gistfile1.go

/**
 * Created with IntelliJ IDEA.
 * User: dacc
 * Date: 8/27/13
 * Time: 10:01 AM
 * To change this template use File | Settings | File Templates.
 */
package main

import (
	"io/ioutil"
	"strings"
	"errors"
	"fmt"
	"strconv"
	"log"
)

func panicIfNotNil(err error) {
	if err != nil { panic(err) }
}

func parseUint32s(strings []string) (uints []uint32, err error) {
	uints = make([]uint32, len(strings))
	for i, s := range strings {
		result, err := strconv.ParseUint(s, 10, 32)
		if err != nil { break }
		uints[i] = uint32(result)
	}
	return
}

func readLines(file string) (lines []string, err error) {
	log.Println("reading in", file, "...")
	bytes, err := ioutil.ReadFile(file)
	lines_raw := strings.Split(string(bytes), "\n")
	lines = lines_raw[0:len(lines_raw) - 1]
	return
}

func loadGraph() (titles []string, adjacencies map[uint32][]uint32, err error) {
	titles, err = readLines("titles-sorted.txt")
	if err != nil { return }

	adjacency_lines, err := readLines("links-simple-sorted.txt")
	if err != nil { return }

	var origin uint64
	var targets []uint32

	adjacencies = make(map[uint32][]uint32)
	for i, line := range adjacency_lines {
		components := strings.Split(line, " ")
		origin_component := components[0]

		if ! strings.HasSuffix(origin_component, ":") {
			err = errors.New(fmt.Sprintf("Bad adjacency line at %d: '%s'", i, line))
			if err != nil { return }
		}

		origin, err = strconv.ParseUint(strings.TrimSuffix(origin_component, ":"), 10, 32)
		if err != nil { return }

		targets, err = parseUint32s(components[1:])
		if err != nil { return }

		adjacencies[uint32(origin)] = targets

		if i % 100000 == 0 {
			log.Println("parsed", i, "/", len(adjacency_lines), "adjacency lists...")
		}
	}
	return
}

func main() {
	titles, adjacencies, err := loadGraph()
	if err != nil { panic(err) }

	log.Println("titles:", len(titles), "adjacencies", len(adjacencies))
}