Merge branch 'nd/pack-objects-threading-doc'
[gitweb.git] / diff.c
diff --git a/diff.c b/diff.c
index 28ee05c9ad57313139906e6734b0be2221babfa1..f830afac79134219e706730c05d98f81d4a1d8cd 100644 (file)
--- a/diff.c
+++ b/diff.c
@@ -13,6 +13,7 @@
 #include "attr.h"
 #include "run-command.h"
 #include "utf8.h"
+#include "object-store.h"
 #include "userdiff.h"
 #include "submodule-config.h"
 #include "submodule.h"
@@ -22,6 +23,7 @@
 #include "argv-array.h"
 #include "graph.h"
 #include "packfile.h"
+#include "help.h"
 
 #ifdef NO_FAST_WORKING_DIRECTORY
 #define FAST_WORKING_DIRECTORY 0
@@ -35,6 +37,7 @@ static int diff_rename_limit_default = 400;
 static int diff_suppress_blank_empty;
 static int diff_use_color_default = -1;
 static int diff_color_moved_default;
+static int diff_color_moved_ws_default;
 static int diff_context_default = 3;
 static int diff_interhunk_context_default;
 static const char *diff_word_regex_cfg;
@@ -69,46 +72,37 @@ static char diff_colors[][COLOR_MAXLEN] = {
        GIT_COLOR_FAINT_ITALIC, /* NEW_MOVED_ALTERNATIVE_DIM */
 };
 
+static const char *color_diff_slots[] = {
+       [DIFF_CONTEXT]                = "context",
+       [DIFF_METAINFO]               = "meta",
+       [DIFF_FRAGINFO]               = "frag",
+       [DIFF_FILE_OLD]               = "old",
+       [DIFF_FILE_NEW]               = "new",
+       [DIFF_COMMIT]                 = "commit",
+       [DIFF_WHITESPACE]             = "whitespace",
+       [DIFF_FUNCINFO]               = "func",
+       [DIFF_FILE_OLD_MOVED]         = "oldMoved",
+       [DIFF_FILE_OLD_MOVED_ALT]     = "oldMovedAlternative",
+       [DIFF_FILE_OLD_MOVED_DIM]     = "oldMovedDimmed",
+       [DIFF_FILE_OLD_MOVED_ALT_DIM] = "oldMovedAlternativeDimmed",
+       [DIFF_FILE_NEW_MOVED]         = "newMoved",
+       [DIFF_FILE_NEW_MOVED_ALT]     = "newMovedAlternative",
+       [DIFF_FILE_NEW_MOVED_DIM]     = "newMovedDimmed",
+       [DIFF_FILE_NEW_MOVED_ALT_DIM] = "newMovedAlternativeDimmed",
+};
+
 static NORETURN void die_want_option(const char *option_name)
 {
        die(_("option '%s' requires a value"), option_name);
 }
 
+define_list_config_array_extra(color_diff_slots, {"plain"});
+
 static int parse_diff_color_slot(const char *var)
 {
-       if (!strcasecmp(var, "context") || !strcasecmp(var, "plain"))
+       if (!strcasecmp(var, "plain"))
                return DIFF_CONTEXT;
-       if (!strcasecmp(var, "meta"))
-               return DIFF_METAINFO;
-       if (!strcasecmp(var, "frag"))
-               return DIFF_FRAGINFO;
-       if (!strcasecmp(var, "old"))
-               return DIFF_FILE_OLD;
-       if (!strcasecmp(var, "new"))
-               return DIFF_FILE_NEW;
-       if (!strcasecmp(var, "commit"))
-               return DIFF_COMMIT;
-       if (!strcasecmp(var, "whitespace"))
-               return DIFF_WHITESPACE;
-       if (!strcasecmp(var, "func"))
-               return DIFF_FUNCINFO;
-       if (!strcasecmp(var, "oldmoved"))
-               return DIFF_FILE_OLD_MOVED;
-       if (!strcasecmp(var, "oldmovedalternative"))
-               return DIFF_FILE_OLD_MOVED_ALT;
-       if (!strcasecmp(var, "oldmoveddimmed"))
-               return DIFF_FILE_OLD_MOVED_DIM;
-       if (!strcasecmp(var, "oldmovedalternativedimmed"))
-               return DIFF_FILE_OLD_MOVED_ALT_DIM;
-       if (!strcasecmp(var, "newmoved"))
-               return DIFF_FILE_NEW_MOVED;
-       if (!strcasecmp(var, "newmovedalternative"))
-               return DIFF_FILE_NEW_MOVED_ALT;
-       if (!strcasecmp(var, "newmoveddimmed"))
-               return DIFF_FILE_NEW_MOVED_DIM;
-       if (!strcasecmp(var, "newmovedalternativedimmed"))
-               return DIFF_FILE_NEW_MOVED_ALT_DIM;
-       return -1;
+       return LOOKUP_CONFIG(color_diff_slots, var);
 }
 
 static int parse_dirstat_params(struct diff_options *options, const char *params_string,
@@ -271,14 +265,54 @@ static int parse_color_moved(const char *arg)
                return COLOR_MOVED_NO;
        else if (!strcmp(arg, "plain"))
                return COLOR_MOVED_PLAIN;
+       else if (!strcmp(arg, "blocks"))
+               return COLOR_MOVED_BLOCKS;
        else if (!strcmp(arg, "zebra"))
                return COLOR_MOVED_ZEBRA;
        else if (!strcmp(arg, "default"))
                return COLOR_MOVED_DEFAULT;
+       else if (!strcmp(arg, "dimmed-zebra"))
+               return COLOR_MOVED_ZEBRA_DIM;
        else if (!strcmp(arg, "dimmed_zebra"))
                return COLOR_MOVED_ZEBRA_DIM;
        else
-               return error(_("color moved setting must be one of 'no', 'default', 'zebra', 'dimmed_zebra', 'plain'"));
+               return error(_("color moved setting must be one of 'no', 'default', 'blocks', 'zebra', 'dimmed-zebra', 'plain'"));
+}
+
+static int parse_color_moved_ws(const char *arg)
+{
+       int ret = 0;
+       struct string_list l = STRING_LIST_INIT_DUP;
+       struct string_list_item *i;
+
+       string_list_split(&l, arg, ',', -1);
+
+       for_each_string_list_item(i, &l) {
+               struct strbuf sb = STRBUF_INIT;
+               strbuf_addstr(&sb, i->string);
+               strbuf_trim(&sb);
+
+               if (!strcmp(sb.buf, "ignore-space-change"))
+                       ret |= XDF_IGNORE_WHITESPACE_CHANGE;
+               else if (!strcmp(sb.buf, "ignore-space-at-eol"))
+                       ret |= XDF_IGNORE_WHITESPACE_AT_EOL;
+               else if (!strcmp(sb.buf, "ignore-all-space"))
+                       ret |= XDF_IGNORE_WHITESPACE;
+               else if (!strcmp(sb.buf, "allow-indentation-change"))
+                       ret |= COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE;
+               else
+                       error(_("ignoring unknown color-moved-ws mode '%s'"), sb.buf);
+
+               strbuf_release(&sb);
+       }
+
+       if ((ret & COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE) &&
+           (ret & XDF_WHITESPACE_FLAGS))
+               die(_("color-moved-ws: allow-indentation-change cannot be combined with other white space modes"));
+
+       string_list_clear(&l, 0);
+
+       return ret;
 }
 
 int git_diff_ui_config(const char *var, const char *value, void *cb)
@@ -294,6 +328,13 @@ int git_diff_ui_config(const char *var, const char *value, void *cb)
                diff_color_moved_default = cm;
                return 0;
        }
+       if (!strcmp(var, "diff.colormovedws")) {
+               int cm = parse_color_moved_ws(value);
+               if (cm < 0)
+                       return -1;
+               diff_color_moved_ws_default = cm;
+               return 0;
+       }
        if (!strcmp(var, "diff.context")) {
                diff_context_default = git_config_int(var, value);
                if (diff_context_default < 0)
@@ -705,16 +746,116 @@ struct moved_entry {
        struct hashmap_entry ent;
        const struct emitted_diff_symbol *es;
        struct moved_entry *next_line;
+       struct ws_delta *wsd;
+};
+
+/**
+ * The struct ws_delta holds white space differences between moved lines, i.e.
+ * between '+' and '-' lines that have been detected to be a move.
+ * The string contains the difference in leading white spaces, before the
+ * rest of the line is compared using the white space config for move
+ * coloring. The current_longer indicates if the first string in the
+ * comparision is longer than the second.
+ */
+struct ws_delta {
+       char *string;
+       unsigned int current_longer : 1;
 };
+#define WS_DELTA_INIT { NULL, 0 }
+
+static int compute_ws_delta(const struct emitted_diff_symbol *a,
+                            const struct emitted_diff_symbol *b,
+                            struct ws_delta *out)
+{
+       const struct emitted_diff_symbol *longer =  a->len > b->len ? a : b;
+       const struct emitted_diff_symbol *shorter = a->len > b->len ? b : a;
+       int d = longer->len - shorter->len;
+
+       out->string = xmemdupz(longer->line, d);
+       out->current_longer = (a == longer);
+
+       return !strncmp(longer->line + d, shorter->line, shorter->len);
+}
+
+static int cmp_in_block_with_wsd(const struct diff_options *o,
+                                const struct moved_entry *cur,
+                                const struct moved_entry *match,
+                                struct moved_entry *pmb,
+                                int n)
+{
+       struct emitted_diff_symbol *l = &o->emitted_symbols->buf[n];
+       int al = cur->es->len, cl = l->len;
+       const char *a = cur->es->line,
+                  *b = match->es->line,
+                  *c = l->line;
+
+       int wslen;
+
+       /*
+        * We need to check if 'cur' is equal to 'match'.
+        * As those are from the same (+/-) side, we do not need to adjust for
+        * indent changes. However these were found using fuzzy matching
+        * so we do have to check if they are equal.
+        */
+       if (strcmp(a, b))
+               return 1;
+
+       if (!pmb->wsd)
+               /*
+                * No white space delta was carried forward? This can happen
+                * when we exit early in this function and do not carry
+                * forward ws.
+                */
+               return 1;
+
+       /*
+        * The indent changes of the block are known and carried forward in
+        * pmb->wsd; however we need to check if the indent changes of the
+        * current line are still the same as before.
+        *
+        * To do so we need to compare 'l' to 'cur', adjusting the
+        * one of them for the white spaces, depending which was longer.
+        */
+
+       wslen = strlen(pmb->wsd->string);
+       if (pmb->wsd->current_longer) {
+               c += wslen;
+               cl -= wslen;
+       } else {
+               a += wslen;
+               al -= wslen;
+       }
+
+       if (strcmp(a, c))
+               return 1;
+
+       return 0;
+}
 
-static int moved_entry_cmp(const struct diff_options *diffopt,
-                          const struct moved_entry *a,
-                          const struct moved_entry *b,
+static int moved_entry_cmp(const void *hashmap_cmp_fn_data,
+                          const void *entry,
+                          const void *entry_or_key,
                           const void *keydata)
 {
+       const struct diff_options *diffopt = hashmap_cmp_fn_data;
+       const struct moved_entry *a = entry;
+       const struct moved_entry *b = entry_or_key;
+       unsigned flags = diffopt->color_moved_ws_handling
+                        & XDF_WHITESPACE_FLAGS;
+
+       if (diffopt->color_moved_ws_handling &
+           COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE)
+               /*
+                * As there is not specific white space config given,
+                * we'd need to check for a new block, so ignore all
+                * white space. The setup of the white space
+                * configuration for the next block is done else where
+                */
+               flags |= XDF_IGNORE_WHITESPACE;
+
        return !xdiff_compare_lines(a->es->line, a->es->len,
                                    b->es->line, b->es->len,
-                                   diffopt->xdl_opts);
+                                   flags);
 }
 
 static struct moved_entry *prepare_entry(struct diff_options *o,
@@ -722,10 +863,12 @@ static struct moved_entry *prepare_entry(struct diff_options *o,
 {
        struct moved_entry *ret = xmalloc(sizeof(*ret));
        struct emitted_diff_symbol *l = &o->emitted_symbols->buf[line_no];
+       unsigned flags = o->color_moved_ws_handling & XDF_WHITESPACE_FLAGS;
 
-       ret->ent.hash = xdiff_hash_string(l->line, l->len, o->xdl_opts);
+       ret->ent.hash = xdiff_hash_string(l->line, l->len, flags);
        ret->es = l;
        ret->next_line = NULL;
+       ret->wsd = NULL;
 
        return ret;
 }
@@ -762,6 +905,56 @@ static void add_lines_to_move_detection(struct diff_options *o,
        }
 }
 
+static void pmb_advance_or_null(struct diff_options *o,
+                               struct moved_entry *match,
+                               struct hashmap *hm,
+                               struct moved_entry **pmb,
+                               int pmb_nr)
+{
+       int i;
+       for (i = 0; i < pmb_nr; i++) {
+               struct moved_entry *prev = pmb[i];
+               struct moved_entry *cur = (prev && prev->next_line) ?
+                               prev->next_line : NULL;
+               if (cur && !hm->cmpfn(o, cur, match, NULL)) {
+                       pmb[i] = cur;
+               } else {
+                       pmb[i] = NULL;
+               }
+       }
+}
+
+static void pmb_advance_or_null_multi_match(struct diff_options *o,
+                                           struct moved_entry *match,
+                                           struct hashmap *hm,
+                                           struct moved_entry **pmb,
+                                           int pmb_nr, int n)
+{
+       int i;
+       char *got_match = xcalloc(1, pmb_nr);
+
+       for (; match; match = hashmap_get_next(hm, match)) {
+               for (i = 0; i < pmb_nr; i++) {
+                       struct moved_entry *prev = pmb[i];
+                       struct moved_entry *cur = (prev && prev->next_line) ?
+                                       prev->next_line : NULL;
+                       if (!cur)
+                               continue;
+                       if (!cmp_in_block_with_wsd(o, cur, match, pmb[i], n))
+                               got_match[i] |= 1;
+               }
+       }
+
+       for (i = 0; i < pmb_nr; i++) {
+               if (got_match[i]) {
+                       /* Carry the white space delta forward */
+                       pmb[i]->next_line->wsd = pmb[i]->wsd;
+                       pmb[i] = pmb[i]->next_line;
+               } else
+                       pmb[i] = NULL;
+       }
+}
+
 static int shrink_potential_moved_blocks(struct moved_entry **pmb,
                                         int pmb_nr)
 {
@@ -779,6 +972,10 @@ static int shrink_potential_moved_blocks(struct moved_entry **pmb,
 
                if (lp < pmb_nr && rp > -1 && lp < rp) {
                        pmb[lp] = pmb[rp];
+                       if (pmb[rp]->wsd) {
+                               free(pmb[rp]->wsd->string);
+                               FREE_AND_NULL(pmb[rp]->wsd);
+                       }
                        pmb[rp] = NULL;
                        rp--;
                        lp++;
@@ -836,19 +1033,18 @@ static void mark_color_as_moved(struct diff_options *o,
                struct moved_entry *key;
                struct moved_entry *match = NULL;
                struct emitted_diff_symbol *l = &o->emitted_symbols->buf[n];
-               int i;
 
                switch (l->s) {
                case DIFF_SYMBOL_PLUS:
                        hm = del_lines;
                        key = prepare_entry(o, n);
-                       match = hashmap_get(hm, key, o);
+                       match = hashmap_get(hm, key, NULL);
                        free(key);
                        break;
                case DIFF_SYMBOL_MINUS:
                        hm = add_lines;
                        key = prepare_entry(o, n);
-                       match = hashmap_get(hm, key, o);
+                       match = hashmap_get(hm, key, NULL);
                        free(key);
                        break;
                default:
@@ -867,17 +1063,11 @@ static void mark_color_as_moved(struct diff_options *o,
                if (o->color_moved == COLOR_MOVED_PLAIN)
                        continue;
 
-               /* Check any potential block runs, advance each or nullify */
-               for (i = 0; i < pmb_nr; i++) {
-                       struct moved_entry *p = pmb[i];
-                       struct moved_entry *pnext = (p && p->next_line) ?
-                                       p->next_line : NULL;
-                       if (pnext && !hm->cmpfn(o, pnext, match, NULL)) {
-                               pmb[i] = p->next_line;
-                       } else {
-                               pmb[i] = NULL;
-                       }
-               }
+               if (o->color_moved_ws_handling &
+                   COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE)
+                       pmb_advance_or_null_multi_match(o, match, hm, pmb, pmb_nr, n);
+               else
+                       pmb_advance_or_null(o, match, hm, pmb, pmb_nr);
 
                pmb_nr = shrink_potential_moved_blocks(pmb, pmb_nr);
 
@@ -888,7 +1078,17 @@ static void mark_color_as_moved(struct diff_options *o,
                         */
                        for (; match; match = hashmap_get_next(hm, match)) {
                                ALLOC_GROW(pmb, pmb_nr + 1, pmb_alloc);
-                               pmb[pmb_nr++] = match;
+                               if (o->color_moved_ws_handling &
+                                   COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE) {
+                                       struct ws_delta *wsd = xmalloc(sizeof(*match->wsd));
+                                       if (compute_ws_delta(l, match->es, wsd)) {
+                                               match->wsd = wsd;
+                                               pmb[pmb_nr++] = match;
+                                       } else
+                                               free(wsd);
+                               } else {
+                                       pmb[pmb_nr++] = match;
+                               }
                        }
 
                        flipped_block = (flipped_block + 1) % 2;
@@ -899,7 +1099,7 @@ static void mark_color_as_moved(struct diff_options *o,
 
                block_length++;
 
-               if (flipped_block)
+               if (flipped_block && o->color_moved != COLOR_MOVED_BLOCKS)
                        l->flags |= DIFF_SYMBOL_MOVED_LINE_ALT;
        }
        adjust_last_block(o, n, block_length);
@@ -1871,8 +2071,8 @@ static void init_diff_words_data(struct emit_callback *ecbdata,
                if (regcomp(ecbdata->diff_words->word_regex,
                            o->word_regex,
                            REG_EXTENDED | REG_NEWLINE))
-                       die ("Invalid regular expression: %s",
-                            o->word_regex);
+                       die("invalid regular expression: %s",
+                           o->word_regex);
        }
        for (i = 0; i < ARRAY_SIZE(diff_words_styles); i++) {
                if (o->word_diff == diff_words_styles[i].type) {
@@ -3472,7 +3672,7 @@ static int reuse_worktree_file(const char *name, const struct object_id *oid, in
         * objects however would tend to be slower as they need
         * to be individually opened and inflated.
         */
-       if (!FAST_WORKING_DIRECTORY && !want_file && has_sha1_pack(oid->hash))
+       if (!FAST_WORKING_DIRECTORY && !want_file && has_object_pack(oid))
                return 0;
 
        /*
@@ -3840,7 +4040,7 @@ static const char *diff_abbrev_oid(const struct object_id *oid, int abbrev)
                char *hex = oid_to_hex(oid);
                if (abbrev < 0)
                        abbrev = FALLBACK_DEFAULT_ABBREV;
-               if (abbrev > GIT_SHA1_HEXSZ)
+               if (abbrev > the_hash_algo->hexsz)
                        BUG("oid abbreviation out of range: %d", abbrev);
                if (abbrev)
                        hex[abbrev] = '\0';
@@ -3898,13 +4098,14 @@ static void fill_metainfo(struct strbuf *msg,
                *must_show_header = 0;
        }
        if (one && two && oidcmp(&one->oid, &two->oid)) {
-               int abbrev = o->flags.full_index ? 40 : DEFAULT_ABBREV;
+               const unsigned hexsz = the_hash_algo->hexsz;
+               int abbrev = o->flags.full_index ? hexsz : DEFAULT_ABBREV;
 
                if (o->flags.binary) {
                        mmfile_t mf;
                        if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
                            (!fill_mmfile(&mf, two) && diff_filespec_is_binary(two)))
-                               abbrev = 40;
+                               abbrev = hexsz;
                }
                strbuf_addf(msg, "%s%sindex %s..%s", line_prefix, set,
                            diff_abbrev_oid(&one->oid, abbrev),
@@ -4131,6 +4332,7 @@ void diff_setup(struct diff_options *options)
        }
 
        options->color_moved = diff_color_moved_default;
+       options->color_moved_ws_handling = diff_color_moved_ws_default;
 }
 
 void diff_setup_done(struct diff_options *options)
@@ -4139,6 +4341,11 @@ void diff_setup_done(struct diff_options *options)
                              DIFF_FORMAT_NAME_STATUS |
                              DIFF_FORMAT_CHECKDIFF |
                              DIFF_FORMAT_NO_OUTPUT;
+       /*
+        * This must be signed because we're comparing against a potentially
+        * negative value.
+        */
+       const int hexsz = the_hash_algo->hexsz;
 
        if (options->set_default)
                options->set_default(options);
@@ -4219,8 +4426,8 @@ void diff_setup_done(struct diff_options *options)
                         */
                        read_cache();
        }
-       if (40 < options->abbrev)
-               options->abbrev = 40; /* full */
+       if (hexsz < options->abbrev)
+               options->abbrev = hexsz; /* full */
 
        /*
         * It does not make sense to show the first hit we happened
@@ -4705,6 +4912,8 @@ int diff_opt_parse(struct diff_options *options,
                if (cm < 0)
                        die("bad --color-moved argument: %s", arg);
                options->color_moved = cm;
+       } else if (skip_prefix(arg, "--color-moved-ws=", &arg)) {
+               options->color_moved_ws_handling = parse_color_moved_ws(arg);
        } else if (skip_to_optional_arg_default(arg, "--color-words", &options->word_regex, NULL)) {
                options->use_color = 1;
                options->word_diff = DIFF_WORDS_COLOR;
@@ -4798,8 +5007,8 @@ int diff_opt_parse(struct diff_options *options,
                options->abbrev = strtoul(arg, NULL, 10);
                if (options->abbrev < MINIMUM_ABBREV)
                        options->abbrev = MINIMUM_ABBREV;
-               else if (40 < options->abbrev)
-                       options->abbrev = 40;
+               else if (the_hash_algo->hexsz < options->abbrev)
+                       options->abbrev = the_hash_algo->hexsz;
        }
        else if ((argcount = parse_long_opt("src-prefix", av, &optarg))) {
                options->a_prefix = optarg;
@@ -4949,7 +5158,7 @@ const char *diff_aligned_abbrev(const struct object_id *oid, int len)
        const char *abbrev;
 
        /* Do we want all 40 hex characters? */
-       if (len == GIT_SHA1_HEXSZ)
+       if (len == the_hash_algo->hexsz)
                return oid_to_hex(oid);
 
        /* An abbreviated value is fine, possibly followed by an ellipsis. */
@@ -4979,7 +5188,7 @@ const char *diff_aligned_abbrev(const struct object_id *oid, int len)
         * the automatic sizing is supposed to give abblen that ensures
         * uniqueness across all objects (statistically speaking).
         */
-       if (abblen < GIT_SHA1_HEXSZ - 3) {
+       if (abblen < the_hash_algo->hexsz - 3) {
                static char hex[GIT_MAX_HEXSZ + 1];
                if (len < abblen && abblen <= len + 2)
                        xsnprintf(hex, sizeof(hex), "%s%.*s", abbrev, len+3-abblen, "..");
@@ -5535,10 +5744,12 @@ static void diff_flush_patch_all_file_pairs(struct diff_options *o)
                if (o->color_moved) {
                        struct hashmap add_lines, del_lines;
 
-                       hashmap_init(&del_lines,
-                                    (hashmap_cmp_fn)moved_entry_cmp, o, 0);
-                       hashmap_init(&add_lines,
-                                    (hashmap_cmp_fn)moved_entry_cmp, o, 0);
+                       if (o->color_moved_ws_handling &
+                           COLOR_MOVED_WS_ALLOW_INDENTATION_CHANGE)
+                               o->color_moved_ws_handling |= XDF_IGNORE_WHITESPACE;
+
+                       hashmap_init(&del_lines, moved_entry_cmp, o, 0);
+                       hashmap_init(&add_lines, moved_entry_cmp, o, 0);
 
                        add_lines_to_move_detection(o, &add_lines, &del_lines);
                        mark_color_as_moved(o, &add_lines, &del_lines);