preprocessor: add ellipsis sanitiser

Some commands have long output that is largely uninteresting to the
user. In such situations, it is typical to trim that output and
"summarise" with '...'.

The ellipsis sanitiser does just that. It takes an optional Start field
which indicates from which line number the ellipsis should start.

Preprocessor-No-Write-Cache: true
Signed-off-by: Paul Jolly <[email protected]>
Change-Id: Ia285f810eb880b2e9b09833d82ab3b91bd0710a9
Dispatch-Trailer: {"type":"trybot","CL":1176701,"patchset":7,"ref":"refs/changes/01/1176701/7","targetBranch":"alpha"}

internal/cmd/preprocessor/cmd/page.go

@@ -177,14 +177,23 @@ func (p *page) parseSanitiser(m json.RawMessage) (sanitiserMatcher, error) {
 
 	switch kind {
 	case "patternSanitiser":
-		var c patternSanitiserMatcher
-		if err := json.Unmarshal(m, &c); err != nil {
+		var ps patternSanitiserMatcher
+		if err := json.Unmarshal(m, &ps); err != nil {
 			return nil, fmt.Errorf("failed to unmarshal %s: %v", kind, err)
 		}
-		if err := c.init(); err != nil {
+		if err := ps.init(); err != nil {
 			return nil, fmt.Errorf("failed to init %s: %v", kind, err)
 		}
-		return &c, nil
+		return &ps, nil
+	case "ellipsisSanitiser":
+		var es ellipsisSanitiserMatcher
+		if err := json.Unmarshal(m, &es); err != nil {
+			return nil, fmt.Errorf("failed to unmarshal %s: %v", kind, err)
+		}
+		if err := es.init(); err != nil {
+			return nil, fmt.Errorf("failed to init %s: %v", kind, err)
+		}
+		return &es, nil
 	default:
 		return nil, fmt.Errorf("unknown sanitiser: %q", kind)
 	}

internal/cmd/preprocessor/cmd/sanitisers.go

@@ -77,6 +77,7 @@ func (m *matchSpec) init() error {
 	if m.CommandPrefix != "" {
 		cmd = m.CommandPrefix
 	}
+	// k.Step is only != "" when we are matching based on step
 	file, err := syntax.NewParser(syntax.KeepComments(true)).Parse(strings.NewReader(cmd), " ")
 	if err != nil {
 		return fmt.Errorf("failed to parse %q: %v", cmd, err)
@@ -187,3 +188,26 @@ func (p *patternSanitiserMatcher) init() error {
 	}
 	return nil
 }
+
+// An ellipsisSanitiser allows very long output to be removed and replaced with
+// the canonical '...' which is intended to indicate "and there is is more not
+// shown here".
+type ellipsisSanitiser struct {
+	Start int `json:"start"`
+}
+
+func (e *ellipsisSanitiser) sanitise(cmd *commandStmt) error {
+	if strings.Count(cmd.Output, "\n") <= e.Start {
+		return nil
+	}
+	lines := strings.Split(cmd.Output, "\n")
+	lines = append(lines[:e.Start], "...")
+	cmd.Output = strings.Join(lines, "\n") + "\n" // re-add trailing newline
+	return nil
+}
+
+type ellipsisSanitiserMatcher struct {
+	kind
+	ellipsisSanitiser
+	matchSpec
+}

internal/cmd/preprocessor/cmd/schema.cue

@@ -20,7 +20,10 @@ package preprocessor
 		}
 	}
 
-	#sanitiser: _kind & #patternSanitiser // build up a disjunction of all sanitisers
+	#sanitiser: _kind & {
+		#patternSanitiser |
+		#ellipsisSanitiser
+	}
 	#matchingSanitiser: {
 		_matcher
 		#sanitiser
@@ -50,6 +53,15 @@ package preprocessor
 		#patternSanitiser
 	}
 
+	// Instances of #ellipsisSanitiser summarise long output with '...'
+	#ellipsisSanitiser: {
+		_kind
+		kind: "ellipsisSanitiser"
+
+		// start defines the line from which to consider an ellipsis
+		start?: int
+	}
+
 	#comparator: _kind & (#patternComparator | #unstableLineOrderComparator)
 	#matchingComparator: {
 		_matcher

internal/cmd/preprocessor/cmd/testdata/execute_ellipsis_sanitiser.txtar

@@ -0,0 +1,61 @@
+# Ensure that the ellipsis sanitiser works.
+
+unquote content/dir/en.md
+
+# Run the preprocessor
+exec preprocessor execute
+
+# Verify that the target files matche the source contents
+cmp hugo/content/en/dir/index.md golden/hugo/content/en/dir/index.md
+
+-- hugo/.keep --
+-- content/dir/site.cue --
+package site
+
+content: dir: page: {
+	leftDelim:  "{{{"
+	rightDelim: "}}}"
+
+	sanitisers: [
+		{
+			kind: "ellipsisSanitiser"
+			command: "seq 1 5"
+		},
+		{
+			kind:  "ellipsisSanitiser"
+			start: 5
+			command: "seq 1 10"
+		},
+	]
+}
+-- content/dir/en.md --
+>---
+>title: JSON Superset
+>---
+>
+>{{{with script "en" "zero"}}}
+>seq 1 5
+>{{{end}}}
+>
+>{{{with script "en" "five"}}}
+>seq 1 10
+>{{{end}}}
+-- golden/hugo/content/en/dir/index.md --
+---
+title: JSON Superset
+---
+
+```text { title="TERMINAL" codeToCopy="c2VxIDEgNQo=" }
+$ seq 1 5
+...
+```
+
+```text { title="TERMINAL" codeToCopy="c2VxIDEgMTAK" }
+$ seq 1 10
+1
+2
+3
+4
+5
+...
+```