Make separate, more adapted single-page version (#339)

Fixes #297

Author: ISSOtm

Cargo.lock

@@ -35,7 +35,8 @@ site-url = "/pandocs/"
 follow-web-links = true
 traverse-parent-directories = false
 optional = true
-# Do not check the "this was generated from commit N" link, as the commit may not have been pushed yet
 exclude = [
-	'https://github\.com/gbdev/pandocs/tree/'
+	'https://github\.com/gbdev/pandocs/tree/', # Do not check the "this was generated from commit N" link, as the commit may not have been pushed yet
+  'print.html', # Do not check the print page, as it is generated separately
+  'single.html', # Do not check the single page, as it is generated separately
 ]

book.toml

@@ -9,7 +9,9 @@ edition = "2018"
 [dependencies]
 anyhow = "1.0.40"
 globwalk = "0.8.1"
+lazy_static = "1.4.0"
 # mdbook here is only used as a lib, so no need for the extra features
 mdbook = { version = "0.4.8", default-features = false, features = [ "search" ] }
 regex = "1.5"
 termcolor = "1.1.2"
+url = "2.2.2"

renderer/Cargo.toml

@@ -9,16 +9,19 @@
 
 use anyhow::Context;
 use globwalk::{FileType, GlobWalkerBuilder};
+use lazy_static::lazy_static;
 use mdbook::book::BookItem;
 use mdbook::errors::{Error, Result};
 use mdbook::renderer::{HtmlHandlebars, RenderContext, Renderer};
+use regex::Regex;
 use std::fs::{self, File};
 use std::io::{self, Write};
+use std::io::{BufRead, BufReader, BufWriter};
 use std::path::PathBuf;
 use std::process::Command;
 use std::str::FromStr;
-
 use termcolor::{Color, ColorChoice, ColorSpec, StandardStream, WriteColor};
+use url::Url;
 
 fn main() -> Result<()> {
     let mut stdin = io::stdin();
@@ -78,11 +81,15 @@ impl Renderer for Pandocs {
         path.set_file_name("print.html");
         render(&mut path, "<print>", usize::MAX).context("Failed to render print page")?;
 
+        // Generate the single-page version
+        let base_url = Url::parse("http://localhost/").unwrap();
+        gen_single_page(&mut path, &base_url).context("Failed to render single-page version")?;
+
         // Generate the graphs in `imgs/src/` by shelling out to Python
         let working_dir = ctx.destination.join("imgs");
         let src_dir = working_dir.join("src");
         let python = if cfg!(windows) { "py3" } else { "python3" };
-        let render = |file_name, title| {
+        let gen_graph = |file_name, title| {
             let mut file_name = PathBuf::from_str(file_name).unwrap(); // Can't fail
             let output = File::create(working_dir.join(&file_name))?;
 
@@ -106,8 +113,8 @@ impl Renderer for Pandocs {
                 )))
             }
         };
-        render("MBC5_Rumble_Mild.svg", "Mild Rumble")?;
-        render("MBC5_Rumble_Strong.svg", "Strong Rumble")?;
+        gen_graph("MBC5_Rumble_Mild.svg", "Mild Rumble")?;
+        gen_graph("MBC5_Rumble_Strong.svg", "Strong Rumble")?;
         fs::remove_dir_all(&src_dir).context(format!("Failed to remove {}", src_dir.display()))?;
 
         // Scrub off files that need not be published
@@ -230,3 +237,89 @@ fn render(path: &mut PathBuf, name: &str, index: usize) -> Result<()> {
 
     Ok(())
 }
+
+/// This generates `single.html` from `print.html`.
+/// This does not properly parse HTML, instead relying on crude assumptions about mdBook's output.
+/// This is for the sake of performance, as we can afford to update this from time to time.
+/// Such assumptions are marked by `HACK:` comments in the function, to at least ease tracability.
+fn gen_single_page(path: &mut PathBuf, base_url: &Url) -> Result<()> {
+    let print_page = BufReader::new(
+        File::open(&path)
+            .with_context(|| format!("Failed to open print page \"{}\"", path.display()))?,
+    );
+    path.set_file_name("single.html");
+    let mut single_page = BufWriter::new(File::create(path)?);
+    // HACK: this almost certainly forgets a bunch of HTML edge cases
+    lazy_static! {
+        static ref LINK_RE: Regex =
+            Regex::new(r#"<a(?:\s+(?:href="([^"]*)"|\w+="[^"]*"))*\s*>"#).unwrap();
+    }
+
+    // HACK: this assumes all link tags span a single line
+    let mut lines = print_page.lines();
+    while let Some(line) = lines.next().transpose()? {
+        let mut i = 0;
+        for url_match in LINK_RE.captures_iter(&line).filter_map(|caps| caps.get(1)) {
+            let url = &line[url_match.range()];
+
+            match Url::parse(url) {
+                Ok(_) => continue, // If not a relative URL, skip
+                Err(url::ParseError::RelativeUrlWithoutBase) => (),
+                Err(e) => return Err(e).with_context(|| format!("Bad link URL \"{}\"", url)), // Return other errors
+            }
+
+            let url = base_url
+                .join(url)
+                .with_context(|| format!("Bad total URL \"{}\"", url))?;
+            if let Some(frag) = url.fragment() {
+                // Write everything up to the match
+                single_page.write_all(line[i..url_match.start()].as_bytes())?;
+                // Write the replaced match
+                single_page.write_all("#".as_bytes())?;
+                single_page.write_all(frag.as_bytes())?;
+                // Start copying after the match
+                i = url_match.end();
+            }
+        }
+
+        // Write rest of line
+        single_page.write_all(line[i..].as_bytes())?;
+        single_page.write_all("\n".as_bytes())?;
+
+        // Remove the automatic print trigger code.
+        // HACK: this assumes the location of the script, the comment's format,
+        // and that there is no content between this comment and the script that cannot be
+        // passed through.
+        if line.trim_start() == "<!-- Custom JS scripts -->" {
+            // Pass extra scripts through unchanged
+            // HACK: we filter scripts by assuming that mdBook's "additional JS" refs are one-line
+            // and the printer script tag isn't.
+            lines
+                .by_ref()
+                .take_while(|line| {
+                    line.as_ref().map_or(true, |line| {
+                        // HACK: this relies on the exact formatting of the auto-printer's script tag
+                        line.trim_start() != "<script type=\"text/javascript\">"
+                    })
+                })
+                .try_for_each(|line| {
+                    single_page.write_all(line?.as_bytes())?;
+                    single_page.write_all("\n".as_bytes())
+                })?;
+
+            // Discard lines until the end of the script.
+            // Also, check if this does discard lines; if this discards none, we have a problem.
+            let mut auto_printer_script_lines = lines.by_ref().take_while(|line| {
+                line.as_ref().map_or(true, |line| {
+                    // HACK: this relies on the exact formatting of the auto-printer's script end tag
+                    line.trim_start() != "</script>"
+                })
+            });
+            if auto_printer_script_lines.next().is_none() {
+                panic!("Warning: unterminated auto-printer script tag??");
+            }
+        }
+    }
+
+    Ok(())
+}

renderer/src/main.rs

@@ -11,7 +11,7 @@ Emulator developers may be also interested in the [Game Boy: Complete Technical
 
 :::
 
-### Contributing
+## Contributing
 
 This project is open source, released under the [CC0 license](https://raw.githubusercontent.com/gbdev/pandocs/master/LICENSE). Everyone is welcome to help, provide feedback and propose additions or improvements. The git repository is hosted at [github.com/gbdev/pandocs](https://github.com/gbdev/pandocs), where you can learn more about how you can [contribute](https://github.com/gbdev/pandocs/blob/master/README.MD), find detailed contribution guidelines and procedures, file Issues and send Pull Requests. 
 
@@ -20,7 +20,7 @@ There is a [Discord chat](https://gbdev.io/chat) dedicated to the gbdev communit
 Finally, you can also contact us and send patches via email: `pandocs (at) gbdev.io`.
 
 
-### Using this document
+## Using this document
 
 In the top navigation bar, you will find a series of icons.
 
@@ -32,6 +32,6 @@ You can search anywhere by pressing <kbd>s</kbd> on your keyboard or clicking th
 
 The <i class="fa fa-edit"></i> icon allows you to suggest an edit on the current page by directly opening the source file in the git repository.
 
-You can find a one-page printable version of Pan Docs [here](https://gbdev.io/pandocs/print.html). 
+Also available are [one-page](single.html) and [printable](print.html) versions of this document.
 
 <br><br>