X-Git-Url: https://fleuret.org/cgi-bin/gitweb/gitweb.cgi?a=blobdiff_plain;f=mymail.c;h=2617d02a71c11603c5d0af165e8053aa0822c682;hb=17557f7799a9b15adab76086f1f22481f1d10bd1;hp=241d6359e8504c605642bcd20da2697db5ddc5ec;hpb=335e2ce30616f0f2a59b2ad908261053d26f7aad;p=mymail.git

diff --git a/mymail.c b/mymail.c
index 241d635..2617d02 100644
--- a/mymail.c
+++ b/mymail.c
@@ -21,8 +21,13 @@
 
 /*
 
-  To use it as a super-history-search for bash:
-  mymail --bash <(history)
+  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.
 
 */
 
@@ -30,21 +35,42 @@
 
 #include <stdio.h>
 #include <stdlib.h>
-#include <unistd.h>
 #include <string.h>
 #include <errno.h>
-#include <ncurses.h>
 #include <fcntl.h>
-#include <sys/ioctl.h>
-#include <termios.h>
-#include <regex.h>
 #include <locale.h>
 #include <getopt.h>
 #include <limits.h>
+#include <dirent.h>
+#include <regex.h>
+
+#define VERSION "0.1"
+
+#define BUFFER_SIZE 65536
 
-#define VERSION "1.1.7"
+struct parsable_field {
+  char *name;
+  char *regexp_string;
+  regex_t regexp;
+};
+
+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;
+}
 
-#define BUFFER_SIZE 16384
+void remove_eof(char *c) {
+  while(*c && *c != '\n' && *c != '\r') c++;
+  *c = '\0';
+}
 
 /********************************************************************/
 
@@ -62,63 +88,275 @@ void *safe_malloc(size_t n) {
 
 /*********************************************************************/
 
-void usage(FILE *out) {
-
+void print_version(FILE *out) {
   fprintf(out, "mymail version %s (%s)\n", VERSION, UNAME);
+}
+
+void print_usage(FILE *out) {
+  print_version(out);
   fprintf(out, "Written by Francois Fleuret <francois@fleuret.org>.\n");
   fprintf(out, "\n");
   fprintf(out, "Usage: mymail [options] [<filename1> [<filename2> ...]]\n");
   fprintf(out, "\n");
+  fprintf(out, " -h, --help\n");
+  fprintf(out, "         show this help\n");
+  fprintf(out, " -v, --version\n");
+  fprintf(out, "         print the version number\n");
+  fprintf(out, " -i, --index\n");
+  fprintf(out, "         index mails\n");
+  fprintf(out, " -d <db filename>, --db-file <db filename>\n");
+  fprintf(out, "         set the data-base file\n");
+  fprintf(out, " -s <search pattern>, --search <search pattern>\n");
+  fprintf(out, "         search for matching mails in the data-base file\n");
 }
 
-void read_file(const char *input_filename,
-               int nb_lines_max, int *nb_lines) {
+/*********************************************************************/
 
+void search_in_db(const char *search_name, const char *search_regexp_string,
+                  FILE *db_file) {
   char raw_line[BUFFER_SIZE];
-  char *s;
+  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,
+             REG_ICASE)) {
+    fprintf(stderr,
+            "mymail: Syntax error in regexp \"%s\" for field \"%s\".\n",
+            search_regexp_string,
+            search_name);
+    exit(EXIT_FAILURE);
+  }
+
+  current_position_in_mail = 0;
+  already_written = 0;
+
+  while(fgets(raw_line, BUFFER_SIZE, db_file)) {
+    name = raw_line;
+    value = segment_next_field(raw_line);
+
+    if(strcmp("mail", name) == 0) {
+      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);
+}
+
+/*********************************************************************/
+
+void index_one_mbox_line(int nb_fields_to_parse, struct parsable_field *fields_to_parse,
+                         char *raw_line, FILE *db_file) {
+  regmatch_t matches;
+  int f;
+  for(f = 0; f < nb_fields_to_parse; f++) {
+    if(regexec(&fields_to_parse[f].regexp, raw_line, 1, &matches, 0) == 0) {
+      fprintf(db_file, "%s %s\n",
+              fields_to_parse[f].name,
+              raw_line + matches.rm_eo);
+    }
+  }
+}
+
+void index_mbox(const char *input_filename,
+                int nb_fields_to_parse, struct parsable_field *fields_to_parse,
+                FILE *db_file) {
+  char raw_line[BUFFER_SIZE], full_line[BUFFER_SIZE];
+  char *end_of_full_line;
   FILE *file;
-  int l;
+  int in_header, new_header;
+  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);
+    fprintf(stderr, "mymail: Can not open '%s'.\n", input_filename);
+    if(paranoid) { exit(EXIT_FAILURE); }
+    return;
   }
 
-  while(*nb_lines < nb_lines_max && fgets(raw_line, BUFFER_SIZE, file)) {
-    l = strlen(raw_line);
-    fgets(raw_line + l, BUFFER_SIZE - l, file);
-    for(s = raw_line + strlen(raw_line) - 1; s > raw_line && *s == '\n'; s--) {
-      *s = '\0';
+  in_header = 0;
+  new_header = 0;
+
+  position_in_file = 0;
+  end_of_full_line = 0;
+  full_line[0] = '\0';
+
+  while(fgets(raw_line, BUFFER_SIZE, file)) {
+    if(strncmp(raw_line, "From ", 5) == 0) {
+      if(in_header) {
+        fprintf(stderr,
+                "Got a ^\"From \" in the header in %s:%lu.\n",
+                input_filename, position_in_file);
+        fprintf(stderr, "%s", raw_line);
+        if(paranoid) { exit(EXIT_FAILURE); }
+      }
+      in_header = 1;
+      new_header = 1;
+    } else if(strncmp(raw_line, "\n", 1) == 0) {
+      if(in_header) { in_header = 0; }
     }
-    /* store_line(hash_table, raw_line, nb_lines, lines); */
+
+    /* printf("PARSE %d %s", in_header, raw_line); */
+
+    if(in_header) {
+      if(new_header) {
+        fprintf(db_file, "mail %lu %s\n", position_in_file, input_filename);
+        new_header = 0;
+      }
+
+      if(raw_line[0] == ' ' || raw_line[0] == '\t') {
+        char *start = raw_line;
+        while(*start == ' ' || *start == '\t') start++;
+        *(end_of_full_line++) = ' ';
+        strcpy(end_of_full_line, start);
+        while(*end_of_full_line && *end_of_full_line != '\n') {
+          end_of_full_line++;
+        }
+        *end_of_full_line = '\0';
+      }
+
+      else {
+        /* if(!((raw_line[0] >= 'a' && raw_line[0] <= 'z') || */
+             /* (raw_line[0] >= 'A' && raw_line[0] <= 'Z'))) { */
+          /* fprintf(stderr, */
+                  /* "Header line syntax error %s:%lu.\n", */
+                  /* input_filename, position_in_file); */
+          /* fprintf(stderr, "%s", raw_line); */
+        /* } */
+
+        if(full_line[0]) {
+          index_one_mbox_line(nb_fields_to_parse, fields_to_parse, full_line, db_file);
+        }
+
+        end_of_full_line = full_line;
+        strcpy(end_of_full_line, raw_line);
+        while(*end_of_full_line && *end_of_full_line != '\n') {
+          end_of_full_line++;
+        }
+        *end_of_full_line = '\0';
+      }
+
+    }
+
+    position_in_file += strlen(raw_line);
   }
 
   fclose(file);
 }
 
+int ignore_entry(const char *name) {
+  return
+    /* strcmp(name, ".") == 0 || */
+    /* strcmp(name, "..") == 0 || */
+    (name[0] == '.' && name[1] != '/');
+}
+
+void process_entry(const char *dir_name,
+                   int nb_fields_to_parse, struct parsable_field *fields_to_parse,
+                   FILE *db_file) {
+  DIR *dir;
+  struct dirent *dir_e;
+  struct stat sb;
+  char subname[PATH_MAX + 1];
+
+  if(lstat(dir_name, &sb) != 0) {
+    fprintf(stderr,
+            "mymail: Can not stat \"%s\": %s\n",
+            dir_name,
+            strerror(errno));
+    exit(EXIT_FAILURE);
+  }
+
+  dir = opendir(dir_name);
+
+  if(dir) {
+    printf("Processing directory '%s'.\n", dir_name);
+    while((dir_e = readdir(dir))) {
+      if(!ignore_entry(dir_e->d_name)) {
+        snprintf(subname, PATH_MAX, "%s/%s", dir_name, dir_e->d_name);
+        process_entry(subname, nb_fields_to_parse, fields_to_parse, db_file);
+      }
+    }
+    closedir(dir);
+  } else {
+    index_mbox(dir_name, nb_fields_to_parse, fields_to_parse, db_file);
+  }
+}
+
 /*********************************************************************/
 
 /* For long options that have no equivalent short option, use a
    non-character as a pseudo short option, starting with CHAR_MAX + 1.  */
-enum
-{
+enum {
   OPT_BASH_MODE = CHAR_MAX + 1
 };
 
 static struct option long_options[] = {
   { "help", no_argument, 0, 'h' },
+  { "version", no_argument, 0, 'v' },
+  { "db-file", 1, 0, 'd' },
+  { "search-pattern", 1, 0, 's' },
+  { "index", 0, 0, 'i' },
   { 0, 0, 0, 0 }
 };
 
+static struct parsable_field fields_to_parse[] = {
+  {
+    "from",
+    "^\\([Ff][Rr][Oo][Mm]:\\|From\\) *",
+    { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }
+  },
+
+  {
+    "dest",
+    "^\\([Tt][Oo]\\|[Cc][Cc]\\|[Bb][Cc][Cc]\\): *",
+    { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }
+  },
+};
+
 int main(int argc, char **argv) {
   int error = 0, show_help = 0;
+  const int nb_fields_to_parse = sizeof(fields_to_parse) / sizeof(struct parsable_field);
   char c;
+  int f;
+
+  paranoid = 0;
+  action_index = 0;
+  search_pattern = 0;
 
   setlocale(LC_ALL, "");
 
-  while ((c = getopt_long(argc, argv, "o:s:x:vwmqf:ibzdeajyunt:r:l:c:-h",
+  while ((c = getopt_long(argc, argv, "hvip:s:",
                           long_options, NULL)) != -1) {
 
     switch(c) {
@@ -127,21 +365,115 @@ int main(int argc, char **argv) {
       show_help = 1;
       break;
 
+    case 'v':
+      print_version(stdout);
+      break;
+
+    case 'i':
+      action_index = 1;
+      break;
+
+    case 'p':
+      db_filename = strdup(optarg);
+      /* printf("db_filename=\"%s\"\n", db_filename); */
+      break;
+
+    case 's':
+      if(search_pattern) {
+        fprintf(stderr, "mymail: Search pattern already defined.\n");
+        exit(EXIT_FAILURE);
+      }
+      search_pattern = strdup(optarg);
+      break;
+
     default:
       error = 1;
       break;
     }
   }
 
+  if(!db_filename) {
+    char *default_db_filename = getenv("MYMAIL_DB_FILE");
+    if(!default_db_filename) { default_db_filename = "/tmp/mymail.db"; }
+    db_filename = strdup(default_db_filename);
+  }
+
   if(error) {
-    usage(stderr);
+    print_usage(stderr);
     exit(EXIT_FAILURE);
   }
 
   if(show_help) {
-    usage(stdout);
+    print_usage(stdout);
     exit(EXIT_SUCCESS);
   }
 
+  if(action_index) {
+    FILE *db_file = fopen(db_filename, "w");
+    if(!db_file) {
+      fprintf(stderr,
+              "mymail: Can not open \"%s\" for writing: %s\n",
+              db_filename,
+              strerror(errno));
+      exit(EXIT_FAILURE);
+    }
+
+    for(f = 0; f < nb_fields_to_parse; f++) {
+      if(regcomp(&fields_to_parse[f].regexp,
+                 fields_to_parse[f].regexp_string,
+                 REG_ICASE)) {
+        fprintf(stderr,
+                "mymail: Syntax error in regexp \"%s\" for field \"%s\".\n",
+                fields_to_parse[f].regexp_string,
+                fields_to_parse[f].name);
+        exit(EXIT_FAILURE);
+      }
+    }
+
+    while(optind < argc) {
+      process_entry(argv[optind],
+                    nb_fields_to_parse, fields_to_parse, db_file);
+      optind++;
+    }
+
+    fclose(db_file);
+
+    for(f = 0; f < nb_fields_to_parse; f++) {
+      regfree(&fields_to_parse[f].regexp);
+    }
+  }
+
+  else {
+    if(search_pattern) {
+      FILE *db_file;
+      char *search_name;
+      char *search_regexp_string;
+      search_name = search_pattern;
+      search_regexp_string = segment_next_field(search_pattern);
+      if(!*search_regexp_string) {
+        fprintf(stderr,
+                "Syntax error in the search pattern.\n");
+        exit(EXIT_FAILURE);
+      }
+
+      db_file = fopen(db_filename, "r");
+
+      if(!db_file) {
+        fprintf(stderr,
+                "mymail: Can not open \"%s\" for reading: %s\n",
+                db_filename,
+                strerror(errno));
+        exit(EXIT_FAILURE);
+      }
+
+      search_in_db(search_name, search_regexp_string, db_file);
+
+      fclose(db_file);
+      free(search_pattern);
+    }
+  }
+
+  free(db_filename);
+
   exit(EXIT_SUCCESS);
 }