Merge branch 'sb/submodule-core-worktree'
[gitweb.git] / grep.c
diff --git a/grep.c b/grep.c
index 56ef0ecbfffd8531a5b88585efc0c9097e182ed6..7e2f440955a01ae305664df37f77ed87a55c2b06 100644 (file)
--- a/grep.c
+++ b/grep.c
@@ -1,22 +1,41 @@
 #include "cache.h"
+#include "config.h"
 #include "grep.h"
+#include "object-store.h"
 #include "userdiff.h"
 #include "xdiff-interface.h"
 #include "diff.h"
 #include "diffcore.h"
 #include "commit.h"
 #include "quote.h"
+#include "help.h"
 
 static int grep_source_load(struct grep_source *gs);
 static int grep_source_is_binary(struct grep_source *gs);
 
 static struct grep_opt grep_defaults;
 
+static const char *color_grep_slots[] = {
+       [GREP_COLOR_CONTEXT]        = "context",
+       [GREP_COLOR_FILENAME]       = "filename",
+       [GREP_COLOR_FUNCTION]       = "function",
+       [GREP_COLOR_LINENO]         = "lineNumber",
+       [GREP_COLOR_MATCH_CONTEXT]  = "matchContext",
+       [GREP_COLOR_MATCH_SELECTED] = "matchSelected",
+       [GREP_COLOR_SELECTED]       = "selected",
+       [GREP_COLOR_SEP]            = "separator",
+};
+
 static void std_output(struct grep_opt *opt, const void *buf, size_t size)
 {
        fwrite(buf, size, 1, stdout);
 }
 
+static void color_set(char *dst, const char *color_bytes)
+{
+       xsnprintf(dst, COLOR_MAXLEN, "%s", color_bytes);
+}
+
 /*
  * Initialize the grep_defaults template with hardcoded defaults.
  * We could let the compiler do this, but without C99 initializers
@@ -34,18 +53,16 @@ void init_grep_defaults(void)
        memset(opt, 0, sizeof(*opt));
        opt->relative = 1;
        opt->pathname = 1;
-       opt->regflags = REG_NEWLINE;
        opt->max_depth = -1;
        opt->pattern_type_option = GREP_PATTERN_TYPE_UNSPECIFIED;
-       opt->extended_regexp_option = 0;
-       color_set(opt->color_context, "");
-       color_set(opt->color_filename, "");
-       color_set(opt->color_function, "");
-       color_set(opt->color_lineno, "");
-       color_set(opt->color_match_context, GIT_COLOR_BOLD_RED);
-       color_set(opt->color_match_selected, GIT_COLOR_BOLD_RED);
-       color_set(opt->color_selected, "");
-       color_set(opt->color_sep, GIT_COLOR_CYAN);
+       color_set(opt->colors[GREP_COLOR_CONTEXT], "");
+       color_set(opt->colors[GREP_COLOR_FILENAME], "");
+       color_set(opt->colors[GREP_COLOR_FUNCTION], "");
+       color_set(opt->colors[GREP_COLOR_LINENO], "");
+       color_set(opt->colors[GREP_COLOR_MATCH_CONTEXT], GIT_COLOR_BOLD_RED);
+       color_set(opt->colors[GREP_COLOR_MATCH_SELECTED], GIT_COLOR_BOLD_RED);
+       color_set(opt->colors[GREP_COLOR_SELECTED], "");
+       color_set(opt->colors[GREP_COLOR_SEP], GIT_COLOR_CYAN);
        opt->color = -1;
        opt->output = std_output;
 }
@@ -65,6 +82,8 @@ static int parse_pattern_type_arg(const char *opt, const char *arg)
        die("bad %s argument: %s", opt, arg);
 }
 
+define_list_config_array_extra(color_grep_slots, {"match"});
+
 /*
  * Read the configuration file once and store it in
  * the grep_defaults template.
@@ -72,16 +91,13 @@ static int parse_pattern_type_arg(const char *opt, const char *arg)
 int grep_config(const char *var, const char *value, void *cb)
 {
        struct grep_opt *opt = &grep_defaults;
-       char *color = NULL;
+       const char *slot;
 
        if (userdiff_config(var, value) < 0)
                return -1;
 
        if (!strcmp(var, "grep.extendedregexp")) {
-               if (git_config_bool(var, value))
-                       opt->extended_regexp_option = 1;
-               else
-                       opt->extended_regexp_option = 0;
+               opt->extended_regexp_option = git_config_bool(var, value);
                return 0;
        }
 
@@ -102,32 +118,18 @@ int grep_config(const char *var, const char *value, void *cb)
 
        if (!strcmp(var, "color.grep"))
                opt->color = git_config_colorbool(var, value);
-       else if (!strcmp(var, "color.grep.context"))
-               color = opt->color_context;
-       else if (!strcmp(var, "color.grep.filename"))
-               color = opt->color_filename;
-       else if (!strcmp(var, "color.grep.function"))
-               color = opt->color_function;
-       else if (!strcmp(var, "color.grep.linenumber"))
-               color = opt->color_lineno;
-       else if (!strcmp(var, "color.grep.matchcontext"))
-               color = opt->color_match_context;
-       else if (!strcmp(var, "color.grep.matchselected"))
-               color = opt->color_match_selected;
-       else if (!strcmp(var, "color.grep.selected"))
-               color = opt->color_selected;
-       else if (!strcmp(var, "color.grep.separator"))
-               color = opt->color_sep;
-       else if (!strcmp(var, "color.grep.match")) {
-               int rc = 0;
-               if (!value)
-                       return config_error_nonbool(var);
-               rc |= color_parse(value, opt->color_match_context);
-               rc |= color_parse(value, opt->color_match_selected);
-               return rc;
-       }
-
-       if (color) {
+       if (!strcmp(var, "color.grep.match")) {
+               if (grep_config("color.grep.matchcontext", value, cb) < 0)
+                       return -1;
+               if (grep_config("color.grep.matchselected", value, cb) < 0)
+                       return -1;
+       } else if (skip_prefix(var, "color.grep.", &slot)) {
+               int i = LOOKUP_CONFIG(color_grep_slots, slot);
+               char *color;
+
+               if (i < 0)
+                       return -1;
+               color = opt->colors[i];
                if (!value)
                        return config_error_nonbool(var);
                return color_parse(value, color);
@@ -143,6 +145,7 @@ int grep_config(const char *var, const char *value, void *cb)
 void grep_init(struct grep_opt *opt, const char *prefix)
 {
        struct grep_opt *def = &grep_defaults;
+       int i;
 
        memset(opt, 0, sizeof(*opt));
        opt->prefix = prefix;
@@ -156,48 +159,60 @@ void grep_init(struct grep_opt *opt, const char *prefix)
        opt->linenum = def->linenum;
        opt->max_depth = def->max_depth;
        opt->pathname = def->pathname;
-       opt->regflags = def->regflags;
        opt->relative = def->relative;
        opt->output = def->output;
 
-       color_set(opt->color_context, def->color_context);
-       color_set(opt->color_filename, def->color_filename);
-       color_set(opt->color_function, def->color_function);
-       color_set(opt->color_lineno, def->color_lineno);
-       color_set(opt->color_match_context, def->color_match_context);
-       color_set(opt->color_match_selected, def->color_match_selected);
-       color_set(opt->color_selected, def->color_selected);
-       color_set(opt->color_sep, def->color_sep);
+       for (i = 0; i < NR_GREP_COLORS; i++)
+               color_set(opt->colors[i], def->colors[i]);
 }
 
 static void grep_set_pattern_type_option(enum grep_pattern_type pattern_type, struct grep_opt *opt)
 {
+       /*
+        * When committing to the pattern type by setting the relevant
+        * fields in grep_opt it's generally not necessary to zero out
+        * the fields we're not choosing, since they won't have been
+        * set by anything. The extended_regexp_option field is the
+        * only exception to this.
+        *
+        * This is because in the process of parsing grep.patternType
+        * & grep.extendedRegexp we set opt->pattern_type_option and
+        * opt->extended_regexp_option, respectively. We then
+        * internally use opt->extended_regexp_option to see if we're
+        * compiling an ERE. It must be unset if that's not actually
+        * the case.
+        */
+       if (pattern_type != GREP_PATTERN_TYPE_ERE &&
+           opt->extended_regexp_option)
+               opt->extended_regexp_option = 0;
+
        switch (pattern_type) {
        case GREP_PATTERN_TYPE_UNSPECIFIED:
                /* fall through */
 
        case GREP_PATTERN_TYPE_BRE:
-               opt->fixed = 0;
-               opt->pcre = 0;
-               opt->regflags &= ~REG_EXTENDED;
                break;
 
        case GREP_PATTERN_TYPE_ERE:
-               opt->fixed = 0;
-               opt->pcre = 0;
-               opt->regflags |= REG_EXTENDED;
+               opt->extended_regexp_option = 1;
                break;
 
        case GREP_PATTERN_TYPE_FIXED:
                opt->fixed = 1;
-               opt->pcre = 0;
-               opt->regflags &= ~REG_EXTENDED;
                break;
 
        case GREP_PATTERN_TYPE_PCRE:
-               opt->fixed = 0;
-               opt->pcre = 1;
-               opt->regflags &= ~REG_EXTENDED;
+#ifdef USE_LIBPCRE2
+               opt->pcre2 = 1;
+#else
+               /*
+                * It's important that pcre1 always be assigned to
+                * even when there's no USE_LIBPCRE* defined. We still
+                * call the PCRE stub function, it just dies with
+                * "cannot use Perl-compatible regexes[...]".
+                */
+               opt->pcre1 = 1;
+#endif
                break;
        }
 }
@@ -209,6 +224,11 @@ void grep_commit_pattern_type(enum grep_pattern_type pattern_type, struct grep_o
        else if (opt->pattern_type_option != GREP_PATTERN_TYPE_UNSPECIFIED)
                grep_set_pattern_type_option(opt->pattern_type_option, opt);
        else if (opt->extended_regexp_option)
+               /*
+                * This branch *must* happen after setting from the
+                * opt->pattern_type_option above, we don't want
+                * grep.extendedRegexp to override grep.patternType!
+                */
                grep_set_pattern_type_option(GREP_PATTERN_TYPE_ERE, opt);
 }
 
@@ -324,8 +344,32 @@ static NORETURN void compile_regexp_failed(const struct grep_pat *p,
        die("%s'%s': %s", where, p->pattern, error);
 }
 
-#ifdef USE_LIBPCRE
-static void compile_pcre_regexp(struct grep_pat *p, const struct grep_opt *opt)
+static int is_fixed(const char *s, size_t len)
+{
+       size_t i;
+
+       for (i = 0; i < len; i++) {
+               if (is_regex_special(s[i]))
+                       return 0;
+       }
+
+       return 1;
+}
+
+static int has_null(const char *s, size_t len)
+{
+       /*
+        * regcomp cannot accept patterns with NULs so when using it
+        * we consider any pattern containing a NUL fixed.
+        */
+       if (memchr(s, 0, len))
+               return 1;
+
+       return 0;
+}
+
+#ifdef USE_LIBPCRE1
+static void compile_pcre1_regexp(struct grep_pat *p, const struct grep_opt *opt)
 {
        const char *error;
        int erroffset;
@@ -333,23 +377,36 @@ static void compile_pcre_regexp(struct grep_pat *p, const struct grep_opt *opt)
 
        if (opt->ignore_case) {
                if (has_non_ascii(p->pattern))
-                       p->pcre_tables = pcre_maketables();
+                       p->pcre1_tables = pcre_maketables();
                options |= PCRE_CASELESS;
        }
        if (is_utf8_locale() && has_non_ascii(p->pattern))
                options |= PCRE_UTF8;
 
-       p->pcre_regexp = pcre_compile(p->pattern, options, &error, &erroffset,
-                                     p->pcre_tables);
-       if (!p->pcre_regexp)
+       p->pcre1_regexp = pcre_compile(p->pattern, options, &error, &erroffset,
+                                     p->pcre1_tables);
+       if (!p->pcre1_regexp)
                compile_regexp_failed(p, error);
 
-       p->pcre_extra_info = pcre_study(p->pcre_regexp, 0, &error);
-       if (!p->pcre_extra_info && error)
+       p->pcre1_extra_info = pcre_study(p->pcre1_regexp, GIT_PCRE_STUDY_JIT_COMPILE, &error);
+       if (!p->pcre1_extra_info && error)
                die("%s", error);
+
+#ifdef GIT_PCRE1_USE_JIT
+       pcre_config(PCRE_CONFIG_JIT, &p->pcre1_jit_on);
+       if (p->pcre1_jit_on == 1) {
+               p->pcre1_jit_stack = pcre_jit_stack_alloc(1, 1024 * 1024);
+               if (!p->pcre1_jit_stack)
+                       die("Couldn't allocate PCRE JIT stack");
+               pcre_assign_jit_stack(p->pcre1_extra_info, NULL, p->pcre1_jit_stack);
+       } else if (p->pcre1_jit_on != 0) {
+               BUG("The pcre1_jit_on variable should be 0 or 1, not %d",
+                   p->pcre1_jit_on);
+       }
+#endif
 }
 
-static int pcrematch(struct grep_pat *p, const char *line, const char *eol,
+static int pcre1match(struct grep_pat *p, const char *line, const char *eol,
                regmatch_t *match, int eflags)
 {
        int ovector[30], ret, flags = 0;
@@ -357,8 +414,19 @@ static int pcrematch(struct grep_pat *p, const char *line, const char *eol,
        if (eflags & REG_NOTBOL)
                flags |= PCRE_NOTBOL;
 
-       ret = pcre_exec(p->pcre_regexp, p->pcre_extra_info, line, eol - line,
-                       0, flags, ovector, ARRAY_SIZE(ovector));
+#ifdef GIT_PCRE1_USE_JIT
+       if (p->pcre1_jit_on) {
+               ret = pcre_jit_exec(p->pcre1_regexp, p->pcre1_extra_info, line,
+                                   eol - line, 0, flags, ovector,
+                                   ARRAY_SIZE(ovector), p->pcre1_jit_stack);
+       } else
+#endif
+       {
+               ret = pcre_exec(p->pcre1_regexp, p->pcre1_extra_info, line,
+                               eol - line, 0, flags, ovector,
+                               ARRAY_SIZE(ovector));
+       }
+
        if (ret < 0 && ret != PCRE_ERROR_NOMATCH)
                die("pcre_exec failed with error code %d", ret);
        if (ret > 0) {
@@ -370,55 +438,191 @@ static int pcrematch(struct grep_pat *p, const char *line, const char *eol,
        return ret;
 }
 
-static void free_pcre_regexp(struct grep_pat *p)
+static void free_pcre1_regexp(struct grep_pat *p)
 {
-       pcre_free(p->pcre_regexp);
-       pcre_free(p->pcre_extra_info);
-       pcre_free((void *)p->pcre_tables);
+       pcre_free(p->pcre1_regexp);
+#ifdef GIT_PCRE1_USE_JIT
+       if (p->pcre1_jit_on) {
+               pcre_free_study(p->pcre1_extra_info);
+               pcre_jit_stack_free(p->pcre1_jit_stack);
+       } else
+#endif
+       {
+               pcre_free(p->pcre1_extra_info);
+       }
+       pcre_free((void *)p->pcre1_tables);
 }
-#else /* !USE_LIBPCRE */
-static void compile_pcre_regexp(struct grep_pat *p, const struct grep_opt *opt)
+#else /* !USE_LIBPCRE1 */
+static void compile_pcre1_regexp(struct grep_pat *p, const struct grep_opt *opt)
 {
        die("cannot use Perl-compatible regexes when not compiled with USE_LIBPCRE");
 }
 
-static int pcrematch(struct grep_pat *p, const char *line, const char *eol,
+static int pcre1match(struct grep_pat *p, const char *line, const char *eol,
                regmatch_t *match, int eflags)
 {
        return 1;
 }
 
-static void free_pcre_regexp(struct grep_pat *p)
+static void free_pcre1_regexp(struct grep_pat *p)
 {
 }
-#endif /* !USE_LIBPCRE */
+#endif /* !USE_LIBPCRE1 */
 
-static int is_fixed(const char *s, size_t len)
+#ifdef USE_LIBPCRE2
+static void compile_pcre2_pattern(struct grep_pat *p, const struct grep_opt *opt)
 {
-       size_t i;
+       int error;
+       PCRE2_UCHAR errbuf[256];
+       PCRE2_SIZE erroffset;
+       int options = PCRE2_MULTILINE;
+       const uint8_t *character_tables = NULL;
+       int jitret;
+       int patinforet;
+       size_t jitsizearg;
 
-       /* regcomp cannot accept patterns with NULs so we
-        * consider any pattern containing a NUL fixed.
-        */
-       if (memchr(s, 0, len))
-               return 1;
+       assert(opt->pcre2);
 
-       for (i = 0; i < len; i++) {
-               if (is_regex_special(s[i]))
-                       return 0;
+       p->pcre2_compile_context = NULL;
+
+       if (opt->ignore_case) {
+               if (has_non_ascii(p->pattern)) {
+                       character_tables = pcre2_maketables(NULL);
+                       p->pcre2_compile_context = pcre2_compile_context_create(NULL);
+                       pcre2_set_character_tables(p->pcre2_compile_context, character_tables);
+               }
+               options |= PCRE2_CASELESS;
+       }
+       if (is_utf8_locale() && has_non_ascii(p->pattern))
+               options |= PCRE2_UTF;
+
+       p->pcre2_pattern = pcre2_compile((PCRE2_SPTR)p->pattern,
+                                        p->patternlen, options, &error, &erroffset,
+                                        p->pcre2_compile_context);
+
+       if (p->pcre2_pattern) {
+               p->pcre2_match_data = pcre2_match_data_create_from_pattern(p->pcre2_pattern, NULL);
+               if (!p->pcre2_match_data)
+                       die("Couldn't allocate PCRE2 match data");
+       } else {
+               pcre2_get_error_message(error, errbuf, sizeof(errbuf));
+               compile_regexp_failed(p, (const char *)&errbuf);
+       }
+
+       pcre2_config(PCRE2_CONFIG_JIT, &p->pcre2_jit_on);
+       if (p->pcre2_jit_on == 1) {
+               jitret = pcre2_jit_compile(p->pcre2_pattern, PCRE2_JIT_COMPLETE);
+               if (jitret)
+                       die("Couldn't JIT the PCRE2 pattern '%s', got '%d'\n", p->pattern, jitret);
+
+               /*
+                * The pcre2_config(PCRE2_CONFIG_JIT, ...) call just
+                * tells us whether the library itself supports JIT,
+                * but to see whether we're going to be actually using
+                * JIT we need to extract PCRE2_INFO_JITSIZE from the
+                * pattern *after* we do pcre2_jit_compile() above.
+                *
+                * This is because if the pattern contains the
+                * (*NO_JIT) verb (see pcre2syntax(3))
+                * pcre2_jit_compile() will exit early with 0. If we
+                * then proceed to call pcre2_jit_match() further down
+                * the line instead of pcre2_match() we'll either
+                * segfault (pre PCRE 10.31) or run into a fatal error
+                * (post PCRE2 10.31)
+                */
+               patinforet = pcre2_pattern_info(p->pcre2_pattern, PCRE2_INFO_JITSIZE, &jitsizearg);
+               if (patinforet)
+                       BUG("pcre2_pattern_info() failed: %d", patinforet);
+               if (jitsizearg == 0) {
+                       p->pcre2_jit_on = 0;
+                       return;
+               }
+
+               p->pcre2_jit_stack = pcre2_jit_stack_create(1, 1024 * 1024, NULL);
+               if (!p->pcre2_jit_stack)
+                       die("Couldn't allocate PCRE2 JIT stack");
+               p->pcre2_match_context = pcre2_match_context_create(NULL);
+               if (!p->pcre2_match_context)
+                       die("Couldn't allocate PCRE2 match context");
+               pcre2_jit_stack_assign(p->pcre2_match_context, NULL, p->pcre2_jit_stack);
+       } else if (p->pcre2_jit_on != 0) {
+               BUG("The pcre2_jit_on variable should be 0 or 1, not %d",
+                   p->pcre1_jit_on);
+       }
+}
+
+static int pcre2match(struct grep_pat *p, const char *line, const char *eol,
+               regmatch_t *match, int eflags)
+{
+       int ret, flags = 0;
+       PCRE2_SIZE *ovector;
+       PCRE2_UCHAR errbuf[256];
+
+       if (eflags & REG_NOTBOL)
+               flags |= PCRE2_NOTBOL;
+
+       if (p->pcre2_jit_on)
+               ret = pcre2_jit_match(p->pcre2_pattern, (unsigned char *)line,
+                                     eol - line, 0, flags, p->pcre2_match_data,
+                                     NULL);
+       else
+               ret = pcre2_match(p->pcre2_pattern, (unsigned char *)line,
+                                 eol - line, 0, flags, p->pcre2_match_data,
+                                 NULL);
+
+       if (ret < 0 && ret != PCRE2_ERROR_NOMATCH) {
+               pcre2_get_error_message(ret, errbuf, sizeof(errbuf));
+               die("%s failed with error code %d: %s",
+                   (p->pcre2_jit_on ? "pcre2_jit_match" : "pcre2_match"), ret,
+                   errbuf);
        }
+       if (ret > 0) {
+               ovector = pcre2_get_ovector_pointer(p->pcre2_match_data);
+               ret = 0;
+               match->rm_so = (int)ovector[0];
+               match->rm_eo = (int)ovector[1];
+       }
+
+       return ret;
+}
+
+static void free_pcre2_pattern(struct grep_pat *p)
+{
+       pcre2_compile_context_free(p->pcre2_compile_context);
+       pcre2_code_free(p->pcre2_pattern);
+       pcre2_match_data_free(p->pcre2_match_data);
+       pcre2_jit_stack_free(p->pcre2_jit_stack);
+       pcre2_match_context_free(p->pcre2_match_context);
+}
+#else /* !USE_LIBPCRE2 */
+static void compile_pcre2_pattern(struct grep_pat *p, const struct grep_opt *opt)
+{
+       /*
+        * Unreachable until USE_LIBPCRE2 becomes synonymous with
+        * USE_LIBPCRE. See the sibling comment in
+        * grep_set_pattern_type_option().
+        */
+       die("cannot use Perl-compatible regexes when not compiled with USE_LIBPCRE");
+}
 
+static int pcre2match(struct grep_pat *p, const char *line, const char *eol,
+               regmatch_t *match, int eflags)
+{
        return 1;
 }
 
+static void free_pcre2_pattern(struct grep_pat *p)
+{
+}
+#endif /* !USE_LIBPCRE2 */
+
 static void compile_fixed_regexp(struct grep_pat *p, struct grep_opt *opt)
 {
        struct strbuf sb = STRBUF_INIT;
        int err;
-       int regflags;
+       int regflags = 0;
 
        basic_regex_quote_buf(&sb, p->pattern);
-       regflags = opt->regflags & ~REG_EXTENDED;
        if (opt->ignore_case)
                regflags |= REG_ICASE;
        err = regcomp(&p->regexp, sb.buf, regflags);
@@ -428,19 +632,18 @@ static void compile_fixed_regexp(struct grep_pat *p, struct grep_opt *opt)
        if (err) {
                char errbuf[1024];
                regerror(err, &p->regexp, errbuf, sizeof(errbuf));
-               regfree(&p->regexp);
                compile_regexp_failed(p, errbuf);
        }
 }
 
 static void compile_regexp(struct grep_pat *p, struct grep_opt *opt)
 {
-       int icase, ascii_only;
+       int ascii_only;
        int err;
+       int regflags = REG_NEWLINE;
 
        p->word_regexp = opt->word_regexp;
        p->ignore_case = opt->ignore_case;
-       icase          = opt->regflags & REG_ICASE || p->ignore_case;
        ascii_only     = !has_non_ascii(p->pattern);
 
        /*
@@ -455,13 +658,13 @@ static void compile_regexp(struct grep_pat *p, struct grep_opt *opt)
         * simple string match using kws.  p->fixed tells us if we
         * want to use kws.
         */
-       if (opt->fixed || is_fixed(p->pattern, p->patternlen))
-               p->fixed = !icase || ascii_only;
-       else
-               p->fixed = 0;
+       if (opt->fixed ||
+           has_null(p->pattern, p->patternlen) ||
+           is_fixed(p->pattern, p->patternlen))
+               p->fixed = !p->ignore_case || ascii_only;
 
        if (p->fixed) {
-               p->kws = kwsalloc(icase ? tolower_trans_tbl : NULL);
+               p->kws = kwsalloc(p->ignore_case ? tolower_trans_tbl : NULL);
                kwsincr(p->kws, p->pattern, p->patternlen);
                kwsprep(p->kws);
                return;
@@ -475,16 +678,24 @@ static void compile_regexp(struct grep_pat *p, struct grep_opt *opt)
                return;
        }
 
-       if (opt->pcre) {
-               compile_pcre_regexp(p, opt);
+       if (opt->pcre2) {
+               compile_pcre2_pattern(p, opt);
                return;
        }
 
-       err = regcomp(&p->regexp, p->pattern, opt->regflags);
+       if (opt->pcre1) {
+               compile_pcre1_regexp(p, opt);
+               return;
+       }
+
+       if (p->ignore_case)
+               regflags |= REG_ICASE;
+       if (opt->extended_regexp_option)
+               regflags |= REG_EXTENDED;
+       err = regcomp(&p->regexp, p->pattern, regflags);
        if (err) {
                char errbuf[1024];
                regerror(err, &p->regexp, errbuf, 1024);
-               regfree(&p->regexp);
                compile_regexp_failed(p, errbuf);
        }
 }
@@ -700,10 +911,10 @@ static struct grep_expr *prep_header_patterns(struct grep_opt *opt)
 
        for (p = opt->header_list; p; p = p->next) {
                if (p->token != GREP_PATTERN_HEAD)
-                       die("BUG: a non-header pattern in grep header list.");
+                       BUG("a non-header pattern in grep header list.");
                if (p->field < GREP_HEADER_FIELD_MIN ||
                    GREP_HEADER_FIELD_MAX <= p->field)
-                       die("BUG: unknown header field %d", p->field);
+                       BUG("unknown header field %d", p->field);
                compile_regexp(p, opt);
        }
 
@@ -716,7 +927,7 @@ static struct grep_expr *prep_header_patterns(struct grep_opt *opt)
 
                h = compile_pattern_atom(&pp);
                if (!h || pp != p->next)
-                       die("BUG: malformed header expr");
+                       BUG("malformed header expr");
                if (!header_group[p->field]) {
                        header_group[p->field] = h;
                        continue;
@@ -832,8 +1043,10 @@ void free_grep_patterns(struct grep_opt *opt)
                case GREP_PATTERN_BODY:
                        if (p->kws)
                                kwsfree(p->kws);
-                       else if (p->pcre_regexp)
-                               free_pcre_regexp(p);
+                       else if (p->pcre1_regexp)
+                               free_pcre1_regexp(p);
+                       else if (p->pcre2_pattern)
+                               free_pcre2_pattern(p);
                        else
                                regfree(&p->regexp);
                        free(p->pattern);
@@ -881,12 +1094,12 @@ static void output_sep(struct grep_opt *opt, char sign)
        if (opt->null_following_name)
                opt->output(opt, "\0", 1);
        else
-               output_color(opt, &sign, 1, opt->color_sep);
+               output_color(opt, &sign, 1, opt->colors[GREP_COLOR_SEP]);
 }
 
 static void show_name(struct grep_opt *opt, const char *name)
 {
-       output_color(opt, name, strlen(name), opt->color_filename);
+       output_color(opt, name, strlen(name), opt->colors[GREP_COLOR_FILENAME]);
        opt->output(opt, opt->null_following_name ? "\0" : "\n", 1);
 }
 
@@ -912,8 +1125,10 @@ static int patmatch(struct grep_pat *p, char *line, char *eol,
 
        if (p->fixed)
                hit = !fixmatch(p, line, eol, match);
-       else if (p->pcre_regexp)
-               hit = !pcrematch(p, line, eol, match, eflags);
+       else if (p->pcre1_regexp)
+               hit = !pcre1match(p, line, eol, match, eflags);
+       else if (p->pcre2_pattern)
+               hit = !pcre2match(p, line, eol, match, eflags);
        else
                hit = !regexec_buf(&p->regexp, line, eol - line, 1, match,
                                   eflags);
@@ -1151,28 +1366,28 @@ static void show_line(struct grep_opt *opt, char *bol, char *eol,
        } else if (opt->pre_context || opt->post_context || opt->funcbody) {
                if (opt->last_shown == 0) {
                        if (opt->show_hunk_mark) {
-                               output_color(opt, "--", 2, opt->color_sep);
+                               output_color(opt, "--", 2, opt->colors[GREP_COLOR_SEP]);
                                opt->output(opt, "\n", 1);
                        }
                } else if (lno > opt->last_shown + 1) {
-                       output_color(opt, "--", 2, opt->color_sep);
+                       output_color(opt, "--", 2, opt->colors[GREP_COLOR_SEP]);
                        opt->output(opt, "\n", 1);
                }
        }
        if (opt->heading && opt->last_shown == 0) {
-               output_color(opt, name, strlen(name), opt->color_filename);
+               output_color(opt, name, strlen(name), opt->colors[GREP_COLOR_FILENAME]);
                opt->output(opt, "\n", 1);
        }
        opt->last_shown = lno;
 
        if (!opt->heading && opt->pathname) {
-               output_color(opt, name, strlen(name), opt->color_filename);
+               output_color(opt, name, strlen(name), opt->colors[GREP_COLOR_FILENAME]);
                output_sep(opt, sign);
        }
        if (opt->linenum) {
                char buf[32];
-               snprintf(buf, sizeof(buf), "%d", lno);
-               output_color(opt, buf, strlen(buf), opt->color_lineno);
+               xsnprintf(buf, sizeof(buf), "%d", lno);
+               output_color(opt, buf, strlen(buf), opt->colors[GREP_COLOR_LINENO]);
                output_sep(opt, sign);
        }
        if (opt->color) {
@@ -1182,15 +1397,15 @@ static void show_line(struct grep_opt *opt, char *bol, char *eol,
                int eflags = 0;
 
                if (sign == ':')
-                       match_color = opt->color_match_selected;
+                       match_color = opt->colors[GREP_COLOR_MATCH_SELECTED];
                else
-                       match_color = opt->color_match_context;
+                       match_color = opt->colors[GREP_COLOR_MATCH_CONTEXT];
                if (sign == ':')
-                       line_color = opt->color_selected;
+                       line_color = opt->colors[GREP_COLOR_SELECTED];
                else if (sign == '-')
-                       line_color = opt->color_context;
+                       line_color = opt->colors[GREP_COLOR_CONTEXT];
                else if (sign == '=')
-                       line_color = opt->color_function;
+                       line_color = opt->colors[GREP_COLOR_FUNCTION];
                *eol = '\0';
                while (next_match(opt, bol, eol, ctx, &match, eflags)) {
                        if (match.rm_so == match.rm_eo)
@@ -1286,31 +1501,52 @@ static void show_funcname_line(struct grep_opt *opt, struct grep_source *gs,
        }
 }
 
+static int is_empty_line(const char *bol, const char *eol);
+
 static void show_pre_context(struct grep_opt *opt, struct grep_source *gs,
                             char *bol, char *end, unsigned lno)
 {
-       unsigned cur = lno, from = 1, funcname_lno = 0;
-       int funcname_needed = !!opt->funcname;
-
-       if (opt->funcbody && !match_funcname(opt, gs, bol, end))
-               funcname_needed = 2;
+       unsigned cur = lno, from = 1, funcname_lno = 0, orig_from;
+       int funcname_needed = !!opt->funcname, comment_needed = 0;
 
        if (opt->pre_context < lno)
                from = lno - opt->pre_context;
        if (from <= opt->last_shown)
                from = opt->last_shown + 1;
+       orig_from = from;
+       if (opt->funcbody) {
+               if (match_funcname(opt, gs, bol, end))
+                       comment_needed = 1;
+               else
+                       funcname_needed = 1;
+               from = opt->last_shown + 1;
+       }
 
        /* Rewind. */
-       while (bol > gs->buf &&
-              cur > (funcname_needed == 2 ? opt->last_shown + 1 : from)) {
+       while (bol > gs->buf && cur > from) {
+               char *next_bol = bol;
                char *eol = --bol;
 
                while (bol > gs->buf && bol[-1] != '\n')
                        bol--;
                cur--;
+               if (comment_needed && (is_empty_line(bol, eol) ||
+                                      match_funcname(opt, gs, bol, eol))) {
+                       comment_needed = 0;
+                       from = orig_from;
+                       if (cur < from) {
+                               cur++;
+                               bol = next_bol;
+                               break;
+                       }
+               }
                if (funcname_needed && match_funcname(opt, gs, bol, eol)) {
                        funcname_lno = cur;
                        funcname_needed = 0;
+                       if (opt->funcbody)
+                               comment_needed = 1;
+                       else
+                               from = orig_from;
                }
        }
 
@@ -1403,14 +1639,14 @@ static int fill_textconv_grep(struct userdiff_driver *driver,
         */
        df = alloc_filespec(gs->path);
        switch (gs->type) {
-       case GREP_SOURCE_SHA1:
+       case GREP_SOURCE_OID:
                fill_filespec(df, gs->identifier, 1, 0100644);
                break;
        case GREP_SOURCE_FILE:
-               fill_filespec(df, null_sha1, 0, 0100644);
+               fill_filespec(df, &null_oid, 0, 0100644);
                break;
        default:
-               die("BUG: attempt to textconv something without a path?");
+               BUG("attempt to textconv something without a path?");
        }
 
        /*
@@ -1506,7 +1742,7 @@ static int grep_source_1(struct grep_opt *opt, struct grep_source *gs, int colle
                case GREP_BINARY_TEXT:
                        break;
                default:
-                       die("BUG: unknown binary handling mode");
+                       BUG("unknown binary handling mode");
                }
        }
 
@@ -1576,7 +1812,7 @@ static int grep_source_1(struct grep_opt *opt, struct grep_source *gs, int colle
                        if (binary_match_only) {
                                opt->output(opt, "Binary file ", 12);
                                output_color(opt, gs->name, strlen(gs->name),
-                                            opt->color_filename);
+                                            opt->colors[GREP_COLOR_FILENAME]);
                                opt->output(opt, " matches\n", 9);
                                return 1;
                        }
@@ -1631,7 +1867,7 @@ static int grep_source_1(struct grep_opt *opt, struct grep_source *gs, int colle
                return 0;
 
        if (opt->status_only)
-               return 0;
+               return opt->unmatch_name_only;
        if (opt->unmatch_name_only) {
                /* We did not see any hit, so we want to show this */
                show_name(opt, gs->name);
@@ -1650,10 +1886,10 @@ static int grep_source_1(struct grep_opt *opt, struct grep_source *gs, int colle
                char buf[32];
                if (opt->pathname) {
                        output_color(opt, gs->name, strlen(gs->name),
-                                    opt->color_filename);
+                                    opt->colors[GREP_COLOR_FILENAME]);
                        output_sep(opt, ':');
                }
-               snprintf(buf, sizeof(buf), "%u\n", count);
+               xsnprintf(buf, sizeof(buf), "%u\n", count);
                opt->output(opt, buf, strlen(buf));
                return 1;
        }
@@ -1737,19 +1973,8 @@ void grep_source_init(struct grep_source *gs, enum grep_source_type type,
        case GREP_SOURCE_FILE:
                gs->identifier = xstrdup(identifier);
                break;
-       case GREP_SOURCE_SUBMODULE:
-               if (!identifier) {
-                       gs->identifier = NULL;
-                       break;
-               }
-               /*
-                * FALL THROUGH
-                * If the identifier is non-NULL (in the submodule case) it
-                * will be a SHA1 that needs to be copied.
-                */
-       case GREP_SOURCE_SHA1:
-               gs->identifier = xmalloc(20);
-               hashcpy(gs->identifier, identifier);
+       case GREP_SOURCE_OID:
+               gs->identifier = oiddup(identifier);
                break;
        case GREP_SOURCE_BUF:
                gs->identifier = NULL;
@@ -1759,12 +1984,9 @@ void grep_source_init(struct grep_source *gs, enum grep_source_type type,
 
 void grep_source_clear(struct grep_source *gs)
 {
-       free(gs->name);
-       gs->name = NULL;
-       free(gs->path);
-       gs->path = NULL;
-       free(gs->identifier);
-       gs->identifier = NULL;
+       FREE_AND_NULL(gs->name);
+       FREE_AND_NULL(gs->path);
+       FREE_AND_NULL(gs->identifier);
        grep_source_clear_data(gs);
 }
 
@@ -1772,10 +1994,8 @@ void grep_source_clear_data(struct grep_source *gs)
 {
        switch (gs->type) {
        case GREP_SOURCE_FILE:
-       case GREP_SOURCE_SHA1:
-       case GREP_SOURCE_SUBMODULE:
-               free(gs->buf);
-               gs->buf = NULL;
+       case GREP_SOURCE_OID:
+               FREE_AND_NULL(gs->buf);
                gs->size = 0;
                break;
        case GREP_SOURCE_BUF:
@@ -1784,18 +2004,18 @@ void grep_source_clear_data(struct grep_source *gs)
        }
 }
 
-static int grep_source_load_sha1(struct grep_source *gs)
+static int grep_source_load_oid(struct grep_source *gs)
 {
        enum object_type type;
 
        grep_read_lock();
-       gs->buf = read_sha1_file(gs->identifier, &type, &gs->size);
+       gs->buf = read_object_file(gs->identifier, &type, &gs->size);
        grep_read_unlock();
 
        if (!gs->buf)
                return error(_("'%s': unable to read %s"),
                             gs->name,
-                            sha1_to_hex(gs->identifier));
+                            oid_to_hex(gs->identifier));
        return 0;
 }
 
@@ -1841,14 +2061,12 @@ static int grep_source_load(struct grep_source *gs)
        switch (gs->type) {
        case GREP_SOURCE_FILE:
                return grep_source_load_file(gs);
-       case GREP_SOURCE_SHA1:
-               return grep_source_load_sha1(gs);
+       case GREP_SOURCE_OID:
+               return grep_source_load_oid(gs);
        case GREP_SOURCE_BUF:
                return gs->buf ? 0 : -1;
-       case GREP_SOURCE_SUBMODULE:
-               break;
        }
-       die("BUG: invalid grep_source type to load");
+       BUG("invalid grep_source type to load");
 }
 
 void grep_source_load_driver(struct grep_source *gs)