_single_pages']: exporter.export_singlePages(tree.pages, config['output_path'] + '_pages', config['export_format']) if config['export_pages_tree']: exporter.export_pages_tree(tree.pages.values(), config['output_path'] + "_pages") reporter.print_report(console=True) logging.info('Finished')
def execute_texla_mediawiki(config): logging.info('######## STARTING PARSING ########') p = Parser(config) a = open(config['input_path'], 'r').read() tree = p.parse(a) f = open(config['output_path'] + '.tree', 'w') json_tree = tree.to_json(0) n_blocks = tree.n_blocks() logging.info('PARSED %i Blocks', n_blocks) f.write(json_tree) logging.info('######## STARTING RENDERING ########') #rendering rend = MediaWikiRenderer(config) rend.start_rendering(tree) o = open(config['output_path'] + '.json', 'w') o.write(json.dumps(rend.tree.get_tree_json(), indent=3)) p = open(config['output_path'] + '.debug', 'w') p.write(json.dumps(rend.used_tags, indent=2)) #print page tree before POST-PROCESSING logging.info('PageTree:\n'+rend.tree.get_tree_debug()) #collpasing logging.info('######## STARTING POST-PROCESSING ########') tree = rend.tree tree.collapse_tree(config['collapse_content_level'], config['collapse_pages_level']) #printing tree after POST-PROCESSING logging.info('PageTree:\n'+rend.tree.get_tree_debug()) oc = open(config['output_path'] + '-coll.json', 'w') oc.write(json.dumps(rend.tree.get_tree_json(), indent=3)) logging.info('######## STARTING EXPORTING ########') if config['create_index']: tree.create_indexes() exporter.exportPages(tree.pages, config['output_path'] + '.mw', config['export_format']) if config['export_single_pages']: exporter.export_singlePages(tree.pages, config['output_path'] + '_pages', config['export_format']) if config['export_pages_tree']: exporter.export_pages_tree(tree.pages.values(), config['output_path'] + "_pages") logging.info('Finished')