Neural search in one* line of code

one_line_neural_search.py

exec("""import os; from pprint import pprint; import pretty_errors; from jina.types.arrays.memmap import DocumentArrayMemmap; from jina import Flow, Document, DocumentArray; from jina.parsers.helloworld import set_hw_chatbot_parser; import csv; from appstore_config import my_port, my_workdir, my_datafile; from executors import MyTransformer, MyIndexer;\ndef trim_string(input_string, word_count=50, sep=" "): sanitized_string = input_string.replace("\\n", sep).replace("\\u2022", sep); words = sanitized_string.split(sep)[:word_count]; output = " ".join(words); return output;\ndef prep_docs(input_file, max_docs=1000):\n\tdocs = DocumentArray();\n\twith open(input_file, "r") as csv_file:\n\t\tcsv_reader = csv.DictReader(csv_file); input_field = "Description";\n\t\tfor row in csv_reader: input_data = trim_string(row[input_field]); doc = Document(text=input_data); doc.tags = row; docs.extend([doc]);\n\treturn docs;\ndef run_appstore(inputs, args):\n\tf = Flow().add(uses=MyTransformer, parallel=args.parallel).add(uses=MyIndexer, workspace=args.workdir);\n\twith f: f.post(on="/index", inputs=inputs, on_done=print); f.use_rest_gateway(args.port_expose); f.block();\nargs = set_hw_chatbot_parser().parse_args(); args.port_expose = my_port; args.workdir = my_workdir; docs = prep_docs(input_file=my_datafile); run_appstore(inputs=docs, args=args)\n""")

This technically works and is a great brag for how simple it is to use Jina

Or it would be a great brag if it were decent code. Reader, this code is not decent. This is not a place of honor, no highly esteemed deed is commemorated here