process links and headings

README.md

@@ -7,18 +7,19 @@ state machines.
 
 <!-- testing markdown, this should be deleted, below merged -->
 See the [gemini
-protocol](https://gemini.circumlunar.space/).
+protocol](https://gemini.circumlunar.space/) and the [protocol
+spec](https://gemini.circumlunar.space/docs/specification.gmi).
 
 Internally md2gmi does a 1st pass that constructs the core layout for gemtext. This is then streamed
 to the 2nd pass line by line. The 2nd pass will convert links and stream line by line to the output.
 
-### Usage
+###Usage
 
 ```plain
 Usage of ./md2gmi:
-  -in string
+  -f string
         specify a .md (Markdown) file to read from, otherwise stdin (default)
-  -out string
+  -o string
         specify a .gmi (gemtext) file to write to, otherwise stdout (default)
 ```
 

preproc.go

@@ -80,7 +80,7 @@ func normal(m *fsm, data []byte) stateFn {
 	}
 	if needsFence(data) {
 		m.out <- []byte("```\n")
-		m.out <- append(data, '\n')
+		m.out <- append(data[4:], '\n')
 		m.pending = []byte("```")
 		return toFence
 	}
@@ -105,7 +105,7 @@ func fence(m *fsm, data []byte) stateFn {
 }
 
 func toFence(m *fsm, data []byte) stateFn {
-	m.out <- append(data, '\n')
+	m.out <- append(data[4:], '\n')
 	if needsFence(data) {
 		return toFence
 	}

proc.go

@@ -1,17 +1,14 @@
 package main
 
-// state function
+import (
-type stateFn2 func(*proc, []byte) stateFn2
+	"bytes"
+	"fmt"
+	"regexp"
+)
 
 // state machine
 type proc struct {
-	state stateFn2
+	out chan []byte
-	out   chan []byte
-
-	// combining multiple input lines
-	buffer []byte
-	// if we have a termination rule to abide, e.g. implied code fences
-	pending []byte
 }
 
 func NewProc() *proc {
@@ -21,32 +18,43 @@ func NewProc() *proc {
 func (m *proc) Process(in chan []byte) chan []byte {
 	m.out = make(chan []byte)
 	go func() {
-		for m.state = line; m.state != nil; {
+		for b := range in {
-			b, ok := <-in
+			m.out <- m.process(b)
-			if !ok {
-				m.flush()
-				close(m.out)
-				m.state = nil
-				continue
-			}
-			m.state = m.state(m, b)
 		}
+		close(m.out)
 	}()
 	return m.out
 }
 
-func (m *proc) flush() {
+func (m *proc) process(data []byte) []byte {
-	if len(m.pending) > 0 {
+	// find link name and url
-		m.out <- append(m.pending, '\n')
+	var buffer []byte
-		m.pending = m.pending[:0]
+	re := regexp.MustCompile(`\[([^\]*]*)\]\(([^)]*)\)`)
+	for i, match := range re.FindAllSubmatch(data, -1) {
+		replaceWithIndex := append(match[1], fmt.Sprintf("[%d]", i+1)...)
+		data = bytes.Replace(data, match[0], replaceWithIndex, 1)
+		link := fmt.Sprintf("=> %s %d: %s\n", match[2], i+1, match[1])
+		buffer = append(buffer, link...)
+	}
+	if len(buffer) > 0 {
+		data = append(data, []byte("\n")...)
+		data = append(data, buffer...)
 	}
-}
 
-func line(m *proc, data []byte) stateFn2 {
+	// remove comments
-	// TODO
+	re2 := regexp.MustCompile(`<!--.*-->`)
-	// find links
+	data = re2.ReplaceAll(data, []byte{})
-	// collapse lists
-	m.out <- data
 
-	return line
+	// collapse headings
+	re3 := regexp.MustCompile(`^[#]{4,}`)
+	data = re3.ReplaceAll(data, []byte("###"))
+
+	// heading without spacing
+	re4 := regexp.MustCompile(`^(#+)[^# ]`)
+	sub := re4.FindSubmatch(data)
+	if len(sub) > 0 {
+		data = bytes.Replace(data, sub[1], append(sub[1], []byte(" ")...), 1)
+	}
+
+	return data
 }