Get a state session law's text, current or a past version

The act's text exactly as printed, one page of about 30,000 characters at a time, cut at a paragraph break where one is near. Follow nextCursor while hasMore is true; pageStart / pageEnd / totalChars are character offsets into the whole text, and concatenating the pages in order reproduces it.

Cost: 6 credits per page served; each page is charged on its own, so you can stop whenever you have enough. pagePrice is what this page cost, totalPages the estimated pages in all, and estimatedRemainingCredits the pages still to come times the price.

Cost of reading a law. Reading a whole law costs estimatedBodyPages times the price of one body page (6 credits). estimatedBodyPages is on the record next to charCount and is the most pages the text can take: a page holds up to 30,000 characters and is cut back to a paragraph break, so a long act usually takes fewer. It is null when the text cannot be read (it can be when served, or when withheld for a reason we label and /body reads with includeUnverified). Most acts take a page or two. Each body response carries pagePrice and estimatedRemainingCredits, so you can stop whenever you have enough.

Worked example: the longest act we hold, Ohio's 2025 budget act (HB 96, 9,571,132 characters), takes at most 399 pages, 2,394 credits to read in full.

Where a page starts. Follow nextCursor from the page before, or send startChar, a character offset into the text (0-based, the scale of pageStart and totalChars), to jump to a place you know. Send one or the other, never both (a 422, uncharged). A startChar page costs and sizes the same as any other. startChar has no version binding of its own, so add expectTextSha256 (the textSha256 you read the text under): if the text was re-collected since, the answer is a 409 and nothing is charged. There is no page-size parameter: a page is always about 30,000 characters, so its price is fixed.

Attribution. Every outcome carries citation, jurisdiction, session, approvedDate and sourceUrl (the official copy on a government host, present when the text is withheld or not held too), so a page is attributable on its own.

Id only. This route takes a sessionLawId, not a citation: resolve a citation once with GET /us/session-laws/{sessionLawId} and use the id it returns.

Not served, not charged. When we do not serve the text (textStatus is not held), the answer is a 200 with available: false, a reason (not_held, withheld, pending or empty_text), creditsConsumed: 0, the withheldReason for a withheld law, and the official sourceUrl. A federal id (SAL_...) is a 404 (federal_session_law); read federal session laws with GET /us/statutes/section/{actId}. Errors (unknown id 404, stale cursor or mismatched expectTextSha256 409, malformed cursor or startChar 422) are not charged, and a failure on our side (500, 503) is refunded. The parameters and the 409 response give the exact conditions.

Past versions: version. We keep every version of a law's text that we served or adopted. List them with GET /us/session-laws/{sessionLawId}/versions, then send one of those numbers as version; omit it for the current version. The response then names version and isCurrentVersion, and the page, nextCursor and textSha256 belong to that version. The price is unchanged: 6 credits per page, whichever version you read. A version the law does not have is a 404, refunded. Without version and includeUnverified the response is the plain current-text page.

Text we hold back: includeUnverified. A law whose text has a defect we have not cleared is withheld, and by default /body answers available: false, free. Send includeUnverified=true to read its text when the reason it is held back is one we can label. The page then carries textReliability (unverified: a check has not been completed, or known_issue: we have evidence of the defect), a stable reasonCode and a one-sentence notice to show beside the text: the text may show deleted (struck) or vetoed law as if it were enacted, so check the printed act at sourceUrl before relying on it. The codes that serve the text, flagged, are markup_unverified, strike_unchecked, struck_text_shown_as_current, struck_text_merged, veto_unchecked and licence_unconfirmed. Some text is not served even when you opt in, and answers available: false, free, with the reasonCode and no text: a law whose licence bars use (licence_barred), text that failed our own validation (validation_failed), a session we cannot yet tell apart from another (session_unresolved), a source we have not yet audited (collector_unaudited) and a reason we have not classified (unclassified). A served law with includeUnverified=true reads as usual and says textReliability: verified. Such a page costs 6 credits like any other and can be combined with version. The fields version, isCurrentVersion, textReliability, reasonCode and notice are omitted unless you asked for them.

Amendatory markup. An amending act prints the language it deletes and inserts. Where amendatoryMarkup is preserved, that markup is in text in the convention markupConvention names: wdiff writes deleted language [-like this-] and inserted language {+like this+}; publisher_literal keeps the publisher's own markers as printed (Minnesota's new text begin and deleted text begin); mixed has both. Where it is lost, the publisher printed markup our text does not keep, so deleted words can read as law: check the source before quoting. Only the as-printed text is served; there is no clean "as enacted" view.

Example

curl "https://api.vaquill.ai/api/v1/us/session-laws/SSL_MN_2025S1_G_Y2025_C1/body" \
  -H "Authorization: Bearer $VAQUILL_API_KEY"

# jump to a character offset, pinned to the version of the text you read
curl "https://api.vaquill.ai/api/v1/us/session-laws/SSL_MN_2025S1_G_Y2025_C1/body?startChar=29987&expectTextSha256=02693d5b55d4709b" \
  -H "Authorization: Bearer $VAQUILL_API_KEY"

# an older version (list the versions first at .../versions), or a withheld law's labelled text
curl "https://api.vaquill.ai/api/v1/us/session-laws/SSL_MN_2025S1_G_Y2025_C1/body?version=1" \
  -H "Authorization: Bearer $VAQUILL_API_KEY"
curl "https://api.vaquill.ai/api/v1/us/session-laws/SSL_LA_2025R_G_C2/body?includeUnverified=true" \
  -H "Authorization: Bearer $VAQUILL_API_KEY"

const MY_BUDGET = 100; // the most you will spend to finish reading, in credits
let cursor;
let text = "";
do {
  const url = new URL("https://api.vaquill.ai/api/v1/us/session-laws/SSL_MN_2025S1_G_Y2025_C1/body");
  if (cursor) url.searchParams.set("cursor", cursor);
  const res = await fetch(url, {
    headers: { Authorization: `Bearer ${process.env.VAQUILL_API_KEY}` },
  });
  const page = await res.json();
  if (!page.available) break; // withheld or not held: page.sourceUrl is the official copy
  text += page.text;
  if (page.estimatedRemainingCredits > MY_BUDGET) break; // stop whenever you have enough
  cursor = page.nextCursor;
} while (cursor);

Related endpoints

• GET /us/statutes/resolve answers a citation that names a state session law rather than a code section with a sessionLaw block (matched, or ambiguous with every candidate), pointing here. It never changes resolved or section.
• GET /us/statutes/coverage lists, per jurisdiction under sessionLaws, the sessions held and whether each is complete or partial. A law missing from a partial session may still exist.
• Federal session laws (the Statutes at Large, SAL_ ids) are read with GET /us/statutes/section/{actId}, not here.

Nulls. Every response field is present: one with no value is null ([] for a list). The examples here leave the nulls out. publisher, provenance and session.coverageStatus on a law, and coverage on a list page with rows, are absent when they do not apply.

Rate limits. Calls count against your key's per-minute budget like every Data API route; a 429 carries Retry-After.