Hi Andreas,
On Wed, 28 Feb 2024 18:36:04 +0200,
Andreas Beckmann wrote:
> On Wed, 20 Dec 2023 21:18:20 +0200 Yavor Doganov <ya...@gnu.org> wrote:
> > Please find attached a patch;
>
> Thanks for the patch, I uploaded it to Debian and so far noone
> complained ;-)
Thanks! Complaints usually come a bit later...
> But the patch doesn't apply cleanly on newer virtuoso-opensource
> versions, there are actually changes in pcre usage in
> libsrc/Wi/sqlbif.h that require adjustments.
Right; there's a new function.
> Could you take a look again and update the patch?
Attached is a patch (commit made to the try-7.2.12 branch) that
updates pcre2.patch so that it applies cleanly and restores the
build-dependency on libpcre2-dev.
> I've never worked with (any version of) pcre (from the programmer
> persepective,
Likewise, I'm a complete novice here. It would be nice to finish this
transition, though.
>From 82b97264413540aa72d96297a93a6fd24f56adc2 Mon Sep 17 00:00:00 2001
From: Yavor Doganov <ya...@gnu.org>
Date: Mon, 4 Mar 2024 16:29:34 +0200
Subject: [PATCH] pcre2.patch: Update for the new upstream release
---
debian/changelog | 4 +
debian/control | 2 +-
debian/patches/pcre2.patch | 152 ++++++++++++++++++++++++-------------
debian/patches/series | 2 +-
4 files changed, 107 insertions(+), 53 deletions(-)
diff --git a/debian/changelog b/debian/changelog
index 907e89fc6..a566f67f2 100644
--- a/debian/changelog
+++ b/debian/changelog
@@ -1,8 +1,12 @@
virtuoso-opensource (7.2.12+dfsg-0.1) UNRELEASED; urgency=medium
+ [ Andreas Beckmann ]
* Non-maintainer upload.
* New upstream release.
+ [ Yavor Doganov ]
+ * debian/patches/pcre2.patch: Update for the new upstream release.
+
-- Andreas Beckmann <a...@debian.org> Wed, 28 Feb 2024 15:19:59 +0100
virtuoso-opensource (7.2.5.1+dfsg1-0.6) unstable; urgency=medium
diff --git a/debian/control b/debian/control
index 5fd05fab9..db9b573c4 100644
--- a/debian/control
+++ b/debian/control
@@ -14,7 +14,7 @@ Build-Depends: debhelper-compat (= 13),
gperf,
libldap2-dev,
libmagickwand-dev,
- libpcre3-dev,
+ libpcre2-dev,
libreadline-dev,
libssl-dev,
libtirpc-dev,
diff --git a/debian/patches/pcre2.patch b/debian/patches/pcre2.patch
index 98c06be51..104c1abc9 100644
--- a/debian/patches/pcre2.patch
+++ b/debian/patches/pcre2.patch
@@ -2,12 +2,12 @@ Description: Port to PCRE2.
Bug-Debian: https://bugs.debian.org/999931
Author: Yavor Doganov <ya...@gnu.org>
Forwarded: no
-Last-Update: 2023-12-20
+Last-Update: 2024-03-04
---
--- virtuoso-opensource.orig/libsrc/Wi/Makefile.am
+++ virtuoso-opensource/libsrc/Wi/Makefile.am
-@@ -559,7 +559,7 @@
+@@ -563,7 +563,7 @@
$(libwi_base_la_sources)
libwi_la_CFLAGS = $(libwi_base_la_cflags)
@@ -18,7 +18,7 @@ Last-Update: 2023-12-20
libwi_odbc_la_SOURCES += \
--- virtuoso-opensource.orig/libsrc/Wi/bif_regexp.c
+++ virtuoso-opensource/libsrc/Wi/bif_regexp.c
-@@ -30,7 +30,8 @@
+@@ -31,7 +31,8 @@
// Debian maintainer: replaced by external PCRE
// #include "util/pcrelib/pcre.h"
@@ -28,7 +28,7 @@ Last-Update: 2023-12-20
/*
typedef struct rx_query_s {
-@@ -66,15 +67,16 @@
+@@ -65,16 +66,17 @@
typedef struct compiled_regexp_s
{
int refctr;
@@ -42,14 +42,16 @@ Last-Update: 2023-12-20
-int32 c_pcre_match_limit_recursion = 500;
-int32 c_pcre_match_limit = 100000;
+-int32 pcre_max_cache_sz = 20000;
+static pcre2_match_context *match_ctxt = NULL;
+
+uint32 c_pcre_match_limit_recursion = 500;
+uint32 c_pcre_match_limit = 100000;
- int32 pcre_max_cache_sz = 20000;
++uint32 pcre_max_cache_sz = 20000;
int32 pcre_rnd_seed;
-@@ -97,6 +99,23 @@
+ id_hashed_key_t
+@@ -96,6 +98,23 @@
}
void
@@ -73,7 +75,7 @@ Last-Update: 2023-12-20
release_compiled_regexp (id_hash_t *c_r, compiled_regexp_t *data)
{
int delete_data;
-@@ -112,9 +131,7 @@
+@@ -111,9 +130,7 @@
if (!delete_data)
return;
if (NULL != data->code)
@@ -84,7 +86,7 @@ Last-Update: 2023-12-20
dk_free (data, sizeof (compiled_regexp_t));
}
-@@ -137,10 +154,11 @@
+@@ -136,10 +153,11 @@
}
static compiled_regexp_t *
@@ -99,9 +101,9 @@ Last-Update: 2023-12-20
regexp_key_t key;
compiled_regexp_t **val = NULL;
compiled_regexp_t tmp, *new_val;
-@@ -156,46 +174,18 @@
+@@ -155,46 +173,18 @@
}
- HT_LEAVE (c_r);
+ HT_UNLOCK (c_r);
dbg_printf (("regex compiling (%s) with options %x ...\n", pattern,
options));
- tmp.code = pcre_compile (pattern, options, &error, &erroff, 0);
+ tmp.code = pcre2_compile ((PCRE2_SPTR) pattern, strlen (pattern),
@@ -149,9 +151,9 @@ Last-Update: 2023-12-20
new_val->code = tmp.code;
- new_val->code_x = tmp.code_x;
new_val->refctr = 1;
- HT_ENTER (c_r);
+ HT_WRLOCK (c_r);
pcre_cache_check (c_r);
-@@ -302,18 +292,18 @@
+@@ -301,18 +291,18 @@
}
@@ -175,7 +177,7 @@ Last-Update: 2023-12-20
/*
#define PCRE_EXTENDED 0x0008
#define PCRE_ANCHORED 0x0010
-@@ -324,7 +314,7 @@
+@@ -323,7 +313,7 @@
#define PCRE_UNGREEDY 0x0200
#define PCRE_NOTEMPTY 0x0400
*/
@@ -184,7 +186,7 @@ Last-Update: 2023-12-20
/*
#define PCRE_NO_AUTO_CAPTURE 0x1000
*/
-@@ -338,15 +328,16 @@
+@@ -337,15 +327,16 @@
bif_regexp_match (caddr_t * qst, caddr_t * err_ret, state_slot_t ** args)
{
int utf8_mode;
@@ -204,7 +206,7 @@ Last-Update: 2023-12-20
int result;
utf8_mode = 0;
-@@ -364,7 +355,7 @@
+@@ -363,7 +354,7 @@
if (*err_ret) goto done;
if (utf8_mode)
@@ -213,7 +215,7 @@ Last-Update: 2023-12-20
if (!pattern || !str)
goto done;
-@@ -373,10 +364,14 @@
+@@ -372,10 +363,14 @@
if (err_ret[0])
goto done;
@@ -230,7 +232,7 @@ Last-Update: 2023-12-20
ret_str = dk_alloc_box (offvect[1] - offvect[0] + 1, DV_SHORT_STRING);
strncpy (ret_str, str + offvect[0], offvect[1] - offvect[0]);
ret_str[offvect[1] - offvect[0]] = 0;
-@@ -417,6 +412,7 @@
+@@ -416,6 +411,7 @@
done:
release_compiled_regexp (compiled_regexps, cd_info);
@@ -238,7 +240,7 @@ Last-Update: 2023-12-20
if (*err_ret)
dk_free_box (ret_str);
dk_free_tree (p_to_free);
-@@ -428,15 +424,15 @@
+@@ -427,15 +423,15 @@
bif_rdf_regex_impl (caddr_t * qst, caddr_t * err_ret, state_slot_t ** args)
{
int utf8_mode = 1;
@@ -257,7 +259,7 @@ Last-Update: 2023-12-20
switch ((BOX_ELEMENTS (args)))
{
default:
-@@ -449,7 +445,7 @@
+@@ -448,7 +444,7 @@
if (err) goto done;
if (utf8_mode)
@@ -266,7 +268,7 @@ Last-Update: 2023-12-20
if (!pattern || !str)
goto done;
-@@ -457,11 +453,15 @@
+@@ -456,11 +452,15 @@
cd_info = get_compiled_regexp (compiled_regexps, pattern, c_opts, err_ret);
if (err_ret[0])
goto done;
@@ -284,7 +286,7 @@ Last-Update: 2023-12-20
if (err)
dk_free_tree (err);
dk_free_tree (p_to_free);
-@@ -477,13 +477,14 @@
+@@ -476,13 +476,14 @@
char *pattern;
char *str;
int offset;
@@ -302,7 +304,7 @@ Last-Update: 2023-12-20
utf8_mode = 0;
pattern = bif_regexp_str_arg (qst, args, 0, "regexp_substr", REGEXP_BF,
&utf8_mode, &p_to_free, err_ret);
-@@ -504,14 +505,20 @@
+@@ -503,14 +504,20 @@
if (NULL != err_ret[0])
goto done;
@@ -326,7 +328,7 @@ Last-Update: 2023-12-20
if (rc < 0)
{
*err_ret = srv_make_new_error ("2201B", "SR097",
-@@ -522,15 +529,21 @@
+@@ -521,15 +528,21 @@
{
if (utf8_mode)
{
@@ -349,7 +351,7 @@ Last-Update: 2023-12-20
if (*err_ret)
dk_free_box (ret_str);
dk_free_tree (p_to_free);
-@@ -539,7 +552,7 @@
+@@ -538,10 +551,10 @@
}
ptrlong *
@@ -357,8 +359,21 @@ Last-Update: 2023-12-20
+regexp_offvect_to_array_of_long (utf8char *str, PCRE2_SIZE *offvect, int
result, int utf8_mode)
{
int i, idx_to_fill;
- int prev_ofs, ofs, prev_wide_len;
-@@ -648,11 +661,13 @@
+- int prev_ofs, ofs, prev_wide_len;
++ PCRE2_SIZE prev_ofs, ofs, prev_wide_len;
+ dk_set_t skipped_i = NULL;
+ ptrlong *ret_vec;
+ virt_mbstate_t mb;
+@@ -576,7 +589,7 @@
+ int next_nonnegative_ofs_i = i + 1;
+ while (next_nonnegative_ofs_i < result)
+ {
+- if (0 > offvect[next_nonnegative_ofs_i])
++ if (PCRE2_UNSET == offvect[next_nonnegative_ofs_i])
+ {
+ next_nonnegative_ofs_i++;
+ continue;
+@@ -647,11 +660,13 @@
int utf8_mode, utf8_mode2;
char *pattern = NULL;
char *str = NULL;
@@ -375,7 +390,7 @@ Last-Update: 2023-12-20
ptrlong *ret_vec = NULL;
dk_set_t ret_revlist = NULL;
-@@ -680,13 +695,16 @@
+@@ -679,13 +694,16 @@
if (NULL != err_ret[0])
goto done;
@@ -395,7 +410,7 @@ Last-Update: 2023-12-20
if (0 >= result)
break;
ret_vec = regexp_offvect_to_array_of_long ((utf8char *)str,
offvect, result, utf8_mode);
-@@ -699,13 +717,15 @@
+@@ -698,13 +716,15 @@
}
else
{
@@ -413,7 +428,7 @@ Last-Update: 2023-12-20
dk_free_tree (p_to_free);
dk_free_tree (str_to_free);
if (*err_ret)
-@@ -990,7 +1010,10 @@
+@@ -989,7 +1009,10 @@
static caddr_t
bif_regexp_version (caddr_t * qst, caddr_t * err_ret, state_slot_t ** args)
{
@@ -424,8 +439,36 @@ Last-Update: 2023-12-20
+ return box_dv_short_string(ver);
}
- void
-@@ -1013,21 +1036,26 @@
+ /*
+@@ -1199,10 +1222,10 @@
+ {
+ static compiled_regexp_t *cd_iri, *cd_abs, *cd_ref;
+ compiled_regexp_t *cd_info = NULL;
+- int r_opts = 0, c_opts = PCRE_UTF8 | PCRE_CASELESS;
+- int offvect[NOFFSETS];
++ uint32_t r_opts = 0, c_opts = PCRE2_UTF | PCRE2_CASELESS;
++ pcre2_match_data *md;
+ int result;
+- int str_len = (int) strlen (str);
++ PCRE2_SIZE str_len = strlen (str);
+ caddr_t pattern;
+
+ cd_info = (RX_IRI_REF == what ? cd_ref : (RX_IRI == what ? cd_iri :
cd_abs));
+@@ -1226,8 +1249,11 @@
+ break;
+ }
+ }
+- memset (offvect, -1, NOFFSETS * sizeof (int));
+- result = pcre_exec (cd_info->code, cd_info->code_x, str, str_len, 0,
r_opts, offvect, NOFFSETS);
++ create_match_context ();
++ md = pcre2_match_data_create (NOFFSETS, NULL);
++ result = pcre2_match (cd_info->code, (PCRE2_SPTR) str, str_len, 0, r_opts,
++ md, match_ctxt);
++ pcre2_match_data_free (md);
+ return (result != -1 ? 1 : 0);
+ }
+
+@@ -1268,21 +1294,26 @@
/* internal functions for internal usage in Virtuoso */
caddr_t
@@ -457,7 +500,7 @@ Last-Update: 2023-12-20
release_compiled_regexp (compiled_regexps, cd_info);
if (result != -1)
{
-@@ -1041,14 +1069,15 @@
+@@ -1296,14 +1327,15 @@
caddr_t
@@ -477,7 +520,7 @@ Last-Update: 2023-12-20
if (NULL == cd_info)
{
cd_info = get_compiled_regexp (compiled_regexps, pattern, c_opts, &err);
-@@ -1056,8 +1085,12 @@
+@@ -1311,8 +1343,12 @@
sqlr_resignal (err);
ret[0] = cd_info;
}
@@ -492,7 +535,7 @@ Last-Update: 2023-12-20
if (result != -1)
return box_dv_short_nchars (str + offvect[0], offvect[1] - offvect[0]);
return NULL;
-@@ -1068,21 +1101,21 @@
+@@ -1323,26 +1359,26 @@
char mc;
int opt;
} regexp_mode_table[] = {
@@ -520,7 +563,13 @@ Last-Update: 2023-12-20
if (!mode)
return 0;
while (mode_char[0])
-@@ -1117,18 +1150,24 @@
+ {
+- int i;
++ size_t i;
+ for (i=0;i<regexp_mode_table_l;i++)
+ {
+ if (regexp_mode_table[i].mc == mode_char[0])
+@@ -1372,18 +1408,24 @@
*/
int
@@ -550,7 +599,7 @@ Last-Update: 2023-12-20
release_compiled_regexp (compiled_regexps, cd_info);
return result;
}
-@@ -1142,11 +1181,12 @@
+@@ -1397,11 +1439,12 @@
*/
caddr_t
@@ -567,7 +616,7 @@ Last-Update: 2023-12-20
int result;
caddr_t ret_str;
caddr_t err = NULL;
-@@ -1154,15 +1194,19 @@
+@@ -1409,15 +1452,19 @@
if (err)
sqlr_resignal (err);
@@ -590,7 +639,7 @@ Last-Update: 2023-12-20
}
else
{
-@@ -1171,5 +1215,6 @@
+@@ -1426,5 +1473,6 @@
next[0] = -1;
}
release_compiled_regexp (compiled_regexps, cd_info);
@@ -599,7 +648,7 @@ Last-Update: 2023-12-20
}
--- virtuoso-opensource.orig/libsrc/Wi/srvstat.c
+++ virtuoso-opensource/libsrc/Wi/srvstat.c
-@@ -414,8 +414,8 @@
+@@ -419,8 +419,8 @@
extern int dbf_assert_on_malformed_data;
extern int dbf_max_itc_samples;
@@ -608,21 +657,22 @@ Last-Update: 2023-12-20
+extern uint32 c_pcre_match_limit;
+extern uint32 c_pcre_match_limit_recursion;
extern int32 pcre_max_cache_sz;
-
- extern int32 shcompo_max_cache_sz;
+ extern int64 users_cache_sz;
+ extern int32 enable_cpt_rb_ck;
--- virtuoso-opensource.orig/libsrc/Wi/sqlbif.h
+++ virtuoso-opensource/libsrc/Wi/sqlbif.h
-@@ -314,11 +314,11 @@
+@@ -324,12 +324,12 @@
extern caddr_t file_native_name_from_iri_path_nchars (const char *iri_path,
size_t iri_path_len);
caddr_t get_ssl_error_text (char *buf, int len);
-caddr_t regexp_match_01 (const char *pattern, const char *str, int c_opts);
-caddr_t regexp_match_01_const (const char* pattern, const char* str, int
c_opts, void ** compiled_ret);
++caddr_t regexp_match_01 (const char *pattern, const char *str, uint32 c_opts);
++caddr_t regexp_match_01_const (const char* pattern, const char* str, uint32_t
c_opts, void ** compiled_ret);
+ int regexp_match_iri_const (int what, const char* str, caddr_t *err_ret);
-caddr_t regexp_split_match (const char* pattern, const char* str, int* next,
int c_opts);
-int regexp_make_opts (const char* mode);
-int regexp_split_parse (const char* pattern, const char* str, int* offvect,
int offvect_sz, int c_opts);
-+caddr_t regexp_match_01 (const char *pattern, const char *str, uint32 c_opts);
-+caddr_t regexp_match_01_const (const char* pattern, const char* str, uint32_t
c_opts, void ** compiled_ret);
+caddr_t regexp_split_match (const char* pattern, const char* str, int* next,
uint32_t c_opts);
+uint32_t regexp_make_opts (const char* mode);
+int regexp_split_parse (const char* pattern, const char* str, size_t*
offvect, uint32_t offvect_sz, uint32_t c_opts);
@@ -641,7 +691,7 @@ Last-Update: 2023-12-20
#define ecm_isname(c) \
( ((c) & ~0xFF) ? (ecm_utf8props[(c)] & ECM_ISNAME) : \
-@@ -1540,7 +1541,7 @@
+@@ -1558,7 +1559,7 @@
static void
@@ -650,7 +700,7 @@ Last-Update: 2023-12-20
{
caddr_t pre_res;
int next;
-@@ -1552,13 +1553,13 @@
+@@ -1570,13 +1571,13 @@
sqlr_new_error ("42001", "XRQ??", "invalid regular expression");
}
@@ -667,7 +717,7 @@ Last-Update: 2023-12-20
return c_opts;
}
-@@ -1570,7 +1571,7 @@
+@@ -1588,7 +1589,7 @@
int next = 1;
caddr_t str_inx = str;
dk_set_t res_set = 0;
@@ -676,7 +726,7 @@ Last-Update: 2023-12-20
c_opts=xqf_make_regexp_modes (flag);
xqf_check_regexp (pattern, c_opts);
-@@ -1628,7 +1629,7 @@
+@@ -1646,7 +1647,7 @@
xqf_matches (xp_instance_t * xqi, XT * tree, xml_entity_t * ctx_xe)
{
caddr_t val1, val2, val3 = NULL;
@@ -685,7 +735,7 @@ Last-Update: 2023-12-20
if (tree->_.xp_func.argcount)
{
-@@ -2003,7 +2004,7 @@
+@@ -2021,7 +2022,7 @@
#define XQF_REPL_OK 0
static int
@@ -694,7 +744,7 @@ Last-Update: 2023-12-20
{
int repl_sz = box_length (replacement) - 1;
int idx = 0;
-@@ -2042,7 +2043,7 @@
+@@ -2060,7 +2061,7 @@
caddr_t pattern = xpf_arg (xqi, tree, ctx_xe, DV_STRING, 1);
caddr_t replacement = xpf_arg (xqi, tree, ctx_xe, DV_STRING, 2);
caddr_t flag = 0;
@@ -703,7 +753,7 @@ Last-Update: 2023-12-20
if (tree->_.xp_func.argcount > 3)
flag = xpf_arg (xqi, tree, ctx_xe, DV_STRING, 3);
-@@ -2050,7 +2051,7 @@
+@@ -2068,7 +2069,7 @@
xqf_check_regexp (pattern, c_opts);
{
diff --git a/debian/patches/series b/debian/patches/series
index 395a224ed..0c9e456f6 100644
--- a/debian/patches/series
+++ b/debian/patches/series
@@ -21,4 +21,4 @@ cil.patch
wbxml2.patch
tirpc.patch
reproducible.patch
-#pcre2.patch
+pcre2.patch
--
2.43.0
