feat(bible): corpus metadata sidecars + parser

7 files changed, 75 insertions, 1 deletions

diff --git a/internal/bible/bible.go b/internal/bible/bible.go
index 1603188..1bc8983 100644
--- a/internal/bible/bible.go
+++ b/internal/bible/bible.go
@@ -8,7 +8,7 @@ import (
 	"sync"
 )
 
-//go:embed corpora/wuj.tsv corpora/vul.tsv corpora/grb.tsv corpora/drb.tsv
+//go:embed corpora
 var corporaFS embed.FS
 
 // Verse is a single verse.
diff --git a/internal/bible/corpora/drb.ini b/internal/bible/corpora/drb.ini
new file mode 100644
index 0000000..fc5ac52
--- /dev/null
+++ b/internal/bible/corpora/drb.ini
@@ -0,0 +1,4 @@
+; lectio corpus metadata
+lang = en
+name = Douay-Rheims (English)
+psalm_system = drb
diff --git a/internal/bible/corpora/grb.ini b/internal/bible/corpora/grb.ini
new file mode 100644
index 0000000..0d297ff
--- /dev/null
+++ b/internal/bible/corpora/grb.ini
@@ -0,0 +1,4 @@
+; lectio corpus metadata
+lang = el
+name = Greek
+psalm_system = vulgate
diff --git a/internal/bible/corpora/vul.ini b/internal/bible/corpora/vul.ini
new file mode 100644
index 0000000..4efb6e5
--- /dev/null
+++ b/internal/bible/corpora/vul.ini
@@ -0,0 +1,4 @@
+; lectio corpus metadata
+lang = la
+name = Vulgate (Latin)
+psalm_system = vulgate
diff --git a/internal/bible/corpora/wuj.ini b/internal/bible/corpora/wuj.ini
new file mode 100644
index 0000000..8364923
--- /dev/null
+++ b/internal/bible/corpora/wuj.ini
@@ -0,0 +1,4 @@
+; lectio corpus metadata
+lang = pl
+name = Wujek (Polish)
+psalm_system = vulgate
diff --git a/internal/bible/corpusmeta.go b/internal/bible/corpusmeta.go
new file mode 100644
index 0000000..04b9dcb
--- /dev/null
+++ b/internal/bible/corpusmeta.go
@@ -0,0 +1,45 @@
+package bible
+
+import "github.com/lukaszkasprzak/lectio/internal/ini"
+
+// CorpusMeta is the sidecar metadata for a bible corpus (<code>.ini).
+type CorpusMeta struct {
+	Code, Lang, Name, PsalmSystem, Sigla string
+}
+
+// parseCorpusMeta reads a section-less sidecar. ini.Parse returns
+// []ini.Section{Name, Pairs}; keys before any [section] land in the section
+// whose Name == "" (verified against internal/ini).
+func parseCorpusMeta(code string, data []byte) CorpusMeta {
+	m := CorpusMeta{Code: code}
+	secs, err := ini.Parse(data)
+	if err != nil {
+		return m
+	}
+	for _, s := range secs {
+		if s.Name != "" {
+			continue
+		}
+		for _, p := range s.Pairs {
+			switch p.Key {
+			case "lang":
+				m.Lang = p.Val
+			case "name":
+				m.Name = p.Val
+			case "psalm_system":
+				m.PsalmSystem = p.Val
+			case "sigla":
+				m.Sigla = p.Val
+			}
+		}
+	}
+	return m
+}
+
+func embeddedCorpusMeta(code string) (CorpusMeta, bool) {
+	data, err := corporaFS.ReadFile("corpora/" + code + ".ini")
+	if err != nil {
+		return CorpusMeta{}, false
+	}
+	return parseCorpusMeta(code, data), true
+}
diff --git a/internal/bible/corpusmeta_test.go b/internal/bible/corpusmeta_test.go
new file mode 100644
index 0000000..2a089bd
--- /dev/null
+++ b/internal/bible/corpusmeta_test.go
@@ -0,0 +1,13 @@
+package bible
+
+import "testing"
+
+func TestEmbeddedCorpusMeta(t *testing.T) {
+	m, ok := embeddedCorpusMeta("drb")
+	if !ok {
+		t.Fatal("drb sidecar not found")
+	}
+	if m.Lang != "en" || m.PsalmSystem != "drb" || m.Name == "" {
+		t.Fatalf("bad drb meta: %+v", m)
+	}
+}