From fb3616e06ca20ffe78dfb00b962a2599a46b2a5f Mon Sep 17 00:00:00 2001
From: Dirk Engling <erdgeist@erdgeist.org>
Date: Sat, 29 Jun 2019 18:51:00 +0200
Subject: Clean up 1992 post processing

---
 src/postprocess/map_plz.c           |  4 ++--
 src/postprocess/merge_entries.c     |  5 +++--
 src/postprocess/postprocess-1992.sh | 15 ++++++---------
 3 files changed, 11 insertions(+), 13 deletions(-)

(limited to 'src/postprocess')

diff --git a/src/postprocess/map_plz.c b/src/postprocess/map_plz.c
index ab0db71..9dec6bb 100644
--- a/src/postprocess/map_plz.c
+++ b/src/postprocess/map_plz.c
@@ -18,7 +18,7 @@ int main(int argc, char **args) {
   FILE *bfile, *streetfile_out;
   char *ptr, *input = malloc(65335);
   char *ort = malloc(65335), vorwahl_block[16];
-  int i, brutes_count = 0, report = 0;
+  unsigned int i, brutes_count = 0, report = 0;
   brute_t *brutes = malloc(200000*sizeof(brute_t));
 
   /* prepare io */
@@ -63,7 +63,7 @@ int main(int argc, char **args) {
   g_book_by_name = (entry_t*)malloc(g_book_size * sizeof(entry_t));
 
   /* Split pointers into input files into our arrays */
-  for (i = 0, ptr = (char*)tbuch->addr; i < g_book_size; ++i) {
+  for (i=0, ptr=(char*)tbuch->addr; i<g_book_size; ++i) {
     g_book[i].vorwahl = ptr; ptr += strlen(ptr) + 1;
     g_book[i].ort     = ptr; ptr += strlen(ptr) + 1;
     g_book[i].strasse = ptr; ptr += strlen(ptr) + 1;
diff --git a/src/postprocess/merge_entries.c b/src/postprocess/merge_entries.c
index f9ee67d..1dd7d50 100644
--- a/src/postprocess/merge_entries.c
+++ b/src/postprocess/merge_entries.c
@@ -148,9 +148,10 @@ static int sort_me(const void *f_a, const void *f_b) {
     outvec_t *oa_row = oa + row * COLUMNS;
     outvec_t *ob_row = ob + row * COLUMNS;
 
+    if ((res = STRCMP_n(oa_row[ 2].ptr, ob_row[ 2].ptr))) return res; /* PLZ */
+    if ((res = STRCMP_n(oa_row[ 9].ptr, ob_row[ 9].ptr))) return res; /* Ort */
     if ((res = STRCMP_n(oa_row[10].ptr, ob_row[10].ptr))) return res; /* Vorwahl */
     if ((res = STRCMP_n(oa_row[11].ptr, ob_row[11].ptr))) return res; /* Rufnummer */
-    if ((res = STRCMP_n(oa_row[ 2].ptr, ob_row[ 2].ptr))) return res; /* PLZ */
     if ((res = STRCMP_n(oa_row[ 6].ptr, ob_row[ 6].ptr))) return res; /* Strasse */
     if ((res = STRCMP_n(oa_row[ 7].ptr, ob_row[ 7].ptr))) return res; /* Hausnummer */
     if ((res = STRCMP_n(oa_row[ 3].ptr, ob_row[ 3].ptr))) return res; /* Nachname */
@@ -200,7 +201,7 @@ int main(int argc, char **args) {
   unsigned long current = 0, i, flag;
   uint64_t year_list = 0, revflag_list = 0, bizflag_list = 0;
 
-  if (argc != 1) exit(1);
+  if (argc != 2) exit(1);
   tbuch = map_file(args[1], 1);
 
   /* Estimate upper bound for amount of lines */
diff --git a/src/postprocess/postprocess-1992.sh b/src/postprocess/postprocess-1992.sh
index 1e685d2..6720991 100644
--- a/src/postprocess/postprocess-1992.sh
+++ b/src/postprocess/postprocess-1992.sh
@@ -4,19 +4,16 @@ paste 1992_Q2/{01_Flags,12_Vorwahl,12_Vorwahl_block,11_Ort,07_Strasse,08_Hausnum
 # Generate lookup file from 1995
 paste 1995_Q0/{12_Vorwahl,11_Ort,07_Strasse,08_Hausnummer,02_Nachname,03_Vorname,13_Rufnummer,10_Postleitzahl} | tr '\n\t' '\0' > 1995-voshnvrp.bin
 
-# To debug in lldb
-process launch -i 1992_testfile.txt -- 1995-vorwahl-ort-strasse-hnr-name-vorname-rufnummer-plz.bin
-
 # Compile plz mapper
 cc -O3 -o map_plz map_plz.c -I ../src/export/ ../src/export/mystdlib.c
 
+# To debug in lldb
+# cc -O0 -g -o map_plz map_plz.c -I ../src/export/ ../src/export/mystdlib.c
+# process launch -i 1992_testfile.txt -- 1995-vorwahl-ort-strasse-hnr-name-vorname-rufnummer-plz.bin
+
 # outputs mapped plz, generates brutemap.txt
 touch brutemap_input.bin zip_simple_map.bin
-./map_plz 1995-voshnvrp.bin zip_simple_map.bin brutemap_input.bin < 1992-fvvoshnvrpp.txt > 10_Postleitzahl
-
-# generate street name translation table from brutemap,
-# only taking into account similar street names
-# cut -f 3,4 brutemap.txt | tr '[:upper:]' '[:lower:]' | paste brutemap.txt - | cut -f 1-4,6,7 | ./jaro | cut -f 1-5 > brutemap_filtered.txt
+./map_plz 1995-voshnvrp.bin zip_simple_map.bin brutemap_input.bin < 1992-fvvoshnvrpp.txt > 1992_Q2/10_Postleitzahl
 
 # generate street name translation table from brutemap,
 # only taking into account similar street names, new style
@@ -29,4 +26,4 @@ cut -f 1-5 brutemap_simifiltered.txt | sort | uniq -c | sed -E $'s:^ *([[:digit:
 sort -u zip_mapfile.txt | tr '\n' '\0' > zip_simple_map.bin
 
 # Redo the mapping with the data from brutemap and zipmap
-./map_plz 1995-voshnvrp.bin zip_simple_map.bin brutemap_input.bin < 1992-fvvoshnvrpp.txt > 10_Postleitzahl
+./map_plz 1995-voshnvrp.bin zip_simple_map.bin brutemap_input.bin < 1992-fvvoshnvrpp.txt > 1992_Q2/10_Postleitzahl
-- 
cgit v1.2.3