krino: 0.0.5 — keyword cache, t and d in reviewv0.0.5

1 files changed, 28 insertions, 30 deletions

diff --git a/internal/engine/engine.go b/internal/engine/engine.go
index 8cffbd1..b50728e 100644
--- a/internal/engine/engine.go
+++ b/internal/engine/engine.go
@@ -8,6 +8,7 @@ package engine
 import (
 	"fmt"
 	"os"
+	"sort"
 	"strings"
 	"time"
 
@@ -26,6 +27,11 @@ type Engine struct {
 	Extract  *extract.Extractor
 	Now      func() time.Time // time.Now; tests replace it
 	MainFile string
+
+	// CacheDir holds each directory's keyword cache (spec §6.1), as
+	// NAME.cache; "" means no cache is read or written. Load leaves it
+	// empty: the command line sets it.
+	CacheDir string
 }
 
 // Dir is one configured directory, with its ignore matcher and rules
@@ -38,13 +44,11 @@ type Dir struct {
 	Ignore   *ignore.Matcher
 	Rules    []*Rule
 
-	// ContentVariants is the distinct (IgnoreCase, Fold) pairs any of
-	// Rules' content tests evaluate under, in first-seen order. B2: when
-	// this holds exactly one variant, facts.Content releases a file's raw
-	// extracted text once that variant's normalised copy exists, since no
-	// other variant will ever be asked for; with more than one, both must
-	// stay memoised, as before.
-	ContentVariants []cond.Options
+	// ContentKeywords is every content keyword the directory's excludes
+	// and rules test, once each, sorted by Key. When a file's text is
+	// extracted, every one of them is answered at once, so one extraction
+	// serves every content test and fills the keyword cache.
+	ContentKeywords []cond.Keyword
 
 	// Excludes are the (exclude ...) forms that apply here, compiled with the
 	// directory's settings: krino.conf's first, then the directory's own. A
@@ -135,7 +139,7 @@ func Load(mainFile string, names ...string) (*Engine, []*config.Diag) {
 				}
 			}
 		}
-		dir.ContentVariants = contentVariants(dir.Rules, dir.Excludes, dirOpt)
+		dir.ContentKeywords = contentKeywords(dir.Rules, dir.Excludes)
 		dir.DupScopes = dupScopes(dir.Rules)
 		dirs = append(dirs, dir)
 	}
@@ -216,32 +220,26 @@ func dedupeNames(names []string) []string {
 	return out
 }
 
-// contentVariants returns the distinct (IgnoreCase, Fold) pairs any content
-// test evaluates under - each exclude's (under the directory's settings,
-// dirOpt) and each rule's - in first-seen order — B2's per-Dir
-// ContentVariants. A rule whose condition has no content test at all
-// (Cond.UsesContent false) never calls facts.Content, so its resolved
-// case/fold settings contribute no variant here.
-func contentVariants(rules []*Rule, excludes []*Exclude, dirOpt cond.Options) []cond.Options {
-	var out []cond.Options
-	seen := map[cond.Options]bool{}
-	for _, x := range excludes {
-		if x.Cond.UsesContent && !seen[dirOpt] {
-			seen[dirOpt] = true
-			out = append(out, dirOpt)
+// contentKeywords returns every content keyword excludes and rules test,
+// each once, sorted by Key: Dir.ContentKeywords.
+func contentKeywords(rules []*Rule, excludes []*Exclude) []cond.Keyword {
+	seen := map[string]bool{}
+	var out []cond.Keyword
+	add := func(c *cond.Cond) {
+		for _, k := range c.Keywords {
+			if !seen[k.Key()] {
+				seen[k.Key()] = true
+				out = append(out, k)
+			}
 		}
 	}
+	for _, x := range excludes {
+		add(x.Cond)
+	}
 	for _, r := range rules {
-		if !r.Cond.UsesContent {
-			continue
-		}
-		opt := cond.Options{IgnoreCase: r.Settings.Case == config.CaseIgnore, Fold: r.Settings.Fold}
-		if seen[opt] {
-			continue
-		}
-		seen[opt] = true
-		out = append(out, opt)
+		add(r.Cond)
 	}
+	sort.Slice(out, func(i, j int) bool { return out[i].Key() < out[j].Key() })
 	return out
 }