Which part of the archiving standard to claim

In plain words

Written for anybody, with nothing taken for granted. It says what the rule is for and what it changes on the document you end up with.

The family is called PDF/A, and each of its members is called a part. A part is a full set of rules; a level, written as a letter after it, says how far that part goes. All of them promise the same thing first: everything the document needs to be read is inside the document.

Two questions settle the choice. Does the document have to be read out loud by software, or searched by paragraph rather than by word? Then it needs the level that carries the structure of its own text. Does it have to carry another file inside it — the data of an invoice, a spreadsheet, a photograph? Then it needs a part that allows that.

If neither question applies, the plainest claim is the right one: the page will look the same for as long as the file exists, and that is all that was asked.

For a developer

The same thing without the detour: the objects, the constraints, and where the line falls between what is produced and what is declared.

Four claims: part 3 at level B, part 3 at level A, part 4, and part 4 at the level that carries files. Part 3 rests on PDF 1.7 and part 4 on PDF 2.0, and the claim settles the version rather than following it: a document whose version and claim disagree is refused before a byte is written.

Level A of part 3 asks for a structure tree; the engine refuses to write the file without one. Part 4 drops the levels altogether — being readable by a machine became a standard of its own, PDF/UA, and a document that wants both claims both. Part 4 carries no information dictionary at all. The level of part 4 that carries files asks for at least one, and refuses the file that has none.

Through the rendering service, archive takes two of the four: pdfa3b and pdfa4. Level A and the level carrying files are planned for a future version of the request format — the first because a request does not yet describe a structure tree, the second because the one file a request carries today is an invoice, which part 3 already covers.

Clause by clause

The published text behind each requirement, what the engine measures the document against, and what it leaves to a validator.

Parts 1 and 2 of the family are not written, and neither is the engineering level of part 4. An archive that names one of those is asking for something this is not yet the tool for, and it is better heard here than found out afterwards.

What the engine does on its own

Nothing on this list has to be asked for. It comes out of a document that claims the standard, whatever else the calling program says.

• Writes the claim into the document's own description: the part, the level where the part has one, and the year of the edition.
• Refuses the document when the PDF version and the claimed part disagree, naming both.
• Writes no information dictionary at all under part 4, and drops the entries it would otherwise have derived.

What you supply

Nothing on this list can be guessed. The engine states what it was given and declines to invent the rest, which is why the document is refused rather than written with a value nobody chose.

• The claim itself: no document claims a standard by accident.
• The PDF version that goes with the part: 1.7 for part 3, 2.0 for part 4.
• A structure tree, for level A of part 3, and at least one carried file, for the level of part 4 that carries them.

The published texts behind it

Glossary

byte

The unit a file's weight is counted in, the way a parcel is counted in grams. A thousand of them make a kilobyte, and a million make a megabyte — the size of one photograph taken with a telephone. A one-page PDF here weighs between six thousand and a hundred and twenty thousand, so a hundred of them fit in the space of five photographs.

standard

A rule argued out in committee, published under a number anybody may buy and read, and identical for every firm claiming it. A claim to follow one can therefore be checked against the text. A way of working that merely spread because it worked is a habit of the trade: useful, widespread, and answerable to no text at all.

PDF/A

A PDF made to be readable in twenty years' time. Everything it needs is inside the file — the fonts, the colours, the description of what it is — so nothing it depends on can go missing later. Public bodies and archives ask for it by name.

PDF/UA

A PDF a blind reader can use. The file itself says which text is a heading, which is a paragraph, and in what order the software that reads a page aloud should say it all. Without that, the software reads the page in the order it was drawn, which is rarely the right one.

attachment

A whole file tucked inside a PDF and carried along with it, the way a paper clip holds a sheet to a folder. Opening the document is enough to get it back out. An electronic invoice works this way: the page a person reads carries the same amounts as a file accounting software reads.

structure tree

The list, held inside a tagged PDF, that says what each part of the page is and what it belongs to: this block is a section, this line its heading, this run of cells one row of that table. It is the shape of the document, kept apart from the ink.

request

One call to the service: you send what the document should say, and get the document back. Your bill counts these calls, one for each document. The number of pages inside a document is never counted.

Every word the site explains