/* Copyright 2018 Martin Dosch
Licensed under the "MIT License" */

package main

import (
	"encoding/json"
	"hash/fnv"
	"log"
	"os"
	"os/user"
	"strconv"
	"strings"
	"time"

	"github.com/mmcdole/gofeed"
	"jaytaylor.com/html2text"
)

// Get new articles for specified feed.
func getArticles(feedURL string, max int, noExcerpt bool) (string, error) {

	type feedCache struct {
		LastChange string
	}

	var output, cachePath string
	var last time.Time
	var lastUpdate feedCache
	var file *os.File
	var updateTime time.Time

	// Get systems user cache path.
	osCacheDir := os.Getenv("$XDG_CACHE_HOME")
	if osCacheDir != "" {
		// Create configPath if not yet existing.
		cachePath = osCacheDir + "/feed-to-muc/"
		if _, err := os.Stat(cachePath); os.IsNotExist(err) {
			err = os.MkdirAll(cachePath, 0700)
			if err != nil {
				log.Fatal("Error: ", err)
			}
		}

	} else { // Get the current user.
		curUser, err := user.Current()
		if err != nil {
			log.Fatal("Error: ", err)
			return "", err
		}
		// Get home directory.
		home := curUser.HomeDir

		if home == "" {
			log.Fatal("Error: No home directory available.")
			return "", err
		}

		// Create cachePath if not yet existing.
		cachePath = home + "/.cache/feed-to-muc/"
		if _, err := os.Stat(cachePath); os.IsNotExist(err) {
			err = os.MkdirAll(cachePath, 0700)
			if err != nil {
				log.Fatal("Error: ", err)
			}
		}

	}

	// Create a hash as identifier for the feed.
	// The identifier will be used as filename for caching the update time.
	h := fnv.New32a()
	h.Write([]byte(feedURL))
	if _, err := os.Stat(cachePath); os.IsNotExist(err) {
		err = os.MkdirAll(cachePath, 0700)
		if err != nil {
			log.Fatal("Error: ", err)
		}
	}

	cacheFile := cachePath + strconv.Itoa(int(h.Sum32()))

	if _, err := os.Stat(cacheFile); os.IsNotExist(err) {
		file, err = os.Create(cacheFile)
		if err != nil {
			log.Fatal("Error: ", err)
		}
		defer file.Close()

		last = time.Now()

		lastUpdate.LastChange = last.Format(time.RFC3339)

		lastUpdateJSON, _ := json.MarshalIndent(lastUpdate, "", " ")
		_, err = file.Write(lastUpdateJSON)
		if err != nil {
			log.Fatal("Error: ", err)
		}

	} else {

		file, err = os.OpenFile(cacheFile, os.O_RDWR, 0600)
		if err != nil {
			log.Fatal("Error: ", err)
		}
		defer file.Close()

		decoder := json.NewDecoder(file)
		lastUpdate := feedCache{}
		if err := decoder.Decode(&lastUpdate); err != nil {
			log.Fatal("Error: ", err)
		}

		last, err = time.Parse(time.RFC3339, string(lastUpdate.LastChange))
		if err != nil {
			log.Fatal("Error: ", err)
		}
	}

	fp := gofeed.NewParser()
	feed, err := fp.ParseURL(feedURL)
	if err != nil {
		// Don't return an error, but log a message as the
		// bot should not crash when the feed is not available.
		log.Println(feedURL, ": Feed not available.")
		return "", nil
	}

	// If no publish date is offered try update date.
	// If both is not offered give up.
	if feed.Items[0].PublishedParsed == nil {
		if feed.Items[0].UpdatedParsed == nil {
			return "", err
		}
		// If cached timestamp is newer than the one of
		// the last article return.
		if last.After(*feed.Items[0].UpdatedParsed) {
			return "", err
		}
	} else {
		// If cached timestamp is newer than the one of
		// the last article return.
		if last.After(*feed.Items[0].PublishedParsed) {
			return "", err
		}
	}

	// Check last n (defined in config) articles for new ones.
	for i := max - 1; i >= 0; i-- {
		// Stop processing for article i if there are not so
		// many articles in the feed.
		if len(feed.Items) < i+1 {
			continue
		}
		article := *feed.Items[i]
		if err != nil {
			return "", err
		}

		if article.PublishedParsed == nil {
			updateTime = *article.UpdatedParsed
		} else {
			updateTime = *article.PublishedParsed
		}

		// If cached timestamp is not older than the article stop processing.
		// Note: Checking for cached timestamp being newer, instead of not older
		// lead to duplicate messages for the same article. Probably a corner
		// case when the time is identical.
		if last.Before(updateTime) == false {
			continue
		}

		last = updateTime
		lastUpdate.LastChange = updateTime.Format(time.RFC3339)

		// Remove file with cached timestamp and create it
		// again with updated timestamp.
		// ToDo: Replace timestamp without deleting.
		err = os.Remove(cacheFile)
		if err != nil {
			log.Fatal("Error: ", err)
		}

		file, err = os.Create(cacheFile)
		if err != nil {
			log.Fatal("Error: ", err)
		}
		defer file.Close()

		lastUpdateJSON, _ := json.MarshalIndent(lastUpdate, "", " ")
		_, err = file.Write(lastUpdateJSON)
		if err != nil {
			log.Fatal("Error: ", err)
		}

		// Remove redirects and tracking parameters from URL.
		cleanURL, _ := removeTracking(article.Link)

		// Only send title and link if option noExcerpt is set,
		// otherwise add the description.
		if noExcerpt == true {
			output = output + feed.Title + ": *" + article.Title + "*\n" +
				cleanURL
		} else {
			var description string

			// Some feeds don't provide a description, let's use the content
			// in this case, otherwise use the shorter description.
			if article.Description != "" {
				// Strip HTML as we want to get plain text.
				description, err = html2text.FromString(article.Description)
				if err != nil {
					return "", err
				}
			} else {
				// Strip HTML as we want to get plain text.
				description, err = html2text.FromString(article.Content)
				if err != nil {
					return "", err
				}
			}

			// Remove lines only consisting of "> "; thank you reddit.
			description = strings.Replace(description, "> \n", "", -1)

			// To make the message look not so bloated we remove double newlines.
			// Split the article description/content into fragments between double newlines.
			fragments := strings.Split(description, "\n\n")
			// Empty article description/content
			description = ""
			// Fill article description/content with the fragments separated by one newline.
			for _, line := range fragments {
				// Only if the only content is not empty.
				if line != "" {
					description = description + line + "\n"
				}
			}

			// Only append article link if it is not yet contained in description (e.g. read more: URL).
			if strings.Contains(description, article.Link) == true {
				// Replace article link with URL cleaned from redirects and trackers.
				description = strings.Replace(description, article.Link, cleanURL, -1)
				output = output + feed.Title + ": *" + article.Title + "*\n\n" + description
			} else {
				output = output + feed.Title + ": *" + article.Title + "*\n\n" + description + cleanURL
			}
		}

		if i > 0 {
			output = output + "\n\n---\n\n"
		}
	}

	return output, err
}