sed Simulator Guide

Replacing the first occurrence on each line differs from replacing every occurrence in the file. sed Simulator shows that distinction on a separate output, without modifying the pasted source.

First occurrence versus global

Use pattern cat, replacement dog, and this sample:

cat cat
Cat and dog
nothing to replace

With global replacement enabled and case sensitivity unchanged, the output is:

dog dog
Cat and dog
nothing to replace

Turn global off and the first line becomes dog cat. The second line remains unchanged until Ignore case is enabled. Global means all non-overlapping occurrences within each input line, not an instruction to process only one line or the entire input as a single pattern space.

Enable Only print lines with a substitution to suppress lines on which no substitution occurs. A matching replacement that inserts the same text still counts as a substitution.

Insert the complete match

With extended regex on, use [[:digit:]]+ and replacement [&]:

attempt=12 elapsed=345

The global result is:

attempt=[12] elapsed=[345]

& refers to the complete match. \& inserts a literal ampersand, \\ a literal backslash, and \n or \t insert a newline or tab. Replacement group references such as \1 are rejected. Groups in the pattern are useful for matching structure and alternation, but captures are not exposed.

Whole matches use leftmost-longest selection: cat|cater matches cater in cater, rather than choosing cat just because it is the first alternative.

Delete and print operations

Select Delete matching lines with ^DEBUG to remove entire lines that start with DEBUG. It does not remove only the matching substring. Select Print matching lines only with the same pattern to keep those lines instead. Replacement, global, and substitution-print options are hidden for these operations because they do not apply.

Deleting every line can legitimately produce empty output. It is not a parser error. The source remains intact, and changing the sample or options invalidates the previous result until you simulate again.

Supported regex, not every sed feature

The simulator accepts ASCII patterns with literals, ., branch-start/end ^ and $, bracket expressions, common POSIX classes such as [[:digit:]], groups, alternation, and *, +, ?, or bounded intervals. With extended regex disabled, groups and intervals use BRE escapes, and GNU \+, \?, and \| are supported. Empty groups/alternatives, collating/equivalence classes, backslashes inside bracket expressions, and GNU shorthand or word-boundary escapes are not supported.

Matching uses ASCII/C-locale rules. Unicode and binary/NUL input are rejected; do not assume the same behavior as a machine using a different locale. The simulator processes lines separately, preserving whether the source ended with a newline. It does not simulate address ranges, hold space, loops, multiple commands, in-place edits, or reads/writes to files.

For a real terminal command, sed Replacement Builder remains a separate tool. Test generated commands on disposable data before editing actual files. For finding lines rather than changing them, use the separate grep Simulator.

References

• GNU sed manual
• GNU sed: regular expressions