From: Francois Fleuret <francois@fleuret.org>
Date: Thu, 31 Jan 2013 21:47:32 +0000 (+0100)
Subject: Starting to convert the ID from strings to integers to speed things up.
X-Git-Url: https://fleuret.org/cgi-bin/gitweb/gitweb.cgi?a=commitdiff_plain;h=e40b536f28d3d6af2e49a48dc830f15a2fd88ab0;p=mymail.git

Starting to convert the ID from strings to integers to speed things up.
---

diff --git a/mymail.c b/mymail.c
index 313b8d5..dcb44bf 100644
--- a/mymail.c
+++ b/mymail.c
@@ -51,8 +51,25 @@
 
 #define BUFFER_SIZE 65536
 
+enum {
+  ID_MAIL,
+  ID_FROM,
+  ID_DEST,
+  ID_SUBJECT,
+  ID_FROMDEST,
+  MAX_ID
+};
+
+static char *field_names[] = {
+  "mail",
+  "from",
+  "dest",
+  "subj",
+  "fromdest"
+};
+
 struct parsable_field {
-  char *name;
+  int id;
   char *regexp_string;
   regex_t regexp;
 };
@@ -124,26 +141,32 @@ int ignore_entry(const char *name) {
     (name[0] == '.' && name[1] != '/');
 }
 
+int mbox_line_match_search(int search_id, regex_t *search_regexp,
+                           int mbox_id, char *mbox_value) {
+  return search_id == mbox_id && regexec(search_regexp, mbox_value, 0, 0, 0) == 0;
+}
+
 void search_in_db(int nb_search_patterns,
-                  char **search_name, char **search_regexp_string,
+                  int *search_ids, char **search_regexp_strings,
                   FILE *db_file) {
   int hits[MAX_NB_SEARCH_PATTERNS];
   char raw_db_line[BUFFER_SIZE];
   char raw_mbox_line[BUFFER_SIZE];
   char current_mail_filename[PATH_MAX + 1];
   unsigned long int current_position_in_mail;
-  char *name, *value;
-  regex_t regexp[MAX_NB_SEARCH_PATTERNS];
-  int already_written, n;
+  char *mbox_name, *mbox_value;
+  int mbox_id;
+  regex_t search_regexps[MAX_NB_SEARCH_PATTERNS];
+  int already_written, m, n;
 
   for(n = 0; n < nb_search_patterns; n++) {
-    if(regcomp(&regexp[n],
-               search_regexp_string[n],
+    if(regcomp(&search_regexps[n],
+               search_regexp_strings[n],
                REG_ICASE)) {
       fprintf(stderr,
               "mymail: Syntax error in regexp \"%s\" for field \"%s\".\n",
-              search_regexp_string[n],
-              search_name[n]);
+              search_regexp_strings[n],
+              field_names[search_ids[n]]);
       exit(EXIT_FAILURE);
     }
   }
@@ -154,10 +177,10 @@ void search_in_db(int nb_search_patterns,
   for(n = 0; n < nb_search_patterns; n++) { hits[n] = 0; }
 
   while(fgets(raw_db_line, BUFFER_SIZE, db_file)) {
-    name = raw_db_line;
-    value = segment_next_field(raw_db_line);
+    mbox_name = raw_db_line;
+    mbox_value = segment_next_field(raw_db_line);
 
-    if(strcmp("mail", name) == 0) {
+    if(strcmp("mail", mbox_name) == 0) {
       char *position_in_file_string;
       char *mail_filename;
 
@@ -165,12 +188,16 @@ void search_in_db(int nb_search_patterns,
 
       if(n == nb_search_patterns) {
         FILE *mail_file;
+
         mail_file = fopen(current_mail_filename, "r");
+
         if(!mail_file) {
           fprintf(stderr, "mymail: Cannot open mbox '%s'.\n", current_mail_filename);
           exit(EXIT_FAILURE);
         }
+
         fseek(mail_file, current_position_in_mail, SEEK_SET);
+
         if(fgets(raw_mbox_line, BUFFER_SIZE, mail_file)) {
           printf("%s", raw_mbox_line);
           while(fgets(raw_mbox_line, BUFFER_SIZE, mail_file) &&
@@ -178,13 +205,14 @@ void search_in_db(int nb_search_patterns,
             printf("%s", raw_mbox_line);
           }
         }
+
         fclose(mail_file);
       }
 
       for(n = 0; n < nb_search_patterns; n++) { hits[n] = 0; }
 
-      position_in_file_string = value;
-      mail_filename = segment_next_field(value);
+      position_in_file_string = mbox_value;
+      mail_filename = segment_next_field(mbox_value);
       current_position_in_mail = atol(position_in_file_string);
       strcpy(current_mail_filename, mail_filename);
 
@@ -193,22 +221,27 @@ void search_in_db(int nb_search_patterns,
     }
 
     else {
+      mbox_id = -1;
+      for(m = 0; (m < MAX_ID) && mbox_id == -1; m++) {
+        if(strncmp(field_names[m], mbox_name, strlen(mbox_name)) == 0) {
+          mbox_id = m;
+        }
+      }
       for(n = 0; n < nb_search_patterns; n++) {
-        hits[n] |=
-          (strncmp(search_name[n], name, strlen(search_name[n])) == 0 &&
-           regexec(&regexp[n], value, 0, 0, 0) == 0);
+        hits[n] |= mbox_line_match_search(search_ids[n], &search_regexps[n],
+                                          mbox_id, mbox_value);
       }
     }
   }
 
   for(n = 0; n < nb_search_patterns; n++) {
-    regfree(&regexp[n]);
+    regfree(&search_regexps[n]);
   }
 }
 
 void recursive_search_in_db(const char *entry_name,
                             int nb_search_patterns,
-                            char **search_name, char **search_regexp_string) {
+                            int *search_ids, char **search_regexp_strings) {
   DIR *dir;
   struct dirent *dir_e;
   struct stat sb;
@@ -231,7 +264,7 @@ void recursive_search_in_db(const char *entry_name,
         snprintf(subname, PATH_MAX, "%s/%s", entry_name, dir_e->d_name);
         recursive_search_in_db(subname,
                                nb_search_patterns,
-                               search_name, search_regexp_string);
+                               search_ids, search_regexp_strings);
       }
     }
     closedir(dir);
@@ -264,7 +297,7 @@ void recursive_search_in_db(const char *entry_name,
         exit(EXIT_FAILURE);
       }
 
-      search_in_db(nb_search_patterns, search_name, search_regexp_string,
+      search_in_db(nb_search_patterns, search_ids, search_regexp_strings,
                    db_file);
 
       fclose(db_file);
@@ -281,7 +314,7 @@ void index_one_mbox_line(int nb_fields_to_parse, struct parsable_field *fields_t
   for(f = 0; f < nb_fields_to_parse; f++) {
     if(regexec(&fields_to_parse[f].regexp, raw_mbox_line, 1, &matches, 0) == 0) {
       fprintf(db_file, "%s %s\n",
-              fields_to_parse[f].name,
+              field_names[fields_to_parse[f].id],
               raw_mbox_line + matches.rm_eo);
     }
   }
@@ -344,13 +377,15 @@ void index_mbox(const char *mbox_filename,
       }
 
       else {
-        /* if(!((raw_mbox_line[0] >= 'a' && raw_mbox_line[0] <= 'z') || */
-             /* (raw_mbox_line[0] >= 'A' && raw_mbox_line[0] <= 'Z'))) { */
-          /* fprintf(stderr, */
-                  /* "Header line syntax error %s:%lu.\n", */
-                  /* mbox_filename, position_in_file); */
-          /* fprintf(stderr, "%s", raw_mbox_line); */
-        /* } */
+        /*
+          if(!((raw_mbox_line[0] >= 'a' && raw_mbox_line[0] <= 'z') ||
+          (raw_mbox_line[0] >= 'A' && raw_mbox_line[0] <= 'Z'))) {
+          fprintf(stderr,
+          "Header line syntax error %s:%lu.\n",
+          mbox_filename, position_in_file);
+          fprintf(stderr, "%s", raw_mbox_line);
+          }
+        */
 
         if(full_line[0]) {
           index_one_mbox_line(nb_fields_to_parse, fields_to_parse, full_line, db_file);
@@ -423,19 +458,19 @@ static struct option long_options[] = {
 
 static struct parsable_field fields_to_parse[] = {
   {
-    "from",
+    ID_FROM,
     "^\\([Ff][Rr][Oo][Mm]:\\|From\\) *",
     { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }
   },
 
   {
-    "dest",
+    ID_DEST,
     "^\\([Tt][Oo]\\|[Cc][Cc]\\|[Bb][Cc][Cc]\\): *",
     { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }
   },
 
   {
-    "subj",
+    ID_SUBJECT,
     "^[Ss][Uu][Bb][Jj][Ee][Cc][Tt]: *",
     { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }
   },
@@ -559,7 +594,7 @@ int main(int argc, char **argv) {
         fprintf(stderr,
                 "mymail: Syntax error in regexp \"%s\" for field \"%s\".\n",
                 fields_to_parse[f].regexp_string,
-                fields_to_parse[f].name);
+                field_names[fields_to_parse[f].id]);
         exit(EXIT_FAILURE);
       }
     }
@@ -583,23 +618,28 @@ int main(int argc, char **argv) {
   else {
 
     if(nb_search_patterns > 0) {
-      char *search_name[MAX_NB_SEARCH_PATTERNS];
-      char *search_regexp_string[MAX_NB_SEARCH_PATTERNS];
-      int n;
+      int search_ids[MAX_NB_SEARCH_PATTERNS];
+      char *search_regexp_strings[MAX_NB_SEARCH_PATTERNS];
+      int m, n;
 
       for(n = 0; n < nb_search_patterns; n++) {
-        search_name[n] = search_pattern[n];
-        search_regexp_string[n] = segment_next_field(search_pattern[n]);
+        search_regexp_strings[n] = segment_next_field(search_pattern[n]);
+        search_ids[n] = -1;
+        for(m = 0; (m < MAX_ID) && search_ids[n] == -1; m++) {
+          if(strncmp(field_names[m], search_pattern[n], strlen(search_pattern[n])) == 0) {
+            search_ids[n] = m;
+          }
+        }
       }
 
-      if(!*search_regexp_string) {
+      if(!*search_regexp_strings) {
         fprintf(stderr,
                 "Syntax error in the search pattern.\n");
         exit(EXIT_FAILURE);
       }
 
       recursive_search_in_db(db_root_path,
-                             nb_search_patterns, search_name, search_regexp_string);
+                             nb_search_patterns, search_ids, search_regexp_strings);
 
       for(n = 0; n < nb_search_patterns; n++) {
         free(search_pattern[n]);