Changeset 0.17.3 (#78)

server/notelinks/extract.go

@@ -0,0 +1,50 @@
+package notelinks
+
+import (
+	"regexp"
+	"strings"
+
+	"git.gobha.me/xcaliber/chat-switchboard/models"
+)
+
+// wikiRe matches [[Title]], [[Title|Display]], ![[Title]], and ![[Title|Display]].
+// Title first char rejects [ and ] to avoid matching inside [[[triple]]].
+var wikiRe = regexp.MustCompile(`(!?)\[\[([^\[\]|][^\]|]*?)(?:\|([^\]]+?))?\]\]`)
+
+// ExtractWikilinks parses [[Title]], [[Title|Display]], and ![[Title]] from markdown content.
+// Returns deduplicated links preserving first occurrence.
+// Ignores matches preceded by [ (e.g. [[[triple]]]) since Go regexp lacks lookbehind.
+func ExtractWikilinks(content string) []models.NoteLink {
+	indices := wikiRe.FindAllStringSubmatchIndex(content, -1)
+	seen := make(map[string]bool)
+	var links []models.NoteLink
+
+	for _, idx := range indices {
+		matchStart := idx[0] // start of full match (including optional !)
+
+		// Skip if preceded by [ — rejects [[[triple]]] patterns
+		if matchStart > 0 && content[matchStart-1] == '[' {
+			continue
+		}
+
+		// Extract submatches by index pairs: idx[2*n], idx[2*n+1]
+		title := strings.TrimSpace(content[idx[4]:idx[5]]) // group 2: title
+
+		key := strings.ToLower(title)
+		if key == "" || seen[key] {
+			continue
+		}
+		seen[key] = true
+
+		link := models.NoteLink{
+			TargetTitle:    title,
+			IsTransclusion: content[idx[2]:idx[3]] == "!", // group 1
+		}
+		// group 3: display text (optional — idx[6:7] may be -1)
+		if idx[6] >= 0 && idx[7] >= 0 {
+			link.DisplayText = strings.TrimSpace(content[idx[6]:idx[7]])
+		}
+		links = append(links, link)
+	}
+	return links
+}

server/notelinks/extract_test.go

@@ -0,0 +1,122 @@
+package notelinks
+
+import (
+	"testing"
+)
+
+func TestExtractWikilinks(t *testing.T) {
+	tests := []struct {
+		name    string
+		content string
+		want    int
+		checks  func(t *testing.T, links []struct{ title, display string; transclusion bool })
+	}{
+		{
+			name:    "no links",
+			content: "Just some plain text with [single brackets]",
+			want:    0,
+		},
+		{
+			name:    "single link",
+			content: "See [[Project Notes]] for details",
+			want:    1,
+			checks: func(t *testing.T, links []struct{ title, display string; transclusion bool }) {
+				if links[0].title != "Project Notes" {
+					t.Errorf("title = %q, want %q", links[0].title, "Project Notes")
+				}
+				if links[0].transclusion {
+					t.Error("should not be transclusion")
+				}
+			},
+		},
+		{
+			name:    "link with display text",
+			content: "Check the [[Architecture Doc|arch doc]] here",
+			want:    1,
+			checks: func(t *testing.T, links []struct{ title, display string; transclusion bool }) {
+				if links[0].title != "Architecture Doc" {
+					t.Errorf("title = %q", links[0].title)
+				}
+				if links[0].display != "arch doc" {
+					t.Errorf("display = %q", links[0].display)
+				}
+			},
+		},
+		{
+			name:    "transclusion",
+			content: "Embed this: ![[Meeting Notes]]",
+			want:    1,
+			checks: func(t *testing.T, links []struct{ title, display string; transclusion bool }) {
+				if links[0].title != "Meeting Notes" {
+					t.Errorf("title = %q", links[0].title)
+				}
+				if !links[0].transclusion {
+					t.Error("should be transclusion")
+				}
+			},
+		},
+		{
+			name:    "multiple links deduplicated",
+			content: "See [[Alpha]] and [[Beta]] and [[Alpha]] again",
+			want:    2,
+		},
+		{
+			name:    "case-insensitive dedup",
+			content: "See [[Project]] and [[project]]",
+			want:    1,
+			checks: func(t *testing.T, links []struct{ title, display string; transclusion bool }) {
+				if links[0].title != "Project" {
+					t.Errorf("title = %q, want first occurrence", links[0].title)
+				}
+			},
+		},
+		{
+			name:    "mixed links and transclusions",
+			content: "Link to [[Alpha]], embed ![[Beta|b]], and [[Gamma]]",
+			want:    3,
+		},
+		{
+			name:    "whitespace trimmed",
+			content: "See [[ Spaced Title ]] here",
+			want:    1,
+			checks: func(t *testing.T, links []struct{ title, display string; transclusion bool }) {
+				if links[0].title != "Spaced Title" {
+					t.Errorf("title = %q, want trimmed", links[0].title)
+				}
+			},
+		},
+		{
+			name:    "nested brackets ignored",
+			content: "Not a link: [[[triple]]] but [[Valid]] is",
+			want:    1,
+			checks: func(t *testing.T, links []struct{ title, display string; transclusion bool }) {
+				if links[0].title != "Valid" {
+					t.Errorf("title = %q, want Valid", links[0].title)
+				}
+			},
+		},
+		{
+			name:    "empty brackets ignored",
+			content: "Empty [[]] should not match",
+			want:    0,
+		},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			links := ExtractWikilinks(tt.content)
+			if len(links) != tt.want {
+				t.Fatalf("got %d links, want %d", len(links), tt.want)
+			}
+			if tt.checks != nil {
+				simplified := make([]struct{ title, display string; transclusion bool }, len(links))
+				for i, l := range links {
+					simplified[i] = struct{ title, display string; transclusion bool }{
+						title: l.TargetTitle, display: l.DisplayText, transclusion: l.IsTransclusion,
+					}
+				}
+				tt.checks(t, simplified)
+			}
+		})
+	}
+}