timetable-mgr/corpus/parse.go

package corpus

import (
	"encoding/json"
	"errors"
	"io"

	"git.fjla.uk/owlboard/go-types/pkg/database"
	"git.fjla.uk/owlboard/timetable-mgr/log"
	"go.uber.org/zap"
)

// Accepts CORPUS data as a byte array and formats it ready for database insertion
func parseCorpusData(stream io.ReadCloser) (*[]database.CorpusEntry, error) {
	defer stream.Close()

	log.Msg.Debug("Starting CORPUS Data parsing")

	var corpusEntries []database.CorpusEntry
	decoder := json.NewDecoder(stream)

	// Expect an object at the root of the JSON stream
	if _, err := decoder.Token(); err != nil {
		log.Msg.Error("Error parsing CORPUS Data", zap.Error(err))
		return nil, err
	}

	// Search for the "TIPLOCDATA" key
	for decoder.More() {
		// Decode the next JSON token
		if tok, err := decoder.Token(); err != nil {
			log.Msg.Error("Error parsing CORPUS Data", zap.Error(err))
			return nil, err
		} else if tok == "TIPLOCDATA" {
			// Found the "TIPLOCDATA" key, expect the associated array
			if !decoder.More() {
				err := errors.New("missing array after TIPLOCDATA key")
				log.Msg.Error("Error parsing CORPUS Data", zap.Error(err))
				return nil, err
			}

			// Start reading the array associated with the "TIPLOCDATA" key
			if _, err := decoder.Token(); err != nil {
				log.Msg.Error("Error parsing CORPUS Data", zap.Error(err))
				return nil, err
			}

			// Iterate over the JSON array
			for decoder.More() {
				var corpusEntry database.CorpusEntry
				if err := decoder.Decode(&corpusEntry); err != nil {
					log.Msg.Error("Error parsing CORPUS Data", zap.Error(err))
					return nil, err
				}
				corpusEntries = append(corpusEntries, corpusEntry)
			}
			break // Exit loop after processing "TIPLOCDATA" array
		}
	}

	log.Msg.Debug("CORPUS parsing complete")

	return &corpusEntries, nil
}

// Removes empty fields from CORPUS entries
func pruneCorpusEntries(corpusEntries []database.CorpusEntry) *[]database.CorpusEntry {
	for i := range corpusEntries {
		if corpusEntries[i].CRS == " " {
			corpusEntries[i].CRS = ""
		}
		if corpusEntries[i].TIPLOC == " " {
			corpusEntries[i].TIPLOC = ""
		}
		if corpusEntries[i].NLCDESC16 == " " {
			corpusEntries[i].NLCDESC16 = ""
		}
		if corpusEntries[i].STANOX == " " {
			corpusEntries[i].STANOX = ""
		}
		if corpusEntries[i].UIC == " " {
			corpusEntries[i].UIC = ""
		}
	}

	return &corpusEntries
}