scanner

Documentation

Overview

Package scanner is a tiny combined lexer+parser that emits lines of printer output, page breaks, and end-of-job indications. It specifically handles reading from Hercules sockdev printer output of JES2 jobs in MVS 3.8J, although if there is a reliable way to detect the last line of a job (which must immediately be followed by a form feed character) with a regular expression, the code can be easily adapted for other operating systems.

We tolerate a variety of combinations of CR, LF, and FF. A bare CR will cause the next line to overtype the current line. Bare LF, CR+LF, or LF+CR have the effect of CR+LF.

Lines will be trimmed to 132 bytes; additional bytes on a line will be discarded.

The implementation is a state machine that reads one byte at a time, updates internal state as necessary, performs actions (emit lines, pages, EOJ) as necessary, then sets the function to handle the next byte. Thus the current state of the state machine is represented by the value of the state function on the scanner struct.

The state machine is approximately an implementation of the following diagram in the dot language:

digraph {
	"get next byte" -> "have lf" [label="b=lf"];
	"get next byte" -> "have cr" [label="b=cr"];
	"get next byte" -> "emit line and page" [label="b=ff"];
	"emit line and page" [shape=box];
	"emit line and page" -> "last line match EOJ pattern?";
	"last line match EOJ pattern?" [shape=box];
	"last line match EOJ pattern?" -> "emit EOJ" [label="yes"];
	"last line match EOJ pattern?" -> "get next byte" [label="no"];
	"have lf" -> "emit line" [label="b=cr"];
	"have lf" -> "emit line and await lf" [label="b=lf"];
	"have lf" -> "emit line and page" [label="b=ff"];
	"have lf" -> "get next byte" [label="b != lf | cr | ff"];
	"emit line and await lf" [shape=box];
	"emit line and await lf" -> "have lf";
	"have cr" -> "emit line" [label="b=lf"];
	"have cr" -> "emit line and await cr" [label="b=cr"];
	"have cr" -> "emit line and page" [label="b=ff"];
	"have cr" -> "get next byte" [label="b != lf | cr | ff"];
	"emit line and await cr" [shape=box];
	"emit line and await cr" -> "have cr";
	"emit line" [shape=box];
	"emit line" -> "get next byte";
	"get next byte" -> "add to current line";
	"add to current line" [shape=box];
	"add to current line" -> "dispose of bytes" [label="n>=132"];
	"add to current line" -> "get next byte" [label="n<132"];
	"dispose of bytes" -> "have lf" [label="b=lf"];
	"dispose of bytes" -> "have cr" [label="b=cr"];
	"dispose of bytes" -> "emit line and page" [label="b=ff"];
	"dispose of bytes" -> "dispose of bytes" [label="b != lf | cr | ff"];
	"emit EOJ" [shape=box];
	"emit EOJ" -> "get next byte";
}

Is this all slightly over-complicated for our needs? Perhaps.

Index

• func Scan(conn net.Conn, handler PrinterHandler, trace bool) error
• func ScanWithLogTag(conn net.Conn, handler PrinterHandler, trace bool, tag string) error
• type PrinterHandler

Functions

func Scan

func Scan(conn net.Conn, handler PrinterHandler, trace bool) error

Scan will read from a net.Conn, conn, which should be sent data from Hercules printer output. It will output lines (trimmed to 132 characters if necessary) and page breaks and identify the end of jobs in the printer data stream.

This function exists for backwards-compatibility and just calls ScanWithLogTag with the tag "default"

func ScanWithLogTag

func ScanWithLogTag(conn net.Conn, handler PrinterHandler, trace bool,
	tag string) error

Scan will read from a net.Conn, conn, which should be sent data from Hercules printer output. It will output lines (trimmed to 132 characters if necessary) and page breaks and identify the end of jobs in the printer data stream.

Types

type PrinterHandler

type PrinterHandler interface {
	AddLine(line string, linefeed bool)
	PageBreak()
	EndOfJob(jobinfo string)
}

PrinterHandler interface receives the output of printer output parsing.