From f6fc0fc424a35c67d257012c9574bc60537f63b4 Mon Sep 17 00:00:00 2001
From: Dirk Engling <erdgeist@erdgeist.org>
Date: Sun, 18 Aug 2024 12:50:20 +0200
Subject: Spamming entries seem to have broken the exporter, so we're dealing
 with a lot of corrupt files. Add helpers that fix these entries up

---
 makecolumns.sh | 36 ++++++++++++++++++++++++++++++------
 1 file changed, 30 insertions(+), 6 deletions(-)

diff --git a/makecolumns.sh b/makecolumns.sh
index 1a0e271..cef88e0 100755
--- a/makecolumns.sh
+++ b/makecolumns.sh
@@ -373,6 +373,18 @@ handle_format_version_3() {
     rm ??_unknown
 }
 
+fix_broken_v4_file() {
+    name=$1
+    offset=$2
+    nullbytes=$3
+    gap=$4
+
+    head -c $(( offset - 1 )) file_${name} > file_${name}_
+    [ "${nullbytes}" -gt 0 ] && printf '\x00' >> file_${name}_
+    tail -c +$(( offset + gap )) file_${name} >> file_${name}_
+    mv file_${name}_ file_${name}
+}
+
 handle_format_version_4() {
     if [ -f "$1/branchcodes.tl" ]; then
       is_yp=true
@@ -392,12 +404,24 @@ handle_format_version_4() {
     printf "Extracting phonebook.db ... "
     extract_version_4 "$1/phonebook.db"
 
-    # Fixup broken file in 2023_Q1
-    case $1 in *2023_Q1*)
-        head -c 442 file_17F0A > file_17F0A_
-        printf '\0' >> file_17F0A_
-        tail -c +536 file_17F0A >> file_17F0A_
-        mv file_17F0A_ file_17F0A
+    case $1 in
+    *2023_Q1*)
+        # Fixup broken file in 2023_Q1
+        fix_broken_v4_file 17F0A 0x01BB 1 93
+        ;;
+    *2023_Q3*)
+        # Fixup broken file in 2023_Q3
+        fix_broken_v4_file 15DB2 0x3D05 0 373
+        fix_broken_v4_file 15DBD 0x8CA8 0 391
+        fix_broken_v4_file 15DD3 0x72B7 0 120
+        fix_broken_v4_file 15E2B 0x0549 0 2051
+        fix_broken_v4_file 15EE6 0x0569 1 144
+        fix_broken_v4_file 15F28 0x2EF8 0 103
+        fix_broken_v4_file 15F3E 0x3C1C 0 164
+        fix_broken_v4_file 16348 0x0942 0 221
+        fix_broken_v4_file 16419 0x0CF8 0 140
+        fix_broken_v4_file 16471 0x0681 0 106
+        ;;
     esac
 
     rows=`find . -name file_\* | wc -l`
-- 
cgit v1.2.3