/*
* Copyright 2000 HelixCode (http://www.helixcode.com).
*
* Author :
* Michael Zucchi <notzed@helixcode.com>
* This program is free software; you can redistribute it and/or
* modify it under the terms of the GNU General Public License as
* published by the Free Software Foundation; either version 2 of the
* License, or (at your option) any later version.
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU General Public License for more details.
*
* You should have received a copy of the GNU General Public License
* along with this program; if not, write to the Free Software
* Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307
* USA
*/
#include <glib.h>
#include <stdio.h>
#include <time.h>
#include <string.h>
#include <camel/gmime-utils.h>
#include <camel/camel-log.h>
#include "camel/camel-mime-message.h"
#include "camel/camel-mime-part.h"
#include "camel/camel-stream.h"
#include "camel/camel-stream-fs.h"
#include "camel/camel.h"
#include "camel-mbox-folder.h"
#define HAVE_IBEX
#ifdef HAVE_IBEX
#include "ibex.h"
#endif
#define p(x) /* parse debug */
#define r(x) /* run debug */
#define d(x) /* general debug */
/*
This is not yet complete.
The following s-exp's are supported:
list = (and list*)
perform an intersection of a number of lists, and return that.
bool = (and bool*)
perform a boolean AND of boolean values.
list = (or list*)
perform a union of a number of lists, returning the new list.
bool = (or bool*)
perform a boolean OR of boolean values.
Comparison operators:
bool = (lt int int)
bool = (gt int int)
bool = (eq int int)
bool = (lt string string)
bool = (gt string string)
bool = (eq string string)
Perform a comparision of 2 integers, or 2 string values.
Matching operators:
list = (contains string)
Returns a list of all messages containing the string in their body.
list = (match-all bool-expr)
Returns a list of all messages for which the bool expression is true.
The bool-expr is evaluated for each message in turn.
It is more efficient not to perform body-content comparisons inside a
match-all operator.
int = (date-sent)
Returns a time_t of the date-sent of the message.
bool = (header-contains string1 string2)
Returns true if the current message (inside a match-all operator)
has a header 'string1', which contains 'string2'
*/
static GScannerConfig scanner_config =
{
(
" \t\r\n"
) /* cset_skip_characters */,
(
G_CSET_a_2_z
"_"
G_CSET_A_2_Z
) /* cset_identifier_first */,
(
G_CSET_a_2_z
"_0123456789-"
G_CSET_A_2_Z
G_CSET_LATINS
G_CSET_LATINC
) /* cset_identifier_nth */,
( "#\n" ) /* cpair_comment_single */,
FALSE /* case_sensitive */,
TRUE /* skip_comment_multi */,
TRUE /* skip_comment_single */,
TRUE /* scan_comment_multi */,
TRUE /* scan_identifier */,
FALSE /* scan_identifier_1char */,
FALSE /* scan_identifier_NULL */,
TRUE /* scan_symbols */,
FALSE /* scan_binary */,
TRUE /* scan_octal */,
TRUE /* scan_float */,
TRUE /* scan_hex */,
FALSE /* scan_hex_dollar */,
TRUE /* scan_string_sq */,
TRUE /* scan_string_dq */,
TRUE /* numbers_2_int */,
FALSE /* int_2_float */,
FALSE /* identifier_2_string */,
TRUE /* char_2_token */,
FALSE /* symbol_2_token */,
FALSE /* scope_0_fallback */,
};
enum _searchtermtype_t {
SEARCH_AND,
SEARCH_OR,
SEARCH_LT,
SEARCH_GT,
SEARCH_EQ,
SEARCH_CONTAINS,
SEARCH_DATESENT,
SEARCH_STRING,
SEARCH_INT,
SEARCH_FUNC,
};
struct _searchterm {
enum _searchtermtype_t type;
union {
char *string;
int number;
struct {
struct _searchterm_symbol *sym;
struct _searchterm **terms;
int termcount;
} func;
} value;
};
struct _searchcontext {
int whatever;
CamelFolder *folder;
#ifdef HAVE_IBEX
ibex *index; /* index of content for this folder */
#endif
CamelFolderSummary *summary;
const GArray *message_info;
CamelMessageInfo *message_current; /* when performing a (match operation */
};
enum _searchresulttype_t {
RESULT_ARRAY_PTR=0,
RESULT_INT,
RESULT_STRING,
RESULT_BOOL,
RESULT_UNDEFINED
};
struct _searchresult {
enum _searchresulttype_t type;
union {
GPtrArray *ptrarray;
int number;
char *string;
int bool;
} value;
};
/* function callbacks */
static struct _searchresult *search_contains(struct _searchcontext *ctx, struct _searchterm *t);
static struct _searchresult *search_matches(struct _searchcontext *ctx, struct _searchterm *t);
static struct _searchresult *search_date_sent(struct _searchcontext *ctx, struct _searchterm *t);
static struct _searchresult *header_contains(struct _searchcontext *ctx, struct _searchterm *t);
struct _searchterm_symbol {
char *name;
int type;
int argtype;
struct _searchresult * (*func)(struct _searchcontext *ctx, struct _searchterm *t);
} symbols[] = {
{ "and", SEARCH_AND, 0, NULL },
{ "or", SEARCH_OR, 0, NULL },
{ "lt", SEARCH_LT, 1, NULL },
{ "gt", SEARCH_GT, 1, NULL },
{ "eq", SEARCH_EQ, 1, NULL },
{ "contains", SEARCH_FUNC, 1, search_contains },
{ "match-all", SEARCH_FUNC, 1, search_matches },
{ "date-sent", SEARCH_FUNC, 1, search_date_sent },
{ "header-contains", SEARCH_FUNC, 1, header_contains },
};
static struct _searchterm * parse_list(GScanner *gs, int gotbrace);
static struct _searchterm * parse_value(GScanner *gs);
static struct _searchresult *term_eval(struct _searchcontext *ctx, struct _searchterm *t);
static void parse_dump_term(struct _searchterm *t, int depth);
/* can you tell, i dont like glib? */
struct _glib_sux_donkeys {
int count;
GPtrArray *uids;
};
/* ok, store any values that are in all sets */
static void
g_lib_sux_htand(char *key, int value, struct _glib_sux_donkeys *fuckup)
{
if (value == fuckup->count) {
g_ptr_array_add(fuckup->uids, key);
}
}
/* or, store all unique values */
static void
g_lib_sux_htor(char *key, int value, struct _glib_sux_donkeys *fuckup)
{
g_ptr_array_add(fuckup->uids, key);
}
static struct _searchresult *
result_new(int type)
{
struct _searchresult *r = g_malloc0(sizeof(*r));
r->type = type;
return r;
}
static void
result_free(struct _searchresult *t)
{
switch(t->type) {
case RESULT_ARRAY_PTR:
g_ptr_array_free(t->value.ptrarray, TRUE);
break;
case RESULT_BOOL:
case RESULT_INT:
break;
case RESULT_STRING:
g_free(t->value.string);
break;
case RESULT_UNDEFINED:
break;
}
g_free(t);
}
static struct _searchresult *search_contains(struct _searchcontext *ctx, struct _searchterm *t)
{
struct _searchresult *r, *r1;
r = result_new(RESULT_UNDEFINED);
if (t->value.func.termcount>0) {
if (t->value.func.termcount!=1) {
printf("warning, only looking for first string in contains clause\n");
}
r1 = term_eval(ctx, t->value.func.terms[0]);
if (r1->type == RESULT_STRING) {
if (ctx->message_current) {
int truth = FALSE;
#ifdef HAVE_IBEX
int i;
GPtrArray *array;
if (ctx->index) {
array = ibex_find(ctx->index, r1->value.string);
for (i=0;i<array->len;i++) {
if (!strcmp(g_ptr_array_index(array, i), ctx->message_current->uid)) {
truth = TRUE;
break;
}
}
g_ptr_array_free(array, TRUE);
}
#endif
r->type = RESULT_BOOL;
r->value.bool = truth;
} else {
r->type = RESULT_ARRAY_PTR;
#ifdef HAVE_IBEX
if (ctx->index) {
/* blah, this should probably copy the index strings? */
r->value.ptrarray = ibex_find(ctx->index, r1->value.string);
} else {
r->value.ptrarray = g_ptr_array_new();
}
#endif
}
} else {
printf("you can't search for a contents of a non-string, fool\n");
}
result_free(r1);
}
return r;
}
/* run a sub-tree of commands which match on header fields etc */
static struct _searchresult *search_matches(struct _searchcontext *ctx, struct _searchterm *t)
{
int i;
struct _searchresult *r, *r1;
if (t->value.func.termcount == 1) {
r = result_new(RESULT_ARRAY_PTR);
r->value.ptrarray = g_ptr_array_new();
for (i=0;i<ctx->message_info->len;i++) {
ctx->message_current = &g_array_index(ctx->message_info, CamelMessageInfo, i);
r1 = term_eval(ctx, t->value.func.terms[0]);
if (r1->type == RESULT_BOOL) {
if (r1->value.bool) {
r(printf("adding message %s\n", ctx->message_current->uid));
g_ptr_array_add(r->value.ptrarray, ctx->message_current->uid);
}
} else {
printf("invalid syntax, matches require a single bool result\n");
}
result_free(r1);
}
ctx->message_current = NULL;
} else {
r = result_new(RESULT_UNDEFINED);
printf("invalid syntax, matches only allows a single bool arg\n");
}
return r;
}
/* these variable-getting things could be put into 1 function */
static struct _searchresult *search_date_sent(struct _searchcontext *ctx, struct _searchterm *t)
{
struct _searchresult *r;
if (ctx->message_current) {
r = result_new(RESULT_INT);
r->value.number = time(0);
/* r->value.number = ctx->current_message->date_sent;*/
} else {
r = result_new(RESULT_UNDEFINED);
}
return r;
}
/* header contains - can only be used inside a match-all construct */
/* all headers should be inside a lookup table, so this can search
all header types */
static struct _searchresult *header_contains(struct _searchcontext *ctx, struct _searchterm *t)
{
struct _searchresult *r;
r(printf("executing header-contains\n"));
/* are we inside a match-all? */
if (ctx->message_current
&& t->value.func.termcount == 2) {
char *header, *substring;
int truth = FALSE;
struct _searchresult *r1, *r2;
r1 = term_eval(ctx, t->value.func.terms[0]);
r2 = term_eval(ctx, t->value.func.terms[1]);
if (r1->type == RESULT_STRING
&& r2->type == RESULT_STRING) {
header = r1->value.string;
substring = r2->value.string;
if (!strcasecmp(header, "subject")) {
r(printf("comparing subject: %s\n", ctx->message_current->subject));
if (ctx->message_current->subject)
truth = (strstr(ctx->message_current->subject, substring)) != NULL;
else
printf("Warning: no subject line in message?\n");
}
r(printf("header-contains %s %s = %s\n", header, substring, truth?"TRUE":"FALSE"));
}
result_free(r1);
result_free(r2);
r = result_new(RESULT_BOOL);
r->value.number = truth;
} else {
r = result_new(RESULT_UNDEFINED);
}
return r;
}
static struct _searchresult *
term_eval(struct _searchcontext *ctx, struct _searchterm *t)
{
struct _searchresult *r, *r1, *r2;
int i;
r(printf("eval term :\n"));
r(parse_dump_term(t, 0));
r = g_malloc0(sizeof(*r));
r->type = RESULT_UNDEFINED;
switch (t->type) {
case SEARCH_AND: {
GHashTable *ht = g_hash_table_new(g_str_hash, g_str_equal);
struct _glib_sux_donkeys lambdafoo;
int type=-1;
int bool = TRUE;
r(printf("( and\n"));
for (i=0;bool && i<t->value.func.termcount;i++) {
r1 = term_eval(ctx, t->value.func.terms[i]);
if (type == -1)
type = r1->type;
if (type != r1->type) {
printf("invalid types in and operation, all types must be the same\n");
} else if ( r1->type == RESULT_ARRAY_PTR ) {
char **a1;
int l1, j;
a1 = (char **)r1->value.ptrarray->pdata;
l1 = r1->value.ptrarray->len;
for (j=0;i<l1;j++) {
int n;
n = (int)g_hash_table_lookup(ht, a1[i]);
g_hash_table_insert(ht, a1[i], (void *)n+1);
}
} else if ( r1->type == RESULT_BOOL ) {
bool &= r1->value.bool;
}
result_free(r1);
}
if (type == RESULT_ARRAY_PTR) {
lambdafoo.count = t->value.func.termcount;
lambdafoo.uids = g_ptr_array_new();
g_hash_table_foreach(ht, (GHFunc)g_lib_sux_htand, &lambdafoo);
r->type = RESULT_ARRAY_PTR;
r->value.ptrarray = lambdafoo.uids;
} else if (type == RESULT_BOOL) {
r->type = RESULT_BOOL;
r->value.bool = bool;
}
g_hash_table_destroy(ht);
break; }
case SEARCH_OR: {
GHashTable *ht = g_hash_table_new(g_str_hash, g_str_equal);
struct _glib_sux_donkeys lambdafoo;
int type = -1;
int bool = FALSE;
r(printf("(or \n"));
for (i=0;!bool && i<t->value.func.termcount;i++) {
r1 = term_eval(ctx, t->value.func.terms[i]);
if (type == -1)
type = r1->type;
if (r1->type != type) {
printf("wrong types in or operation\n");
} else if (r1->type == RESULT_ARRAY_PTR) {
char **a1;
int l1, j;
a1 = (char **)r1->value.ptrarray->pdata;
l1 = r1->value.ptrarray->len;
for (j=0;i<l1;j++) {
g_hash_table_insert(ht, a1[j], (void *)1);
}
} else if (r1->type == RESULT_BOOL) {
bool |= r1->value.bool;
}
result_free(r1);
}
if (type == RESULT_ARRAY_PTR) {
lambdafoo.count = t->value.func.termcount;
lambdafoo.uids = g_ptr_array_new();
g_hash_table_foreach(ht, (GHFunc)g_lib_sux_htor, &lambdafoo);
r->type = RESULT_ARRAY_PTR;
r->value.ptrarray = lambdafoo.uids;
} else if (type == RESULT_BOOL) {
r->type = RESULT_BOOL;
r->value.bool = bool;
}
g_hash_table_destroy(ht);
break; }
case SEARCH_LT:
r(printf("(lt \n"));
if (t->value.func.termcount == 2) {
r1 = term_eval(ctx, t->value.func.terms[0]);
r2 = term_eval(ctx, t->value.func.terms[1]);
if (r1->type != r2->type) {
printf("error, invalid types in compare\n");
} else if (r1->type == RESULT_INT) {
r->type = RESULT_BOOL;
r->value.bool = r1->value.number < r2->value.number;
} else if (r1->type == RESULT_STRING) {
r->type = RESULT_BOOL;
r->value.bool = strcmp(r1->value.string, r2->value.string) < 0;
}
}
break;
case SEARCH_GT:
r(printf("(gt \n"));
if (t->value.func.termcount == 2) {
r1 = term_eval(ctx, t->value.func.terms[0]);
r2 = term_eval(ctx, t->value.func.terms[1]);
if (r1->type != r2->type) {
printf("error, invalid types in compare\n");
} else if (r1->type == RESULT_INT) {
r->type = RESULT_BOOL;
r->value.bool = r1->value.number > r2->value.number;
} else if (r1->type == RESULT_STRING) {
r->type = RESULT_BOOL;
r->value.bool = strcmp(r1->value.string, r2->value.string) > 0;
}
}
break;
case SEARCH_STRING:
r(printf(" (string \"%s\")\n", t->value.string));
r->type = RESULT_STRING;
/* erk, this shoul;dn't need to strdup this ... */
r->value.string = g_strdup(t->value.string);
break;
case SEARCH_INT:
r(printf(" (int %d)\n", t->value.number));
r->type = RESULT_INT;
r->value.number = t->value.number;
break;
case SEARCH_FUNC:
g_free(r); /* <---- FIXME: ICK !! */
r(printf("function '%s'\n", t->value.func.sym->name));
return t->value.func.sym->func(ctx, t);
default:
printf("Warning: Unknown type encountered in parse tree: %d\n", t->type);
r->type = RESULT_UNDEFINED;
}
return r;
}
static void
parse_dump_term(struct _searchterm *t, int depth)
{
int dumpvals = FALSE;
int i;
if (t==NULL) {
printf("null term??\n");
return;
}
for (i=0;i<depth;i++)
printf(" ");
switch (t->type) {
case SEARCH_AND:
printf("(and \n");
dumpvals = 1;
break;
case SEARCH_OR:
printf("(or \n");
dumpvals = 1;
break;
case SEARCH_LT:
printf("(lt \n");
dumpvals = 1;
break;
case SEARCH_GT:
printf("(gt \n");
dumpvals = 1;
break;
case SEARCH_STRING:
printf(" \"%s\"", t->value.string);
break;
case SEARCH_INT:
printf(" %d", t->value.number);
break;
case SEARCH_FUNC:
printf(" (function %s", t->value.func.sym->name);
dumpvals = 1;
break;
default:
printf("unknown type: %d\n", t->type);
}
if (dumpvals) {
/*printf(" [%d] ", t->value.func.termcount);*/
for (i=0;i<t->value.func.termcount;i++) {
parse_dump_term(t->value.func.terms[i], depth+1);
}
for (i=0;i<depth;i++)
printf(" ");
printf(")\n");
}
printf("\n");
}
/*
PARSER
*/
static struct _searchterm *
parse_new_term(int type)
{
struct _searchterm *s = g_malloc0(sizeof(*s));
s->type = type;
return s;
}
static void
parse_term_free(struct _searchterm *t)
{
int i;
if (t==NULL) {
return;
}
switch (t->type) {
case SEARCH_AND:
case SEARCH_OR:
case SEARCH_LT:
case SEARCH_GT:
case SEARCH_FUNC:
for (i=0;i<t->value.func.termcount;i++) {
parse_term_free(t->value.func.terms[i]);
}
g_free(t->value.func.terms);
break;
case SEARCH_STRING:
g_free(t->value.string);
break;
case SEARCH_INT:
break;
default:
printf("parse_term_free: unknown type: %d\n", t->type);
}
g_free(t);
}
static struct _searchterm **
parse_lists(GScanner *gs, int *len)
{
int token;
struct _searchterm **terms;
int i=0;
p(printf("parsing lists\n"));
terms = g_malloc0(20*sizeof(*terms));
while ( (token = g_scanner_peek_next_token(gs)) != G_TOKEN_EOF
&& token != ')') {
terms[i]=parse_list(gs, FALSE);
i++;
}
if (len)
*len = i;
p(printf("found %d subterms\n", i));
p(printf("done parsing lists, token= %d %c\n", token, token));
return terms;
}
static struct _searchterm **
parse_values(GScanner *gs, int *len)
{
int token;
struct _searchterm **terms;
int i=0;
p(printf("parsing values\n"));
terms = g_malloc0(20*sizeof(*terms));
while ( (token = g_scanner_peek_next_token(gs)) != G_TOKEN_EOF
&& token != ')') {
terms[i]=parse_value(gs);
i++;
}
p(printf("found %d subterms\n", i));
*len = i;
p(printf("dont parsing values\n"));
return terms;
}
static struct _searchterm *
parse_value(GScanner *gs)
{
int token;
struct _searchterm *t = NULL;
p(printf("parsing value\n"));
token = g_scanner_get_next_token(gs);
switch(token) {
case G_TOKEN_LEFT_PAREN:
p(printf("got brace, its a list!\n"));
return parse_list(gs, TRUE);
case G_TOKEN_STRING:
p(printf("got string\n"));
t = parse_new_term(SEARCH_STRING);
t->value.string = g_strdup(g_scanner_cur_value(gs).v_string);
break;
case G_TOKEN_INT:
t = parse_new_term(SEARCH_INT);
t->value.number = g_scanner_cur_value(gs).v_int;
p(printf("got int\n"));
break;
default:
printf("Innvalid token trying to parse a list of values\n");
}
p(printf("done parsing value\n"));
return t;
}
/* FIXME: this needs some robustification */
static struct _searchterm *
parse_list(GScanner *gs, int gotbrace)
{
int token;
struct _searchterm *t = NULL;
p(printf("parsing list\n"));
if (gotbrace)
token = '(';
else
token = g_scanner_get_next_token(gs);
if (token =='(') {
token = g_scanner_get_next_token(gs);
if (token == G_TOKEN_SYMBOL) {
struct _searchterm_symbol *s;
s = g_scanner_cur_value(gs).v_symbol;
p(printf("got funciton: %s\n", s->name));
t = parse_new_term(s->type);
t->value.func.sym = s;
p(printf("created new list %p\n", t));
switch(s->argtype) {
case 0: /* it MUST be a list of lists */
t->value.func.terms = parse_lists(gs, &t->value.func.termcount);
break;
case 1:
t->value.func.terms = parse_values(gs, &t->value.func.termcount);
break;
default:
printf("Error, internal error parsing symbols\n");
}
} else {
printf("unknown sequence encountered, type = %d\n", token);
}
token = g_scanner_get_next_token(gs);
if (token != ')') {
printf("Error, expected ')' not found\n");
}
} else {
printf("Error, list term without opening (\n");
}
p(printf("returning list %p\n", t));
return t;
}
GList *
camel_mbox_folder_search_by_expression(CamelFolder *folder, char *expression, CamelException *ex)
{
GScanner *gs;
int i;
struct _searchterm *t;
struct _searchcontext *ctx;
struct _searchresult *r;
GList *matches = NULL;
gs = g_scanner_new(&scanner_config);
for(i=0;i<sizeof(symbols)/sizeof(symbols[0]);i++)
g_scanner_scope_add_symbol(gs, 0, symbols[i].name, &symbols[i]);
g_scanner_input_text(gs, expression, strlen(expression));
t = parse_list(gs, 0);
if (t) {
ctx = g_malloc0(sizeof(*ctx));
ctx->folder = folder;
ctx->summary = camel_folder_get_summary(folder, ex);
ctx->message_info = camel_folder_summary_get_message_info_list(ctx->summary);
#ifdef HAVE_IBEX
ctx->index = ibex_open(CAMEL_MBOX_FOLDER(folder)->index_file_path, FALSE);
if (!ctx->index) {
perror("Cannot open index file, body searches will be ignored\n");
}
#endif
r = term_eval(ctx, t);
/* now create a folder summary to return?? */
if (r
&& r->type == RESULT_ARRAY_PTR) {
d(printf("got result ...\n"));
for (i=0;i<r->value.ptrarray->len;i++) {
d(printf("adding match: %s\n", (char *)g_ptr_array_index(r->value.ptrarray, i)));
matches = g_list_prepend(matches, g_strdup(g_ptr_array_index(r->value.ptrarray, i)));
}
result_free(r);
}
if (ctx->index)
ibex_close(ctx->index);
gtk_object_unref((GtkObject *)ctx->summary);
g_free(ctx);
parse_term_free(t);
} else {
printf("Warning, Could not parse expression!\n %s\n", expression);
}
g_scanner_destroy(gs);
return matches;
}
