sed

Home > Books > NetKernel API and Services Reference > Accessor Listing > Utility and System Accessors > sed

Module

The sed accessor is exported by the urn:org:ten60:netkernel:ext:sys module. Import this module to gain access to the accessor.

Syntax

URI

active:sed

Argument	Rules	Description
operand	Mandatory	resource to edit
operator	Mandatory	edit instructions

Example Usage

DPML

<instr>
  <type>sed</type>
  <operand>somefile.txt</operand>
  <operator>editcommands.sed</operator>
  <target>var:edited</target>
</instr>

NetKernel Foundation API

req=context.createSubRequest("active:sed");
req.addArgument("operand", [resource representation, aspect, or URI] );
req.addArgument("operator", [resource representation, aspect, or URI] );
result=context.issueSubRequest(req);

Purpose

The sed accessor provides a stream editor similar in intent to Unix sed. The operand provides the stream to be edited. The operator identifies the edit instructions to use - edit instructions are specified in a declarative XML syntax which can be static or generated dynamically. Dynamic editing patterns can be useful as a value substitution stage in, for example, an XRL pipeline process.

Processing Model

This accessor uses the Unix line-delimited ASCII processing model. The operand is treated as a line delimited resource. Each line (marked by either CR or LF or CRLF) is selected. All editing rules are applied in order to each line. The processed line is written to the output resource before the next line is selected. Processing terminates when all lines have been read.

Operator Edit Instruction Syntax

The syntax for the operator edit instructions takes the following form:

<sed>
  <pattern>
    <regex>Some Text</regex>
    <replace>Some Other Text</replace>
  </pattern>
            ...
        
</sed>

Any number of patterns may be specified. The editor applies the patterns in specified order to each line of the operand document.

Example

If the following resource is processed:

Mary had a little REPLACE1,
Its REPLACE2 was white as REPLACE3.

with this edit specification:

<sed>
  <pattern>
    <regex>REPLACE1</regex>
    <replace>lamb</replace>
  </pattern>
  <pattern>
    <regex>REPLACE2</regex>
    <replace>fleece</replace>
  </pattern>
  <pattern>
    <regex>REPLACE3</regex>
    <replace>snow</replace>
  </pattern>
</sed>

the following resource is produced:

Mary had a little lamb,
Its fleece was white as snow.

Regex

The regular expression value specified in the pattern is a full Java regex. Therefore just as with Unix sed, captured groups may be reinserted by using $<int> syntax in the replace statement. Here's an example:

If the following resource is processed:

This text is for editing.

with this edit specification:

<sed>
  <pattern>
    <regex>(.*?) is for (edit).*</regex>
    <replace>$2: $1 has been $2ed.</replace>
  </pattern>
</sed>

the following resource is produced:

edit: This text has been edited.