historian/src/lib.rs

static DEFAULT_INDEX_FILENAME: &'static str = "README.md";
static DEFAULT_TEMPLATES_PATH: &'static str = "templates";
static DEFAULT_TOML_FILENAME: &'static str = "Historian.toml";
static MD_EXTENSION: &'static str = ".md";

extern crate pulldown_cmark;
extern crate tera;
extern crate serde;
extern crate git2;
extern crate regex;
extern crate pathdiff;
extern crate chrono;
extern crate grep;
#[macro_use] extern crate toml;

use std::fs;
use std::path::{Path, PathBuf};
use std::io::{Read, Write};
use std::collections::{HashMap, VecDeque};
use std::convert::From;
use std::sync::Mutex;
use std::ffi::OsStr;

use git2::{Repository, Signature, Commit};

use rocket::form::validate::Contains;
use tera::Tera;
use toml::value::Table;
use serde::Serialize;

use regex::Regex;
use pathdiff::diff_paths;

use chrono::{DateTime, MappedLocalTime, TimeZone, Local};

use pulldown_cmark::{Event, Tag, TagEnd};

pub struct Historian {
    source_root: PathBuf,
    index_filename: String,
    pub repository: Option<Mutex<Repository>>,
    pub site_config: Table
}

impl Historian {
    pub fn new(root: String) -> Historian {
        let source_root: PathBuf = root.into();
        let toml_path = source_root.join(DEFAULT_TOML_FILENAME);
        let site_config = match fs::metadata(&toml_path) {
            Err(_) => Table::new(),
            Ok(_) => {
                let toml_data = fs::read_to_string(&toml_path).unwrap();
                toml::de::from_str(&toml_data).unwrap()
            }
        };

        Historian {
            source_root: source_root.clone(),
            index_filename: DEFAULT_INDEX_FILENAME.to_owned(),
            repository: Repository::open(source_root).ok().map(Mutex::new),
            site_config
        }
    }

    pub fn resolve_from_file(file_path_str: &str) -> Option<(Historian, Page)> {
        let file_path: PathBuf = std::path::absolute(file_path_str).unwrap();
        match fs::metadata(&file_path) {
            Err(_) => None,
            Ok(metadata) => {
                if !metadata.is_file() {
                    None
                } else {
                    let mut wiki_path = file_path.parent().unwrap();
                    while wiki_path.parent() != None {
                        let toml_path = wiki_path.join(DEFAULT_TOML_FILENAME);
                        match fs::metadata(&toml_path) {
                            Err(_) => {
                                wiki_path = wiki_path.parent().unwrap();
                            },
                            Ok(_) => {
                                let historian = Historian::new(wiki_path.to_str().unwrap().to_owned());
                                let relative_page_path = diff_paths(&file_path, &wiki_path).unwrap();
                                return relative_page_path.to_str()
                                    .and_then(|page_path| historian.resolve_to_page(page_path))
                                    .map(|page| (historian, page));
                            }
                        };
                    }
                    None
                }
            }
        }
    }

    pub fn resolve_to_page(&self, name: &str) -> Option<Page> {
        let mut file_path = self.source_root.clone().join(Path::new(name));
        match fs::metadata(&file_path) {
            Err(_) => None,
            Ok(metadata) => {
                let mut children = vec![];
                let mut attachments = vec![];
                if metadata.is_dir() {
                    for entry in fs::read_dir(&file_path).unwrap() {
                        let entry_file = entry.unwrap();
                        let child = entry_file.file_name().into_string().unwrap();
                        let entry_metadata = entry_file.metadata().unwrap();
                        let entry_path = entry_file.path();

                        if entry_metadata.is_file() && !child.ends_with(MD_EXTENSION) {
                            attachments.push(child);
                        } else if !(child.starts_with(".") || child == self.index_filename || child == DEFAULT_TOML_FILENAME) {
                            let mut full_name = format!("{}/{}", name, child);
                            if full_name.starts_with('/') {
                                full_name.remove(0);
                            }

                            children.push(Child {
                                name: child.to_owned(),
                                full_name: full_name,
                                title: if entry_metadata.is_file() {
                                    entry_path.file_stem().unwrap().to_str().unwrap().to_owned()
                                } else {
                                    child.to_owned()
                                },
                                path: entry_path
                            });
                        }
                    }

                    file_path.push(&self.index_filename);
                }

                let mut split_path = name.rsplitn(2, "/");
                let base_name = split_path.next().unwrap().to_owned();
                let parent_page = if name != "" {
                    self.resolve_to_page(split_path.next().unwrap_or(""))
                        .map(Box::new)
                } else {
                    None
                };

                let mut url = name.to_owned();
                if url != "" && metadata.is_dir() {
                    url.push('/');
                }
                Some(Page {
                    title: if metadata.is_file() {
                        file_path.file_stem().unwrap().to_str().unwrap().to_owned()
                    } else {
                        base_name.to_owned()
                    },
                    full_name: name.to_owned(),
                    name: base_name,
                    url,
                    parent: parent_page,
                    path: file_path,
                    is_directory: metadata.is_dir(),
                    children,
                    attachments
                })
            }
        }
    }

    pub fn resolve_to_attachment(&self, name: &str) -> Option<PathBuf> {
        if name.ends_with(MD_EXTENSION) {
            return None;
        }

        let file_path = self.source_root.clone().join(Path::new(name));
        match fs::metadata(&file_path) {
            Err(_) => None,
            Ok(metadata) => {
                if metadata.is_dir() {
                    None
                } else {
                    Some(file_path)
                }
            }
        }
    }

    // Creates a struct for a new page.
    // The page is not saved until an edit is submitted.
    pub fn start_page(&self, name: &str) -> Page {
        let mut file_path = self.source_root.clone().join(Path::new(name));
        let is_directory = file_path.extension().is_none();
        let mut url = name.to_owned();

        if is_directory {
            file_path.push(&self.index_filename);
            url.push('/');
        }

        let mut split_path = name.rsplitn(2, "/");
        let base_name = split_path.next().unwrap().to_owned();

        Page {
            title: name.to_owned(),
            full_name: name.to_owned(),
            name: base_name,
            url,
            parent: None,
            path: file_path,
            is_directory,
            children: vec![],
            attachments: vec![]
        }
    }

    pub fn submit_edit(&self, page: &Page, change: &Edit) {
        // Create parent directories if necessary
        if let Some(directory) = &page.path.parent() {
            fs::create_dir_all(directory);
        }

        // write contents of file
        let mut page_html_file = fs::File::create(&page.path).unwrap();
        page_html_file.write_all(change.content.as_bytes()).unwrap();
        
        // commit file to git repository
        if let Some(repository_mutex) = &self.repository {
            if let Ok(repository) = repository_mutex.lock() {
                // add file to index
                let mut index = repository.index().unwrap();
                index.add_path(&page.path.strip_prefix(&self.source_root).unwrap()).unwrap();
                index.write().unwrap();
                let tree_oid = index.write_tree().unwrap();
                let tree = repository.find_tree(tree_oid).unwrap();

                // find parent commit
                let parent_commit = if let Ok(commit) = repository.revparse_single("HEAD") {
                    commit.into_commit().ok()
                } else {
                    None
                };

                let mut parents = Vec::new();
                if parent_commit.is_some() {
                    parents.push(parent_commit.as_ref().unwrap());
                }

                // create commit
                let signature = Signature::now(change.author.as_deref().unwrap_or("Historian"), "historian@local").unwrap();
                repository.commit(
                    Some("HEAD"),
                    &signature,
                    &signature,
                    &change.summary,
                    &tree,
                    &parents[..],
                ).unwrap();
            }
        }
    }

    pub fn get_history_of_page(&self, page: &Page) -> Vec<Revision> {
        let mut changes = Vec::new();
        let page_path_in_repo = &page.path.strip_prefix(&self.source_root).unwrap();
        if let Some(repository_mutex) = &self.repository {
            if let Ok(repository) = repository_mutex.lock() {
                let mut revwalk = repository.revwalk().unwrap();
                revwalk.push_head().unwrap();
                for result in revwalk {
                    if let Ok(rev) = result {
                        let commit = repository.find_commit(rev).unwrap();
                        if commit_includes_file(&repository, &commit, page_path_in_repo) {
                            changes.push(Revision {
                                id: commit.id().to_string(),
                                author: commit.author().name().unwrap().to_owned(),
                                summary: commit.summary().unwrap().to_owned(),
                                datetime: Local.timestamp_opt(commit.time().seconds(), 0).unwrap()
                            });
                        }
                    }
                }
            }
        }
        changes
    }

    pub fn get_revision_by_id(&self, id: &str) -> Option<Revision> {
        if let Some(repository_mutex) = &self.repository {
            if let Ok(repository) = repository_mutex.lock() {
                let object = repository.revparse_single(id).ok()?;
                let commit = object.as_commit()?;
                return Some(Revision {
                    id: commit.id().to_string(),
                    author: commit.author().name().unwrap().to_owned(),
                    summary: commit.summary().unwrap().to_owned(),
                    datetime: Local.timestamp_opt(commit.time().seconds(), 0).unwrap()
                });
            }
        }
        None
    }

    // ref: https://github.com/rust-lang/git2-rs/issues/996
    pub fn get_page_text_of_revision(&self, page: &Page, revision: &Revision) -> Option<String> {
        let page_path_in_repo = &page.path.strip_prefix(&self.source_root).ok()?;
        if let Some(repository_mutex) = &self.repository {
            if let Ok(repository) = repository_mutex.lock() {
                let object = repository.revparse_single(&revision.id).ok()?;
                let tree = object.peel_to_tree().ok()?;
                let rev_path = tree.get_path(page_path_in_repo).ok()?;
                let path_object = rev_path.to_object(&repository).ok()?;
                let blob = path_object.into_blob().ok()?;
                return Some(std::str::from_utf8(blob.content()).ok()?.to_owned());
            }
        }
        None
    }
}

fn commit_includes_file(repository: &Repository, commit: &Commit, path: &Path) -> bool {
    if commit.parent_count() != 1 {
        return false;
    }
    
    let prev_commit = commit.parent(0).unwrap();
    let tree = commit.tree().unwrap();
    let prev_tree = prev_commit.tree().unwrap();
    let diff = repository.diff_tree_to_tree(Some(&prev_tree), Some(&tree), None).unwrap();
    for delta in diff.deltas() {
        let file_path = delta.new_file().path().unwrap();
        if file_path == path {
            println!(" -- file path {:?}", file_path);
            return true;
        }
    }
    false
}

#[derive(Serialize)]
pub struct Page {
    pub title: String,
    pub full_name: String,
    pub name: String,
    pub path: PathBuf,
    pub url: String,
    pub is_directory: bool,
    pub parent: Option<Box<Page>>,
    pub children: Vec<Child>,
    pub attachments: Vec<String>
}

#[derive(Serialize)]
pub struct Child {
    pub title: String,
    pub full_name: String,
    pub name: String,
    pub path: PathBuf
}

#[derive(Serialize)]
pub struct Edit {
    pub author: Option<String>,
    pub content: String,
    pub summary: String
}

#[derive(Serialize)]
pub struct Revision {
    pub id: String,
    pub author: String,
    pub summary: String,
    pub datetime: DateTime<Local>
}

pub struct PageRenderer {
    template_root: PathBuf,
    tera: Tera
}

fn parse_markdown (content: &str, metadata: &mut Table) -> String {
    let mut metadata_str = String::new();
    let mut in_metadata = false;
    let parser = pulldown_cmark::Parser::new_ext(content, pulldown_cmark::Options::ENABLE_YAML_STYLE_METADATA_BLOCKS).filter(|event| {
        match &event {
            Event::Start(tag) => {
                if let Tag::MetadataBlock(_) = tag {
                    in_metadata = true;
                }
            }
            Event::Text(text) => {
                if in_metadata {
                    metadata_str.push_str(text);
                }
            }
            Event::Html(html) => {
                if html.starts_with("<!---") {
                    in_metadata = true;
                } else if html.starts_with("--->") {
                    in_metadata = false;
                    return false;
                } else if in_metadata {
                    metadata_str.push_str(html);
                }
            }
            Event::End(tag) => {
                if !metadata_str.is_empty() {
                    if let Ok(parsed_metadata) = toml::de::from_str::<Table>(&metadata_str) {
                        println!("{:?}", parsed_metadata);
                        for (key, value) in parsed_metadata {
                            metadata.insert(key, value);
                        }
                    }

                    metadata_str.clear();
                    in_metadata = false;
                    return false;
                }
            }
            _ => {}
        }
        !in_metadata
    });
    let mut html_output = String::new();
    pulldown_cmark::html::push_html(&mut html_output, parser);
    html_output
}

fn render_markdown (content: &tera::Value, args: &HashMap<String, tera::Value>) -> tera::Result<tera::Value> {
    Ok(tera::Value::String(parse_markdown(content.as_str().unwrap(), &mut Table::new())))
}

impl PageRenderer {
    pub fn new() -> PageRenderer {
        Self::with_template_path(DEFAULT_TEMPLATES_PATH)
    }

    pub fn with_template_path(template_path: &str) -> PageRenderer {
        let mut tera = tera::Tera::new(&format!("{template_path}/**/*.html")).unwrap();
        tera.register_filter("markdown", render_markdown);
        PageRenderer {
            template_root: template_path.into(),
            tera
        }
    }
    
    pub fn render_page(&self, historian: &Historian, page: &Page, options: &Table) -> String {
        self.render_page_template("page.html", &historian, &page, &options)
    }

    pub fn render_page_template(&self, template: &str, historian: &Historian, page: &Page, options: &Table) -> String {
        self.template(template)
            .with_page(page)
            .with_historian(historian)
            .insert("options", options)
            .render()
    }

    pub fn render_page_template_with_variables(&self, template: &str, historian: &Historian, page: &Page, variables: &Table) -> String {
        self.template(template)
            .with_page(page)
            .with_historian(historian)
            .insert_all(variables)
            .render()
    }

    pub fn render_template(&self, template: &str, historian: &Historian, variables: &Table, options: &Table) -> String {
        self.template(template)
            .with_url("/")
            .with_historian(historian)
            .insert_all(variables)
            .insert("options", options)
            .render()
    }

    pub fn resource_names(&self) -> Vec<String> {
        let mut children = vec![];
        for entry in fs::read_dir(&self.template_root).unwrap() {
            let child = entry.unwrap().file_name().into_string().unwrap();
            if !child.ends_with(".html") {
                children.push(child);
            }
        }

        children
    }

    pub fn resolve_to_resource(&self, name: &str) -> Option<PathBuf> {
        let file_path = self.template_root.clone().join(Path::new(name));
        match fs::metadata(&file_path) {
            Err(_) => None,
            Ok(metadata) => {
                if metadata.is_dir() {
                    None
                } else {
                    Some(file_path)
                }
            }
        }
    }

    pub fn template<'a>(&'a self, template: &'a str) -> Render<'a> {
        Render::new(&self.tera, template)
    }
}

pub struct Render<'a> {
    tera: &'a Tera,
    context: tera::Context,
    template: &'a str
}

fn make_relative_root (url: &str) -> String {
    let mut slash_count = 0;
    for character in url.chars() {
        if character == '/' {
            slash_count = slash_count + 1;
        }
    }
    "../".repeat(slash_count)
}

impl<'a> Render<'a> {
    fn new(tera: &'a Tera, template: &'a str) -> Render<'a> {
        Render {
            tera,
            context: tera::Context::new(),
            template
        }
    }

    pub fn insert<T: Serialize + ?Sized, S: Into<String>>(mut self, key: S, value: &'a T) -> Render {
        self.context.insert(key, value);
        self
    }

    pub fn insert_all(mut self, variables: &'a Table) -> Render {
        for (key, value) in variables {
            self.context.insert(key, value);
        }
        self
    }

    pub fn with_url(mut self, url: &'a str) -> Render {
        self.context.insert("relative_root", &make_relative_root(url));
        self
    }

    pub fn with_historian(mut self, historian: &'a Historian) -> Render {
        self.context.insert("site", &historian.site_config);
        self.context.insert("has_git", &historian.repository.is_some());
        self
    }

    pub fn with_page(mut self, page: &'a Page) -> Render {
        self.context.insert("page", &page);

        let mut content = String::new();
        if let Ok(mut file) = fs::File::open(&page.path) {
            file.read_to_string(&mut content).unwrap();
        }
        self.context.insert("content", &content);

        let mut metadata = Table::new();
        let html_content = parse_markdown(&content, &mut metadata);
        self.context.insert("html_content", &html_content);
        self.context.insert("metadata", &metadata);

        let mut ancestors: VecDeque<&Page> = VecDeque::new();
        let mut ancestor = page.parent.as_ref();
        while ancestor.is_some() {
            ancestors.push_front(ancestor.unwrap().as_ref());
            ancestor = ancestor.unwrap().parent.as_ref();
        }
        self.context.insert("ancestors", &ancestors);

        let mut page_url = page.full_name.clone();
        if page.full_name != "" && page.is_directory {
            page_url.push('/');
        }
        self.context.insert("relative_root", &make_relative_root(&page.url));
        self
    }

    pub fn render(self) -> String {
        self.tera.render(&self.template, &self.context).unwrap()
    }
}

impl From<Render<'_>> for String {
    fn from(render: Render) -> Self {
        render.render()
    }
}

pub fn export_wiki(historian: &Historian, renderer: &PageRenderer, output_path: &str) {
    fs::create_dir_all(&output_path).unwrap();
    for resource_name in renderer.resource_names() {
        let mut resource_output_path: PathBuf = output_path.into();
        resource_output_path.push(&resource_name);
        export_resource(renderer.resolve_to_resource(&resource_name).unwrap(), output_path);
    }

    export_wiki_page(historian, renderer, "", output_path);
}

fn export_resource(resource_path: PathBuf, output_path: &str) {
    let mut resource_output_path: PathBuf = output_path.into();
    let resource_name = resource_path.file_name().unwrap().to_str().unwrap();
    resource_output_path.push(&resource_name);
    println!("export resource {} from {:?} to {:?}", resource_name, resource_path, resource_output_path);
    fs::copy(resource_path, resource_output_path).unwrap();
}

fn export_wiki_page(historian: &Historian, renderer: &PageRenderer, name: &str, output_path: &str) {
    println!("resolve page {} {:?}", name, historian.resolve_to_page(name).is_some());    
    if let Some(page) = historian.resolve_to_page(name) {
        let page_path: PathBuf = page.full_name.to_owned().replace(".md", ".html").into();
        let mut page_output_path: PathBuf = output_path.into();
        page_output_path.push(&page_path);

        let page_html_output_path = if page.is_directory {
            println!("create directory {:?}", page_output_path);
            fs::create_dir_all(&page_output_path);
            page_output_path.join("index.html")
        } else {
            page_output_path.clone()
        };

        println!("export: {} to {:?}", page.full_name, page_html_output_path);
        let page_html = renderer.render_page_template("page.html", historian, &page, &toml! {
            dynamic = false
        }).replace(".md", ".html");

        let mut page_html_file = fs::File::create(page_html_output_path).unwrap();
        page_html_file.write_all(page_html.as_bytes());

        for attachment in page.attachments {
            let attachment_path = page.path.parent().unwrap().join(attachment);
            export_resource(attachment_path, page_output_path.to_str().unwrap());
        }

        for child in page.children {
            println!("child {} {} {:?} {:?}", child.full_name, child.name, child.path, output_path);
            export_wiki_page(historian, renderer, &child.full_name, output_path);
        }
    }
}

pub struct Linker<'a> {
    historian: &'a Historian,
    link_regex: Regex
}

impl<'a> Linker<'a> {
    pub fn new(historian: &Historian) -> Linker {
        Linker {
            historian,
            link_regex: Regex::new(r"\[\[(?<link>[\w\s\-]+)(?:\|(?<label>[\w\s\-]+))?\]\]").unwrap()
        }
    }

    pub fn resolve_link(&self, link: &str) -> Option<String> {
        let root = self.historian.resolve_to_page("")?;
        let mut page_names: Vec<String> = root.children.iter().map(|child| child.name.to_owned()).collect();

        loop {
            let mut next_page_names: Vec<String> = vec![];

            // loop through current list of page names to find match
            for page_name in page_names {
                let page = self.historian.resolve_to_page(&page_name)?;
                if self.link_matches_path(link, &page_name) {
                    return Some(page.full_name.to_owned());
                }

                // Collect list of (fully qualified) subpages
                for child in page.children {
                    let mut child_path = page.full_name.to_owned();
                    child_path.push('/');
                    child_path.push_str(&child.name);
                    next_page_names.push(child_path);
                }
            }

            if next_page_names.is_empty() {
                break;
            }

            //
            page_names = next_page_names;
        }

        None
    }

    pub fn resolve_links(&self, page: &Page) -> String {
        let mut content = String::new();
        if let Ok(mut file) = fs::File::open(&page.path) {
            file.read_to_string(&mut content).unwrap();
        }

        self.resolve_links_for_edit(&page, &Edit {
            author: None,
            content,
            summary: String::new()
        })
    }

    pub fn resolve_links_for_edit(&self, page: &Page, edit: &Edit) -> String {
        let mut content = edit.content.to_owned();
        for capture in self.link_regex.captures_iter(&content.to_owned()) {
            let link = capture.name("link").unwrap();
            if let Some(resolved_link) = self.resolve_link(link.as_str()) {
                let link_full = capture.get(0).unwrap().as_str();
                let label = capture.name("label").unwrap_or(link).as_str();
                let mut absolute_link_path = self.historian.source_root.to_owned();
                absolute_link_path.push(&resolved_link);
                let relative_link_path = diff_paths(&absolute_link_path, &page.path.parent().unwrap()).unwrap();
                content = content.replace(link_full, &format!("[{}](<{}>)", label, relative_link_path.display()));
            }
        }

        content
    }

    fn link_matches_path(&self, link: &str, path: &str) -> bool {
        if let Some(page_name) = path.split('/').last() {
            page_name == link || Path::new(page_name).file_stem() == Some(OsStr::new(link))
        } else {
            false
        }
    }
}

pub struct Searcher<'a> {
    historian: &'a Historian
}

#[derive(Serialize)]
pub struct SearchResult {
    pub page: Page,
    pub matches: Vec<String>
}

impl<'a> Searcher<'a> {
    pub fn new(historian: &Historian) -> Searcher {
        Searcher {
            historian
        }
    }

    pub fn search(&self, root: &Page, query: &str) -> Vec<SearchResult> {
        let mut searcher = grep::searcher::SearcherBuilder::new().build();
        let matcher = grep::regex::RegexMatcherBuilder::new()
            .fixed_strings(true)
            .case_insensitive(true)
            .build(query)
            .unwrap();

        let mut results: Vec<SearchResult> = vec![];
        self.do_search(root, &mut results, &mut searcher, &matcher);
        results
    }

    pub fn tag_search(&self, root: &Page, tag: &str) -> Vec<SearchResult> {
        let mut results: Vec<SearchResult> = vec![];
        for child in &root.children {
            if let Some(child_page) = self.historian.resolve_to_page(&child.full_name) {
                for result in self.tag_search(&child_page, tag) {
                    results.push(result);
                }

                let mut content = String::new();
                if let Ok(mut file) = fs::File::open(&child_page.path) {
                    file.read_to_string(&mut content).unwrap();
                }

                let mut metadata = Table::new();
                parse_markdown(&content, &mut metadata);

                if metadata.contains_key("tags") {
                    let tags = metadata.get("tags").unwrap();
                    if tags.is_array() && tags.as_array().contains(toml::Value::String(tag.to_owned())) {
                        results.push(SearchResult {
                            page: child_page,
                            matches: vec![]
                        });
                    }
                }
            }
        }
        results
    }

    fn do_search(&self, root: &Page, results: &mut Vec<SearchResult>, searcher: &mut grep::searcher::Searcher, matcher: &grep::regex::RegexMatcher) {
        for child in &root.children {
            if let Some(child_page) = self.historian.resolve_to_page(&child.full_name) {
                let mut matches: Vec<String> = vec![];
                searcher.search_path(matcher, &child.path, grep::searcher::sinks::UTF8(|lnum, line| {
                    matches.push(line.to_owned());
                    Ok(true)
                }));
                
                self.do_search(&child_page, results, searcher, matcher);

                if !matches.is_empty() {
                    results.push(SearchResult {
                        page: child_page,
                        matches
                    });
                }
            }
        }
    }
}