api/internal/importers/forumlog/logs.go

package forumlog

import (
	"bufio"
	"fmt"
	"strings"
	"time"

	"git.aiterp.net/rpdata/api/internal/importers/mirclike"

	"git.aiterp.net/rpdata/api/models"
)

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

// ParseLogs parses the logs from the data.
func ParseLogs(data string, tz *time.Location) ([]ParsedLog, error) {
	metadata := ParseMetadata(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 := mirclike.ParsePost(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
}