heuristic assignment transaction <-> vcard

[confetti] / ledgers / iban_assign.awk

diff --git a/ledgers/iban_assign.awk b/ledgers/iban_assign.awk
new file mode 100755 (executable)
index 0000000..090d950
--- /dev/null
+++ b/ledgers/iban_assign.awk
@@ -0,0 +1,91 @@
+#!/bin/awk -f
+
+function STRING( inp ) {
+  gsub(/\\/, "\\\\", inp);
+  gsub(/\n/, "\\n",  inp);
+  gsub(/\r/, "\\r",  inp);
+  gsub(/\t/, "\\t",  inp);
+  gsub(/\+/, "\\+",  inp);
+  gsub(/ /,  "+",    inp);
+  return inp ? inp : "\\";
+}
+
+function UNSTRING( inp,    out, tmp ) {
+  while ( inp ) {
+         if ( inp ~ /^\\\\/) { out = out "\\"; sub(/^\\\\/, "", inp); }
+    else if ( inp ~ /^\\n/)  { out = out "\n"; sub(/^\\n/,  "", inp); }
+    else if ( inp ~ /^\\r/)  { out = out "\r"; sub(/^\\r/,  "", inp); }
+    else if ( inp ~ /^\\t/)  { out = out "\t"; sub(/^\\t/,  "", inp); }
+    else if ( inp ~ /^\\+/)  { out = out "+";  sub(/^\\+/,  "", inp); }
+    else if ( inp ~ /^\+/)   { out = out " ";  sub(/^\+/,   "", inp); }
+    else if ( inp ~ /^\\/)   { out = out "";   sub(/^\+/,   "", inp); }
+    else { tmp = inp; sub(/[\\+].*$/, "", tmp); out = out tmp; sub(/^[^\\+]*/, "", inp); }
+  }
+  return out;
+}
+
+function rx( regex ) {
+  gsub(/[].*+?^${}()|\\[]/, "\\\\&", regex);
+  return regex;
+}
+
+BEGIN {
+  FS = "       "; ledger = 0;
+  fn = n = uid = iban = tmp = "";
+  split("", uid_n); split("", uid_fn); split("", uid_iban);
+  split("", iban_uid); split("", ibans); split("", uids);
+
+  split("", sure); split("", unsure); split("", unknown);
+  split("", unsure_rec)
+}
+
+/^BEGIN;:VCARD$/ { fn = n = id = iban = tmp = ""; }
+
+   /^UID;[^:]*:/ { uid = $0; sub(/^[^;]+;[^:]*:/, "", uid); }
+    /^FN;[^:]*:/ {  fn = $0; sub(/^[^;]+;[^:]*:/, "",  fn); }
+     /^N;[^:]*:/ {   n = $0; sub(/^[^;]+;[^:]*:/, "",   n); sub(/;.*$/, "", n); }
+/^X-IBAN;[^:]*:/ { tmp = $0; sub(/^[^;]+;[^:]*:/, "", tmp); iban = iban ? iban " " tmp : tmp; }
+
+/^END;:VCARD$/ {
+  uid_n[uid] = n; uid_fn[uid] = fn; uid_iban[uid] = iban;
+  split(iban, ibans, / /);
+  for (iban in ibans) iban_uid[iban] = iban_uid[iban] ? iban_uid[iban] " " uid : uid;
+  fn = n = uid = iban = tmp = "";
+}
+
+strftime("%Y-%m-%d", $1, "UTC") == $2 && strftime("%Y-%m-%d", $3, "UTC") == $4 { ledger = 1; }
+
+ledger && strftime("%Y-%m-%d", $2, "UTC") == $1 {
+  if ($3 in iban_uid) {
+    sure[$3] = iban_uid[$3];
+  } else {
+    for (uid in uid_fn) if ( match(UNSTRING($5), rx(uid_fn[uid])) ) {
+      if (! match(unsure[$3], rx(uid))) unsure[$3] = unsure[$3] ? unsure[$3] " " uid : uid;
+      unsure_rec[$3] = $0
+    }
+    for (uid in uid_n) if ( uid_n[uid] && match(UNSTRING($4), rx(uid_n[uid])) ) {
+      if (! match(unsure[$3], rx(uid))) unsure[$3] = unsure[$3] ? unsure[$3] " " uid : uid;
+      unsure_rec[$3] = $0
+    }
+  }
+  if (!($3 in sure) && !($3 in unsure)) unknown[$3] = $0;
+}
+
+END {
+  for (iban in sure) {
+    line = "sure       " iban;
+    split(iban_uid[iban], uids, / /);
+    for (uid in uids) line = line "    " STRING(uid "/" uid_fn[uid]);
+    print line;
+  }
+  for (iban in unsure) {
+    line = "guess      " iban " " STRING(unsure_rec[iban]);
+    split(unsure[iban], uids, / /);
+    for (k in uids) line = line "      " STRING(uids[k] "/" uid_fn[uids[k]]);
+    print line;
+  }
+  for (iban in unknown) {
+    line = "unknown    " iban " " unknown[iban];
+    print line;
+  }
+}