Update.

[mymail.git] / mymail.c

diff --git a/mymail.c b/mymail.c
index 877215c..5b821bc 100644 (file)
--- a/mymail.c
+++ b/mymail.c
@@ -19,6 +19,18 @@
  *
  */
 
+/*
+
+   This command is a dumb mail indexer. It can either (1) scan
+   directories containing mbox files, and create a db file containing
+   for each mail a list of fields computed from the header, or (2)
+   read such a db file and get all the mails matching regexp-defined
+   conditions on the fields.
+
+   It is low-tech, simple, light and fast.
+
+*/
+
 #define _GNU_SOURCE
 
 #include <stdio.h>
@@ -45,8 +57,21 @@ struct parsable_field {
 char *db_filename;
 char *search_pattern;
 
+int paranoid;
 int action_index;
 
+char *segment_next_field(char *current) {
+  while(*current && *current != ' ') current++;
+  *current = '\0'; current++;
+  while(*current && *current == ' ') current++;
+  return current;
+}
+
+void remove_eof(char *c) {
+  while(*c && *c != '\n' && *c != '\r') c++;
+  *c = '\0';
+}
+
 /********************************************************************/
 
 /* malloc with error checking.  */
@@ -76,9 +101,11 @@ void usage(FILE *out) {
 void search_in_db(const char *search_name, const char *search_regexp_string,
                   FILE *db_file) {
   char raw_line[BUFFER_SIZE];
-  char current_mail[BUFFER_SIZE];
+  char current_mail_filename[BUFFER_SIZE];
+  unsigned long int current_position_in_mail;
   char *name, *value;
   regex_t regexp;
+  int already_written;
 
   if(regcomp(&regexp,
              search_regexp_string,
@@ -90,26 +117,46 @@ void search_in_db(const char *search_name, const char *search_regexp_string,
     exit(EXIT_FAILURE);
   }
 
+  current_position_in_mail = 0;
+  already_written = 0;
+
   while(fgets(raw_line, BUFFER_SIZE, db_file)) {
     name = raw_line;
-    value = raw_line;
-    while(*value && *value != ' ') value++;
-    *value = '\0'; value++;
-    while(*value && *value == ' ') value++;
-
-    /* printf("LINE [%s] %s", name, value); */
+    value = segment_next_field(raw_line);
 
     if(strcmp("mail", name) == 0) {
-      strcpy(current_mail, value);
-      /* printf("READING [%s]\n", current_mail); */
-    } else {
-      if(strcmp(search_name, name) == 0 && regexec(&regexp, value, 0, 0, 0) == 0) {
-        printf("%s", current_mail);
+      char *position_in_file_string = value;
+      char *mail_filename = segment_next_field(value);
+      current_position_in_mail = atol(position_in_file_string);
+      strcpy(current_mail_filename, mail_filename);
+      remove_eof(current_mail_filename);
+      already_written = 0;
+    }
+
+    else if(!already_written) {
+     if(strcmp(search_name, name) == 0 && regexec(&regexp, value, 0, 0, 0) == 0) {
+        FILE *mail_file;
+        mail_file = fopen(current_mail_filename, "r");
+        if(!mail_file) {
+          fprintf(stderr, "mymail: Can not open `%s'.\n", current_mail_filename);
+          exit(EXIT_FAILURE);
+        }
+        fseek(mail_file, current_position_in_mail, SEEK_SET);
+        if(fgets(raw_line, BUFFER_SIZE, mail_file)) {
+          printf("%s", raw_line);
+          while(fgets(raw_line, BUFFER_SIZE, mail_file) &&
+                strncmp(raw_line, "From ", 5)) {
+            printf("%s", raw_line);
+          }
+        }
+        fclose(mail_file);
+        already_written = 1;
       }
     }
   }
-}
 
+  regfree(&regexp);
+}
 
 /*********************************************************************/
 
@@ -119,13 +166,14 @@ void read_file(const char *input_filename,
   char raw_line[BUFFER_SIZE];
   FILE *file;
   int in_header, new_header;
-  unsigned int position_in_file;
+  unsigned long int position_in_file;
 
   file = fopen(input_filename, "r");
 
   if(!file) {
     fprintf(stderr, "mymail: Can not open `%s'.\n", input_filename);
-    exit(EXIT_FAILURE);
+    if(paranoid) { exit(EXIT_FAILURE); }
+    return;
   }
 
   in_header = 0;
@@ -137,10 +185,10 @@ void read_file(const char *input_filename,
     if(strncmp(raw_line, "From ", 5) == 0) {
       if(in_header) {
         fprintf(stderr,
-                "Got a 'From ' in the header in %s:%u.\n",
+                "Got a ^\"From \" in the header in %s:%lu.\n",
                 input_filename, position_in_file);
         fprintf(stderr, "%s", raw_line);
-        exit(EXIT_FAILURE);
+        if(paranoid) { exit(EXIT_FAILURE); }
       }
       in_header = 1;
       new_header = 1;
@@ -148,17 +196,11 @@ void read_file(const char *input_filename,
       if(in_header) { in_header = 0; }
     }
 
-    /* if(in_header) { */
-    /* printf("LINE.H %s", raw_line); */
-    /* } else { */
-    /* printf("LINE.B %s", raw_line); */
-    /* } */
-
     if(in_header) {
       int f;
       regmatch_t matches;
       if(new_header) {
-        fprintf(db_file, "mail %s:%d\n", input_filename, position_in_file);
+        fprintf(db_file, "mail %lu %s\n", position_in_file, input_filename);
         new_header = 0;
       }
       for(f = 0; f < nb_fields_to_parse; f++) {
@@ -178,8 +220,8 @@ void read_file(const char *input_filename,
 
 int ignore_entry(const char *name) {
   return
-    strcmp(name, ".") == 0 ||
-    strcmp(name, "..") == 0 ||
+    /* strcmp(name, ".") == 0 || */
+    /* strcmp(name, "..") == 0 || */
     (name[0] == '.' && name[1] != '/');
 }
 
@@ -197,11 +239,6 @@ void process_entry(const char *dir_name,
             dir_name,
             strerror(errno));
     exit(EXIT_FAILURE);
-  } else {
-  }
-
-  if(S_ISLNK(sb.st_mode)) {
-    return;
   }
 
   dir = opendir(dir_name);
@@ -216,10 +253,7 @@ void process_entry(const char *dir_name,
     }
     closedir(dir);
   } else {
-    if(S_ISREG(sb.st_mode)) {
-      /* printf("Processing regular file '%s'.\n", dir_name); */
-      read_file(dir_name, nb_fields_to_parse, fields_to_parse, db_file);
-    }
+    read_file(dir_name, nb_fields_to_parse, fields_to_parse, db_file);
   }
 }
 
@@ -260,6 +294,7 @@ int main(int argc, char **argv) {
   char c;
   int f;
 
+  paranoid = 0;
   action_index = 0;
   search_pattern = 0;
 
@@ -284,8 +319,7 @@ int main(int argc, char **argv) {
 
     case 's':
       if(search_pattern) {
-        fprintf(stderr,
-                "mymail: Search pattern already defined.\n");
+        fprintf(stderr, "mymail: Search pattern already defined.\n");
         exit(EXIT_FAILURE);
       }
       search_pattern = strdup(optarg);
@@ -352,20 +386,17 @@ int main(int argc, char **argv) {
       char *search_name;
       char *search_regexp_string;
       search_name = search_pattern;
-      search_regexp_string = search_pattern;
-      while(*search_regexp_string && *search_regexp_string != ' ') search_regexp_string++;
-      *search_regexp_string = '\0'; search_regexp_string++;
-      while(*search_regexp_string && *search_regexp_string == ' ') search_regexp_string++;
+      search_regexp_string = segment_next_field(search_pattern);
       if(!*search_regexp_string) {
         fprintf(stderr,
                 "Syntax error in the search pattern.\n");
         exit(EXIT_FAILURE);
       }
 
-      printf("Starting search in %s for field \"%s\" matching \"%s\".\n",
-             db_filename,
-             search_name,
-             search_regexp_string);
+      /* printf("Starting search in %s for field \"%s\" matching \"%s\".\n", */
+             /* db_filename, */
+             /* search_name, */
+             /* search_regexp_string); */
 
       db_file = fopen(db_filename, "r");
 
@@ -376,6 +407,7 @@ int main(int argc, char **argv) {
                 strerror(errno));
         exit(EXIT_FAILURE);
       }
+
       search_in_db(search_name, search_regexp_string, db_file);
 
       fclose(db_file);
@@ -383,5 +415,7 @@ int main(int argc, char **argv) {
     }
   }
 
+  free(db_filename);
+
   exit(EXIT_SUCCESS);
 }