Now deals properly with multi-lines header lines. Does not deal with multipart mails.

[mymail.git] / mymail.c

diff --git a/mymail.c b/mymail.c
index 99a76d8..125366a 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>
@@ -45,6 +57,7 @@ struct parsable_field {
 char *db_filename;
 char *search_pattern;
 
+int paranoid;
 int action_index;
 
 char *segment_next_field(char *current) {
@@ -121,7 +134,7 @@ void search_in_db(const char *search_name, const char *search_regexp_string,
     }
 
     else if(!already_written) {
-     if(strcmp(search_name, name) == 0 && regexec(&regexp, value, 0, 0, 0) == 0) {
+      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) {
@@ -145,13 +158,26 @@ void search_in_db(const char *search_name, const char *search_regexp_string,
   regfree(&regexp);
 }
 
-
 /*********************************************************************/
 
-void read_file(const char *input_filename,
-               int nb_fields_to_parse, struct parsable_field *fields_to_parse,
-               FILE *db_file) {
-  char raw_line[BUFFER_SIZE];
+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 in_header, new_header;
   unsigned long int position_in_file;
@@ -160,7 +186,8 @@ void read_file(const char *input_filename,
 
   if(!file) {
     fprintf(stderr, "mymail: Can not open `%s'.\n", input_filename);
-    exit(EXIT_FAILURE);
+    if(paranoid) { exit(EXIT_FAILURE); }
+    return;
   }
 
   in_header = 0;
@@ -172,10 +199,10 @@ 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:%lu.\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;
@@ -183,26 +210,46 @@ 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); */
-    /* } */
+    /* printf("PARSE %d %s", in_header, raw_line); */
 
     if(in_header) {
-      int f;
-      regmatch_t matches;
       if(new_header) {
         fprintf(db_file, "mail %lu %s\n", position_in_file, input_filename);
         new_header = 0;
       }
-      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",
-                  fields_to_parse[f].name,
-                  raw_line + matches.rm_eo);
+
+      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);
@@ -234,10 +281,6 @@ void process_entry(const char *dir_name,
     exit(EXIT_FAILURE);
   }
 
-  /* if(S_ISLNK(sb.st_mode)) { */
-    /* return; */
-  /* } */
-
   dir = opendir(dir_name);
 
   if(dir) {
@@ -250,9 +293,7 @@ void process_entry(const char *dir_name,
     }
     closedir(dir);
   } else {
-    /* if(S_ISREG(sb.st_mode)) { */
-      read_file(dir_name, nb_fields_to_parse, fields_to_parse, db_file);
-    /* } */
+    index_mbox(dir_name, nb_fields_to_parse, fields_to_parse, db_file);
   }
 }
 
@@ -260,8 +301,7 @@ void process_entry(const char *dir_name,
 
 /* 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
 };
 
@@ -293,6 +333,7 @@ int main(int argc, char **argv) {
   char c;
   int f;
 
+  paranoid = 0;
   action_index = 0;
   search_pattern = 0;
 
@@ -317,8 +358,7 @@ 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);
@@ -393,9 +433,9 @@ int main(int argc, char **argv) {
       }
 
       /* printf("Starting search in %s for field \"%s\" matching \"%s\".\n", */
-             /* db_filename, */
-             /* search_name, */
-             /* search_regexp_string); */
+      /* db_filename, */
+      /* search_name, */
+      /* search_regexp_string); */
 
       db_file = fopen(db_filename, "r");