Make sort_plz work on single files instead of creating all the columns

1 files changed, 76 insertions, 45 deletions

diff --git a/src/postprocess/sort_plz.c b/src/postprocess/sort_plz.c
index f44cec7..26ac9d0 100644
--- a/src/postprocess/sort_plz.c
+++ b/src/postprocess/sort_plz.c
@@ -4,7 +4,7 @@
     <PLZ> is either [0-9_]{5} or _____ or brken
   opens files in source directory:
     01_Flags 02_Nachname 03_Vorname 04_Zusaetze 07_Strasse 08_Hausnummer 09_Verweise 10_Postleitzahl
-    11_Ort 12_Vorwahl 13_Rufnummer 16_Koordinaten
+    11_Ort 12_Vorwahl 13_Rufnummer 14_15_Email_Webadresse 16_Koordinaten
   appends to all of the above dirs plus
     00_Jahr
 */
@@ -14,12 +14,21 @@
 #include <stdio.h>
 #include <string.h>
 #include <errno.h>
+#include <unistd.h>
 #include <err.h>
 
-enum { F_00, F_01, F_10, F_02, F_03, F_04, F_07, F_08, F_09, F_11, F_12, F_13, F_16, F_COUNT };
+enum { F_00, F_01, F_10, F_02, F_03, F_04, F_07, F_08, F_09, F_11, F_12, F_13, F_14, F_16, F_COUNT };
 
 static char *g_filenames[] = {
- "00_Jahr", "01_Flags", "10_Postleitzahl", "02_Nachname", "03_Vorname", "04_Zusaetze", "07_Strasse", "08_Hausnummer", "09_Verweise", "11_Ort", "12_Vorwahl", "13_Rufnummer", "16_Koordinaten" };
+ "00_Jahr", "01_Flags", "10_Postleitzahl", "02_Nachname", "03_Vorname", "04_Zusaetze", "07_Strasse", "08_Hausnummer", "09_Verweise", "11_Ort", "12_Vorwahl", "13_Rufnummer", "14_15_Email_Webadresse", "16_Koordinaten" };
+
+typedef struct {
+  char plz[8];
+  FILE * file;
+} outhandle;
+
+static outhandle g_outhandles[32*1024];
+static int g_outhandle_count;
 
 FILE * fopen_prefix(char *prefix, int file_id, int readonly) {
   char filename[1024];
@@ -27,10 +36,50 @@ FILE * fopen_prefix(char *prefix, int file_id, int readonly) {
   return fopen(filename, readonly ? "r" : "a");
 }
 
+/* This function gives us a binary search that returns a pointer, even if
+   no exact match is found. In that case it sets exactmatch 0 and gives
+   calling functions the chance to insert data
+*/
+void *binary_search( const void * const key, const void * base, const size_t member_count, const size_t member_size,
+                     size_t compare_size, int *exactmatch ) {
+  size_t interval = member_count;
+
+  while( interval ) {
+    uint8_t *lookat = ((uint8_t*)base) + member_size * ( interval / 2 );
+    int cmp = memcmp( lookat, key, compare_size );
+    if(cmp == 0 ) {
+      base = lookat;
+      break;
+    }
+    if(cmp < 0) {
+      base = lookat + member_size;
+      interval--;
+    }
+    interval /= 2;
+  }
+
+  *exactmatch = interval;
+  return (void*)base;
+}
+
+FILE * get_file_for_postleitzahl(char *plz) {
+  int exactmatch = 0;
+  outhandle * oh = (outhandle *)binary_search(plz, g_outhandles, g_outhandle_count, sizeof(outhandle), 5, &exactmatch);
+  if (!exactmatch) {
+    size_t s = (g_outhandles + g_outhandle_count) - oh;
+    memmove(oh + 1, oh, s * sizeof(outhandle));
+    oh->file = fopen(plz, "a");
+    if (!oh->file) errx( 1, "Couldn't open file %s for writing\n", plz);
+    memcpy(oh->plz, plz, 5);
+    g_outhandle_count++;
+  }
+  return oh->file;
+}
+
 int main(int argc, char **args) {
   FILE * in_handles[F_COUNT] = { NULL };
-  FILE * out_handles[F_COUNT] = { NULL };
-  char flags[4];
+  FILE * out_handle = NULL;
+  char flags[4], outfile[6];
   int i, in_multi = 0;
   char *input = malloc(1024);
   size_t input_size = 1024;
@@ -42,33 +91,26 @@ int main(int argc, char **args) {
       errx( 1, "Couldn't open file %s\n", g_filenames[i]);
   }
 
-  mkdir( "multi", 0755);
-  mkdir( "single", 0755);
+  mkdir("output", 0755);
+  chdir("output");
 
   /* Get Flags to check if we're processing a continuation */
   while (fgets(flags, 4, in_handles[F_01])) {
-    char out_dir[32];
     ssize_t linelen;
     char flag = strtoul(flags, 0, 16);
-    char *type = flag & 1 ? "multi/" : "single/";
 
     /* If we're in multiline mode, we just copy lines as long as we see continuations */
     if (in_multi) {
       if (flag & 0x2) {
-        fputs(args[1], out_handles[F_00]);      // write Jahr
-        fputc(10, out_handles[F_00]);
-        fwrite(flags, 3, 1, out_handles[F_01]); // copy Flags verbatim
+        fputs(args[1], out_handle);      // write Jahr
+        fputc(10, out_handle);
+        fwrite(flags, 3, 1, out_handle); // copy Flags verbatim
         for (i=F_10; i<F_COUNT; ++i) {          // process the rest of entries
           ssize_t linelen = getline(&input, &input_size, in_handles[i]);
-          fwrite(input, linelen, 1, out_handles[i]);
+          fwrite(input, linelen, 1, out_handle);
         }
         continue;
       }
-      /* If the entry is not a continuation, close all output files and switch off multi mode */
-      for (i=0; i<F_COUNT; ++i) {
-        fclose(out_handles[i]);
-        out_handles[i] = NULL;
-      }
       in_multi = 0;
     }
 
@@ -83,50 +125,39 @@ int main(int argc, char **args) {
     }
 
     if (linelen == 0) // empty PLZ
-      strcpy(out_dir, in_multi ? "multi/_____" : "single/_____");
+      strcpy(outfile, "_____");
     else if (linelen == 5) { // potentially normal
       int broken = 0;
-      char * dest = out_dir + sprintf(out_dir, in_multi ? "multi/" : "single/");
       for (i=0; i<5; ++i) {
         if ( (input[i] < '0' || input[i] > '9') && input[i] != '.') {
           broken = 1;
           break;
         }
-        dest[i] = input[i];
-        if (dest[i] == '.') dest[i] = '_';
+        outfile[i] = input[i];
+        if (outfile[i] == '.') outfile[i] = '_';
       }
-      dest[5] = 0;
+      outfile[5] = 0;
       if (broken)
-        strcpy(out_dir, in_multi ? "multi/broken" : "single/broken");
+        strcpy(outfile, "brken");
     } else
-      strcpy(out_dir, in_multi ? "multi/broken" : "single/broken");
+      strcpy(outfile, "brken");
 
-    if (mkdir(out_dir, 0755) == -1 && errno != EEXIST)
-      errx( 1, "Couldn't create directory %s %d\n", out_dir, errno);
+    out_handle = get_file_for_postleitzahl(outfile);
 
-    for (i=F_00; i<F_COUNT; ++i) {
-      out_handles[i] = fopen_prefix(out_dir, i, 0);
-      if (!out_handles[i])
-        errx( 1, "Couldn't open file %s\n", g_filenames[i]);
-    }
-
-    fputs(args[1], out_handles[F_00]);            // write Jahr
-    fputc(10, out_handles[F_00]);
-    fwrite(flags, 3, 1, out_handles[F_01]);       // copy Flags verbatim
-    fputs(input, out_handles[F_10]); // copy Postleitzahl verbatim
-    fputc(10, out_handles[F_10]);
+    fputs(args[1], out_handle);            // write Jahr
+    fputc(10, out_handle);
+    fwrite(flags, 3, 1, out_handle);       // copy Flags verbatim
+    fputs(input, out_handle); // copy Postleitzahl verbatim
+    fputc(10, out_handle);
 
     for (i=F_02; i<F_COUNT; ++i) {                // process the rest of entries
       ssize_t linelen = getline(&input, &input_size, in_handles[i]);
-      fwrite(input, linelen, 1, out_handles[i]);
+      fwrite(input, linelen, 1, out_handle);
     }
-
-    if (!in_multi)
-      for (i=0; i<F_COUNT; ++i) {
-        fclose(out_handles[i]);
-        out_handles[i] = NULL;
-      }
   }
 
+  for (i=0; i<g_outhandle_count; ++i)
+    fclose(g_outhandles[i].file);
+
   return 0;
 }