{# The PDF's document information (#195). WeasyPrint reads Author, Subject and
Keywords off these tags, and the title and language are the skeleton's
own. Every value is a fact the document owns, escaped as an attribute.
Deliberately no dcterms.created or dcterms.modified: a date makes two
renders of one document differ, and the goldens rest on determinism.
Keywords split on commas in the PDF, which is why the list joins on one. #}
{% if firm_name %}
{% endif %}
{% if campaign_name or department or date_range %}
{% endif %}
{% if department or issue_label %}
{% endif %}
{% if stamp %}
{% include "common/stamp.html" %}
{% endif %}
{{ cover_html }}
{# Zero-height leaves where a following running box seeds its fallback.
A leaf, because a string set on an element spanning sheets -- the body,
the document table -- wins on every sheet it spans and the section
titles below would never show. After the cover, because an element
ahead of a named page opens a blank sheet of its own, which is why the
leaves take a turned page's name when the body opens on one (#341). #}
{{ contents_html }}{{ exhibits_html }}
{# The body table is frame.width, not frame.inner: every container computes
its column widths from the frame and pads inside that, so narrowing the
table here would overflow each split by its own padding. #}
{# One table per run of sheets laid the same way up (#341): a named page
applies to a block in the body's flow and never to a table row, so a
turned page is a table of its own, as wide as its frame. #}
{% for run in body_runs %}
{% endfor %}
{{ back_matter_html }}