api/services/parsers/forumlog.go

package parsers

import (
	"bufio"
	"fmt"
	"git.aiterp.net/rpdata/api/models"
	"strings"
	"time"
)

// A ParsedLog contains the parsed log header and its posts.
type ParsedLog struct {
	Log   models.Log
	Posts []*models.Post
}

// ForumLog parses the logs from the data.
func ForumLog(data string, tz *time.Location) ([]ParsedLog, error) {
	metadata := ForumLogMetadata(data)
	results := make([]ParsedLog, 0, len(metadata["Date"]))
	scanner := bufio.NewScanner(strings.NewReader(data))

	for i, dateStr := range metadata["Date"] {
		// Parse date
		date, err := time.ParseInLocation("January 2, 2006", dateStr, tz)
		if err != nil {
			return nil, fmt.Errorf("failed to parse date #%d: %#+v is not the in the correct format of \"January 2, 2006\"", i+1, dateStr)
		}

		// Parse posts
		posts := make([]*models.Post, 0, 128)
		parsing := false
		prev := ""
		prevPost := models.Post{}
		for scanner.Scan() {
			line := strings.Trim(scanner.Text(), "\r\t  ")

			// Skip lines with links to other logs using the --> and <-- notation.
			if strings.HasPrefix(line, "-") || strings.HasPrefix(line, "<") {
				prev = line
				continue
			}

			// If parsing and reaching a double empty-line, the session is done.
			if parsing && len(line) < 2 && len(prev) < 2 {
				break
			}

			// Skip empty lines, but record them as thep revious for the above check.
			if len(line) < 2 {
				prev = line
				continue
			}

			// If not parsing, skip until the first mirclike post.
			if !parsing {
				if strings.HasPrefix(line, "[") {
					parsing = true
				} else {
					prev = line
					continue
				}
			}

			// Parse the post.
			post, err := MircPost(line, date, prevPost)
			if err != nil {
				summary := ""
				for _, ru := range line {
					summary += string(ru)
					if len(summary) > 60 {
						summary += "..."
						break
					}
				}

				return nil, fmt.Errorf("failed to parse post: %s", summary)
			}

			posts = append(posts, &post)
			prevPost = post
			prev = line
		}

		// No posts means there's a problem.
		if len(posts) == 0 {
			return nil, fmt.Errorf("session %d (%s) has no posts (too many dates?)", i+1, dateStr)
		}

		// Add it.
		results = append(results, ParsedLog{
			Log:   models.Log{Date: posts[0].Time},
			Posts: posts,
		})
	}

	return results, nil
}

// ForumLogMetadata parses metadata, discards the broken parts, and returns the
// parsed data as a map (`m`) and the position of the first IRC post (`n`)
func ForumLogMetadata(data string) map[string][]string {
	result := make(map[string][]string)
	key := ""

	scanner := bufio.NewScanner(strings.NewReader(data))
	for scanner.Scan() {
		line := strings.Trim(scanner.Text(), "\r\t  ")

		if strings.HasPrefix(line, "[") {
			break
		}
		if len(line) < 1 {
			key = ""
			continue
		}

		if key == "" {
			split := strings.Split(line, ":")
			key = split[0]
		} else {
			result[key] = append(result[key], line)
		}
	}

	return result
}