Record what published decisions say, one row per case

This is a research aid, not legal advice, and it does not give an opinion on anything. It records what a page says and where it said it. Every entry must be checked against the primary source before it is relied on for anything.

Once a week it runs your standing search against the one website you nominate, opens each decision it finds there, and appends a row per case: case name, citation, court, decision date, the outcome as the page states it, a passage copied from the page, the address it was read from, and a note of what the page did not say.

Three rules hold it to the source. It only ever opens pages on the host you named, checked address by address rather than trusted to the search engine. Anything the page does not state is left as an empty cell and named in the last column - never guessed, and never filled in from what a model already knows about a case. And the quoted passage is checked against the text that was actually fetched: if those words are not in the page, the quote and the outcome are both emptied and the row says so, because a model that is not copying cannot be trusted to have summarised either. An empty row that points you at the source beats a plausible invented holding.

It never says what a case means, what it establishes, or how it applies to anyone. Those are your job.

Set up: paste your spreadsheet address into 'sheetUrl', describe what you are looking for in 'researchQuestion', and put the one site to read in 'sourceHost' - just the host, exactly as it appears in that site's own web addresses, for example 'www.bailii.org'. Rows are appended to a tab named Case notes in this column order: Recorded on, Case name, Citation, Court, Decision date, Outcome as stated on the page, Quoted passage, Source, What the page did not state. The Source column is also how it knows what it has already recorded, so leave it alone. It reads at most 5 pages per run; change 'maxCasesPerRun' to move that.