From aa23984bc21f99e00b8552e928d23ef02bb745f3 Mon Sep 17 00:00:00 2001
From: tjpcc <tjp@ctrl-c.club>
Date: Fri, 13 Jan 2023 10:50:30 -0700
Subject: Initial gemtext package.

Contains:
- gemtext AST (Document and line types)
- Parse from an io.Reader
- ParseLine a []byte
- doc comments on everything
- ParseLine tests for every line type

Still needs tests for Parse & Document.
---
 gemtext/parse.go | 154 +++++++++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 154 insertions(+)
 create mode 100644 gemtext/parse.go

(limited to 'gemtext/parse.go')

diff --git a/gemtext/parse.go b/gemtext/parse.go
new file mode 100644
index 0000000..4a8c641
--- /dev/null
+++ b/gemtext/parse.go
@@ -0,0 +1,154 @@
+package gemtext
+
+import (
+	"bufio"
+	"bytes"
+	"io"
+)
+
+// Parse parses the full contents of an io.Reader into a gemtext.Document.
+func Parse(input io.Reader) (Document, error) {
+	rdr := bufio.NewReader(input)
+
+	var lines []Line
+	inPFT := false
+
+	for {
+		raw, err := rdr.ReadBytes('\n')
+		if err != io.EOF && err != nil {
+			return nil, err
+		}
+
+		var line Line
+
+		if inPFT && (len(raw) < 3 || raw[0] != '`' || raw[1] != '`' || raw[2] != '`') {
+				line = PreformattedTextLine{raw: raw}
+		} else {
+			line = ParseLine(raw)
+		}
+
+		if line.Type() == LineTypePreformatToggle {
+			if inPFT {
+				toggle := line.(PreformatToggleLine)
+				(&toggle).clearAlt()
+				line = toggle
+			}
+
+			inPFT = !inPFT
+		}
+
+		if line != nil {
+			lines = append(lines, line)
+		}
+
+		if err == io.EOF {
+			break
+		}
+	}
+
+	return Document(lines), nil
+}
+
+// ParseLine parses a single line (including the trailing \n) into a gemtext.Line.
+func ParseLine(line []byte) Line {
+	if len(line) == 0 {
+		return nil
+	}
+
+	switch line[0] {
+	case '=':
+		if len(line) == 1 || line[1] != '>' {
+			break
+		}
+		return parseLinkLine(line)
+	case '`':
+		if len(line) < 3 || line[1] != '`' || line[2] != '`' {
+			break
+		}
+		return parsePreformatToggleLine(line)
+	case '#':
+		level := 1
+		if len(line) > 1 && line[1] == '#' {
+			level += 1
+			if len(line) > 2 && line[2] == '#' {
+				level += 1
+			}
+		}
+		return parseHeadingLine(level, line)
+	case '*':
+		if len(line) == 1 || line[1] != ' ' {
+			break
+		}
+		return parseListItemLine(line)
+	case '>':
+		return parseQuoteLine(line)
+	}
+
+	return TextLine{raw: line}
+}
+
+func parseLinkLine(raw []byte) LinkLine {
+	line := LinkLine{raw: raw}
+
+	// move past =>[<whitespace>]
+	raw = bytes.TrimLeft(raw[2:], " \t")
+
+	// find the next space or tab
+	spIdx := bytes.IndexByte(raw, ' ')
+	tbIdx := bytes.IndexByte(raw, '\t')
+	idx := spIdx
+	if idx == -1 {
+		idx = tbIdx
+	}
+	if tbIdx >= 0 && tbIdx < idx {
+		idx = tbIdx
+	}
+
+	if idx < 0 {
+		line.URL = bytes.TrimRight(raw, "\r\n")
+		return line
+	}
+
+	line.URL = raw[:idx]
+	raw = raw[idx+1:]
+
+	label := bytes.TrimRight(bytes.TrimLeft(raw, " \t"), "\r\n")
+	if len(label) > 0 {
+		line.Label = label
+	}
+
+	return line
+}
+
+func parsePreformatToggleLine(raw []byte) PreformatToggleLine {
+	line := PreformatToggleLine{raw: raw}
+
+	raw = bytes.TrimRight(raw[3:], "\r\n")
+	if len(raw) > 0 {
+		line.AltText = raw
+	}
+
+	return line
+}
+
+func parseHeadingLine(level int, raw []byte) HeadingLine {
+	return HeadingLine{
+		raw:      raw,
+		lineType: LineTypeHeading1 - 1 + LineType(level),
+		Body:     bytes.TrimRight(bytes.TrimLeft(raw[level:], " \t"), "\r\n"),
+	}
+}
+
+func parseListItemLine(raw []byte) ListItemLine {
+	return ListItemLine{
+		raw:  raw,
+		Body: bytes.TrimRight(raw[2:], "\r\n"),
+	}
+}
+
+func parseQuoteLine(raw []byte) QuoteLine {
+	return QuoteLine{
+		raw:  raw,
+		Body: bytes.TrimRight(raw[1:], "\r\n"),
+	}
+}
-- 
cgit v1.2.3