Merge pull request #45 from ranfdev/rewritten_parser

Rewritten parser, refactor pages, put hypertext rendering in Hypertext page

Author: ranfdev

data/resources/ui/download_page.blp

@@ -1,7 +1,7 @@
 using Gtk 4.0;
 using Adw 1;
 
-template DownloadPage: Gtk.Box {
+template Download: Gtk.Box {
   margin-top: 8;
   margin-bottom: 8;
   margin-start: 8;

data/resources/ui/input_page.blp

@@ -1,7 +1,7 @@
 using Gtk 4.0;
 using Adw 1;
 
-template InputPage: Gtk.Box {
+template Input: Gtk.Box {
   margin-top: 8;
   margin-bottom: 8;
   margin-start: 8;

data/resources/ui/tab.blp

@@ -9,16 +9,6 @@ template Tab: Adw.Bin {
     Gtk.Stack stack {
       Gtk.ScrolledWindow scroll_win {
         vexpand: true;
-        Gtk.TextView text_view {
-          top-margin: 40;
-          bottom-margin: 80;
-          left-margin: 20;
-          right-margin: 20;
-          indent: 2;
-          editable: false;
-          cursor-visible: false;
-          wrap-mode: word_char;
-        }
       }
     }
   }

gemini/src/parser.rs

@@ -1,51 +1,101 @@
 use once_cell::sync::Lazy;
 use regex::Regex;
 static R_GEMINI_LINK: Lazy<Regex> =
-    Lazy::new(|| Regex::new(r"^=>\s*(?P<href>\S*)\s*(?P<label>.*)").unwrap());
-
-#[derive(Debug)]
-pub enum PageElement {
-    Heading(String),
-    Quote(String),
-    Preformatted(String),
-    Text(String),
+    Lazy::new(|| Regex::new(r"^=>\s+(?P<href>\S+)(\s+(?P<label>.+))?").unwrap());
+
+// See gemini://gemini.circumlunar.space/docs/cheatsheet.gmi
+
+#[derive(Debug, Clone)]
+pub enum Tag {
+    Paragraph, // Is just a text line
+    Heading(u8),
+    BlockQuote,
+    CodeBlock,
+    UnorderedList,
+    Item,
     Link(String, Option<String>),
-    ListItem(String),
-    Empty,
+}
+
+#[derive(Debug, Clone)]
+pub enum Event<'a> {
+    Start(Tag),
+    End,
+    Text(&'a str),
+    BlankLine,
 }
 
 #[derive(Debug, Clone, Default)]
 pub struct Parser {
-    inside_pre: bool,
+    tag_stack: Vec<Tag>,
 }
 
 impl Parser {
     pub fn new() -> Self {
-        Self { inside_pre: false }
+        Self { tag_stack: vec![] }
     }
-    pub fn parse_line(&mut self, line: &str) -> PageElement {
+
+    /// Returns an `Event` when an event it's ready, else, `None`
+    // TODO: Make this work on text input of any length, don't impose the "line" chunk requirement
+    // some work has already been done, the pushed result is already structured to do so.
+    pub fn parse_line<'a>(&mut self, line: &'a str, res: &mut Vec<Event<'a>>) {
+        let parent_tag = self.tag_stack.last();
+
+        // Close pending multi-line tags
+        if matches!(parent_tag, Some(Tag::BlockQuote)) && !line.starts_with('>')
+            || matches!(parent_tag, Some(Tag::UnorderedList))
+        {
+            res.push(Event::End);
+            self.tag_stack.pop();
+        }
+
+        let parent_tag = self.tag_stack.last();
+
         if line.starts_with("```") {
-            self.inside_pre = !self.inside_pre;
-            PageElement::Empty
-        } else if self.inside_pre {
-            PageElement::Preformatted(line.to_string())
+            let inner_res = if let Some(Tag::CodeBlock) = parent_tag {
+                self.tag_stack.pop();
+                Event::End
+            } else {
+                self.tag_stack.push(Tag::CodeBlock);
+                Event::Start(Tag::CodeBlock)
+            };
+            res.push(inner_res);
+        } else if let Some(Tag::CodeBlock) = parent_tag {
+            res.push(Event::Text(line));
+        } else if line.trim().is_empty() {
+            res.push(Event::BlankLine);
         } else if line.starts_with('#') {
-            PageElement::Heading(line.to_string())
+            let line = line.trim_end();
+            let lvl = line.chars().filter(|c| *c == '#').count();
+            let heading = Tag::Heading(lvl as u8);
+            res.push(Event::Start(heading));
+
+            let text = line.trim_start_matches('#').trim_start();
+            res.push(Event::Text(text));
+            res.push(Event::End);
         } else if line.starts_with('>') {
-            PageElement::Quote(line.to_string())
+            if !matches!(parent_tag, Some(Tag::BlockQuote)) {
+                res.push(Event::Start(Tag::BlockQuote));
+            }
+            res.push(Event::Text(line.trim_start_matches('>')));
         } else if let Some(stripped) = line.strip_prefix("* ") {
-            PageElement::ListItem(stripped.to_string())
-        } else if let Some(captures) = R_GEMINI_LINK.captures(line) {
-            match (captures.name("href"), captures.name("label")) {
-                (Some(m_href), Some(m_label)) if !m_label.as_str().is_empty() => PageElement::Link(
-                    m_href.as_str().to_string(),
-                    Some(m_label.as_str().to_string()),
-                ),
-                (Some(m_href), _) => PageElement::Link(m_href.as_str().to_string(), None),
-                _ => PageElement::Empty,
+            if !matches!(parent_tag, Some(Tag::UnorderedList)) {
+                res.push(Event::Start(Tag::UnorderedList));
             }
+            res.push(Event::Start(Tag::Item));
+            res.push(Event::Text(stripped.trim_end()));
+            res.push(Event::End);
+        } else if let Some(captures) = R_GEMINI_LINK.captures(line.trim_end()) {
+            let href = captures.name("href").unwrap();
+            let label = captures.name("label").map(|x| x.as_str());
+            res.push(Event::Start(Tag::Link(
+                href.as_str().to_string(),
+                label.map(|x| x.to_string()),
+            )));
+            res.push(Event::End);
         } else {
-            PageElement::Text(line.to_string())
+            res.push(Event::Start(Tag::Paragraph));
+            res.push(Event::Text(line.trim_end()));
+            res.push(Event::End);
         }
     }
 }

src/main.rs

@@ -15,7 +15,6 @@ mod common;
 mod config;
 mod lossy_text_read;
 mod macros;
-mod text_extensions;
 mod widgets;
 
 use gtk::prelude::*;

src/text_extensions/gemini.rs

@@ -1,10 +1,7 @@
-mod download_page;
-mod input_page;
+mod pages;
 #[allow(clippy::await_holding_refcell_ref)]
 mod tab;
 mod window;
 
-pub use download_page::DownloadPage;
-pub use input_page::InputPage;
 pub use tab::Tab;
 pub use window::Window;