RISolve

Folder

• ./data
  • cache -> cache for overview tests
  • expected
    • overview -> expected xml links of law_ids

Add new law text

Tests

• Getting paragraphs from law_id (risparser::overview::test::parse())
  • Create file law_id in ./data/expected/overview (then run tests to get current output + save in file)

Features (to be moved to lib.rs one-by-one)

• Text to structured law
  • LawBuilder: Structure law, by specifying (sub-)sections (new_header), its description (new_desc), paragraphs under the current (sub-)section (new_par), and the description of the next paragraph (new_next_para_header). Classifier need to be set.
    • Main output: Properly structured laws (Law)
  • Law: Represents a structured law text. Can be generated with LawBuilder.
    • Main output: properly formatted (md for a start) law text, no need to export Heading/... etc
• RIS Fetcher (to be mocked)
  • all paragraphs of specific law (overview)
  • xml document from url (par/mod.rs fetch_age)
• Parser
  • replace errors w/ config file

Integration test

• Nice test would be to re-create html ris file and compare it (problem with custom fixes, though)

History

• I've created my first parser using RIS API, daily updated. Failed because I tried to do too much automatically (e.g. recognizing headers
• Using print-website, I've extracted stuff w/ regex.
• Tried to create a parser using print-website, proper(-ish) parser

Goals

• I want to have the text of the law.
• I want to see the structure (proper headers) of the law.
• I want to be able to make comments (e.g. Erschöpfung) on certain parts
• I want to see since when this paragraph is in use.
• [~] Lawtext should be updateable

Technical

• I don't want to restrict myself with a parser combinators but code it myself using recursive descent parser.
• Be strict in what I process. Fail if anything unexpected happens. The user should handle this case. It's fine if one decides to ignore the new/unexpected field, but it should be done deliberately.

Progress / Functions

• Parse structure of law into struct using Deserilize trait, pot. multiple requests (if > 100 paragraphs)
• Parse risdok using own RD parser, again strict: fail if anything not expected happens, not sure (yet) if I want to operate on strings, or first parse using off-the-shelve XML reader (prob. 2nd option)

Next step

• Parse ABGB
• Create config files for laws
  • law_id
  • replace stuff
  • headers
• Create argument parse
  • --law mschg.conf

Naming

• Law ("Gesetz"): e.g. UHG, TEG, ABGB
• Section ("Paragraph")
• Subsection ("Absatz")
• Item ("Ziffer")
• Heading-{1,2,3,...}

"Scripts"

• Retrieve overview law: curl -X POST "https://data.bka.gv.at/ris/api/v2.6/Bundesrecht" -H "Content-Type: application/x-www-form-urlencoded" -d "Applikation=BrKons" -d "Gesetzesnummer=10001899" -d "DokumenteProSeite=OneHundred" -d "Seitennummer=1" -d "Fassung.FassungVom=2023-11-03" | jq . > law.json