Add branch name mapper code for v3

4 files changed, 115 insertions, 17 deletions

diff --git a/Makefile b/Makefile
index 089ae06..691675e 100644
--- a/Makefile
+++ b/Makefile
@@ -1,4 +1,4 @@
-BINARIES=bin/extract_version_1 bin/extract_version_2 bin/extract_version_3 bin/extract_version_4 bin/split_version_2 bin/split_version_3 bin/map_coords bin/map_branches bin/convert_coords bin/join
+BINARIES=bin/extract_version_1 bin/extract_version_2 bin/extract_version_3 bin/extract_version_4 bin/split_version_2 bin/split_version_3 bin/map_coords bin/map_branches_v3 bin/map_branches_v4 bin/convert_coords bin/join
 CFLAGS += -W -Wall -Wextra -O3 # -Weverything -Wno-cast-align -Wno-padded
 
 all: $(BINARIES)
@@ -24,8 +24,11 @@ bin/split_version_2: src/export/split_version_2.c src/export/mystdlib.c
 bin/map_coords: src/export/map_coords.c src/export/mystdlib.c
         $(CC) $(CFLAGS) -o $@ src/export/map_coords.c src/export/mystdlib.c
 
-bin/map_branches: src/export/map_branches.c
-        $(CC) $(CFLAGS) -o $@ src/export/map_branches.c
+bin/map_branches_v4: src/export/map_branches_v4.c
+        $(CC) $(CFLAGS) -o $@ src/export/map_branches_v4.c
+
+bin/map_branches_v3: src/export/map_branches_v3.c
+        $(CC) $(CFLAGS) -o $@ src/export/map_branches_v3.c
 
 bin/convert_coords: src/export/convert_coords.c
         $(CC) $(CFLAGS) -o $@ src/export/convert_coords.c -lm
diff --git a/makecolumns.sh b/makecolumns.sh
index edd965c..4f4bebc 100755
--- a/makecolumns.sh
+++ b/makecolumns.sh
@@ -171,9 +171,21 @@ handle_format_version_2() {
 }
 
 handle_format_version_3() {
-    echo "Working on $1. Detected pre-2004 Telefonbuch version."
+    # glob
+    teiln=`printf "%s" "$1"/[Dd][Aa][Tt]/[Tt][Ee][Ii][Ll][Nn].[Dd][Aa][Tt]`
+    braid=`printf "%s" "$1"/[Dd][Aa][Tt]/[Bb][Rr][Aa][Ii][Dd].[Dd][Aa][Tt]`
+    streets=`printf "%s" "$1"/[Dd][Aa][Tt]/[Ss][Tt][Rr][Aa][Ss][Ss][Ee][Nn].[Dd][Aa][Tt]`
+    karto=`printf "%s" "$1"/[Dd][Aa][Tt]/[Kk][Aa][Rr][Tt][Oo].[Dd][Aa][Tt]`
+
+    if [ -f "${braid}" ]; then
+      echo "Working on $1. Detected pre-2004 Yellow Pages version."
+      is_yp=true
+    else
+      echo "Working on $1. Detected pre-2004 Telefonbuch version."
+      unset is_yp
+    fi
     # Extract teiln.dat
-    do_decompress_version_3 "$1"/[Dd][Aa][Tt]/[Tt][Ee][Ii][Ll][Nn].[Dd][Aa][Tt] "teiln.dat"
+    do_decompress_version_3 "${teiln}" "teiln.dat"
 
     # See how long each filename is
     export filename_len=$(( `ls | head -n 1 | wc -c` - 1 ))
@@ -205,8 +217,14 @@ handle_format_version_3() {
 #        set -- `hexdump -n 8 -v -e '" " 1/4 "%u"' ${file}`
 #        tail -c +$(( $2 + 1 )) ${file}
 #    done | tr '\n\0' '\t\n' > 01_02_Flags_Nachname
-    cut -c 1  < 01_unknown > 01_Flags
-    cut -c 2- < 01_unknown > 02_Nachname
+    if [ "${is_yp}" ]; then
+        cut -c 1  < 01_unknown > 01_Flags
+        cut -c 2-7 < 01_unknown > 09_Branchenindex
+        cut -c 8- < 01_unknown > 02_Nachname
+    else
+        cut -c 1  < 01_unknown > 01_Flags
+        cut -c 2- < 01_unknown > 02_Nachname
+    fi
     rm 01_unknown
     printf "done.\n"
 
@@ -259,7 +277,6 @@ handle_format_version_3() {
 
     # If street names come in an extra file, extract
     # street names first
-    streets="$1"/[Dd][Aa][Tt]/[Ss][Tt][Rr][Aa][Ss][Ss][Ee][Nn].[Dd][Aa][Tt]
     [ -f "${streets}" ] && do_processfile_version_3 "${streets}" "street name" 99_Strassenname convert_zeros
 
     # extract street names if 07_unknown contains street indexes
@@ -280,7 +297,6 @@ handle_format_version_3() {
         tidy_streetnames 07_Strasse
     fi
 
-    karto="$1"/[Dd][Aa][Tt]/[Kk][Aa][Rr][Tt][Oo].[Dd][Aa][Tt]
     if [ -f "${karto}" ]; then
       do_processfile_version_3 "${karto}" "geo coordinates" 90_Geokoordinaten_hnr_raw
 
@@ -290,6 +306,14 @@ handle_format_version_3() {
       paste 10_Postleitzahl 11_Ort 07_Strasse 08_Hausnummer | map_coords 90_Geokoordinaten_hnr | convert_coords > 16_Koordinaten
       printf "done.\n"
     fi
+
+    if [ -f "${braid}" ]; then
+      do_processfile_version_3 "${braid}" "branchen name index" 97_Branchenname convert_zeros
+
+      printf "Looking up branch names from codes ... "
+      map_branches_v3 97_Branchenname < 09_Branchenindex > 09_Branchen
+      printf "done.\n"
+    fi
 }
 
 handle_format_version_4() {
@@ -387,17 +411,9 @@ handle_format_version_4() {
       rm file_*
       printf "done.\n"
 
-      printf "Generating branch name index ... "
-      mkdir branchcodes/
-      while read index name; do
-        printf $name > branchcodes/${index}
-      done < 97_Branchenname
-      printf "done.\n"
-
       printf "Looking up branch names from codes ... "
       map_branches 97_Branchenname < 09_Verweise > 09_Branchen
       printf "done.\n"
-      rm -r branchcodes
     fi
 }
 
diff --git a/src/export/map_branches_v3.c b/src/export/map_branches_v3.c
new file mode 100644
index 0000000..22d0036
--- /dev/null
+++ b/src/export/map_branches_v3.c
@@ -0,0 +1,79 @@
+#define _WITH_GETLINE
+#define _GNU_SOURCE
+#include <stdlib.h>
+#include <stdint.h>
+#include <stdio.h>
+#include <string.h>
+#include <ctype.h>
+
+typedef struct {
+  long code;
+  char *name;
+} branchen_code;
+
+enum { MAX_CODES = 128 * 1024 };
+branchen_code g_codes[MAX_CODES];
+long g_code_count;
+
+static int find_code( const void *key, const void *bc)
+{
+  return (long)key - ((branchen_code*)bc)->code;
+}
+
+static int qsort_cmp( const void *a, const void *b )
+{
+  return ((branchen_code*)a)->code - ((branchen_code*)b)->code;
+}
+
+int main( int argc, char ** args )
+{
+  FILE * map_file;
+  char *end_p, *input = malloc(1024);
+  size_t input_length = 1024;
+  ssize_t ll;
+
+  if( argc != 2 ) { fprintf( stderr, "Syntax: %s <branchcodes> < <branches_files>\n", args[0] ); exit(111); }
+
+  map_file = fopen( args[1], "r" );
+  if (!map_file || !input) { fprintf( stderr, "Error allocating resources\n" ); exit( 111 ); }
+
+  /* Fill array with maps */
+  while ( (ll = getline( &input, &input_length, map_file ) ) >= 0 ) {
+    char * r = strchr(input, 10);
+    if (r) *r = 0;
+    g_codes[g_code_count].code = strtoul(input, &end_p, 10);
+
+    if (input == end_p) break;
+    if (*end_p != ';') { fprintf( stderr, "Input error, line: %s\n", input); exit(1); }
+
+    r = strchr(end_p + 1, ';');
+    if (!r) { fprintf( stderr, "Input error, line: %s\n", input); exit(1); }
+    *r = 0;
+
+    asprintf(&g_codes[g_code_count].name, "%s", end_p + 1) ;
+    // printf( "%ld: %s\n", g_codes[g_code_count].code, g_codes[g_code_count].name);
+    g_code_count++;
+  }
+
+  qsort(g_codes, g_code_count, sizeof(branchen_code), qsort_cmp );
+
+  /* Now scan lines from 09_Verweise for semicolon separated branchen codes */
+  while ( (ll = getline( &input, &input_length, stdin ) ) >= 0 ) {
+    char *codes = input;
+    branchen_code *bc;
+    int multiple;
+    for (multiple = 0;; ++multiple) {
+      long code = strtoul(codes, &end_p, 10);
+      if (codes == end_p) break;
+      bc = (branchen_code*)bsearch((void *)(uintptr_t)code, g_codes, g_code_count, sizeof(branchen_code), find_code);
+      if (bc) {
+        if (multiple) putchar(';');
+        printf("%s", bc->name);
+      }
+      if (*end_p != ';') break;
+      codes = end_p + 1;
+    }
+    putchar(10);
+  }
+  return 0;
+}
diff --git a/src/export/map_branches.c b/src/export/map_branches_v4.c
index 160945d..160945d 100644
--- a/src/export/map_branches.c
+++ b/src/export/map_branches_v4.c