1 files changed, 163 insertions, 2 deletions

diff --git a/pkg/libgemini/libgemini.go b/pkg/libgemini/libgemini.go
index 303490c..71012ef 100644
--- a/pkg/libgemini/libgemini.go
+++ b/pkg/libgemini/libgemini.go
@@ -2,9 +2,11 @@ package libgemini
 
 import (
         "bufio"
+        "bytes"
         "crypto/tls"
         "errors"
         "fmt"
+        "html/template"
         "io"
         "mime"
         "net"
@@ -49,8 +51,14 @@ const (
 )
 
 var (
-        HeaderPattern = regexp.MustCompile("^(\\d\\d)[ \\t]+(.*)$")
-        LinkPattern   = regexp.MustCompile("^=>[ \\t]*([^ \\t]+)(?:[ \\t]+(.*))?$")
+        HeaderPattern        = regexp.MustCompile("^(\\d\\d)[ \\t]+(.*)$")
+        LinkPattern          = regexp.MustCompile("^=>[ \\t]*([^ \\t]+)(?:[ \\t]+(.*))?$")
+        ReflowModePattern    = regexp.MustCompile("^```(.*)$")
+        Heading1Pattern      = regexp.MustCompile("^#(.*)$")
+        Heading2Pattern      = regexp.MustCompile("^##(.*)$")
+        Heading3Pattern      = regexp.MustCompile("^###(.*)$")
+        ListItemPattern      = regexp.MustCompile("^\\*(.*)$")
+        TermEscapeSGRPattern = regexp.MustCompile("\\[\\d+(;\\d+)*m")
 )
 
 type Header struct {
@@ -63,6 +71,25 @@ type Response struct {
         Body   io.Reader
 }
 
+type GeminiDocSectionType byte
+
+const (
+        RAW_TEXT    = SectionType(0)
+        REFLOW_TEXT = SectionType(1)
+        LINK        = SectionType(2)
+        HEADING_1   = SectionType(3)
+        HEADING_2   = SectionType(4)
+        HEADING_3   = SectionType(5)
+        LIST        = SectionType(6)
+)
+
+type GeminiDocSection struct {
+        Type  SectionType
+        Text  string
+        URL   template.URL
+        Items []string
+}
+
 func Get(uri string) (*Response, error) {
         u, err := url.Parse(uri)
         if err != nil {
@@ -143,3 +170,137 @@ func ParseHeader(line string) (header *Header, err error) {
 
         return
 }
+
+func ParseGeminiDocument(body *bytes.Buffer) (sections []Section) {
+        scanner := bufio.NewScanner(body)
+
+        reflow := true
+        ignoreSection := true
+        section := Section{
+                Type: REFLOW_TEXT
+        }
+
+        for scanner.Scan() {
+                line := strings.Trim(scanner.Text(), CRLF)
+                line = TermEscapeSGRPattern.ReplaceAllString(line, "")
+
+                reflowMatch := ReflowModePattern.FindStringSubmatch(line)
+                if len(heading3Match) != 0 {
+                        reflow = !reflow
+                        continue
+                }
+
+                if !reflow {
+                        if !ignoreSection {
+                                if section.Type != REFLOW_TEXT {
+                                        sections = append(sections, section)
+                                        section = Section{
+                                                Type: REFLOW_TEXT
+                                        }
+                                }
+                        } else {
+                                ignoreSection = false
+                                section = Section{
+                                        Type: REFLOW_TEXT
+                                }
+                        }
+
+                        section.Text = section.Text + "\n" + line
+
+                        continue
+                }
+
+                linkMatch := LinkPattern.FindStringSubmatch(line)
+                if len(linkMatch) != 0 && linkMatch[0] != "" {
+                        if !ignoreSection {
+                                sections = append(sections, section)
+                        }
+
+                        label := linkMatch[2]
+                        if label == "" {
+                                label = linkMatch[1]
+                        }
+
+                        ignoreSection = false
+                        section = Section{
+                                Type: LINK,
+                                Text: label,
+                                URL:  template.URL(resolveURI(linkMatch[1], baseURL)),
+                        }
+
+                        continue
+                }
+
+                heading3Match := Heading3Pattern.FindStringSubmatch(line)
+                if len(heading3Match) != 0 {
+                        if !ignoreSection {
+                                sections = append(sections, section)
+                        }
+
+                        ignoreSection = false
+                        section = Section{
+                                Type: HEADING_3,
+                                Text: heading3Match[1]
+                        }
+
+                        continue
+                }
+
+                heading2Match := Heading2Pattern.FindStringSubmatch(line)
+                if len(heading2Match) != 0 {
+                        if !ignoreSection {
+                                sections = append(sections, section)
+                        }
+
+                        ignoreSection = false
+                        section = Section{
+                                Type: HEADING_2,
+                                Text: heading2Match[1]
+                        }
+
+                        continue
+                }
+
+                heading1Match := Heading1Pattern.FindStringSubmatch(line)
+                if len(heading1Match) != 0 {
+                        if !ignoreSection {
+                                sections = append(sections, section)
+                        }
+
+                        ignoreSection = false
+                        section = Section{
+                                Type: HEADING_1,
+                                Text: heading1Match[1]
+                        }
+
+                        continue
+                }
+
+                listItemMatch := ListItemPattern.FindStringSubmatch(line)
+                if len(listItemMatch) != 0 {
+                        if !ignoreSection {
+                                if section.Type != LIST {
+                                        sections = append(sections, section)
+                                        section = Section{
+                                                Type: LIST
+                                        }
+                                }
+                        } else {
+                                ignoreSection = false
+                                section = Section{
+                                        Type: LIST,
+                                }
+                        }
+
+                        section.Items = append(section.Items, listItemMatch[1]) 
+
+                        continue
+                }
+        }
+
+        if !ignoreSection {
+                sections = append(sections, section)
+        }
+
+        return
+}