Improve robustness of link rewriting

src/config.rs

@@ -877,8 +877,6 @@ pub struct Language {
     pub description: Option<String>,
 }
 
-impl LanguageConfig {}
-
 /// Allows you to "update" any arbitrary field in a struct by round-tripping via
 /// a `toml::Value`.
 ///

src/preprocess/cmd.rs

@@ -193,6 +193,7 @@ mod tests {
         let md = guide();
         let ctx = PreprocessorContext::new(
             md.root.clone(),
+            None,
             BuildOpts::default(),
             md.config.clone(),
             "some-renderer".to_string(),

src/renderer/html_handlebars/hbs_renderer.rs

@@ -4,7 +4,7 @@ use crate::errors::*;
 use crate::renderer::html_handlebars::helpers;
 use crate::renderer::{RenderContext, Renderer};
 use crate::theme::{self, playground_editor, Theme};
-use crate::utils;
+use crate::utils::{self, RenderMarkdownContext};
 
 use std::borrow::Cow;
 use std::collections::BTreeMap;
@@ -35,14 +35,12 @@ impl HtmlHandlebars {
         match ctx.book {
             LoadedBook::Localized(ref books) => {
                 for (lang_ident, book) in books.0.iter() {
-                    let localized_src_dir = src_dir.join(lang_ident);
                     let localized_destination = ctx.destination.join(lang_ident);
                     let localized_build_dir = ctx.config.build.build_dir.join(lang_ident);
                     self.render_book(
                         ctx,
                         &book,
-                        &localized_src_dir,
+                        src_dir,
-                        &localized_src_dir,
                         &localized_destination,
                         &localized_build_dir,
                         &Some(lang_ident.to_string()),
@@ -53,27 +51,10 @@ impl HtmlHandlebars {
                 }
             }
             LoadedBook::Single(ref book) => {
-                // `src_dir` points to the root source directory. If this book
-                // is actually multilingual and we specified a single language
-                // to build on the command line, then `src_dir` will not be
-                // pointing at the subdirectory with the specified translation's
-                // index/summary files. We have to append the language
-                // identifier to prevent the files from the other translations
-                // from being copied in the final step.
-                let extra_file_dir = match &ctx.build_opts.language_ident {
-                    Some(lang_ident) => {
-                        let mut path = src_dir.clone();
-                        path.push(lang_ident);
-                        path
-                    }
-                    None => src_dir.clone(),
-                };
-
                 self.render_book(
                     ctx,
                     &book,
                     src_dir,
-                    &extra_file_dir,
                     &ctx.destination,
                     &ctx.config.build.build_dir,
                     &ctx.build_opts.language_ident,
@@ -92,10 +73,9 @@ impl HtmlHandlebars {
         ctx: &RenderContext,
         book: &Book,
         src_dir: &PathBuf,
-        extra_file_dir: &PathBuf,
         destination: &PathBuf,
         build_dir: &PathBuf,
-        language_ident: &Option<String>,
+        language: &Option<String>,
         html_config: &HtmlConfig,
         handlebars: &mut Handlebars<'a>,
         theme: &Theme,
@@ -106,7 +86,7 @@ impl HtmlHandlebars {
             &book,
             &ctx.book,
             &ctx.config,
-            language_ident,
+            language,
             &html_config,
             &theme,
         )?;
@@ -127,7 +107,14 @@ impl HtmlHandlebars {
                 html_config: html_config.clone(),
                 edition: ctx.config.rust.edition,
             };
-            self.render_item(item, item_ctx, src_dir, &ctx.config, &mut print_content)?;
+            self.render_item(
+                item,
+                item_ctx,
+                &src_dir,
+                language,
+                &ctx.config,
+                &mut print_content,
+            )?;
             is_index = false;
         }
 
@@ -136,9 +123,9 @@ impl HtmlHandlebars {
             self.render_404(
                 ctx,
                 &html_config,
-                src_dir,
+                &src_dir,
                 destination,
-                language_ident,
+                language,
                 handlebars,
                 &mut data,
             )?;
@@ -178,6 +165,28 @@ impl HtmlHandlebars {
         self.emit_redirects(&ctx.destination, handlebars, &html_config.redirect)
             .context("Unable to emit redirects")?;
 
+        // `src_dir` points to the root source directory. If this book
+        // is actually multilingual and we specified a single language
+        // to build on the command line, then `src_dir` will not be
+        // pointing at the subdirectory with the specified translation's
+        // index/summary files. We have to append the language
+        // identifier to prevent the files from the other translations
+        // from being copied in the final step.
+        let extra_file_dir = match language {
+            Some(lang_ident) => {
+                // my_book/src/ja/
+                let mut path = src_dir.clone();
+                path.push(lang_ident);
+                path
+            }
+            // my_book/src/
+            None => src_dir.clone(),
+        };
+        debug!(
+            "extra file dir {:?} {:?} {:?}",
+            extra_file_dir, language, ctx.config
+        );
+
         // Copy all remaining files, avoid a recursive copy from/to the book build dir
         utils::fs::copy_files_except_ext(
             &extra_file_dir,
@@ -195,6 +204,7 @@ impl HtmlHandlebars {
         item: &BookItem,
         mut ctx: RenderItemContext<'_>,
         src_dir: &PathBuf,
+        language: &Option<String>,
         cfg: &Config,
         print_content: &mut String,
     ) -> Result<()> {
@@ -219,28 +229,32 @@ impl HtmlHandlebars {
                 .insert("git_repository_edit_url".to_owned(), json!(edit_url));
         }
 
-        let fallback_path = cfg.default_language().map(|lang_ident| {
+        let mut md_ctx = match language {
-            let mut fallback = PathBuf::from(utils::fs::path_to_root(&path));
+            Some(lang_ident) => RenderMarkdownContext {
-            fallback.push("../");
+                path: path.clone(),
-            fallback.push(lang_ident.clone());
+                src_dir: src_dir.clone(),
-            fallback
+                language: Some(lang_ident.clone()),
-        });
+                fallback_language: cfg.default_language(),
+                prepend_parent: false,
+            },
+            None => RenderMarkdownContext {
+                path: path.clone(),
+                src_dir: src_dir.clone(),
+                language: None,
+                fallback_language: None,
+                prepend_parent: false,
+            },
+        };
 
         let content = ch.content.clone();
-        let content = utils::render_markdown_with_path(
+        let content =
-            &content,
+            utils::render_markdown_with_path(&content, ctx.html_config.curly_quotes, Some(&md_ctx));
-            ctx.html_config.curly_quotes,
-            Some(&path),
-            Some(&src_dir),
-            &fallback_path,
-        );
 
+        md_ctx.prepend_parent = true;
         let fixed_content = utils::render_markdown_with_path(
             &ch.content,
             ctx.html_config.curly_quotes,
-            Some(&path),
+            Some(&md_ctx),
-            Some(&src_dir),
-            &fallback_path,
         );
         if !ctx.is_index {
             // Add page break between chapters

src/utils/fs.rs

@@ -47,6 +47,9 @@ pub fn path_to_root<P: Into<PathBuf>>(path: P) -> String {
         .fold(String::new(), |mut s, c| {
             match c {
                 Component::Normal(_) => s.push_str("../"),
+                Component::ParentDir => {
+                    s.truncate(s.len() - 3);
+                }
                 _ => {
                     debug!("Other path component... {:?}", c);
                 }
@@ -189,7 +192,7 @@ pub fn get_404_output_file(input_404: &Option<String>) -> String {
 
 #[cfg(test)]
 mod tests {
-    use super::copy_files_except_ext;
+    use super::{copy_files_except_ext, path_to_root};
     use std::{fs, io::Result, path::Path};
 
     #[cfg(target_os = "windows")]
@@ -272,4 +275,10 @@ mod tests {
             panic!("output/symlink.png should exist")
         }
     }
+
+    #[test]
+    fn test_path_to_root() {
+        assert_eq!(path_to_root("some/relative/path"), "../../");
+        assert_eq!(path_to_root("some/relative/other/../path"), "../../");
+    }
 }

src/utils/mod.rs

@@ -10,13 +10,36 @@ use pulldown_cmark::{html, CodeBlockKind, CowStr, Event, Options, Parser, Tag};
 
 use std::borrow::Cow;
 use std::fmt::Write;
-use std::path::{Path, PathBuf};
+use std::path::PathBuf;
 
 pub use self::string::{
     take_anchored_lines, take_lines, take_rustdoc_include_anchored_lines,
     take_rustdoc_include_lines,
 };
 
+/// Context for rendering markdown. This is used for fixing up links in the
+/// output if one is missing in a translation.
+#[derive(Clone, Debug)]
+pub struct RenderMarkdownContext {
+    /// Directory of the file being rendered, relative to the language's directory.
+    /// If the file is "src/en/chapter/README.md", it is "chapter".
+    pub path: PathBuf,
+    /// Absolute path to the source directory of the book being rendered, across
+    /// all languages.
+    /// If the file is "src/en/chapter/README.md", it is "src/".
+    pub src_dir: PathBuf,
+    /// Language of the book being rendered.
+    /// If the file is "src/en/chapter/README.md", it is "en".
+    /// If the book is not multilingual, it is `None`.
+    pub language: Option<String>,
+    /// Fallback language to use if a link is missing. This is configured in
+    /// `book.language` in the config.
+    /// If the book is not multilingual, it is `None`.
+    pub fallback_language: Option<String>,
+    /// If true, prepend the parent path to the link.
+    pub prepend_parent: bool,
+}
+
 lazy_static! {
     static ref SCHEME_LINK: Regex = Regex::new(r"^[a-z][a-z0-9+.-]*:").unwrap();
     static ref MD_LINK: Regex = Regex::new(r"(?P<link>.*)\.md(?P<anchor>#.*)?").unwrap();
@@ -76,32 +99,70 @@ pub fn id_from_content(content: &str) -> String {
     normalize_id(trimmed)
 }
 
-fn md_to_html_link<'a>(dest: &CowStr<'a>, fixed_link: &mut String) {
+fn rewrite_if_missing(
-    if let Some(caps) = MD_LINK.captures(&dest) {
+    fixed_link: &mut String,
-        fixed_link.push_str(&caps["link"]);
+    path_to_dest: &PathBuf,
-        fixed_link.push_str(".html");
+    dest: &str,
-        if let Some(anchor) = caps.name("anchor") {
+    src_dir: &PathBuf,
-            fixed_link.push_str(anchor.as_str());
+    language: &str,
+    fallback_language: &str,
+) {
+    // We are inside a multilingual book.
+    //
+    // `fixed_link` is a string relative to the current language directory, like
+    // "cli/README.md". Prepend the language's source directory (like "src/ja") and see
+    // if the file exists.
+    let mut path_on_disk = src_dir.clone();
+    path_on_disk.push(language);
+    path_on_disk.push(path_to_dest);
+    path_on_disk.push(dest);
+
+    debug!("Checking if {} exists", path_on_disk.display());
+    if !path_on_disk.exists() {
+        // Now see if the file exists in the fallback language directory (like "src/en").
+        let mut fallback_path = src_dir.clone();
+        fallback_path.push(fallback_language);
+        fallback_path.push(path_to_dest);
+        fallback_path.push(dest);
+
+        debug!(
+            "Not found, checking if fallback {} exists",
+            fallback_path.display()
+        );
+        if fallback_path.exists() {
+            // We can fall back to this link. Get enough parent directories to
+            // reach the root source directory, append the fallback language
+            // directory to it, the prepend the whole thing to the link.
+            let mut relative_path = PathBuf::from(path_to_dest);
+            relative_path.push(dest);
+
+            let mut path_to_fallback_src = fs::path_to_root(&relative_path);
+            // One more parent directory out of language folder ("en")
+            write!(path_to_fallback_src, "../{}/", fallback_language).unwrap();
+
+            debug!(
+                "Rewriting link to be under fallback: {}",
+                path_to_fallback_src
+            );
+            fixed_link.insert_str(0, &path_to_fallback_src);
         }
-    } else {
+    }
-        fixed_link.push_str(&dest);
-    };
 }
 
-fn fix<'a, P: AsRef<Path>>(
+fn fix<'a>(dest: CowStr<'a>, ctx: Option<&RenderMarkdownContext>) -> CowStr<'a> {
-    dest: CowStr<'a>,
-    path: Option<&Path>,
-    src_dir: Option<&Path>,
-    fallback_path: &Option<P>,
-) -> CowStr<'a> {
     if dest.starts_with('#') {
         // Fragment-only link.
-        if let Some(path) = path {
+        if let Some(ctx) = ctx {
-            let mut base = path.display().to_string();
+            if ctx.prepend_parent {
-            if base.ends_with(".md") {
+                let mut base = ctx.path.display().to_string();
-                base.replace_range(base.len() - 3.., ".html");
+                if base.ends_with(".md") {
+                    base.replace_range(base.len() - 3.., ".html");
+                }
+                info!("{:?} {:?}", base, dest);
+                return format!("{}{}", base, dest).into();
+            } else {
+                return dest;
             }
-            return format!("{}{}", base, dest).into();
         } else {
             return dest;
         }
@@ -111,62 +172,54 @@ fn fix<'a, P: AsRef<Path>>(
         // This is a relative link, adjust it as necessary.
         let mut fixed_link = String::new();
 
-        // If this link is missing on the filesystem in the current directory,
+        if let Some(ctx) = ctx {
-        // but not in the fallback directory, use the fallback's page.
+            // If the book is multilingual, check if the file actually
-        let mut redirected_path = false;
+            // exists, and if not rewrite the link to the fallback
-        if let Some(src_dir) = src_dir {
+            // language's page.
-            let mut dest_path = src_dir.to_str().unwrap().to_string();
+            if let Some(language) = &ctx.language {
-            write!(dest_path, "/{}", dest).unwrap();
+                if let Some(fallback_language) = &ctx.fallback_language {
-            trace!("Check existing: {:?}", dest_path);
+                    rewrite_if_missing(
-            if !PathBuf::from(dest_path).exists() {
+                        &mut fixed_link,
-                if let Some(fallback_path) = fallback_path {
+                        &ctx.path,
-                    let mut fallback_file = src_dir.to_str().unwrap().to_string();
+                        &dest,
-                    // Check if there is a Markdown or other file in the fallback.
+                        &ctx.src_dir,
-                    write!(
+                        &language,
-                        fallback_file,
+                        &fallback_language,
-                        "/{}/{}",
+                    );
-                        fallback_path.as_ref().display(),
+                }
-                        dest
+            }
-                    )
+
-                    .unwrap();
+            if ctx.prepend_parent {
-                    trace!("Check fallback: {:?}", fallback_file);
+                let base = ctx
-                    if PathBuf::from(fallback_file).exists() {
+                    .path
-                        write!(fixed_link, "{}/", fallback_path.as_ref().display()).unwrap();
+                    .parent()
-                        debug!(
+                    .expect("path can't be empty")
-                            "Redirect link to default translation: {:?} -> {:?}",
+                    .to_str()
-                            dest, fixed_link
+                    .expect("utf-8 paths only");
-                        );
+
-                        redirected_path = true;
+                if !base.is_empty() {
-                    }
+                    write!(fixed_link, "{}/", base).unwrap();
                 }
             }
         }
 
-        if let Some(path) = path {
+        if let Some(caps) = MD_LINK.captures(&dest) {
-            let base = path
+            fixed_link.push_str(&caps["link"]);
-                .parent()
+            fixed_link.push_str(".html");
-                .expect("path can't be empty")
+            if let Some(anchor) = caps.name("anchor") {
-                .to_str()
+                fixed_link.push_str(anchor.as_str());
-                .expect("utf-8 paths only");
-            trace!("Base: {:?}", base);
-
-            if !redirected_path && !base.is_empty() {
-                write!(fixed_link, "{}/", base).unwrap();
             }
-        }
+        } else {
+            fixed_link.push_str(&dest);
+        };
 
-        md_to_html_link(&dest, &mut fixed_link);
+        debug!("Fixed link: {:?}, {:?} => {:?}", dest, ctx, fixed_link);
         return CowStr::from(fixed_link);
     }
     dest
 }
 
-fn fix_html<'a, P: AsRef<Path>>(
+fn fix_html<'a>(html: CowStr<'a>, ctx: Option<&RenderMarkdownContext>) -> CowStr<'a> {
-    html: CowStr<'a>,
-    path: Option<&Path>,
-    src_dir: Option<&Path>,
-    fallback_path: &Option<P>,
-) -> CowStr<'a> {
     // This is a terrible hack, but should be reasonably reliable. Nobody
     // should ever parse a tag with a regex. However, there isn't anything
     // in Rust that I know of that is suitable for handling partial html
@@ -182,7 +235,7 @@ fn fix_html<'a, P: AsRef<Path>>(
 
     HTML_LINK
         .replace_all(&html, move |caps: &regex::Captures<'_>| {
-            let fixed = fix(caps[2].into(), path, src_dir, fallback_path);
+            let fixed = fix(caps[2].into(), ctx);
             format!("{}{}\"", &caps[1], fixed)
         })
         .into_owned()
@@ -198,31 +251,22 @@ fn fix_html<'a, P: AsRef<Path>>(
 /// page go to the original location. Normal page rendering sets `path` to
 /// None. Ideally, print page links would link to anchors on the print page,
 /// but that is very difficult.
-fn adjust_links<'a, P: AsRef<Path>>(
+fn adjust_links<'a>(event: Event<'a>, ctx: Option<&RenderMarkdownContext>) -> Event<'a> {
-    event: Event<'a>,
-    path: Option<&Path>,
-    src_dir: Option<&Path>,
-    fallback_path: &Option<P>,
-) -> Event<'a> {
     match event {
-        Event::Start(Tag::Link(link_type, dest, title)) => Event::Start(Tag::Link(
+        Event::Start(Tag::Link(link_type, dest, title)) => {
-            link_type,
+            Event::Start(Tag::Link(link_type, fix(dest, ctx), title))
-            fix(dest, path, src_dir, fallback_path),
+        }
-            title,
+        Event::Start(Tag::Image(link_type, dest, title)) => {
-        )),
+            Event::Start(Tag::Image(link_type, fix(dest, ctx), title))
-        Event::Start(Tag::Image(link_type, dest, title)) => Event::Start(Tag::Image(
+        }
-            link_type,
+        Event::Html(html) => Event::Html(fix_html(html, ctx)),
-            fix(dest, path, src_dir, fallback_path),
-            title,
-        )),
-        Event::Html(html) => Event::Html(fix_html(html, path, src_dir, fallback_path)),
         _ => event,
     }
 }
 
 /// Wrapper around the pulldown-cmark parser for rendering markdown to HTML.
 pub fn render_markdown(text: &str, curly_quotes: bool) -> String {
-    render_markdown_with_path(text, curly_quotes, None, None, &None::<PathBuf>)
+    render_markdown_with_path(text, curly_quotes, None)
 }
 
 pub fn new_cmark_parser(text: &str) -> Parser<'_> {
@@ -234,19 +278,17 @@ pub fn new_cmark_parser(text: &str) -> Parser<'_> {
     Parser::new_ext(text, opts)
 }
 
-pub fn render_markdown_with_path<P: AsRef<Path>>(
+pub fn render_markdown_with_path(
     text: &str,
     curly_quotes: bool,
-    path: Option<&Path>,
+    ctx: Option<&RenderMarkdownContext>,
-    src_dir: Option<&Path>,
-    fallback_path: &Option<P>,
 ) -> String {
     let mut s = String::with_capacity(text.len() * 3 / 2);
     let p = new_cmark_parser(text);
     let mut converter = EventQuoteConverter::new(curly_quotes);
     let events = p
         .map(clean_codeblock_headers)
-        .map(|event| adjust_links(event, path, src_dir, fallback_path))
+        .map(|event| adjust_links(event, ctx))
         .map(|event| converter.convert(event));
 
     html::push_html(&mut s, events);
@@ -350,7 +392,7 @@ pub fn log_backtrace(e: &Error) {
 #[cfg(test)]
 mod tests {
     mod render_markdown {
-        use super::super::{render_markdown, render_markdown_with_path};
+        use super::super::{fix, render_markdown, RenderMarkdownContext};
 
         #[test]
         fn preserves_external_links() {
@@ -468,72 +510,78 @@ more text with spaces
             assert_eq!(render_markdown(input, true), expected);
         }
 
-        use std::fs::{self, File};
+        use std::fs;
+        use std::fs::File;
         use std::io::Write;
         use std::path::PathBuf;
-        use tempfile::{Builder as TempFileBuilder, TempDir};
+        use tempfile;
-
-        const DUMMY_SRC: &str = "
-# Dummy Chapter
-
-this is some dummy text.
-
-And here is some \
-more text.
-";
-
-        /// Create a dummy `Link` in a temporary directory.
-        fn dummy_link() -> (PathBuf, TempDir) {
-            let temp = TempFileBuilder::new().prefix("book").tempdir().unwrap();
-
-            let chapter_path = temp.path().join("chapter_1.md");
-            File::create(&chapter_path)
-                .unwrap()
-                .write_all(DUMMY_SRC.as_bytes())
-                .unwrap();
-
-            let path = chapter_path.to_path_buf();
-
-            (path, temp)
-        }
 
         #[test]
-        fn links_are_rewritten_to_fallback_for_nonexistent_files() {
+        fn test_link_rewriting() {
-            let input = r#"
+            use pulldown_cmark::CowStr;
-[Link](chapter_1.md)
-"#;
 
-            let (localized_file, localized_dir) = dummy_link();
+            let _ = env_logger::builder().is_test(true).try_init();
-            fs::remove_file(&localized_file).unwrap();
+            let test = |dest, path, exists, expected| {
+                let src_dir = tempfile::tempdir().unwrap();
 
-            let (_, fallback_dir) = dummy_link();
+                let ctx = if exists {
-            let mut relative_fallback_dir =
+                    Some(RenderMarkdownContext {
-                PathBuf::from(super::super::fs::path_to_root(localized_dir.path()));
+                        path: PathBuf::from(path),
-            relative_fallback_dir.push(fallback_dir.path().file_name().unwrap());
+                        src_dir: PathBuf::new(),
+                        language: None,
+                        fallback_language: None,
+                        prepend_parent: false,
+                    })
+                } else {
+                    let localized_dir = src_dir.path().join("ja");
+                    fs::create_dir_all(&localized_dir).unwrap();
 
-            let expected_fallback = format!(
+                    let fallback_dir = src_dir.path().join("en");
-                "<p><a href=\"{}/chapter_1.html\">Link</a></p>\n",
+                    fs::create_dir_all(&fallback_dir).unwrap();
-                relative_fallback_dir.display()
+
+                    let chapter_path = fallback_dir.join(path).join(dest);
+                    fs::create_dir_all(chapter_path.parent().unwrap()).unwrap();
+                    debug!("Create: {}", chapter_path.display());
+                    File::create(&chapter_path)
+                        .unwrap()
+                        .write_all(b"# Chapter")
+                        .unwrap();
+
+                    Some(RenderMarkdownContext {
+                        path: PathBuf::from(path),
+                        src_dir: PathBuf::from(src_dir.path()),
+                        language: Some(String::from("ja")),
+                        fallback_language: Some(String::from("en")),
+                        prepend_parent: false,
+                    })
+                };
+
+                assert_eq!(
+                    fix(CowStr::from(dest), ctx.as_ref()),
+                    CowStr::from(expected)
+                );
+            };
+
+            test("../b/summary.md", "a", true, "../b/summary.html");
+            test("../b/summary.md", "a", false, "../../en/../b/summary.html");
+            test("../c/summary.md", "a/b", true, "../c/summary.html");
+            test(
+                "../c/summary.md",
+                "a/b",
+                false,
+                "../../../en/../c/summary.html",
             );
-            assert_eq!(
+            test(
-                render_markdown_with_path(
+                "#translations",
-                    input,
+                "config.md",
-                    false,
+                true,
-                    None,
+                "#translations",
-                    Some(localized_dir.path()),
-                    &Some(&relative_fallback_dir)
-                ),
-                expected_fallback
             );
-            assert_eq!(
+            test(
-                render_markdown_with_path(
+                "#translations",
-                    input,
+                "config.md",
-                    true,
+                false,
-                    None,
+                "#translations",
-                    Some(localized_dir.path()),
-                    &Some(&relative_fallback_dir)
-                ),
-                expected_fallback
             );
         }
     }

tests/rendered_output.rs

@@ -464,7 +464,7 @@ fn theme_dir_overrides_work_correctly() {
     let md = MDBook::load(book_dir).unwrap();
     md.build().unwrap();
 
-    let built_index = book_dir.join("book").join("index.html");
+    let built_index = book_dir.join("book").join("en").join("index.html");
     dummy_book::assert_contains_strings(built_index, &["This is a modified index.hbs!"]);
 }