git.notmuchmail.org Git - notmuch/blob - bindings/go/src/notmuch-addrlookup/addrlookup.go

   1 package main
   2
   3 // stdlib imports
   4 import "os"
   5 import "path"
   6 import "log"
   7 import "fmt"
   8 import "regexp"
   9 import "strings"
  10 import "sort"
  11
  12 // 3rd-party imports
  13 import "notmuch"
  14 import "github.com/msbranco/goconfig"
  15
  16 type mail_addr_freq struct {
  17         addr  string
  18         count [3]uint
  19 }
  20
  21 type frequencies map[string]uint
  22
  23 /* Used to sort the email addresses from most to least used */
  24 func sort_by_freq(m1, m2 *mail_addr_freq) int {
  25         if m1.count[0] == m2.count[0] &&
  26                 m1.count[1] == m2.count[1] &&
  27                 m1.count[2] == m2.count[2] {
  28                 return 0
  29         }
  30
  31         if m1.count[0] > m2.count[0] ||
  32                 m1.count[0] == m2.count[0] &&
  33                         m1.count[1] > m2.count[1] ||
  34                 m1.count[0] == m2.count[0] &&
  35                         m1.count[1] == m2.count[1] &&
  36                         m1.count[2] > m2.count[2] {
  37                 return -1
  38         }
  39
  40         return 1
  41 }
  42
  43 type maddresses []*mail_addr_freq
  44
  45 func (self *maddresses) Len() int {
  46         return len(*self)
  47 }
  48
  49 func (self *maddresses) Less(i, j int) bool {
  50         m1 := (*self)[i]
  51         m2 := (*self)[j]
  52         v := sort_by_freq(m1, m2)
  53         if v <= 0 {
  54                 return true
  55         }
  56         return false
  57 }
  58
  59 func (self *maddresses) Swap(i, j int) {
  60         (*self)[i], (*self)[j] = (*self)[j], (*self)[i]
  61 }
  62
  63 // find most frequent real name for each mail address
  64 func frequent_fullname(freqs frequencies) string {
  65         var maxfreq uint = 0
  66         fullname := ""
  67         freqs_sz := len(freqs)
  68
  69         for mail, freq := range freqs {
  70                 if (freq > maxfreq && mail != "") || freqs_sz == 1 {
  71                         // only use the entry if it has a real name
  72                         // or if this is the only entry
  73                         maxfreq = freq
  74                         fullname = mail
  75                 }
  76         }
  77         return fullname
  78 }
  79
  80 func addresses_by_frequency(msgs *notmuch.Messages, name string, pass uint, addr_to_realname *map[string]*frequencies) *frequencies {
  81
  82         freqs := make(frequencies)
  83
  84         pattern := `\s*(("(\.|[^"])*"|[^,])*<?(?mail\b\w+([-+.]\w+)*\@\w+[-\.\w]*\.([-\.\w]+)*\w\b)>?)`
  85         // pattern := "\\s*((\\\"(\\\\.|[^\\\\\"])*\\\"|[^,])*" +
  86         //      "<?(?P<mail>\\b\\w+([-+.]\\w+)*\\@\\w+[-\\.\\w]*\\.([-\\.\\w]+)*\\w\\b)>?)"
  87         pattern = `.*` + strings.ToLower(name) + `.*`
  88         var re *regexp.Regexp = nil
  89         var err error = nil
  90         if re, err = regexp.Compile(pattern); err != nil {
  91                 log.Printf("error: %v\n", err)
  92                 return &freqs
  93         }
  94
  95         headers := []string{"from"}
  96         if pass == 1 {
  97                 headers = append(headers, "to", "cc", "bcc")
  98         }
  99
 100         for ; msgs.Valid(); msgs.MoveToNext() {
 101                 msg := msgs.Get()
 102                 //println("==> msg [", msg.GetMessageId(), "]")
 103                 for _, header := range headers {
 104                         froms := strings.ToLower(msg.GetHeader(header))
 105                         //println("  froms: ["+froms+"]")
 106                         for _, from := range strings.Split(froms, ",") {
 107                                 from = strings.Trim(from, " ")
 108                                 match := re.FindString(from)
 109                                 //println("  -> match: ["+match+"]")
 110                                 occ, ok := freqs[match]
 111                                 if !ok {
 112                                         freqs[match] = 0
 113                                         occ = 0
 114                                 }
 115                                 freqs[match] = occ + 1
 116                         }
 117                 }
 118         }
 119         return &freqs
 120 }
 121
 122 func search_address_passes(queries [3]*notmuch.Query, name string) []string {
 123         var val []string
 124         addr_freq := make(map[string]*mail_addr_freq)
 125         addr_to_realname := make(map[string]*frequencies)
 126
 127         var pass uint = 0 // 0-based
 128         for _, query := range queries {
 129                 if query == nil {
 130                         //println("**warning: idx [",idx,"] contains a nil query")
 131                         continue
 132                 }
 133                 msgs := query.SearchMessages()
 134                 ht := addresses_by_frequency(msgs, name, pass, &addr_to_realname)
 135                 for addr, count := range *ht {
 136                         freq, ok := addr_freq[addr]
 137                         if !ok {
 138                                 freq = &mail_addr_freq{addr: addr, count: [3]uint{0, 0, 0}}
 139                         }
 140                         freq.count[pass] = count
 141                         addr_freq[addr] = freq
 142                 }
 143                 msgs.Destroy()
 144                 pass += 1
 145         }
 146
 147         addrs := make(maddresses, len(addr_freq))
 148         {
 149                 iaddr := 0
 150                 for _, freq := range addr_freq {
 151                         addrs[iaddr] = freq
 152                         iaddr += 1
 153                 }
 154         }
 155         sort.Sort(&addrs)
 156
 157         for _, addr := range addrs {
 158                 freqs, ok := addr_to_realname[addr.addr]
 159                 if ok {
 160                         val = append(val, frequent_fullname(*freqs))
 161                 } else {
 162                         val = append(val, addr.addr)
 163                 }
 164         }
 165         //println("val:",val)
 166         return val
 167 }
 168
 169 type address_matcher struct {
 170         // the notmuch database
 171         db *notmuch.Database
 172         // full path of the notmuch database
 173         user_db_path string
 174         // user primary email
 175         user_primary_email string
 176         // user tag to mark from addresses as in the address book
 177         user_addrbook_tag string
 178 }
 179
 180 func new_address_matcher() *address_matcher {
 181         // honor NOTMUCH_CONFIG
 182         home := os.Getenv("NOTMUCH_CONFIG")
 183         if home == "" {
 184                 home = os.Getenv("HOME")
 185         }
 186
 187         cfg, err := goconfig.ReadConfigFile(path.Join(home, ".notmuch-config"))
 188         if err != nil {
 189                 log.Fatalf("error loading config file:", err)
 190         }
 191
 192         db_path, _ := cfg.GetString("database", "path")
 193         primary_email, _ := cfg.GetString("user", "primary_email")
 194         addrbook_tag, err := cfg.GetString("user", "addrbook_tag")
 195         if err != nil {
 196                 addrbook_tag = "addressbook"
 197         }
 198
 199         self := &address_matcher{db: nil,
 200                 user_db_path:       db_path,
 201                 user_primary_email: primary_email,
 202                 user_addrbook_tag:  addrbook_tag}
 203         return self
 204 }
 205
 206 func (self *address_matcher) run(name string) {
 207         queries := [3]*notmuch.Query{}
 208
 209         // open the database
 210         if db, status := notmuch.OpenDatabase(self.user_db_path,
 211                 notmuch.DATABASE_MODE_READ_ONLY); status == notmuch.STATUS_SUCCESS {
 212                 self.db = db
 213         } else {
 214                 log.Fatalf("Failed to open the database: %v\n", status)
 215         }
 216
 217         // pass 1: look at all from: addresses with the address book tag
 218         query := "tag:" + self.user_addrbook_tag
 219         if name != "" {
 220                 query = query + " and from:" + name + "*"
 221         }
 222         queries[0] = self.db.CreateQuery(query)
 223
 224         // pass 2: look at all to: addresses sent from our primary mail
 225         query = ""
 226         if name != "" {
 227                 query = "to:" + name + "*"
 228         }
 229         if self.user_primary_email != "" {
 230                 query = query + " from:" + self.user_primary_email
 231         }
 232         queries[1] = self.db.CreateQuery(query)
 233
 234         // if that leads only to a few hits, we check every from too
 235         if queries[0].CountMessages()+queries[1].CountMessages() < 10 {
 236                 query = ""
 237                 if name != "" {
 238                         query = "from:" + name + "*"
 239                 }
 240                 queries[2] = self.db.CreateQuery(query)
 241         }
 242
 243         // actually retrieve and sort addresses
 244         results := search_address_passes(queries, name)
 245         for _, v := range results {
 246                 if v != "" && v != "\n" {
 247                         fmt.Println(v)
 248                 }
 249         }
 250         return
 251 }
 252
 253 func main() {
 254         //fmt.Println("args:",os.Args)
 255         app := new_address_matcher()
 256         name := ""
 257         if len(os.Args) > 1 {
 258                 name = os.Args[1]
 259         }
 260         app.run(name)
 261 }