grep.h

   1 #ifndef GREP_H
   2 #define GREP_H
   3 #include "color.h"
   4 #ifdef USE_LIBPCRE2
   5 #define PCRE2_CODE_UNIT_WIDTH 8
   6 #include <pcre2.h>
   7 #if (PCRE2_MAJOR >= 10 && PCRE2_MINOR >= 36) || PCRE2_MAJOR >= 11
   8 #define GIT_PCRE2_VERSION_10_36_OR_HIGHER
   9 #endif
  10 #if (PCRE2_MAJOR >= 10 && PCRE2_MINOR >= 35) || PCRE2_MAJOR >= 11
  11 #define GIT_PCRE2_VERSION_10_35_OR_HIGHER
  12 #endif
  13 #if (PCRE2_MAJOR >= 10 && PCRE2_MINOR >= 34) || PCRE2_MAJOR >= 11
  14 #define GIT_PCRE2_VERSION_10_34_OR_HIGHER
  15 #endif
  16 #else
  17 typedef int pcre2_code;
  18 typedef int pcre2_match_data;
  19 typedef int pcre2_compile_context;
  20 typedef int pcre2_general_context;
  21 #endif
  22 #ifndef PCRE2_MATCH_INVALID_UTF
  23 /* PCRE2_MATCH_* dummy also with !USE_LIBPCRE2, for test-pcre2-config.c */
  24 #define PCRE2_MATCH_INVALID_UTF 0
  25 #endif
  26 #include "thread-utils.h"
  27 #include "userdiff.h"
  28
  29 struct repository;
  30
  31 enum grep_pat_token {
  32         GREP_PATTERN,
  33         GREP_PATTERN_HEAD,
  34         GREP_PATTERN_BODY,
  35         GREP_AND,
  36         GREP_OPEN_PAREN,
  37         GREP_CLOSE_PAREN,
  38         GREP_NOT,
  39         GREP_OR
  40 };
  41
  42 enum grep_context {
  43         GREP_CONTEXT_HEAD,
  44         GREP_CONTEXT_BODY
  45 };
  46
  47 enum grep_header_field {
  48         GREP_HEADER_FIELD_MIN = 0,
  49         GREP_HEADER_AUTHOR = GREP_HEADER_FIELD_MIN,
  50         GREP_HEADER_COMMITTER,
  51         GREP_HEADER_REFLOG,
  52
  53         /* Must be at the end of the enum */
  54         GREP_HEADER_FIELD_MAX
  55 };
  56
  57 enum grep_color {
  58         GREP_COLOR_CONTEXT,
  59         GREP_COLOR_FILENAME,
  60         GREP_COLOR_FUNCTION,
  61         GREP_COLOR_LINENO,
  62         GREP_COLOR_COLUMNNO,
  63         GREP_COLOR_MATCH_CONTEXT,
  64         GREP_COLOR_MATCH_SELECTED,
  65         GREP_COLOR_SELECTED,
  66         GREP_COLOR_SEP,
  67         NR_GREP_COLORS
  68 };
  69
  70 struct grep_pat {
  71         struct grep_pat *next;
  72         const char *origin;
  73         int no;
  74         enum grep_pat_token token;
  75         char *pattern;
  76         size_t patternlen;
  77         enum grep_header_field field;
  78         regex_t regexp;
  79         pcre2_code *pcre2_pattern;
  80         pcre2_match_data *pcre2_match_data;
  81         pcre2_compile_context *pcre2_compile_context;
  82         pcre2_general_context *pcre2_general_context;
  83         const uint8_t *pcre2_tables;
  84         uint32_t pcre2_jit_on;
  85         unsigned fixed:1;
  86         unsigned is_fixed:1;
  87         unsigned ignore_case:1;
  88         unsigned word_regexp:1;
  89 };
  90
  91 enum grep_expr_node {
  92         GREP_NODE_ATOM,
  93         GREP_NODE_NOT,
  94         GREP_NODE_AND,
  95         GREP_NODE_TRUE,
  96         GREP_NODE_OR
  97 };
  98
  99 enum grep_pattern_type {
 100         GREP_PATTERN_TYPE_UNSPECIFIED = 0,
 101         GREP_PATTERN_TYPE_BRE,
 102         GREP_PATTERN_TYPE_ERE,
 103         GREP_PATTERN_TYPE_FIXED,
 104         GREP_PATTERN_TYPE_PCRE
 105 };
 106
 107 struct grep_expr {
 108         enum grep_expr_node node;
 109         unsigned hit;
 110         union {
 111                 struct grep_pat *atom;
 112                 struct grep_expr *unary;
 113                 struct {
 114                         struct grep_expr *left;
 115                         struct grep_expr *right;
 116                 } binary;
 117         } u;
 118 };
 119
 120 struct grep_opt {
 121         struct grep_pat *pattern_list;
 122         struct grep_pat **pattern_tail;
 123         struct grep_pat *header_list;
 124         struct grep_pat **header_tail;
 125         struct grep_expr *pattern_expression;
 126
 127         /*
 128          * NEEDSWORK: See if we can remove this field, because the repository
 129          * should probably be per-source. That is, grep.c functions using this
 130          * field should probably start using "repo" in "struct grep_source"
 131          * instead.
 132          *
 133          * This is potentially the cause of at least one bug - "git grep"
 134          * using the textconv attributes from the superproject on the
 135          * submodules. See the failing "git grep --textconv" tests in
 136          * t7814-grep-recurse-submodules.sh for more information.
 137          */
 138         struct repository *repo;
 139
 140         int linenum;
 141         int columnnum;
 142         int invert;
 143         int ignore_case;
 144         int status_only;
 145         int name_only;
 146         int unmatch_name_only;
 147         int count;
 148         int word_regexp;
 149         int all_match;
 150         int no_body_match;
 151         int body_hit;
 152 #define GREP_BINARY_DEFAULT     0
 153 #define GREP_BINARY_NOMATCH     1
 154 #define GREP_BINARY_TEXT        2
 155         int binary;
 156         int allow_textconv;
 157         int use_reflog_filter;
 158         int relative;
 159         int pathname;
 160         int null_following_name;
 161         int only_matching;
 162         int color;
 163         int max_depth;
 164         int funcname;
 165         int funcbody;
 166         int extended_regexp_option;
 167         enum grep_pattern_type pattern_type_option;
 168         int ignore_locale;
 169         char colors[NR_GREP_COLORS][COLOR_MAXLEN];
 170         unsigned pre_context;
 171         unsigned post_context;
 172         unsigned last_shown;
 173         int show_hunk_mark;
 174         int file_break;
 175         int heading;
 176         int max_count;
 177         void *priv;
 178
 179         void (*output)(struct grep_opt *opt, const void *data, size_t size);
 180         void *output_priv;
 181 };
 182
 183 #define GREP_OPT_INIT { \
 184         .relative = 1, \
 185         .pathname = 1, \
 186         .max_depth = -1, \
 187         .max_count = -1, \
 188         .pattern_type_option = GREP_PATTERN_TYPE_UNSPECIFIED, \
 189         .colors = { \
 190                 [GREP_COLOR_CONTEXT] = "", \
 191                 [GREP_COLOR_FILENAME] = GIT_COLOR_MAGENTA, \
 192                 [GREP_COLOR_FUNCTION] = "", \
 193                 [GREP_COLOR_LINENO] = GIT_COLOR_GREEN, \
 194                 [GREP_COLOR_COLUMNNO] = GIT_COLOR_GREEN, \
 195                 [GREP_COLOR_MATCH_CONTEXT] = GIT_COLOR_BOLD_RED, \
 196                 [GREP_COLOR_MATCH_SELECTED] = GIT_COLOR_BOLD_RED, \
 197                 [GREP_COLOR_SELECTED] = "", \
 198                 [GREP_COLOR_SEP] = GIT_COLOR_CYAN, \
 199         }, \
 200         .only_matching = 0, \
 201         .color = -1, \
 202         .output = std_output, \
 203 }
 204
 205 int grep_config(const char *var, const char *value, void *);
 206 void grep_init(struct grep_opt *, struct repository *repo);
 207
 208 void append_grep_pat(struct grep_opt *opt, const char *pat, size_t patlen, const char *origin, int no, enum grep_pat_token t);
 209 void append_grep_pattern(struct grep_opt *opt, const char *pat, const char *origin, int no, enum grep_pat_token t);
 210 void append_header_grep_pattern(struct grep_opt *, enum grep_header_field, const char *);
 211 void compile_grep_patterns(struct grep_opt *opt);
 212 void free_grep_patterns(struct grep_opt *opt);
 213 int grep_buffer(struct grep_opt *opt, const char *buf, unsigned long size);
 214
 215 /* The field parameter is only used to filter header patterns
 216  * (where appropriate). If filtering isn't desirable
 217  * GREP_HEADER_FIELD_MAX should be supplied.
 218  */
 219 int grep_next_match(struct grep_opt *opt,
 220                     const char *bol, const char *eol,
 221                     enum grep_context ctx, regmatch_t *pmatch,
 222                     enum grep_header_field field, int eflags);
 223
 224 struct grep_source {
 225         char *name;
 226
 227         enum grep_source_type {
 228                 GREP_SOURCE_OID,
 229                 GREP_SOURCE_FILE,
 230                 GREP_SOURCE_BUF,
 231         } type;
 232         void *identifier;
 233         struct repository *repo; /* if GREP_SOURCE_OID */
 234
 235         const char *buf;
 236         unsigned long size;
 237
 238         char *path; /* for attribute lookups */
 239         struct userdiff_driver *driver;
 240 };
 241
 242 void grep_source_init_file(struct grep_source *gs, const char *name,
 243                            const char *path);
 244 void grep_source_init_oid(struct grep_source *gs, const char *name,
 245                           const char *path, const struct object_id *oid,
 246                           struct repository *repo);
 247 void grep_source_clear_data(struct grep_source *gs);
 248 void grep_source_clear(struct grep_source *gs);
 249 void grep_source_load_driver(struct grep_source *gs,
 250                              struct index_state *istate);
 251
 252
 253 int grep_source(struct grep_opt *opt, struct grep_source *gs);
 254
 255 struct grep_opt *grep_opt_dup(const struct grep_opt *opt);
 256
 257 /*
 258  * Mutex used around access to the attributes machinery if
 259  * opt->use_threads.  Must be initialized/destroyed by callers!
 260  */
 261 extern int grep_use_locks;
 262 extern pthread_mutex_t grep_attr_mutex;
 263
 264 #endif