Skip to content
polite-crawler

Opening-hours parser

Venues write their hours in many ways. This parser reads the text of a page and gives a week. The week shows the intervals of each day, the closed days, and the days that the text does not state. Select an example, or paste your own text.

The parser is fromText from polite-crawler, compiled into this page. Your text stays in your browser.

Page text

Mon–Thu 16:30–21:00

Mon
Tue
Wed
Thu
Frinot stated
Satnot stated
Sunnot stated

What the parser reads

Section titled “What the parser reads”
  • Days before or after the times. “Mon–Thu 11am–10pm” and “11am–9pm Mon–Thu” give the same week.
  • Days and times on separate lines. A day on one line goes with the time range on the next line.
  • Two columns. Seven day names and then seven time ranges go together in sequence. Columns of different lengths give no result, because the parser cannot be sure.
  • Closed days. “Closed Mondays” closes Monday.
  • Two services. Lunch and dinner with a gap between them stay two intervals.
  • Late nights. A close at or before the opening, for example 11:00 am – 1:00 am, ends on the next day.
  • A heading with one range. One range below “Hours” or after “Open”, with no day, applies to each day. The parser marks the result as a guess.

When a time has no am or pm, for example “11-9”, the parser chooses the likely meaning and marks the result as a guess. A guess does not look like a fact.

Structured hours

Section titled “Structured hours”

The parser also reads two notations:

  • schema.org. The openingHours strings and the openingHoursSpecification objects of JSON-LD.
  • OpenStreetMap. The opening_hours tag, for example Mo-Fr 11:00-21:00, Sa-Su 12:00-20:00; PH off. A later rule replaces an earlier rule for its days. The parser does not guess a rule that it cannot fully read, for example a rule with months or with sunrise.

A day that no source states is unknown, not closed. A venue with hours from more than one source can thus show where the sources disagree.