blgo/parser/frontmatter.go

package parser

import (
	"errors"
	"fmt"
	"regexp"
	"strings"
)

var (
	// ErrInvalidKeyValuePair is an error when parsing a FrontMatter entry
	// without a : delimiter.
	ErrInvalidKeyValuePair = errors.New("the line does not contain a valid entry for FrontMatter")

	// ErrBlankKey is an error when parsing a FrontMatter entry where
	// the key is nothing, when space is trimmed.
	ErrBlankKey = errors.New("FrontMatter key is empty")

	// ErrDuplicateKey is an error when parsing the content FrontMatter, where
	// the same key appears more than once.
	ErrDuplicateKey = errors.New("the provided key has already been defined")

	// ErrEOF is an error when parsing the content FrontMatter, and it reaches
	// the end of the file before the dashes for the end are reached.
	ErrEOF = errors.New("unexpected EOF. expected closing dashes for FrontMatter")
)

// FrontMatter is the key-value structure that is used to contain metadata about
// the MarkDown file, containing information like author and description.
type FrontMatter map[string]string

// ParseKeyValueLine will take a string in the format of "The Key: The Value".
// The input is split on ":", and space is trimmed before returning. If there
// are no colons, ErrInvalidKeyValuePair is returned. If the key is empty,
// ErrBlankKey is returned.
//
// The input will only split on the first colon. More colons will be part
// of the value.
func ParseKeyValueLine(line string) (string, string, error) {
	split := strings.SplitN(line, ":", 2)

	if len(split) != 2 {
		return "", "", ErrInvalidKeyValuePair
	}

	key := strings.TrimSpace(split[0])
	if key == "" {
		return "", "", ErrBlankKey
	}

	return key, strings.TrimSpace(split[1]), nil
}

// delimiterRegex denotes the regex for a line that matches on when the
// FrontMatter section is delimited from the rest of the content.
// It must be a minimum of 3 dashes (-), and no other content.
var delimiterRegex = regexp.MustCompile("^-{3,}$")

// ExtractFrontMatter will take an entire MarkDown file, and return a map that
// contains key-value pairs. The key-value pairs must end with an extra line
// with the content of "---". If this is not found, an ErrEOF is returned.
// Optionally, the FrontMatter can start with a ---. This is to have support
// with older template files, which follow this format.
//
// If the first line is not "---" and is not parsed as a valid FrontMatter
// entry, then the entire file is skipped and interpreted as having an empty
// FrontMatter.
// Duplicate keys will return a ErrDuplicateKey.
// Invalid FrontMatter entries will return a ErrInvalidKeyValuePair.
// Example Front Matter Format:
//
// ---
// Some Key: Some Value
// Another Key:Another Value
// A Key : A Value
// ---
// # Markdown Content
// ...
func ExtractFrontMatter(contents []string) (FrontMatter, []string, error) {
	matter := FrontMatter{}
	if len(contents) == 0 {
		return matter, contents, nil
	}

	for i, line := range contents {
		if i == 0 && delimiterRegex.MatchString(line) {
			continue
		}

		if delimiterRegex.MatchString(line) {
			return matter, contents[i+1:], nil
		}

		key, value, err := ParseKeyValueLine(line)
		if err != nil && i == 0 {
			return matter, contents, nil
		}
		if err != nil {
			return matter, contents[i+1:], fmt.Errorf("error parsing line %d: %w", i+1, err)
		}

		if _, ok := matter[key]; ok {
			return matter, contents[i+1:], fmt.Errorf("error on parsing line %d: %w", i+1, ErrDuplicateKey)
		}
		matter[key] = value
	}
	return matter, contents, fmt.Errorf("error on parsing: %w", ErrEOF)
}