Ticket #4655: remove PCRE and PCRE2 search engines.

GRegex as PCRE wrapper was introduced in GLib-2.14. PCRE was required for MC on OSes with GLib < 2.14. Currently MC requires GLib >= 2.32. Since 2.73.2 (released at 2022-07-08), GRegex is based on PCRE2. We can remove PCRE and PCRE2 from MC and use GLib only. Signed-off-by: Andrew Borodin <[email protected]>
MidnightCommander · Feb 22, 2025 · 18c0c2f · 18c0c2f
1 parent 9f020f7
commit 18c0c2f
Show file tree

Hide file tree

Showing 11 changed files with 8 additions and 462 deletions.
diff --git a/.github/workflows/ci-ubuntu.yml b/.github/workflows/ci-ubuntu.yml
@@ -14,7 +14,7 @@ jobs:
       - name: Install dependencies
         run: |
           sudo apt-get install -y autoconf autopoint check gettext libtool pkg-config
-          sudo apt-get install -y e2fslibs-dev libaspell-dev libglib2.0-dev libgpm-dev libncurses5-dev libpcre2-dev libslang2-dev libssh2-1-dev libx11-dev unzip
+          sudo apt-get install -y e2fslibs-dev libaspell-dev libglib2.0-dev libgpm-dev libncurses5-dev libslang2-dev libssh2-1-dev libx11-dev unzip
 
       - name: Bootstrap build system
         run: ./autogen.sh
@@ -43,15 +43,14 @@ jobs:
           make check
           make install
 
-      - name: Build ncurses & pcre2 configuration
+      - name: Build ncurses configuration
         run: |
           tar -xjf $(ls build-distrib/mc-*.tar.bz2) --one-top-level=build-ncurses
           cd build-ncurses
 
           ../configure \
               --prefix="$(pwd)/install" \
               --with-screen=ncurses \
-              --with-search-engine=pcre2 \
               --enable-werror
 
           make -j$(nproc)

diff --git a/acinclude.m4 b/acinclude.m4
@@ -6,8 +6,6 @@ m4_include([m4.include/gnulib/fsusage.m4])
 m4_include([m4.include/gnulib/mountlist.m4])
 m4_include([m4.include/gnulib/windows-stat-inodes.m4])
 m4_include([m4.include/gnulib/sys_types_h.m4])
-m4_include([m4.include/ax_path_lib_pcre.m4])
-m4_include([m4.include/ax_check_pcre2.m4])
 m4_include([m4.include/dx_doxygen.m4])
 m4_include([m4.include/ax_require_defined.m4])
 m4_include([m4.include/ax_check_compile_flag.m4])
@@ -16,7 +14,6 @@ m4_include([m4.include/ax_append_compile_flags.m4])
 m4_include([m4.include/mc-cflags.m4])
 m4_include([m4.include/mc-stdckdint.m4])
 m4_include([m4.include/ax_gcc_func_attribute.m4])
-m4_include([m4.include/mc-check-search-type.m4])
 m4_include([m4.include/mc-get-fs-info.m4])
 m4_include([m4.include/mc-with-x.m4])
 m4_include([m4.include/mc-use-termcap.m4])

diff --git a/configure.ac b/configure.ac
@@ -228,7 +228,6 @@ dnl ############################################################################
 mc_CHECK_GLIB
 mc_G_MODULE_SUPPORTED
 mc_WITH_SCREEN
-mc_CHECK_SEARCH_TYPE
 dnl X11 support. Used to read keyboard modifiers when running under X11.
 mc_WITH_X
 
@@ -751,7 +750,6 @@ Configuration:
   Internal editor:                ${edit_msg}
   Diff viewer:                    ${diff_msg}
   Support for charset:            ${charset_msg}
-  Search type:                    ${SEARCH_TYPE}
 ])
 
 dnl option checking is disable by default due to AC_CONFIG_SUBDIRS

diff --git a/doc/INSTALL b/doc/INSTALL
@@ -25,8 +25,6 @@ Build requirements for GNU Midnight Commander
   - https://invisible-island.net/ncurses/ncurses.html
 - gettext >= 0.18.2
   - https://www.gnu.org/software/gettext/
-- pcre or pcre2 (as an alternative to glib-regexp)
-  - https://www.pcre.org
 - gpm (general purpose mouse daemon as an alternative to xterm mouse)
   - https://www.nico.schottelius.org/software/gpm/
 - libssh2 >= 1.2.8 (required only for sftp vfs)

diff --git a/lib/search.h b/lib/search.h
@@ -7,14 +7,6 @@
 
 #include <sys/types.h>
 
-#ifdef SEARCH_TYPE_PCRE
-#    ifdef HAVE_PCRE2
-#        define PCRE2_CODE_UNIT_WIDTH 8
-#        include <pcre2.h>
-#    else
-#        include <pcre.h>
-#    endif
-#endif
 /*** typedefs(not structures) and defined constants **********************************************/
 
 typedef enum mc_search_cbret_t mc_search_cbret_t;
@@ -25,19 +17,6 @@ typedef mc_search_cbret_t (*mc_update_fn) (const void *user_data, off_t char_off
 
 #define MC_SEARCH__NUM_REPLACE_ARGS 64
 
-#ifdef SEARCH_TYPE_GLIB
-#    define mc_search_matchinfo_t GMatchInfo
-#else
-#    ifdef HAVE_PCRE2
-/* no pcre_extra in PCRE2. pcre2_jit_compile (equivalent of pcre_study) handles
- * all of this internally. but we can use this to hold the pcre2_matches data
- * until the search is complete */
-#        define mc_search_matchinfo_t pcre2_match_data
-#    else
-#        define mc_search_matchinfo_t pcre_extra
-#    endif
-#endif
-
 /*** enums ***************************************************************************************/
 
 typedef enum
@@ -122,17 +101,8 @@ typedef struct mc_search_struct
     // some data for regexp
     int num_results;
     gboolean is_utf8;
-    mc_search_matchinfo_t *regex_match_info;
+    GMatchInfo *regex_match_info;
     GString *regex_buffer;
-#ifdef SEARCH_TYPE_PCRE
-#    ifdef HAVE_PCRE2
-    // pcre2 will provide a pointer to a match_data structure that can be manipulated like an
-    // iovector
-    size_t *iovector;
-#    else
-    int iovector[MC_SEARCH__NUM_REPLACE_ARGS * 2];
-#    endif
-#endif
 
     // private data
 

diff --git a/lib/search/internal.h b/lib/search/internal.h
@@ -3,16 +3,6 @@
 
 /*** typedefs(not structures) and defined constants **********************************************/
 
-#ifdef SEARCH_TYPE_GLIB
-#    define mc_search_regex_t GRegex
-#else
-#    ifdef HAVE_PCRE2
-#        define mc_search_regex_t pcre2_code
-#    else
-#        define mc_search_regex_t pcre
-#    endif
-#endif
-
 /*** enums ***************************************************************************************/
 
 typedef enum
@@ -32,7 +22,7 @@ typedef struct mc_search_cond_struct
     GString *str;
     GString *upper;
     GString *lower;
-    mc_search_regex_t *regex_handle;
+    GRegex *regex_handle;
     gchar *charset;
 } mc_search_cond_t;
 

diff --git a/lib/search/regex.c b/lib/search/regex.c
@@ -269,7 +269,6 @@ mc_search__cond_struct_new_regex_ci_str (const char *charset, const GString *ast
 
 /* --------------------------------------------------------------------------------------------- */
 
-#ifdef SEARCH_TYPE_GLIB
 /* A thin wrapper above g_regex_match_full that makes sure the string passed
  * to it is valid UTF-8 (unless G_REGEX_RAW compile flag was set), as it is a
  * requirement by glib and it might crash otherwise. See: mc ticket 3449.
@@ -319,15 +318,12 @@ mc_search__g_regex_match_full_safe (const GRegex *regex, const gchar *string, gs
     g_free (string_safe);
     return ret;
 }
-#endif
 
 /* --------------------------------------------------------------------------------------------- */
 
 static mc_search__found_cond_t
-mc_search__regex_found_cond_one (mc_search_t *lc_mc_search, mc_search_regex_t *regex,
-                                 GString *search_str)
+mc_search__regex_found_cond_one (mc_search_t *lc_mc_search, GRegex *regex, GString *search_str)
 {
-#ifdef SEARCH_TYPE_GLIB
     GError *mcerror = NULL;
 
     if (!mc_search__g_regex_match_full_safe (regex, search_str->str, search_str->len, 0,
@@ -348,21 +344,7 @@ mc_search__regex_found_cond_one (mc_search_t *lc_mc_search, mc_search_regex_t *r
         return COND__NOT_FOUND;
     }
     lc_mc_search->num_results = g_match_info_get_match_count (lc_mc_search->regex_match_info);
-#else  // SEARCH_TYPE_GLIB
-
-    lc_mc_search->num_results =
-#    ifdef HAVE_PCRE2
-        pcre2_match (regex, (unsigned char *) search_str->str, search_str->len, 0, 0,
-                     lc_mc_search->regex_match_info, NULL);
-#    else
-        pcre_exec (regex, lc_mc_search->regex_match_info, search_str->str, search_str->len, 0, 0,
-                   lc_mc_search->iovector, MC_SEARCH__NUM_REPLACE_ARGS);
-#    endif
-    if (lc_mc_search->num_results < 0)
-    {
-        return COND__NOT_FOUND;
-    }
-#endif
+
     return COND__FOUND_OK;
 }
 
@@ -443,12 +425,7 @@ mc_search_regex__get_token_by_num (const mc_search_t *lc_mc_search, gsize lc_ind
 {
     int fnd_start = 0, fnd_end = 0;
 
-#ifdef SEARCH_TYPE_GLIB
     g_match_info_fetch_pos (lc_mc_search->regex_match_info, lc_index, &fnd_start, &fnd_end);
-#else  // SEARCH_TYPE_GLIB
-    fnd_start = lc_mc_search->iovector[lc_index * 2 + 0];
-    fnd_end = lc_mc_search->iovector[lc_index * 2 + 1];
-#endif
 
     if (fnd_end == fnd_start)
         return g_strdup ("");
@@ -807,7 +784,6 @@ mc_search__cond_struct_new_init_regex (const char *charset, mc_search_t *lc_mc_s
     }
 
     {
-#ifdef SEARCH_TYPE_GLIB
         GError *mcerror = NULL;
         GRegexCompileFlags g_regex_options = G_REGEX_OPTIMIZE | G_REGEX_DOTALL;
 
@@ -842,67 +818,6 @@ mc_search__cond_struct_new_init_regex (const char *charset, mc_search_t *lc_mc_s
             g_error_free (mcerror);
             return;
         }
-#else  // SEARCH_TYPE_GLIB
-
-#    ifdef HAVE_PCRE2
-        int errcode;
-        char error[BUF_SMALL] = "";
-        size_t erroffset;
-        int pcre_options = PCRE2_MULTILINE;
-#    else
-        const char *error;
-        int erroffset;
-        int pcre_options = PCRE_EXTRA | PCRE_MULTILINE;
-#    endif
-
-        if (str_isutf8 (charset) && mc_global.utf8_display)
-        {
-#    ifdef HAVE_PCRE2
-            pcre_options |= PCRE2_UTF;
-            if (!lc_mc_search->is_case_sensitive)
-                pcre_options |= PCRE2_CASELESS;
-#    else
-            pcre_options |= PCRE_UTF8;
-            if (!lc_mc_search->is_case_sensitive)
-                pcre_options |= PCRE_CASELESS;
-#    endif
-        }
-        else if (!lc_mc_search->is_case_sensitive)
-        {
-            GString *tmp;
-
-            tmp = mc_search_cond->str;
-            mc_search_cond->str = mc_search__cond_struct_new_regex_ci_str (charset, tmp);
-            g_string_free (tmp, TRUE);
-        }
-
-        mc_search_cond->regex_handle =
-#    ifdef HAVE_PCRE2
-            pcre2_compile ((unsigned char *) mc_search_cond->str->str, PCRE2_ZERO_TERMINATED,
-                           pcre_options, &errcode, &erroffset, NULL);
-#    else
-            pcre_compile (mc_search_cond->str->str, pcre_options, &error, &erroffset, NULL);
-#    endif
-        if (mc_search_cond->regex_handle == NULL)
-        {
-#    ifdef HAVE_PCRE2
-            pcre2_get_error_message (errcode, (unsigned char *) error, sizeof (error));
-#    endif
-            mc_search_set_error (lc_mc_search, MC_SEARCH_E_REGEX_COMPILE, "%s", error);
-            return;
-        }
-#    ifdef HAVE_PCRE2
-        if (pcre2_jit_compile (mc_search_cond->regex_handle, PCRE2_JIT_COMPLETE) && *error != '\0')
-#    else
-        lc_mc_search->regex_match_info = pcre_study (mc_search_cond->regex_handle, 0, &error);
-        if (lc_mc_search->regex_match_info == NULL && error != NULL)
-#    endif
-        {
-            mc_search_set_error (lc_mc_search, MC_SEARCH_E_REGEX_COMPILE, "%s", error);
-            MC_PTR_FREE (mc_search_cond->regex_handle);
-            return;
-        }
-#endif
     }
 
     lc_mc_search->is_utf8 = str_isutf8 (charset);
@@ -985,12 +900,7 @@ mc_search__run_regex (mc_search_t *lc_mc_search, const void *user_data, off_t st
         switch (mc_search__regex_found_cond (lc_mc_search, lc_mc_search->regex_buffer))
         {
         case COND__FOUND_OK:
-#ifdef SEARCH_TYPE_GLIB
             g_match_info_fetch_pos (lc_mc_search->regex_match_info, 0, &start_pos, &end_pos);
-#else  // SEARCH_TYPE_GLIB
-            start_pos = lc_mc_search->iovector[0];
-            end_pos = lc_mc_search->iovector[1];
-#endif
             if (found_len != NULL)
                 *found_len = end_pos - start_pos;
             lc_mc_search->normal_offset = lc_mc_search->start_buffer + start_pos;

diff --git a/lib/search/search.c b/lib/search/search.c
@@ -71,10 +71,7 @@ mc_search__cond_struct_new (mc_search_t *lc_mc_search, const GString *str, const
     mc_search_cond = g_malloc0 (sizeof (mc_search_cond_t));
     mc_search_cond->str = mc_g_string_dup (str);
     mc_search_cond->charset = g_strdup (charset);
-#ifdef HAVE_PCRE2
-    lc_mc_search->regex_match_info = pcre2_match_data_create (MC_SEARCH__NUM_REPLACE_ARGS, NULL);
-    lc_mc_search->iovector = pcre2_get_ovector_pointer (lc_mc_search->regex_match_info);
-#endif
+
     switch (lc_mc_search->search_type)
     {
     case MC_SEARCH_T_GLOB:
@@ -111,12 +108,8 @@ mc_search__cond_struct_free (gpointer data)
     g_string_free (mc_search_cond->str, TRUE);
     g_free (mc_search_cond->charset);
 
-#ifdef SEARCH_TYPE_GLIB
     if (mc_search_cond->regex_handle != NULL)
         g_regex_unref (mc_search_cond->regex_handle);
-#else  // SEARCH_TYPE_GLIB
-    g_free (mc_search_cond->regex_handle);
-#endif
 
     g_free (mc_search_cond);
 }
@@ -188,12 +181,8 @@ mc_search_free (mc_search_t *lc_mc_search)
     if (lc_mc_search->prepared.conditions != NULL)
         g_ptr_array_free (lc_mc_search->prepared.conditions, TRUE);
 
-#ifdef SEARCH_TYPE_GLIB
     if (lc_mc_search->regex_match_info != NULL)
         g_match_info_free (lc_mc_search->regex_match_info);
-#else  // SEARCH_TYPE_GLIB
-    g_free (lc_mc_search->regex_match_info);
-#endif
 
     if (lc_mc_search->regex_buffer != NULL)
         g_string_free (lc_mc_search->regex_buffer, TRUE);
@@ -282,13 +271,12 @@ mc_search_run (mc_search_t *lc_mc_search, const void *user_data, off_t start_sea
         mc_search_set_error (lc_mc_search, MC_SEARCH_E_INPUT, "%s", _ (STR_E_UNKNOWN_TYPE));
         return FALSE;
     }
-#ifdef SEARCH_TYPE_GLIB
+
     if (lc_mc_search->regex_match_info != NULL)
     {
         g_match_info_free (lc_mc_search->regex_match_info);
         lc_mc_search->regex_match_info = NULL;
     }
-#endif
 
     mc_search_set_error (lc_mc_search, MC_SEARCH_E_OK, NULL);
 
@@ -455,17 +443,13 @@ mc_search_getstart_result_by_num (mc_search_t *lc_mc_search, int lc_index)
         return 0;
     if (lc_mc_search->search_type == MC_SEARCH_T_NORMAL)
         return 0;
-#ifdef SEARCH_TYPE_GLIB
     {
         gint start_pos;
         gint end_pos;
 
         g_match_info_fetch_pos (lc_mc_search->regex_match_info, lc_index, &start_pos, &end_pos);
         return (int) start_pos;
     }
-#else  // SEARCH_TYPE_GLIB
-    return lc_mc_search->iovector[lc_index * 2];
-#endif
 }
 
 /* --------------------------------------------------------------------------------------------- */
@@ -477,17 +461,13 @@ mc_search_getend_result_by_num (mc_search_t *lc_mc_search, int lc_index)
         return 0;
     if (lc_mc_search->search_type == MC_SEARCH_T_NORMAL)
         return 0;
-#ifdef SEARCH_TYPE_GLIB
     {
         gint start_pos;
         gint end_pos;
 
         g_match_info_fetch_pos (lc_mc_search->regex_match_info, lc_index, &start_pos, &end_pos);
         return (int) end_pos;
     }
-#else  // SEARCH_TYPE_GLIB
-    return lc_mc_search->iovector[lc_index * 2 + 1];
-#endif
 }
 
 /* --------------------------------------------------------------------------------------------- */