* mu-store/mu-msg-fields: special handling for To:/Cc:/From: so we can better

search for them.
This commit is contained in:
Dirk-Jan C. Binnema 2010-09-12 16:28:34 +03:00
parent d9b5f2514a
commit bb3d9cc0b0
3 changed files with 107 additions and 36 deletions

View File

@ -20,13 +20,22 @@
#include <string.h>
#include "mu-msg-fields.h"
/*
* note: the differences for our purposes between a xapian field and a term:
* - there is only a single value for some item in per document (msg), ie.
* one value containing the list of To: addresses
* - there can be multiple terms, each containing e.g. one of the To: addresses
* - searching uses terms, but to display some field, it must be in the
* value (at least when using MuMsgIter)
*/
enum _FieldFlags {
FLAG_GMIME = 1 << 1, /* field retrieved through gmime */
FLAG_XAPIAN_INDEX = 1 << 2, /* field is indexed in xapian */
FLAG_XAPIAN_TERM = 1 << 3, /* field stored as term in xapian */
FLAG_XAPIAN_VALUE = 1 << 4 /* field stored as value in xapian */
FLAG_GMIME = 1 << 1, /* field retrieved through gmime */
FLAG_XAPIAN_INDEX = 1 << 2, /* field is indexed in xapian */
FLAG_XAPIAN_TERM = 1 << 3, /* field stored as term in xapian */
FLAG_XAPIAN_VALUE = 1 << 4, /* field stored as value in xapian */
FLAG_XAPIAN_CONTACT = 1 << 5 /* field contains e-mail address */
};
typedef enum _FieldFlags FieldFlags;
typedef enum _FieldFlags FieldFlags;
/*
* this struct describes the fields of an e-mail
@ -40,7 +49,9 @@ struct _MuMsgField {
FieldFlags _flags; /* the flags that tells us what to do */
};
/* the name and shortcut fields must be lower case, or they might be
* misinterpreted by the query-preprocesser which turns queries into
* lowercase */
static const MuMsgField FIELD_DATA[] = {
{
MU_MSG_FIELD_ID_BODY_TEXT,
@ -60,7 +71,7 @@ static const MuMsgField FIELD_DATA[] = {
MU_MSG_FIELD_ID_CC,
MU_MSG_FIELD_TYPE_STRING,
"cc", "c", "C",
FLAG_GMIME | FLAG_XAPIAN_INDEX | FLAG_XAPIAN_VALUE
FLAG_GMIME | FLAG_XAPIAN_CONTACT | FLAG_XAPIAN_VALUE
},
{
@ -81,7 +92,7 @@ static const MuMsgField FIELD_DATA[] = {
MU_MSG_FIELD_ID_FROM,
MU_MSG_FIELD_TYPE_STRING,
"from", "f", "F",
FLAG_GMIME | FLAG_XAPIAN_INDEX | FLAG_XAPIAN_VALUE
FLAG_GMIME | FLAG_XAPIAN_CONTACT | FLAG_XAPIAN_VALUE
},
{
@ -123,7 +134,7 @@ static const MuMsgField FIELD_DATA[] = {
MU_MSG_FIELD_ID_TO,
MU_MSG_FIELD_TYPE_STRING,
"to", "t", "T",
FLAG_GMIME | FLAG_XAPIAN_INDEX | FLAG_XAPIAN_VALUE
FLAG_GMIME | FLAG_XAPIAN_CONTACT | FLAG_XAPIAN_VALUE
},
{
@ -232,7 +243,12 @@ mu_msg_field_xapian_term (const MuMsgField *field)
return field->_flags & FLAG_XAPIAN_TERM;
}
gboolean
mu_msg_field_xapian_contact (const MuMsgField *field)
{
g_return_val_if_fail (field, FALSE);
return field->_flags & FLAG_XAPIAN_CONTACT;
}
gboolean

View File

@ -66,7 +66,7 @@ static const guint MU_MSG_FIELD_TYPE_NONE = MU_MSG_FIELD_TYPE_NUM + 1;
typedef void (*MuMsgFieldForEachFunc) (const MuMsgField *field,
gconstpointer data);
/**
/**
* iterator over all possible message fields
*
* @param func a function called for each field
@ -75,7 +75,7 @@ typedef void (*MuMsgFieldForEachFunc) (const MuMsgField *field,
void mu_msg_field_foreach (MuMsgFieldForEachFunc func, gconstpointer data);
/**
/**
* get the name of the field -- this a name that can be use in queries,
* ie. 'subject:foo', with 'subject' being the name
*
@ -86,7 +86,7 @@ void mu_msg_field_foreach (MuMsgFieldForEachFunc func, gconstpointer data);
*/
const char* mu_msg_field_name (const MuMsgField *field) G_GNUC_CONST;
/**
/**
* get the shortcut of the field -- this a shortcut that can be use in
* queries, ie. 's:foo', with 's' meaning 'subject' being the name
*
@ -97,7 +97,7 @@ const char* mu_msg_field_name (const MuMsgField *field) G_GNUC_CONST;
*/
const char* mu_msg_field_shortcut (const MuMsgField *field) G_GNUC_CONST;
/**
/**
* get the xapian prefix of the field -- that is, the prefix used in
* the Xapian database to identify the field
*
@ -108,7 +108,7 @@ const char* mu_msg_field_shortcut (const MuMsgField *field) G_GNUC_CONST;
*/
const char* mu_msg_field_xapian_prefix (const MuMsgField *field) G_GNUC_PURE;
/**
/**
* get the numerical ID of the field
*
* @param field a MuMsgField
@ -118,7 +118,7 @@ const char* mu_msg_field_xapian_prefix (const MuMsgField *field) G_GNUC_PURE;
MuMsgFieldId mu_msg_field_id (const MuMsgField *field) G_GNUC_CONST;
/**
/**
* get the type of the field (string, size, time etc.)
*
* @param field a MuMsgField
@ -129,7 +129,7 @@ MuMsgFieldId mu_msg_field_id (const MuMsgField *field) G_GNUC_CONST;
MuMsgFieldType mu_msg_field_type (const MuMsgField *field) G_GNUC_CONST;
/**
/**
* is the field numeric (has type MU_MSG_FIELD_TYPE_(BYTESIZE|TIME_T|INT))?
*
* @param field a MuMsgField
@ -140,7 +140,7 @@ gboolean mu_msg_field_is_numeric (const MuMsgField *field) G_GNUC_CONST
/**
/**
* is the field Xapian-indexable? That is, should this field be
* indexed in the in the Xapian database, so we can use the all the
* phrasing, stemming etc. magic
@ -151,7 +151,7 @@ gboolean mu_msg_field_is_numeric (const MuMsgField *field) G_GNUC_CONST
*/
gboolean mu_msg_field_xapian_index (const MuMsgField *field) G_GNUC_PURE;
/**
/**
* should this field be stored as a xapian term?
*
* @param field a MuMsgField
@ -160,7 +160,7 @@ gboolean mu_msg_field_xapian_index (const MuMsgField *field) G_GNUC_PURE;
*/
gboolean mu_msg_field_xapian_term (const MuMsgField *field) G_GNUC_PURE;
/**
/**
* should this field be stored as a xapian value?
*
* @param field a MuMsgField
@ -169,8 +169,18 @@ gboolean mu_msg_field_xapian_term (const MuMsgField *field) G_GNUC_PURE;
*/
gboolean mu_msg_field_xapian_value (const MuMsgField *field) G_GNUC_PURE;
/**
* should this field be stored as contact information? This means that
* e-mail address will be stored as terms, and names will be indexed
*
* @param field a MuMsgField
*
* @return TRUE if the field should be stored as contact information,
* FALSE otherwise
*/
gboolean mu_msg_field_xapian_contact (const MuMsgField *field) G_GNUC_PURE;
/**
/**
* is the field gmime-enabled? That is, can be field be retrieved
* using GMime?
*
@ -181,7 +191,7 @@ gboolean mu_msg_field_xapian_value (const MuMsgField *field) G_GNUC_PURE;
gboolean mu_msg_field_gmime (const MuMsgField *field) G_GNUC_PURE;
/**
/**
* get the corresponding MuMsgField for a name (as in mu_msg_field_name)
*
* @param str a name
@ -191,7 +201,7 @@ gboolean mu_msg_field_gmime (const MuMsgField *field) G_GNUC_PURE;
const MuMsgField* mu_msg_field_from_name (const char* str) G_GNUC_PURE;
/**
/**
* get the corresponding MuMsgField for a shortcut (as in mu_msg_field_shortcut)
*
* @param kar a shortcut character
@ -200,7 +210,7 @@ const MuMsgField* mu_msg_field_from_name (const char* str) G_GNUC_PURE;
*/
const MuMsgField* mu_msg_field_from_shortcut (char kar) G_GNUC_CONST;
/**
/**
* get the corresponding MuMsgField for an id (as in mu_msg_field_id)
*
* @param id an id

View File

@ -23,8 +23,10 @@
#include <cstdio>
#include <xapian.h>
#include <cstring>
#include "mu-msg.h"
#include "mu-msg-contact.h"
#include "mu-store.h"
#include "mu-util.h"
@ -146,7 +148,7 @@ mu_store_destroy (MuStore *store)
mu_store_flush (store);
MU_WRITE_LOG ("closing xapian database with %d documents",
(int)store->_db->get_doccount());
(int)store->_db->get_doccount());
delete store->_db;
g_free (store);
@ -164,7 +166,6 @@ mu_store_flush (MuStore *store)
store->_db->flush ();
} MU_XAPIAN_CATCH_BLOCK;
}
@ -200,9 +201,8 @@ add_terms_values_string (Xapian::Document& doc, MuMsg *msg,
}
if (mu_msg_field_xapian_term(field))
/* terms can be up to MU_STORE_MAX_TERM_LENGTH
* (240) long; this is a Xapian limit
* */
/* terms can be up to MU_STORE_MAX_TERM_LENGTH (240)
* long; this is a Xapian limit */
doc.add_term (std::string (prefix + value, 0,
MU_STORE_MAX_TERM_LENGTH));
@ -233,21 +233,23 @@ add_terms_values_body (Xapian::Document& doc, MuMsg *msg,
}
struct _MsgDoc {
Xapian::Document *_doc;
MuMsg *_msg;
Xapian::Document *_doc;
MuMsg *_msg;
};
typedef struct _MsgDoc MsgDoc;
typedef struct _MsgDoc MsgDoc;
static void
add_terms_values (const MuMsgField* field, MsgDoc* msgdoc)
{
MuMsgFieldType type;
/* note: contact-stuff (To/Cc/From) will handled in
* add_contact_info, not here */
if (!mu_msg_field_xapian_index(field) &&
!mu_msg_field_xapian_term(field) &&
!mu_msg_field_xapian_value(field))
return;
type = mu_msg_field_type (field);
if (type == MU_MSG_FIELD_TYPE_STRING) {
@ -270,6 +272,45 @@ add_terms_values (const MuMsgField* field, MsgDoc* msgdoc)
g_return_if_reached ();
}
static void
each_contact_info (MuMsgContact *contact, MsgDoc *data)
{
std::string pfx;
static const MuMsgField *to_field =
mu_msg_field_from_id (MU_MSG_FIELD_ID_TO);
static const MuMsgField *from_field =
mu_msg_field_from_id (MU_MSG_FIELD_ID_FROM);
static const MuMsgField *cc_field =
mu_msg_field_from_id (MU_MSG_FIELD_ID_CC);
static const std::string to_pfx (mu_msg_field_xapian_prefix(to_field));
static const std::string from_pfx (mu_msg_field_xapian_prefix(from_field));
static const std::string cc_pfx (mu_msg_field_xapian_prefix(cc_field));
switch (contact->type) {
case MU_MSG_CONTACT_TYPE_TO: pfx = to_pfx; break;
case MU_MSG_CONTACT_TYPE_FROM: pfx = from_pfx; break;
case MU_MSG_CONTACT_TYPE_CC: pfx = cc_pfx; break;
default: return; /* other types (like bcc) are ignored */
}
// g_print ("[%s %s]\n", pfx.c_str(), contact->address);
if (contact->name && strlen(contact->name) > 0) {
Xapian::TermGenerator termgen;
termgen.set_document (*data->_doc);
termgen.index_text_without_positions (contact->name, 1, pfx);
}
if (contact->address && strlen (contact->address))
data->_doc->add_term (std::string (pfx + contact->address, 0,
MU_STORE_MAX_TERM_LENGTH));
}
/* get a unique id for this message */
static std::string
get_message_uid (const char* path)
@ -282,7 +323,7 @@ get_message_uid (const char* path)
return pathprefix + path;
}
static std::string
static const std::string
get_message_uid (MuMsg *msg)
{
return get_message_uid (mu_msg_get_path(msg));
@ -304,10 +345,14 @@ mu_store_store (MuStore *store, MuMsg *msg)
begin_trx_if (store, !store->_in_transaction);
/* we must add a unique term, so we can replace
* matching documents */
* matching documents */
newdoc.add_term (uid);
mu_msg_field_foreach ((MuMsgFieldForEachFunc)add_terms_values,
&msgdoc);
/* also store the contact-info as separate terms */
mu_msg_contact_foreach (msg,
(MuMsgContactForeachFunc)each_contact_info,
&msgdoc);
/* we replace all existing documents for this file */
id = store->_db->replace_document (uid, newdoc);