From: Francois Fleuret <francois@fleuret.org>
Date: Mon, 16 Mar 2009 17:57:58 +0000 (+0100)
Subject: Changed the -r option to removing of all duplicates, wherever they
X-Git-Url: https://www.fleuret.org/cgi-bin/gitweb/gitweb.cgi?a=commitdiff_plain;h=c597e11ab02854d2552c8966d376523902ad6d25;p=selector.git

Changed the -r option to removing of all duplicates, wherever they
are. Needed a hash-table to do it kind of properly. Seems to work so
far. Wonder if the hash_table_size being ten times the maximum number
of lines makes sense.
---

diff --git a/selector.cc b/selector.cc
index af8ef84..fbcd67a 100644
--- a/selector.cc
+++ b/selector.cc
@@ -65,13 +65,13 @@ void inject_into_tty_buffer(char *line) {
   tcgetattr(STDIN_FILENO,&oldtio);
   memset(&newtio, 0, sizeof(newtio));
   // Set input mode (non-canonical, *no echo*,...)
-  tcsetattr(STDIN_FILENO,TCSANOW, &newtio);
+  tcsetattr(STDIN_FILENO, TCSANOW, &newtio);
   // Put the selected line in the tty input buffer
   for(char *k = line; *k; k++) {
     ioctl(STDIN_FILENO, TIOCSTI, k);
   }
   // Restore the old settings
-  tcsetattr(STDIN_FILENO,TCSANOW, &oldtio);
+  tcsetattr(STDIN_FILENO, TCSANOW, &oldtio);
 }
 
 //////////////////////////////////////////////////////////////////////
@@ -306,6 +306,38 @@ void update_screen(int *current_line, int *temporary_line, int motion,
   refresh();
 }
 
+//////////////////////////////////////////////////////////////////////
+// A quick and dirty hash table
+
+int *new_hash_table(int hash_table_size) {
+  int *result;
+  result = new int[hash_table_size];
+  for(int k = 0; k < hash_table_size; k++) {
+    result[k] = -1;
+  }
+  return result;
+}
+
+int test_and_add(char *new_string, int new_index,
+                 char **strings, int *hash_table, int hash_table_size) {
+  unsigned int code = 0;
+
+  for(int k = 0; new_string[k]; k++) {
+    code += int(new_string[k]) << (8 * k%4);
+  }
+
+  code = code % hash_table_size;
+
+  while(hash_table[code] >= 0) {
+    if(strcmp(new_string, strings[hash_table[code]]) == 0) return 1;
+    code = (code + 1) % hash_table_size;
+  }
+
+  hash_table[code] = new_index;
+
+  return 0;
+}
+
 //////////////////////////////////////////////////////////////////////
 
 int main(int argc, char **argv) {
@@ -433,10 +465,18 @@ int main(int argc, char **argv) {
     return 1;
   }
 
+  int hash_table_size = nb_lines_max * 10;
+  int *hash_table = 0;
+
+  if(remove_duplicates) {
+    hash_table = new_hash_table(hash_table_size);
+  }
+
   while(nb_lines < nb_lines_max && !file.eof()) {
     file.getline(buffer, buffer_size);
     if(strcmp(buffer, "") != 0) {
       char *s = buffer;
+
       if(zsh_history && *s == ':') {
         while(*s && *s != ';') s++;
         if(*s == ';') s++;
@@ -446,7 +486,7 @@ int main(int argc, char **argv) {
         while(*s == ' ' || (*s >= '0' && *s <= '9')) s++;
       }
 
-      if(!remove_duplicates || nb_lines == 0 || strcmp(lines[nb_lines - 1], s)) {
+      if(!hash_table || !test_and_add(s, nb_lines, lines, hash_table, hash_table_size)) {
         lines[nb_lines] = new char[strlen(s) + 1];
         strcpy(lines[nb_lines], s);
         nb_lines++;
@@ -454,6 +494,8 @@ int main(int argc, char **argv) {
     }
   }
 
+  delete[] hash_table;
+
   if(inverse_order) {
     for(int i = 0; i < nb_lines/2; i++) {
       char *s = lines[nb_lines - 1 - i];