Copyright © 2004 Gene Michael Stover. All rights reserved. Permission to copy, store, & view this document unmodified & in its entirety is granted.
These are my own notes for programmers (mostly me). These are techniques for making printers do things & for programs which do things to printers.
``Did you said hacking?'' Indeed. I'm usng the definition of hack from The Jargon File as maintained by Eric S. Raymond. [3] I guess I'm specifically using part 6 of that definition. Anyway, if you are looking for a way to crack the passwords on printers, go away. And if you are thinking ``You can break into a printer'', you're pretty much correct.
Many (most?) modern LaserJet & similar printers have built-in web servers. You can point your web browser at them if you know their IP address, & you'll be looking at a web site which shows whether the printer is currently printing, how much paper & toner are loaded into it, & maybe other details such as the page counts, page description languages it understands, & network protocls it speaks. You can often change settings on the printer unless someone has set an administrator password. (By default, they almost never have an administrator password.)
When you tell an application to print a document (or picture or file or whatever), here's what happens:
On Microsloth Winders, the application obtains a Device Context (HDC) to the printer, then draws the image with the same API it used to draw on the screen.
On unix-like systems, the system-specific imaging language is less standardized. Sometimes there isn't one, & the application is responsible for the conversion. In this case, the application will probably convert directly to a page description language.
If the application had to convert directly to a page description language, this might be a null-step. Then again, it might not. For example, an applicaition could have converted its document to PostScript, but the printer driver might convert that to PCL or some other language. Gnu Ghostscript is sometimes used in this way.
It's probable that the printer driver doesn't communicate directly with the printer. In that caes, the printer driver probably gives the page description program (wrapped in job control statements) to a communication layer which is often called a port driver.
We use page description languages to tell printers what to draw. With a printer language, you (or, more likely, a program) can tell a printer to put this much toner of this colour at that location on the paper.
The two main page description languages are PostScript and PCL. In fact, they are the only page description languages of which I'm aware. I'm sure there are many other page description languages, most of them specific to particular printer models. Happily, they are on the wane, leaving us with two standard page description languages.
PostScript is a supremely cool language. It was created by Adobe in the 1980s, & Adobe still controls the language standard.
I love PostScript. [4]
PostScript is sort of like Forth. Implementations can be very efficient. It's portable. It's Turing-complete; with the proper I/O primitives, it could compute anything that higher level programming languages could compute. In fact, the PostScript language standard specifies a fairly versatile & complete set of I/O primitives for files, & many PostScript implementations include those I/O primitives.
If you want to learn PostScript, start with [1]. If you write a few programs with the information in that book & then find yourself wishing you had more complete information about the language, it's time to read [2].
PCL is another page description language. It originated at Hewlet Packard. I know very little about PCL, but from what I've seen, it's not nearly as neat-o as PostScript. From what I've seen, PCL is to PostScript as COBOL is to Lisp. fixme: Where is the PCL language specification?
Interestingly, the installable file for PostScript printer drivers for Microsloth Winders are less than half a megabyte, whereas the installable file for PCL printer drivers for Microsloth Winders area always considerably larger, often being something like five megabytes.
Line printers, dot matrix printers, & their ilk don't speak page description languages. Instead, they work on a simpler principle. They assume you give them data which they print in lines. They have simple control codes, such as carriage return, line feed, form feed, & maybe a back-space. The important thing is that they do not speak a page description language. They assume the data you give htem is the data to print.
If you prefer, you might ocnsider their control codes to be a page description language. If someone told me that's the way it was, I wouldn't disagree, but notice that control codes make a very, very primitive & inflexible page description language when compared to PostScript.
I will not discuss line printers & dot matrx printers any more in this file.
SNMP stands for Simple Network Management Protocol, though I believe a more appropriate name would have been UCBDMP (Unnecessarily Complicated and Badly Described Manglement Protocol) because, where most RFCs are wonderfully understandable & useable, those of SNMP are not.
SNMP views network entities (those that respond to SNMP) as collections of key/value pairs. The keys are lengthy strings of integers, though the integers are sometimes expresed with alphabetic mnemonics.
The keys have a hierarchical structure, much like
directories in a modern file system except that,
where the separator between directory names is ``/'',
``\'', :'', or possibly ``>'', the
separator between the parts of an SNMP key is a dot
(``.''). (By the way, this is not how the SNMP
standard describes things.)
The basic operations of SNMP are:
For more information about SNMP, see the relevant RFCs. There are a ton of them. One location of RFCs is the RFC Editor.
Hewlet-Packard's LaserJets set the standard.
| attribute | note | oid |
| MAC addy | 1.3.6.1.2.1.2.2.1.6.1 |
| attribute | note | oid |
| okidata | oki-model | 1.3.6.1.4.1.2001.1.3.1.1.11.1.0 |
| oki-model | oki-model | 1.3.6.1.4.1.2001.1.3.1.1.11.2.0 |
| sys-model | sys-model | 1.3.6.1.2.1.25.3.2.1.3.1 |
| is-lying-oki | is-lying-oki | 1.3.6.1.4.1.2001.1.3.1.1.11.1.0 |
Okay, so you are wondering how the okidata part could hold so many different values. You're asking why it can't be simple & always hold the same value. And I used to ask the same questions. Then I realized that with Okidata printers, everything is different. Well, not everything, but lotsof thins. Sometimes it's ``Oki Data'', other times it's ``OKIDATA'', still other times, it's empty. With Okis, each new model might ben an adventure.
When the oki-model OIDs don't give you want you want, try the sys-model OID.
Yet again, you're asking when to use which & why must it be so unpredictably different? And again I say, Because that's how the Okidata company works.
You can encode PJL commands in SNMP requests.
fixme: Document how to do tihs, why & when to do it.
You can at least do this for the model & the page count. Can you do it for other data? What models might require you to do this?