From 34b8d8ff1f9d65c08a9156d72f08cf548183c6f4 Mon Sep 17 00:00:00 2001
From: JP Appel <jeanpierre.appel01@gmail.com>
Date: Sun, 27 Apr 2025 00:49:27 -0400
Subject: Large commit; many features

---
 pkg/index/filters.go | 84 ++++++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 84 insertions(+)
 create mode 100644 pkg/index/filters.go

(limited to 'pkg/index/filters.go')

diff --git a/pkg/index/filters.go b/pkg/index/filters.go
new file mode 100644
index 0000000..f59a5d6
--- /dev/null
+++ b/pkg/index/filters.go
@@ -0,0 +1,84 @@
+package index
+
+import (
+	"io"
+	"path/filepath"
+)
+
+// TODO: create excluded path filter factory
+
+type DocFilter func(infoPath, io.ReadSeeker) bool
+
+func NewExtensionFilter(ext string) DocFilter {
+	return func(ip infoPath, _ io.ReadSeeker) bool {
+		return filepath.Ext(ip.path) == ext
+	}
+}
+
+func NewMaxFilesizeFilter(size int64) DocFilter {
+	return func(ip infoPath, _ io.ReadSeeker) bool {
+		return ip.info.Size() <= size
+	}
+}
+
+func YamlHeaderFilter(_ infoPath, r io.ReadSeeker) bool {
+	const bufSize = 4096
+	buf := make([]byte, bufSize)
+
+	carry := make([]byte, 4)
+	cmp := make([]byte, 4)
+	n, err := r.Read(carry)
+	if err != nil || n < 4 || string(carry) != "---\n" {
+		return false
+	}
+
+	headerFound := false
+	readMore := true
+	for readMore {
+		buf = buf[:bufSize]
+		n, err := r.Read(buf)
+		if err == io.EOF {
+			readMore = false
+		} else if err != nil {
+			return false
+		}
+		buf = buf[:n]
+
+		// PERF: the carry doesn't need to be checked on the first loop iteration
+		for i := range min(4, n) {
+			b := carry[i]
+			for j := range 4 {
+				if i+j < 4 {
+					cmp[j] = carry[i+j]
+				} else {
+					cmp[j] = buf[(i+j)%4]
+				}
+			}
+			if b == '\n' && string(cmp) == "\n---\n" {
+				headerFound = true
+				readMore = false
+				break
+			}
+		}
+		for i := range n - 4 {
+			b := buf[i]
+			if b == '\n' && string(buf[i:i+5]) == "\n---\n" {
+				headerFound = true
+				readMore = false
+				break
+			}
+		}
+
+		if readMore {
+			for i := range 4 {
+				carry[i] = buf[n-4+i]
+			}
+		}
+	}
+
+	return headerFound
+}
+
+func DefaultFilters() []DocFilter {
+	return []DocFilter{NewExtensionFilter(".md"), NewMaxFilesizeFilter(200 * 1024)}
+}
-- 
cgit v1.2.3