path: root/src/collect_files.rs

use crate::*;

use std::fmt::Debug;


pub struct Website {
    pub name: Name,
    pub config: Config,
    pub pages: Vec<Page>,
    pub redirects: Vec<Redirect>,
    pub static_files: Vec<StaticFile>,      // !-prefixed-dir contents
    pub feeds: Vec<Feed>,                   // RSS feeds
}

pub struct Page {
    pub name: Name,
    pub url: String,                        // No extension
    pub parents: Vec<Name>,                 // Parent names
    pub parent_url: String,                 // With trailing slash
    pub source_path: PathBuf,               // Absolute path to source file
    pub document: MarkdownDocument,         // File content parsed as markdown
    pub headings: Vec<Heading>,             // Ordered list of all headings in page
    pub last_modified: Option<SystemTime>,  // Last-modified time of source file
}

pub struct Heading {
    pub name: Name,
    pub prefix: Option<Name>,               // Disambiguation
    pub level: Level,
    pub block_id: usize,                    // Pointer to heading element in document
}

impl Heading {
    pub fn slug(&self) -> String {
        match &self.prefix {
            Some(prefix) => format!("{}-{}", prefix.slug(), self.name.slug()),
            None => self.name.slug(),
        }
    }
}

pub struct StaticFile {
    pub url: String,                        // With extension
    pub source_path: PathBuf,               // Absolute path to source file
    pub last_modified: Option<SystemTime>,  // Last-modified time of source file
}

pub struct Feed {
    pub name: Name,                         // Taken from file name
    pub url: String,                        // With extension
    pub parents: Vec<Name>,                 // Parent names
    pub parent_url: String,                 // Base URL for feed pages
    pub source_path: PathBuf,               // Absolute path to source file
    pub last_modified: Option<SystemTime>,  // Last-modified time of source file
}

pub struct Redirect {
    pub name: Name,
    pub url: String,                        // No extension
    pub parents: Vec<Name>,                 // Parent names
    pub parent_url: String,                 // Base URL for relative redirects
    pub target: String,                     // Page to redirect to, internal link
    pub source_path: PathBuf,               // Absolute path to source file, for logging
    pub last_modified: Option<SystemTime>,  // Last-modified time of source file
}


/// Calculate correct relative path from this entity to a specified page.
pub trait LinkFrom: Debug {
    fn name(&self) -> &Name;
    fn parent_url(&self) -> &str;
    fn parents(&self) -> &[Name];
    fn root(&self) -> String {
        let mut root = String::new();
        for _ in self.parents() {
            root.push_str("../");
        }
        return root;
    }
    fn qualified_name(&self) -> String {
        match self.parents().last() {
            Some(parent) => format!("{parent}/{}", self.name()),
            None => format!("/{}", self.name()),
        }
    }
    /// Convert an internal link to a canonical page URL and optional heading,
    /// both as slugs.
    ///
    /// `path` and returned URL have no extension.
    fn canonicalise(&self, path: &str) -> (String, Option<String>) {
        // Remove heading fragment from path.
        let (path, heading) = match path.rsplit_once('#') {
            Some((path, heading)) => match heading.is_empty() {
                true => (path, None),
                false => (path, Some(to_slug(heading))),
            }
            None => (path, None),
        };
        let mut path = path.to_string();
        // Attach parent URL if not an absolute path.
        if !path.starts_with('/') {
            path = format!("{}{path}", self.parent_url());
        }
        // Convert path to a canonical URL.
        path = to_slug(&collapse_path(&path));
        return (path, heading);
    }
}

impl Page {
    pub fn root(&self) -> String {
        let mut root = String::new();
        for _ in &self.parents {
            root.push_str("../");
        }
        return root;
    }
}

impl Debug for Page {
    fn fmt(&self, f: &mut std::fmt::Formatter) -> Result<(), std::fmt::Error> {
        write!(f, "\"{}\"", self.qualified_name())
    }
}

impl Debug for Redirect {
    fn fmt(&self, f: &mut std::fmt::Formatter) -> Result<(), std::fmt::Error> {
        write!(f, "\"{}\"", self.qualified_name())
    }
}

impl Debug for Feed {
    fn fmt(&self, f: &mut std::fmt::Formatter) -> Result<(), std::fmt::Error> {
        write!(f, "\"{}\"", self.qualified_name())
    }
}

impl LinkFrom for Page {
    fn name(&self) -> &Name { &self.name }
    fn parent_url(&self) -> &str { &self.parent_url }
    fn parents(&self) -> &[Name] { &self.parents }
}

impl LinkFrom for Redirect {
    fn name(&self) -> &Name { &self.name }
    fn parent_url(&self) -> &str { &self.parent_url }
    fn parents(&self) -> &[Name] { &self.parents }
}

impl LinkFrom for Feed {
    fn name(&self) -> &Name { &self.name }
    fn parent_url(&self) -> &str { &self.parent_url }
    fn parents(&self) -> &[Name] { &self.parents }
}

pub struct ImagePaths {
    pub thumb: String,
    pub large: String,
}


impl Website {
    pub fn from_path(path: &Path) -> Self {
        let mut new = Self {
            pages: Vec::new(),
            redirects: Vec::new(),
            static_files: Vec::new(),
            feeds: Vec::new(),
            name: match Entry::from_path(path) {
                Ok(entry) => entry.name.into(),
                Err(err) => fatal!("Couldn't open {:?}: {:?}", &path, err),
            },
            config: Config::new(),
        };
        // Recursively collect entire website.
        new.collect_entry(path, path);
        new.parse_hoisted_folders();
        return new;
    }

    /// Read the hoisted_folders config key, make root redirects for each
    /// child of each listed directory.
    fn parse_hoisted_folders(&mut self) {
        for line in self.config.get("hoisted_folders").lines() {
            if line.is_empty() { continue }
            // Turn line into a path
            let path = PathBuf::from(line);
            let prefix: Vec<Name> = path.components()
                .filter(|c| if let std::path::Component::Normal(_) = c {true} else {false})
                .map(|c| c.as_os_str().to_string_lossy().to_string())
                .map(|s| strip_numeric_prefix(&s).into())
                .collect();
            for page in &self.pages {
                if page.parents == prefix {
                    let name = page.name.clone();
                    let url = name.slug();
                    let parents = Vec::new();
                    let parent_url = String::new();
                    let target = page.url.clone();
                    let source_path = "<hoisted child>".into();
                    let last_modified = self.config.last_modified;
                    self.redirects.push(Redirect {
                        name,
                        url,
                        parents,
                        parent_url,
                        target,
                        source_path,
                        last_modified,
                    });
                }
            }
        }
    }

    /// Recursively collect an entry and all children.
    /// `prefix` is the base directory path for the entire website.
    fn collect_entry(&mut self, path: &Path, prefix: &Path) {
        let entry = Entry::from_path(path).unwrap();
        // Ignore dotted entries.
        if entry.name.starts_with('.') { return }
        // Get name, extension, last-modified.
        let (name_raw, extension) = entry.split_name();
        let name: Name = strip_numeric_prefix(&name_raw).into();
        let last_modified = entry.last_modified;
        // Generate parent URL, used only for files.
        let source_path = entry.original_path.clone();
        let relative_path = source_path.strip_prefix(prefix).unwrap_or_else(
            |_| fatal!("Path doesn't start with {prefix:?}: {source_path:?}"));
        let mut parents: Vec<Name> = relative_path.components()
            .map(|c| c.as_os_str().to_string_lossy().to_string())
            .map(|s| strip_numeric_prefix(&s).into())
            .collect();
        parents.pop();  // Remove final (non-parent) segment.

        // Process each entry.
        if entry.is_directory() {
            if let Some(stripped) = entry.name.strip_prefix("!") {
                // Track all files inside the static directory.
                for child in traverse_directory(&entry).unwrap() {
                    let source_path = child.original_path;
                    let relative_path = source_path.strip_prefix(&entry.original_path).unwrap_or_else(
                        |_| fatal!("Path doesn't start with {prefix:?}: {source_path:?}"))
                        .as_os_str().to_string_lossy().to_string();
                    let url = format!("{stripped}/{relative_path}");
                    self.static_files.push(StaticFile { url, source_path, last_modified })
                }
            } else {
                for child in list_directory(entry.original_path).unwrap() {
                    self.collect_entry(&child.original_path, prefix);
                }
            }
        } else if parents.is_empty() && entry.name.to_lowercase() == "toaster.conf" {
            info!("Reading configuration file at {path:?}");
            let content = std::fs::read_to_string(&source_path).unwrap();
            self.config.parse_file(&content, last_modified);
        } else {
            // Used for error messages, to distinguish between pages of the same name.
            let qualified_name = match parents.last() {
                Some(parent) => format!("{parent}/{name}"),
                None => name.to_string(),
            };
            match extension.as_str() {
                "feed" => {
                    let mut url = String::new();
                    for parent in &parents {
                        url.push_str(&parent.slug());
                        url.push('/');
                    }
                    let parent_url = url.clone();
                    url.push_str(&name.plain());
                    self.feeds.push(Feed {
                        name,
                        url,
                        parents,
                        parent_url,
                        source_path,
                        last_modified,
                    });
                }
                "redirect" => {
                    let mut url = String::new();
                    for parent in &parents {
                        url.push_str(&parent.slug());
                        url.push('/');
                    }
                    let parent_url = url.clone();
                    url.push_str(&name.slug());
                    let target = std::fs::read_to_string(&source_path)
                        .unwrap().trim().to_string();
                    self.redirects.push(Redirect {
                        name,
                        url,
                        parents,
                        parent_url,
                        target,
                        source_path,
                        last_modified,
                    });
                }
                "md" => {
                    let markdown = std::fs::read_to_string(&source_path).unwrap();
                    let document = MarkdownDocument::from_str(&markdown);
                    // Collect headings, check for duplicates.
                    let mut names_set = HashSet::new();   // all heading names
                    let mut duplicates = HashSet::new();  // only duplicates
                    let mut headings: Vec<Heading> = document.blocks.iter().enumerate()
                        .filter_map(|(block_id, block)| if let Block::Heading { line, level } = block {
                            let name: Name = line.to_string().into();
                            let level = level.to_owned();
                            let heading = Heading { name, prefix: None, level, block_id };
                            if !names_set.insert(heading.slug()) {
                                duplicates.insert(heading.slug());
                            }
                            Some(heading)
                        } else {
                            None
                        }).collect();

                    // Namespace any duplicate headings to the parent h1 heading.
                    let mut prefix = None;
                    for heading in &mut headings {
                        if let Level::Heading1 = heading.level {
                            prefix = Some(heading.name.clone());
                        } else {
                            if duplicates.contains(&heading.slug()) {
                                heading.prefix = prefix.clone();
                            }
                        }
                    }
                    // Check for duplicates once more, and warn if any.
                    names_set.clear();
                    duplicates.clear();
                    for heading in &headings {
                        if !names_set.insert(heading.slug()) {
                            duplicates.insert(heading.slug());
                        }
                    }
                    for slug in duplicates {
                        warn!("Page {qualified_name:?} contains multiple headings with ID \"#{slug}\"");
                    }

                    if name.slug() == "+index" {
                        if parents.is_empty() {
                            // This is the index file for the whole site.
                            self.pages.push(Page {
                                name: self.name.clone(),
                                url: String::from("index"),
                                parents,
                                parent_url: String::from(""),
                                source_path,
                                document,
                                headings,
                                last_modified,
                            });
                        } else {
                            // This is an index file for a directory.
                            let name = parents.last().unwrap().clone();
                            let mut url = String::new();
                            for parent in &parents {
                                url.push_str(&parent.slug());
                                url.push('/');
                            }
                            let parent_url = url.clone();
                            url.pop();      // Remove the trailing slash
                            parents.pop();  // Remove this directory
                            self.pages.push(Page {
                                name,
                                url,
                                parents,
                                parent_url,
                                source_path,
                                document,
                                headings,
                                last_modified,
                            });
                        }
                    } else {
                        // This is a regular page.
                        let mut url = String::new();
                        for parent in &parents {
                            url.push_str(&parent.slug());
                            url.push('/');
                        }
                        // Children descend from this page, so the parent
                        // url must contain this page.
                        url.push_str(&name.slug());
                        let mut parent_url = url.clone();
                        parent_url.push('/');
                        self.pages.push(Page {
                            name,
                            url,
                            parents,
                            parent_url,
                            source_path,
                            document,
                            headings,
                            last_modified,
                        });
                    }
                },
                _ => {
                    // This is a static file.
                    let mut parent_url = String::new();
                    for parent in &parents {
                        parent_url.push_str(&parent.slug());
                        parent_url.push('/');
                    }
                    let name_slug = name.slug();
                    let url = format!("{parent_url}{name_slug}.{extension}");
                    self.static_files.push(StaticFile { url, source_path, last_modified });
                },
            }
        }
    }

    /// Check if the internal link `path` is valid, pointing to a real internal
    /// page with extension `ext` and heading, relative to the current page (`from`).
    /// Returns a resolved absolute link to the page, with extension.
    pub fn has_page(&self, from: &impl LinkFrom, path: &str, ext: &str) -> Option<String> {
        let original_path = path;
        let (mut path, mut heading) = from.canonicalise(path);
        if let Some(stripped) = path.strip_suffix(&format!(".{ext}")) {
            path = stripped.to_string();
        };

        // Find page with this path in website, resolving any redirect first.
        for redirect in &self.redirects {
            if redirect.url == path {
                let (target_path, target_heading) = redirect.canonicalise(&redirect.target);
                path = target_path;
                if target_heading.is_some() && heading.is_some() {
                    warn!("Page {from:?} contains link {original_path:?} to a redirect that also links to a heading");
                }
                if heading.is_none() {
                    heading = target_heading;
                }
            }
        }
        for page in &self.pages {
            if page.url == path {
                let root = from.root();
                if let Some(heading) = heading {
                    if !page.headings.iter().any(|h| h.slug() == heading) {
                        warn!("Page {from:?} contains link to nonexistent heading {heading:?} on page {path:?}");
                    }
                    return Some(format!("{root}{path}.{ext}#{heading}"));
                } else {
                    return Some(format!("{root}{path}.{ext}"));
                }
            }
        }
        return None;
    }

    /// Check if the external link `path` points to a valid static file or
    /// generated feed. Returns a resolved absolute link to the file.
    pub fn has_static(&self, from: &impl LinkFrom, path: &str) -> Option<String> {
        // Attach parent if not an absolute path.
        // We don't want to canonicalise/sluggify the path.
        let path = match !path.starts_with('/') {
            true => collapse_path(&format!("{}{path}", from.parent_url())),
            false => collapse_path(path),
        };
        let root = from.root();
        for file in &self.static_files {
            if file.url == path {
                return Some(format!("{root}{path}"));
            }
        }
        for feed in &self.feeds {
            if format!("{}.rss", feed.url) == path {
                return Some(format!("{root}{path}"));
            }
        }
        return None;
    }

    /// Check if a particular image exists.
    pub fn has_image(&self, file_name: &str, root: &str) -> Option<ImagePaths> {
        let check = |path: String|
            match self.static_files.iter().any(|s| s.url == path) {
                true => Some(format!("{root}{path}")),
                false => None,
            };
        let thumb_path  = check(format!("images/thumb/{file_name}"));
        let large_path  = check(format!("images/large/{file_name}"));
        let fallback_path = check(format!("images/{file_name}"))
            .or(large_path.clone())
            .or(thumb_path.clone());

        Some(ImagePaths {
            thumb: thumb_path.or(fallback_path.clone())?,
            large: large_path.or(fallback_path.clone())?,
        })
    }
}


fn collapse_path(path: &str) -> String {
    // Iteratively collapse ".." segments.
    let mut segments: Vec<&str> = path.split('/')
        .filter(|s| !s.is_empty() && *s != ".")
        .collect();
    'outer: loop {
        for i in 0..(segments.len().saturating_sub(1)) {
            if segments[i] == ".." {
                if i == 0 {
                    segments.remove(0);
                } else {
                    segments.remove(i-1);
                    segments.remove(i-1);
                }
                continue 'outer;
            }
        }
        return segments.join("/");
    }
}


fn strip_numeric_prefix(name: &str) -> String {
    if let Some((prefix, suffix)) = name.split_once(' ') {
        if prefix.chars().all(|c| "0123456789-".contains(c)) {
            return suffix.to_string();
        }
    }
    return name.to_string();
}