Extract data from SEC EDGAR website (from Form 10-K; 82,955 .txt links provided)

I’m interested in collecting information about employee unionization from public company annual reports (Form 10-K). Each 10-K contains several standardized sections. The labor union info I’m interested in is located in “Item 1. Business” and “Item 1A. Risk Factors”. Step 1: Access the links to Form 10-K .txt files (N=82955) in the attached read file and search Item 1 and Item 1A ONLY for the keywords below… KEYWORDS: collective bargaining, collective-bargaining, CBA, labo(u)r union(s), labo(u)r agreement(s), labo(u)r contract(s), labo(u)r organization(s), union agreement(s), union contract(s), union organization(s), or union(s) Step 2: If one of the above keywords matches the text in Item 1 or 1A, add the entire sentence (or paragraph, whichever is easier) with the match to new field/column in the read file. Maybe the output file could have one field for any Item 1 output and a second field for any Item 1A output. Appendix C and Appendix D of the attached research paper (pg36-37) provide some examples of the union-related text that I’m looking for. Step 3: (If possible) Create 3 different union-related variables (binary, percentage, number) from the extracted Item 1 sentences/paragraphs. Create a separate set of 3 union-related variables from the Item 1A text. First, identify whether the union-related statement is positive or negative (i.e. employees are represented/covered by a union V.S. employees are NOT in a union, none of our employees are represented) with a binary variable (=1 for (some) union representation and =0 for no representation). Second, extract the percentage of employees covered if available. Third, extract the number of employees covered if available. I realize this last part is tricky to do mechanically. I’ll have to check this part manually anyways, so any progress here with a reasonable error rate will be appreciated.