X-Git-Url: https://git.notmuchmail.org/git?p=notmuch;a=blobdiff_plain;f=notmuch-index-message.cc;h=f7ea340f977329d5da7354a2a613ec5766259720;hp=88eb2c52f5754a00c569a0ad9a9d33dc448a5964;hb=1c63ec7031926fcd06ac3293d9e402af069ea6ac;hpb=ed320cb45ba38c5fbf98ca2ed34abb34f5a5b13b;ds=sidebyside diff --git a/notmuch-index-message.cc b/notmuch-index-message.cc index 88eb2c52..f7ea340f 100644 --- a/notmuch-index-message.cc +++ b/notmuch-index-message.cc @@ -208,39 +208,82 @@ skip_re_in_subject (const char *subject) return s; } -Xapian::Document -find_message (Xapian::Database db, const char *message_id) +static void +find_messages_by_term (Xapian::Database db, + const char *prefix_name, + const char *value, + Xapian::PostingIterator *begin, + Xapian::PostingIterator *end) { Xapian::PostingIterator i; char *term; - term = g_strdup_printf ("%s%s", find_prefix ("msgid"), message_id); - i = db.postlist_begin (term); - if (i != db.postlist_end (term)) - return db.get_document (*i); + term = g_strdup_printf ("%s%s", find_prefix (prefix_name), value); + + *begin = db.postlist_begin (term); + + if (end) + *end = db.postlist_end (term); + + free (term); +} + +Xapian::Document +find_message_by_docid (Xapian::Database db, Xapian::docid docid) +{ + return db.get_document (docid); +} + +Xapian::Document +find_message_by_message_id (Xapian::Database db, const char *message_id) +{ + Xapian::PostingIterator i, end; + + find_messages_by_term (db, "msgid", message_id, &i, &end); + + if (i != end) + return find_message_by_docid (db, *i); else return Xapian::Document (); } +static void +insert_thread_id (GHashTable *thread_ids, Xapian::Document doc) +{ + string value_string; + const char *value; + + value_string = doc.get_value (NOTMUCH_VALUE_THREAD); + value = value_string.c_str(); + if (strlen (value)) + g_hash_table_insert (thread_ids, strdup (value), NULL); +} + static char * -find_thread_id (Xapian::Database db, GPtrArray *parents) +find_thread_id (Xapian::Database db, + GPtrArray *parents, + const char *message_id) { + Xapian::PostingIterator child, children_end; Xapian::Document doc; GHashTable *thread_ids; GList *keys, *l; GString *result = NULL; unsigned int i; - string value_string; - const char *value; + const char *parent_message_id; thread_ids = g_hash_table_new (g_str_hash, g_str_equal); + find_messages_by_term (db, "ref", message_id, &child, &children_end); + for ( ; child != children_end; child++) { + doc = find_message_by_docid (db, *child); + insert_thread_id (thread_ids, doc); + } + for (i = 0; i < parents->len; i++) { - doc = find_message (db, (char *) g_ptr_array_index (parents, i)); - value_string = doc.get_value (NOTMUCH_VALUE_THREAD); - value = value_string.c_str(); - if (strlen (value)) - g_hash_table_insert (thread_ids, strdup (value), NULL); + parent_message_id = (char *) g_ptr_array_index (parents, i); + doc = find_message_by_message_id (db, parent_message_id); + insert_thread_id (thread_ids, doc); } keys = g_hash_table_get_keys (thread_ids); @@ -290,89 +333,90 @@ parse_references (GPtrArray *array, } } -/* Generate terms for the body of a message, given the filename of the - * message and the offset at which the headers of the message end, - * (and hence the body begins). */ +/* Given a string representing the body of a message, generate terms + * for it, (skipping quoted portions and signatures). */ static void -gen_terms_body (Xapian::TermGenerator term_gen, - const char * filename, - gint64 body_offset) +gen_terms_body_str (Xapian::TermGenerator term_gen, + char *body) { - GIOChannel *channel; - GIOStatus gio_status; - GError *error = NULL; - char *p, *body_line = NULL, *prev_line = NULL; + char *line, *line_end, *next_line; - channel = g_io_channel_new_file (filename, "r", &error); - if (channel == NULL) { - fprintf (stderr, "Error: %s\n", error->message); - exit (1); - } + if (body == NULL) + return; - gio_status = g_io_channel_seek_position (channel, body_offset, - G_SEEK_SET, &error); - if (gio_status != G_IO_STATUS_NORMAL) { - fprintf (stderr, "Error: %s\n", error->message); - exit (1); - } + next_line = body; while (1) { - if (body_line) - g_free (body_line); - - gio_status = g_io_channel_read_line (channel, &body_line, - NULL, NULL, &error); - if (gio_status == G_IO_STATUS_EOF) + line = next_line; + if (*line == '\0') break; - if (gio_status != G_IO_STATUS_NORMAL) { - fprintf (stderr, "Error: %s\n", error->message); - exit (1); + + next_line = strchr (line, '\n'); + if (next_line == NULL) { + next_line = line + strlen (line); } + line_end = next_line - 1; - if (strlen (body_line) == 0) - continue; + /* Trim whitespace. */ + while (*next_line && isspace (*next_line)) + next_line++; - /* If the line looks like it might be introducing a quote, - * save it until we see if the next line begins a quote. */ - p = body_line + strlen (body_line) - 1; - while (p > body_line and isspace (*p)) - p--; - if (*p == ':') { - prev_line = body_line; - body_line = NULL; + /* Skip lines that are quotes. */ + if (*line == '>') continue; - } - /* Skip quoted lines, (and previous lines that introduced them) */ - if (body_line[0] == '>') { - if (prev_line) { - g_free (prev_line); - prev_line = NULL; - } - continue; - } + /* Also skip lines introducing a quote on the next line. */ + while (line_end > line && isspace (*line_end)) + line_end--; - /* Now that we're not looking at a quote we can add the prev_line */ - if (prev_line) { - gen_terms (term_gen, "body", prev_line); - g_free (prev_line); - prev_line = NULL; - } + if (*line_end == ':' && *next_line == '>') + continue; - /* Skip signatures */ + /* Finally, bail as soon as we see a signature. */ /* XXX: Should only do this if "near" the end of the message. */ - if (strncmp (body_line, "-- ", 3) == 0 || - strncmp (body_line, "----------", 10) == 0 || - strncmp (body_line, "__________", 10) == 0) + if (strncmp (line, "-- ", 3) == 0 || + strncmp (line, "----------", 10) == 0 || + strncmp (line, "__________", 10) == 0) break; - gen_terms (term_gen, "body", body_line); + *(line_end + 1) = '\0'; + gen_terms (term_gen, "body", line); + } +} + + +/* Callback to generate terms for each mime part of a message. */ +static void +gen_terms_part (GMimeObject *parent, + GMimeObject *part, + gpointer user_data) +{ + Xapian::TermGenerator *term_gen = (Xapian::TermGenerator *) user_data; + GMimeStream *stream; + GMimeDataWrapper *wrapper; + GByteArray *byte_array; + char *body; + + if (! GMIME_IS_PART (part)) { + fprintf (stderr, "Warning: Not indexing unknown mime part: %s.\n", + g_type_name (G_OBJECT_TYPE (part))); + return; } - if (body_line) - g_free (body_line); + byte_array = g_byte_array_new (); + + stream = g_mime_stream_mem_new_with_byte_array (byte_array); + g_mime_stream_mem_set_owner (GMIME_STREAM_MEM (stream), FALSE); + wrapper = g_mime_part_get_content_object (GMIME_PART (part)); + g_mime_data_wrapper_write_to_stream (wrapper, stream); - g_io_channel_close (channel); + g_object_unref (stream); + + body = (char *) g_byte_array_free (byte_array, FALSE); + + gen_terms_body_str (*term_gen, body); + + free (body); } static void @@ -431,8 +475,7 @@ index_file (Xapian::WritableDatabase db, gen_terms (term_gen, "subject", subject); gen_terms (term_gen, "body", subject); - gen_terms_body (term_gen, filename, - g_mime_parser_get_headers_end (parser)); + g_mime_message_foreach (message, gen_terms_part, &term_gen); parents = g_ptr_array_new (); @@ -446,7 +489,9 @@ index_file (Xapian::WritableDatabase db, for (i = 0; i < parents->len; i++) add_term (doc, "ref", (char *) g_ptr_array_index (parents, i)); - thread_id = find_thread_id (db, parents); + message_id = g_mime_message_get_message_id (message); + + thread_id = find_thread_id (db, parents, message_id); for (i = 0; i < parents->len; i++) g_free (g_ptr_array_index (parents, i)); @@ -478,7 +523,6 @@ index_file (Xapian::WritableDatabase db, add_term (doc, "type", "mail"); add_term (doc, "source_id", "1"); - message_id = g_mime_message_get_message_id (message); add_term (doc, "msgid", message_id); doc.add_value (NOTMUCH_VALUE_MESSAGE_ID, message_id);