[pcre2pp] allocate match_data on the stack

src/base/ansi_scrubber.cc

@@ -51,8 +51,9 @@ ansi_regex()
 size_t
 erase_ansi_escapes(string_fragment input)
 {
+    static thread_local auto md = lnav::pcre2pp::match_data::unitialized();
+
     const auto& regex = ansi_regex();
-    auto md = regex.create_match_data();
 
     auto matcher = regex.capture_from(input).into(md);
     while (true) {
@@ -113,8 +114,8 @@ erase_ansi_escapes(string_fragment input)
 void
 scrub_ansi_string(std::string& str, string_attrs_t* sa)
 {
+    static thread_local auto md = lnav::pcre2pp::match_data::unitialized();
     const auto& regex = ansi_regex();
-    auto md = regex.create_match_data();
     int64_t origin_offset = 0;
     int last_origin_offset_end = 0;
 

src/log_format.cc

@@ -312,7 +312,8 @@ log_format::log_scanf(uint32_t line_number,
     int pat_index = this->last_pattern_index();
 
     while (!done && next_format(fmt, curr_fmt, pat_index)) {
-        auto md = fmt[curr_fmt].pcre->create_match_data();
+        static thread_local auto md = lnav::pcre2pp::match_data::unitialized();
+
         auto match_res = fmt[curr_fmt]
                              .pcre->capture_from(line)
                              .into(md)
@@ -802,6 +803,8 @@ external_log_format::scan(logfile& lf,
     auto line_sf = sbr.to_string_fragment();
 
     while (::next_format(this->elf_pattern_order, curr_fmt, pat_index)) {
+        static thread_local auto md = lnav::pcre2pp::match_data::unitialized();
+
         auto* fpat = this->elf_pattern_order[curr_fmt].get();
         auto* pat = fpat->p_pcre.pp_value.get();
 
@@ -809,7 +812,6 @@ external_log_format::scan(logfile& lf,
             continue;
         }
 
-        auto md = pat->create_match_data();
         auto match_res = pat->capture_from(line_sf)
                              .into(md)
                              .matches(PCRE2_NO_UTF_CHECK)
@@ -907,13 +909,15 @@ external_log_format::scan(logfile& lf,
                     mod_iter->second.mf_mod_format);
 
                 if (mod_elf) {
+                    static thread_local auto mod_md
+                        = lnav::pcre2pp::match_data::unitialized();
+
                     shared_buffer_ref body_ref;
 
                     body_cap->trim();
 
                     int mod_pat_index = mod_elf->last_pattern_index();
                     auto& mod_pat = *mod_elf->elf_pattern_order[mod_pat_index];
-                    auto mod_md = mod_pat.p_pcre.pp_value->create_match_data();
                     auto match_res = mod_pat.p_pcre.pp_value
                                          ->capture_from(body_cap.value())
                                          .into(mod_md)
@@ -1014,6 +1018,8 @@ external_log_format::module_scan(string_fragment body_cap,
         int curr_fmt = -1, fmt_lock = -1;
 
         while (::next_format(elf->elf_pattern_order, curr_fmt, fmt_lock)) {
+            static thread_local auto md = lnav::pcre2pp::match_data::unitialized();
+
             auto& fpat = elf->elf_pattern_order[curr_fmt];
             auto& pat = fpat->p_pcre;
 
@@ -1021,7 +1027,6 @@ external_log_format::module_scan(string_fragment body_cap,
                 continue;
             }
 
-            auto md = pat.pp_value->create_match_data();
             auto match_res = pat.pp_value->capture_from(body_cap)
                                  .into(md)
                                  .matches(PCRE2_NO_UTF_CHECK)
@@ -1054,6 +1059,8 @@ external_log_format::annotate(uint64_t line_number,
                               logline_value_vector& values,
                               bool annotate_module) const
 {
+    static thread_local auto md = lnav::pcre2pp::match_data::unitialized();
+
     auto& line = values.lvv_sbr;
     struct line_range lr;
 
@@ -1074,7 +1081,6 @@ external_log_format::annotate(uint64_t line_number,
     auto& pat = *this->elf_pattern_order[pat_index];
 
     sa.reserve(pat.p_pcre.pp_value->get_capture_count());
-    auto md = pat.p_pcre.pp_value->create_match_data();
     auto match_res
         = pat.p_pcre.pp_value->capture_from(line.to_string_fragment())
               .into(md)

src/pcrepp/pcre2pp.cc

@@ -55,6 +55,20 @@ quote(const char* unquoted)
     return retval;
 }
 
+matcher
+capture_builder::into(lnav::pcre2pp::match_data& md) &&
+{
+    if (md.get_capacity() < this->mb_code.get_match_data_capacity()) {
+        md = this->mb_code.create_match_data();
+    }
+
+    return matcher{
+        this->mb_code,
+        this->mb_input,
+        md,
+    };
+}
+
 match_data
 code::create_match_data() const
 {

src/pcrepp/pcre2pp.hh

@@ -106,6 +106,8 @@ public:
 
     int get_count() const { return this->md_capture_end; }
 
+    uint32_t get_capacity() const { return this->md_ovector_count; }
+
 private:
     friend matcher;
     friend code;
@@ -195,14 +197,7 @@ struct capture_builder {
         return *this;
     }
 
-    matcher into(match_data& md) &&
-    {
-        return matcher{
-            this->mb_code,
-            this->mb_input,
-            md,
-        };
-    }
+    matcher into(match_data& md) &&;
 
     template<uint32_t Options = 0, typename F>
     Result<string_fragment, matcher::error> for_each(F func) &&;
@@ -274,6 +269,10 @@ public:
 
     std::vector<string_fragment> get_captures() const;
 
+    uint32_t get_match_data_capacity() const {
+        return this->p_match_proto.md_ovector_count;
+    }
+
     match_data create_match_data() const;
 
     capture_builder capture_from(string_fragment in) const

src/regex101.import.cc

@@ -54,7 +54,7 @@ regex101::import(const std::string& url,
 {
     static const auto USER_URL = lnav::pcre2pp::code::from_const(
         R"(^https://regex101.com/r/(\w+)(?:/(\d+))?)");
-    static thread_local auto URL_MATCH_DATA = USER_URL.create_match_data();
+    static thread_local auto md = lnav::pcre2pp::match_data::unitialized();
     static const auto NAME_RE = lnav::pcre2pp::code::from_const(R"(^\w+$)");
 
     if (url.empty()) {
@@ -100,10 +100,8 @@ regex101::import(const std::string& url,
                                .append("^ matched up to here"_comment)));
     }
 
-    auto user_find_res = USER_URL.capture_from(url)
-                             .into(URL_MATCH_DATA)
-                             .matches()
-                             .ignore_error();
+    auto user_find_res
+        = USER_URL.capture_from(url).into(md).matches().ignore_error();
     if (!user_find_res) {
         auto partial_len = USER_URL.match_partial(url);
         return Err(lnav::console::user_message::error(
@@ -118,7 +116,7 @@ regex101::import(const std::string& url,
                                       .append("^ matched up to here"_comment)));
     }
 
-    auto permalink = URL_MATCH_DATA[1]->to_string();
+    auto permalink = md[1]->to_string();
 
     auto format_filename = existing_format
         ? fmt::format(FMT_STRING("{}.regex101-{}.json"), name, permalink)

src/relative_time.cc

@@ -294,8 +294,9 @@ relative_time::from_str(string_fragment str)
 
         bool found = false;
         for (int lpc = 0; lpc < RTT__MAX && !found; lpc++) {
+            static thread_local auto md = lnav::pcre2pp::match_data::unitialized();
+
             token_t token = (token_t) lpc;
-            auto md = MATCHERS[lpc].pcre.create_match_data();
             auto match_res = MATCHERS[lpc]
                                  .pcre.capture_from(remaining)
                                  .into(md)

src/text_anonymizer.cc

@@ -467,9 +467,11 @@ text_anonymizer::next(string_fragment line)
                 } else {
                     static const auto ATTR_RE
                         = lnav::pcre2pp::code::from_const(R"([\w\-]+=)");
+                    static thread_local auto md
+                        = lnav::pcre2pp::match_data::unitialized();
+
                     auto remaining = string_fragment::from_str_range(
                         open_tag, space_index, open_tag.size());
-                    auto md = ATTR_RE.create_match_data();
 
                     retval += open_tag.substr(0, space_index + 1);
                     while (!remaining.empty()) {

src/yajlpp/yajlpp.cc

@@ -224,7 +224,9 @@ json_path_handler_base::gen(yajlpp_gen_context& ygc, yajl_gen handle) const
             ygc.ygc_depth += 1;
 
             if (this->jph_obj_provider) {
-                auto md = this->jph_regex->create_match_data();
+                static thread_local auto md
+                    = lnav::pcre2pp::match_data::unitialized();
+
                 auto find_res = this->jph_regex->capture_from(full_path)
                                     .into(md)
                                     .matches();
@@ -504,7 +506,9 @@ json_path_handler_base::walk(
 
                 ypc.set_path(full_path).with_obj(root).update_callbacks();
                 if (this->jph_obj_provider) {
-                    auto md = this->jph_regex->create_match_data();
+                    static thread_local auto md
+                        = lnav::pcre2pp::match_data::unitialized();
+
                     std::string full_path = lpath + "/";
 
                     if (!this->jph_regex->capture_from(full_path)
@@ -707,7 +711,8 @@ yajlpp_parse_context::update_callbacks(const json_path_container* orig_handlers,
     auto path_frag = string_fragment::from_byte_range(
         this->ypc_path.data(), 1 + child_start, this->ypc_path.size() - 1);
     for (const auto& jph : handlers->jpc_children) {
-        auto md = jph.jph_regex->create_match_data();
+        static thread_local auto md = lnav::pcre2pp::match_data::unitialized();
+
         if (jph.jph_regex->capture_from(path_frag)
                 .into(md)
                 .matches()
@@ -945,8 +950,8 @@ yajlpp_parse_context::handle_unused_or_delete(void* ctx)
     if (!ypc->ypc_handler_stack.empty()
         && ypc->ypc_handler_stack.back()->jph_obj_deleter)
     {
-        auto& jph = ypc->ypc_handler_stack.back();
-        auto md = jph->jph_regex->create_match_data();
+        static thread_local auto md = lnav::pcre2pp::match_data::unitialized();
+
         auto key_start = ypc->ypc_path_index_stack.back();
         auto path_frag = string_fragment::from_byte_range(
             ypc->ypc_path.data(), key_start + 1, ypc->ypc_path.size() - 1);

test/expected/expected.am

@@ -164,6 +164,8 @@ EXPECTED_FILES = \
     $(srcdir)/%reldir%/test_cmds.sh_be1d9628fc447b6f17121d9457ea1602afe8f3f3.out \
     $(srcdir)/%reldir%/test_cmds.sh_be3b7c5874b5f4d86cc230bd2f9802c98909e148.err \
     $(srcdir)/%reldir%/test_cmds.sh_be3b7c5874b5f4d86cc230bd2f9802c98909e148.out \
+    $(srcdir)/%reldir%/test_cmds.sh_bf4e7fad67e281beaa11b6e2b03a00b419c7c9b0.err \
+    $(srcdir)/%reldir%/test_cmds.sh_bf4e7fad67e281beaa11b6e2b03a00b419c7c9b0.out \
     $(srcdir)/%reldir%/test_cmds.sh_c01e10f7cae8d36fa79ae03be887cb5477025f6d.err \
     $(srcdir)/%reldir%/test_cmds.sh_c01e10f7cae8d36fa79ae03be887cb5477025f6d.out \
     $(srcdir)/%reldir%/test_cmds.sh_c2b4431dd0cc36c6201d263b727b3305e8cda6b1.err \
@@ -974,6 +976,8 @@ EXPECTED_FILES = \
     $(srcdir)/%reldir%/test_sql_xml_func.sh_fefeb387ae14d4171225ea06cbbff3ec43990cf0.out \
     $(srcdir)/%reldir%/test_sql_yaml_func.sh_41c6abde708a69e74f5b7fde865d88fa75f91e0a.err \
     $(srcdir)/%reldir%/test_sql_yaml_func.sh_41c6abde708a69e74f5b7fde865d88fa75f91e0a.out \
+    $(srcdir)/%reldir%/test_sql_yaml_func.sh_dc189d02e8979b7ed245d5d750f68b9965984699.err \
+    $(srcdir)/%reldir%/test_sql_yaml_func.sh_dc189d02e8979b7ed245d5d750f68b9965984699.out \
     $(srcdir)/%reldir%/test_text_file.sh_5b51b55dff7332c5bee2c9b797c401c5614d574a.err \
     $(srcdir)/%reldir%/test_text_file.sh_5b51b55dff7332c5bee2c9b797c401c5614d574a.out \
     $(srcdir)/%reldir%/test_text_file.sh_6a24078983cf1b7a80b6fb65d5186cd125498136.err \

test/expected/test_cmds.sh_bf4e7fad67e281beaa11b6e2b03a00b419c7c9b0.err

@@ -0,0 +1,7 @@
+✘ error: invalid timestamp: 2022-06-16Tabc
+ reason: the leading part of the timestamp was matched, however, the trailing text “Tabc” was not
+ --> command-option:1
+ | :goto 2022-06-16Tabc                    
+ |                 ^--^ unrecognized input 
+ = note: input matched time format “%Y-%m-%d”
+ = help: fix the timestamp or remove the trailing text

test/expected/test_sql_yaml_func.sh_dc189d02e8979b7ed245d5d750f68b9965984699.err

@@ -0,0 +1,4 @@
+✘ error: failed to parse YAML content
+ reason: closing ] not found
+ --> command-option:1
+ | ;SELECT yaml_to_json('[abc')