color-words: expand docs with precise semantics
[gitweb.git] / diff.c
diff --git a/diff.c b/diff.c
index 3e6c752bb77109864f8be170f9b0e2504531b2aa..00c661f82e744b3114db6cbd4a94d9c44d10a564 100644 (file)
--- a/diff.c
+++ b/diff.c
@@ -93,12 +93,6 @@ int git_diff_ui_config(const char *var, const char *value, void *cb)
        if (!strcmp(var, "diff.external"))
                return git_config_string(&external_diff_cmd_cfg, var, value);
 
-       switch (userdiff_config_porcelain(var, value)) {
-               case 0: break;
-               case -1: return -1;
-               default: return 0;
-       }
-
        return git_diff_basic_config(var, value, cb);
 }
 
@@ -109,6 +103,12 @@ int git_diff_basic_config(const char *var, const char *value, void *cb)
                return 0;
        }
 
+       switch (userdiff_config(var, value)) {
+               case 0: break;
+               case -1: return -1;
+               default: return 0;
+       }
+
        if (!prefixcmp(var, "diff.color.") || !prefixcmp(var, "color.diff.")) {
                int slot = parse_diff_color_slot(var, 11);
                if (!value)
@@ -123,12 +123,6 @@ int git_diff_basic_config(const char *var, const char *value, void *cb)
                return 0;
        }
 
-       switch (userdiff_config_basic(var, value)) {
-               case 0: break;
-               case -1: return -1;
-               default: return 0;
-       }
-
        return git_color_default_config(var, value, cb);
 }
 
@@ -235,6 +229,8 @@ static void emit_rewrite_diff(const char *name_a,
                              const char *name_b,
                              struct diff_filespec *one,
                              struct diff_filespec *two,
+                             const char *textconv_one,
+                             const char *textconv_two,
                              struct diff_options *o)
 {
        int lc_a, lc_b;
@@ -247,6 +243,8 @@ static void emit_rewrite_diff(const char *name_a,
        const char *reset = diff_get_color(color_diff, DIFF_RESET);
        static struct strbuf a_name = STRBUF_INIT, b_name = STRBUF_INIT;
        const char *a_prefix, *b_prefix;
+       const char *data_one, *data_two;
+       size_t size_one, size_two;
 
        if (diff_mnemonic_prefix && DIFF_OPT_TST(o, REVERSE_DIFF)) {
                a_prefix = o->b_prefix;
@@ -268,8 +266,27 @@ static void emit_rewrite_diff(const char *name_a,
 
        diff_populate_filespec(one, 0);
        diff_populate_filespec(two, 0);
-       lc_a = count_lines(one->data, one->size);
-       lc_b = count_lines(two->data, two->size);
+       if (textconv_one) {
+               data_one = run_textconv(textconv_one, one, &size_one);
+               if (!data_one)
+                       die("unable to read files to diff");
+       }
+       else {
+               data_one = one->data;
+               size_one = one->size;
+       }
+       if (textconv_two) {
+               data_two = run_textconv(textconv_two, two, &size_two);
+               if (!data_two)
+                       die("unable to read files to diff");
+       }
+       else {
+               data_two = two->data;
+               size_two = two->size;
+       }
+
+       lc_a = count_lines(data_one, size_one);
+       lc_b = count_lines(data_two, size_two);
        fprintf(o->file,
                "%s--- %s%s%s\n%s+++ %s%s%s\n%s@@ -",
                metainfo, a_name.buf, name_a_tab, reset,
@@ -279,9 +296,9 @@ static void emit_rewrite_diff(const char *name_a,
        print_line_count(o->file, lc_b);
        fprintf(o->file, " @@%s\n", reset);
        if (lc_a)
-               copy_file_with_prefix(o->file, '-', one->data, one->size, old, reset);
+               copy_file_with_prefix(o->file, '-', data_one, size_one, old, reset);
        if (lc_b)
-               copy_file_with_prefix(o->file, '+', two->data, two->size, new, reset);
+               copy_file_with_prefix(o->file, '+', data_two, size_two, new, reset);
 }
 
 static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
@@ -294,100 +311,146 @@ static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
        else if (diff_populate_filespec(one, 0))
                return -1;
 
-       diff_filespec_load_driver(one);
-       if (one->driver->textconv) {
-               size_t size;
-               mf->ptr = run_textconv(one->driver->textconv, one, &size);
-               if (!mf->ptr)
-                       return -1;
-               mf->size = size;
-       }
-       else {
-               mf->ptr = one->data;
-               mf->size = one->size;
-       }
+       mf->ptr = one->data;
+       mf->size = one->size;
        return 0;
 }
 
 struct diff_words_buffer {
        mmfile_t text;
        long alloc;
-       long current; /* output pointer */
-       int suppressed_newline;
+       struct diff_words_orig {
+               const char *begin, *end;
+       } *orig;
+       int orig_nr, orig_alloc;
 };
 
 static void diff_words_append(char *line, unsigned long len,
                struct diff_words_buffer *buffer)
 {
-       if (buffer->text.size + len > buffer->alloc) {
-               buffer->alloc = (buffer->text.size + len) * 3 / 2;
-               buffer->text.ptr = xrealloc(buffer->text.ptr, buffer->alloc);
-       }
+       ALLOC_GROW(buffer->text.ptr, buffer->text.size + len, buffer->alloc);
        line++;
        len--;
        memcpy(buffer->text.ptr + buffer->text.size, line, len);
        buffer->text.size += len;
+       buffer->text.ptr[buffer->text.size] = '\0';
 }
 
 struct diff_words_data {
        struct diff_words_buffer minus, plus;
+       const char *current_plus;
        FILE *file;
+       regex_t *word_regex;
 };
 
-static void print_word(FILE *file, struct diff_words_buffer *buffer, int len, int color,
-               int suppress_newline)
+static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
 {
-       const char *ptr;
-       int eol = 0;
+       struct diff_words_data *diff_words = priv;
+       int minus_first, minus_len, plus_first, plus_len;
+       const char *minus_begin, *minus_end, *plus_begin, *plus_end;
 
-       if (len == 0)
+       if (line[0] != '@' || parse_hunk_header(line, len,
+                       &minus_first, &minus_len, &plus_first, &plus_len))
                return;
 
-       ptr  = buffer->text.ptr + buffer->current;
-       buffer->current += len;
+       /* POSIX requires that first be decremented by one if len == 0... */
+       if (minus_len) {
+               minus_begin = diff_words->minus.orig[minus_first].begin;
+               minus_end =
+                       diff_words->minus.orig[minus_first + minus_len - 1].end;
+       } else
+               minus_begin = minus_end =
+                       diff_words->minus.orig[minus_first].end;
 
-       if (ptr[len - 1] == '\n') {
-               eol = 1;
-               len--;
+       if (plus_len) {
+               plus_begin = diff_words->plus.orig[plus_first].begin;
+               plus_end = diff_words->plus.orig[plus_first + plus_len - 1].end;
+       } else
+               plus_begin = plus_end = diff_words->plus.orig[plus_first].end;
+
+       if (diff_words->current_plus != plus_begin)
+               fwrite(diff_words->current_plus,
+                               plus_begin - diff_words->current_plus, 1,
+                               diff_words->file);
+       if (minus_begin != minus_end)
+               color_fwrite_lines(diff_words->file,
+                               diff_get_color(1, DIFF_FILE_OLD),
+                               minus_end - minus_begin, minus_begin);
+       if (plus_begin != plus_end)
+               color_fwrite_lines(diff_words->file,
+                               diff_get_color(1, DIFF_FILE_NEW),
+                               plus_end - plus_begin, plus_begin);
+
+       diff_words->current_plus = plus_end;
+}
+
+/* This function starts looking at *begin, and returns 0 iff a word was found. */
+static int find_word_boundaries(mmfile_t *buffer, regex_t *word_regex,
+               int *begin, int *end)
+{
+       if (word_regex && *begin < buffer->size) {
+               regmatch_t match[1];
+               if (!regexec(word_regex, buffer->ptr + *begin, 1, match, 0)) {
+                       char *p = memchr(buffer->ptr + *begin + match[0].rm_so,
+                                       '\n', match[0].rm_eo - match[0].rm_so);
+                       *end = p ? p - buffer->ptr : match[0].rm_eo + *begin;
+                       *begin += match[0].rm_so;
+                       return *begin >= *end;
+               }
+               return -1;
        }
 
-       fputs(diff_get_color(1, color), file);
-       fwrite(ptr, len, 1, file);
-       fputs(diff_get_color(1, DIFF_RESET), file);
+       /* find the next word */
+       while (*begin < buffer->size && isspace(buffer->ptr[*begin]))
+               (*begin)++;
+       if (*begin >= buffer->size)
+               return -1;
 
-       if (eol) {
-               if (suppress_newline)
-                       buffer->suppressed_newline = 1;
-               else
-                       putc('\n', file);
-       }
+       /* find the end of the word */
+       *end = *begin + 1;
+       while (*end < buffer->size && !isspace(buffer->ptr[*end]))
+               (*end)++;
+
+       return 0;
 }
 
-static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
+/*
+ * This function splits the words in buffer->text, stores the list with
+ * newline separator into out, and saves the offsets of the original words
+ * in buffer->orig.
+ */
+static void diff_words_fill(struct diff_words_buffer *buffer, mmfile_t *out,
+               regex_t *word_regex)
 {
-       struct diff_words_data *diff_words = priv;
+       int i, j;
+       long alloc = 0;
 
-       if (diff_words->minus.suppressed_newline) {
-               if (line[0] != '+')
-                       putc('\n', diff_words->file);
-               diff_words->minus.suppressed_newline = 0;
-       }
+       out->size = 0;
+       out->ptr = NULL;
 
-       len--;
-       switch (line[0]) {
-               case '-':
-                       print_word(diff_words->file,
-                                  &diff_words->minus, len, DIFF_FILE_OLD, 1);
-                       break;
-               case '+':
-                       print_word(diff_words->file,
-                                  &diff_words->plus, len, DIFF_FILE_NEW, 0);
-                       break;
-               case ' ':
-                       print_word(diff_words->file,
-                                  &diff_words->plus, len, DIFF_PLAIN, 0);
-                       diff_words->minus.current += len;
-                       break;
+       /* fake an empty "0th" word */
+       ALLOC_GROW(buffer->orig, 1, buffer->orig_alloc);
+       buffer->orig[0].begin = buffer->orig[0].end = buffer->text.ptr;
+       buffer->orig_nr = 1;
+
+       for (i = 0; i < buffer->text.size; i++) {
+               if (find_word_boundaries(&buffer->text, word_regex, &i, &j))
+                       return;
+
+               /* store original boundaries */
+               ALLOC_GROW(buffer->orig, buffer->orig_nr + 1,
+                               buffer->orig_alloc);
+               buffer->orig[buffer->orig_nr].begin = buffer->text.ptr + i;
+               buffer->orig[buffer->orig_nr].end = buffer->text.ptr + j;
+               buffer->orig_nr++;
+
+               /* store one word */
+               ALLOC_GROW(out->ptr, out->size + j - i + 1, alloc);
+               memcpy(out->ptr + out->size, buffer->text.ptr + i, j - i);
+               out->ptr[out->size + j - i] = '\n';
+               out->size += j - i + 1;
+
+               i = j - 1;
        }
 }
 
@@ -398,38 +461,36 @@ static void diff_words_show(struct diff_words_data *diff_words)
        xdemitconf_t xecfg;
        xdemitcb_t ecb;
        mmfile_t minus, plus;
-       int i;
+
+       /* special case: only removal */
+       if (!diff_words->plus.text.size) {
+               color_fwrite_lines(diff_words->file,
+                       diff_get_color(1, DIFF_FILE_OLD),
+                       diff_words->minus.text.size, diff_words->minus.text.ptr);
+               diff_words->minus.text.size = 0;
+               return;
+       }
+
+       diff_words->current_plus = diff_words->plus.text.ptr;
 
        memset(&xpp, 0, sizeof(xpp));
        memset(&xecfg, 0, sizeof(xecfg));
-       minus.size = diff_words->minus.text.size;
-       minus.ptr = xmalloc(minus.size);
-       memcpy(minus.ptr, diff_words->minus.text.ptr, minus.size);
-       for (i = 0; i < minus.size; i++)
-               if (isspace(minus.ptr[i]))
-                       minus.ptr[i] = '\n';
-       diff_words->minus.current = 0;
-
-       plus.size = diff_words->plus.text.size;
-       plus.ptr = xmalloc(plus.size);
-       memcpy(plus.ptr, diff_words->plus.text.ptr, plus.size);
-       for (i = 0; i < plus.size; i++)
-               if (isspace(plus.ptr[i]))
-                       plus.ptr[i] = '\n';
-       diff_words->plus.current = 0;
-
+       diff_words_fill(&diff_words->minus, &minus, diff_words->word_regex);
+       diff_words_fill(&diff_words->plus, &plus, diff_words->word_regex);
        xpp.flags = XDF_NEED_MINIMAL;
-       xecfg.ctxlen = diff_words->minus.alloc + diff_words->plus.alloc;
+       /* as only the hunk header will be parsed, we need a 0-context */
+       xecfg.ctxlen = 0;
        xdi_diff_outf(&minus, &plus, fn_out_diff_words_aux, diff_words,
                      &xpp, &xecfg, &ecb);
        free(minus.ptr);
        free(plus.ptr);
+       if (diff_words->current_plus != diff_words->plus.text.ptr +
+                       diff_words->plus.text.size)
+               fwrite(diff_words->current_plus,
+                       diff_words->plus.text.ptr + diff_words->plus.text.size
+                       - diff_words->current_plus, 1,
+                       diff_words->file);
        diff_words->minus.text.size = diff_words->plus.text.size = 0;
-
-       if (diff_words->minus.suppressed_newline) {
-               putc('\n', diff_words->file);
-               diff_words->minus.suppressed_newline = 0;
-       }
 }
 
 typedef unsigned long (*sane_truncate_fn)(char *line, unsigned long len);
@@ -453,7 +514,10 @@ static void free_diff_words_data(struct emit_callback *ecbdata)
                        diff_words_show(ecbdata->diff_words);
 
                free (ecbdata->diff_words->minus.text.ptr);
+               free (ecbdata->diff_words->minus.orig);
                free (ecbdata->diff_words->plus.text.ptr);
+               free (ecbdata->diff_words->plus.orig);
+               free(ecbdata->diff_words->word_regex);
                free(ecbdata->diff_words);
                ecbdata->diff_words = NULL;
        }
@@ -1324,6 +1388,16 @@ void diff_set_mnemonic_prefix(struct diff_options *options, const char *a, const
                options->b_prefix = b;
 }
 
+static const char *get_textconv(struct diff_filespec *one)
+{
+       if (!DIFF_FILE_VALID(one))
+               return NULL;
+       if (!S_ISREG(one->mode))
+               return NULL;
+       diff_filespec_load_driver(one);
+       return one->driver->textconv;
+}
+
 static void builtin_diff(const char *name_a,
                         const char *name_b,
                         struct diff_filespec *one,
@@ -1338,6 +1412,12 @@ static void builtin_diff(const char *name_a,
        const char *set = diff_get_color_opt(o, DIFF_METAINFO);
        const char *reset = diff_get_color_opt(o, DIFF_RESET);
        const char *a_prefix, *b_prefix;
+       const char *textconv_one = NULL, *textconv_two = NULL;
+
+       if (DIFF_OPT_TST(o, ALLOW_TEXTCONV)) {
+               textconv_one = get_textconv(one);
+               textconv_two = get_textconv(two);
+       }
 
        diff_set_mnemonic_prefix(o, "a/", "b/");
        if (DIFF_OPT_TST(o, REVERSE_DIFF)) {
@@ -1381,8 +1461,11 @@ static void builtin_diff(const char *name_a,
                 */
                if ((one->mode ^ two->mode) & S_IFMT)
                        goto free_ab_and_return;
-               if (complete_rewrite) {
-                       emit_rewrite_diff(name_a, name_b, one, two, o);
+               if (complete_rewrite &&
+                   (textconv_one || !diff_filespec_is_binary(one)) &&
+                   (textconv_two || !diff_filespec_is_binary(two))) {
+                       emit_rewrite_diff(name_a, name_b, one, two,
+                                               textconv_one, textconv_two, o);
                        o->found_changes = 1;
                        goto free_ab_and_return;
                }
@@ -1392,7 +1475,8 @@ static void builtin_diff(const char *name_a,
                die("unable to read files to diff");
 
        if (!DIFF_OPT_TST(o, TEXT) &&
-           (diff_filespec_is_binary(one) || diff_filespec_is_binary(two))) {
+           ( (diff_filespec_is_binary(one) && !textconv_one) ||
+             (diff_filespec_is_binary(two) && !textconv_two) )) {
                /* Quite common confusing case */
                if (mf1.size == mf2.size &&
                    !memcmp(mf1.ptr, mf2.ptr, mf1.size))
@@ -1413,6 +1497,21 @@ static void builtin_diff(const char *name_a,
                struct emit_callback ecbdata;
                const struct userdiff_funcname *pe;
 
+               if (textconv_one) {
+                       size_t size;
+                       mf1.ptr = run_textconv(textconv_one, one, &size);
+                       if (!mf1.ptr)
+                               die("unable to read files to diff");
+                       mf1.size = size;
+               }
+               if (textconv_two) {
+                       size_t size;
+                       mf2.ptr = run_textconv(textconv_two, two, &size);
+                       if (!mf2.ptr)
+                               die("unable to read files to diff");
+                       mf2.size = size;
+               }
+
                pe = diff_funcname_pattern(one);
                if (!pe)
                        pe = diff_funcname_pattern(two);
@@ -1427,6 +1526,7 @@ static void builtin_diff(const char *name_a,
                ecbdata.file = o->file;
                xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
                xecfg.ctxlen = o->context;
+               xecfg.interhunkctxlen = o->interhunkcontext;
                xecfg.flags = XDL_EMIT_FUNCNAMES;
                if (pe)
                        xdiff_set_find_func(&xecfg, pe->pattern, pe->cflags);
@@ -1440,11 +1540,24 @@ static void builtin_diff(const char *name_a,
                        ecbdata.diff_words =
                                xcalloc(1, sizeof(struct diff_words_data));
                        ecbdata.diff_words->file = o->file;
+                       if (o->word_regex) {
+                               ecbdata.diff_words->word_regex = (regex_t *)
+                                       xmalloc(sizeof(regex_t));
+                               if (regcomp(ecbdata.diff_words->word_regex,
+                                               o->word_regex,
+                                               REG_EXTENDED | REG_NEWLINE))
+                                       die ("Invalid regular expression: %s",
+                                                       o->word_regex);
+                       }
                }
                xdi_diff_outf(&mf1, &mf2, fn_out_consume, &ecbdata,
                              &xpp, &xecfg, &ecb);
                if (DIFF_OPT_TST(o, COLOR_DIFF_WORDS))
                        free_diff_words_data(&ecbdata);
+               if (textconv_one)
+                       free(mf1.ptr);
+               if (textconv_two)
+                       free(mf2.ptr);
        }
 
  free_ab_and_return:
@@ -1727,19 +1840,18 @@ int diff_populate_filespec(struct diff_filespec *s, int size_only)
                s->size = xsize_t(st.st_size);
                if (!s->size)
                        goto empty;
-               if (size_only)
-                       return 0;
                if (S_ISLNK(st.st_mode)) {
-                       int ret;
-                       s->data = xmalloc(s->size);
-                       s->should_free = 1;
-                       ret = readlink(s->path, s->data, s->size);
-                       if (ret < 0) {
-                               free(s->data);
+                       struct strbuf sb = STRBUF_INIT;
+
+                       if (strbuf_readlink(&sb, s->path, s->size))
                                goto err_empty;
-                       }
+                       s->size = sb.len;
+                       s->data = strbuf_detach(&sb, NULL);
+                       s->should_free = 1;
                        return 0;
                }
+               if (size_only)
+                       return 0;
                fd = open(s->path, O_RDONLY);
                if (fd < 0)
                        goto err_empty;
@@ -1837,13 +1949,12 @@ static void prepare_temp_file(const char *name,
                if (S_ISLNK(st.st_mode)) {
                        int ret;
                        char buf[PATH_MAX + 1]; /* ought to be SYMLINK_MAX */
-                       size_t sz = xsize_t(st.st_size);
-                       if (sizeof(buf) <= st.st_size)
-                               die("symlink too long: %s", name);
-                       ret = readlink(name, buf, sz);
+                       ret = readlink(name, buf, sizeof(buf));
                        if (ret < 0)
                                die("readlink(%s)", name);
-                       prep_temp_blob(temp, buf, sz,
+                       if (ret == sizeof(buf))
+                               die("symlink too long: %s", name);
+                       prep_temp_blob(temp, buf, ret,
                                       (one->sha1_valid ?
                                        one->sha1 : null_sha1),
                                       (one->sha1_valid ?
@@ -1995,7 +2106,7 @@ static void diff_fill_sha1_info(struct diff_filespec *one)
                        if (lstat(one->path, &st) < 0)
                                die("stat %s", one->path);
                        if (index_path(one->sha1, one->path, &st, 0))
-                               die("cannot hash %s\n", one->path);
+                               die("cannot hash %s", one->path);
                }
        }
        else
@@ -2449,6 +2560,10 @@ int diff_opt_parse(struct diff_options *options, const char **av, int ac)
                DIFF_OPT_CLR(options, COLOR_DIFF);
        else if (!strcmp(arg, "--color-words"))
                options->flags |= DIFF_OPT_COLOR_DIFF | DIFF_OPT_COLOR_DIFF_WORDS;
+       else if (!prefixcmp(arg, "--color-words=")) {
+               options->flags |= DIFF_OPT_COLOR_DIFF | DIFF_OPT_COLOR_DIFF_WORDS;
+               options->word_regex = arg + 14;
+       }
        else if (!strcmp(arg, "--exit-code"))
                DIFF_OPT_SET(options, EXIT_WITH_STATUS);
        else if (!strcmp(arg, "--quiet"))
@@ -2457,6 +2572,10 @@ int diff_opt_parse(struct diff_options *options, const char **av, int ac)
                DIFF_OPT_SET(options, ALLOW_EXTERNAL);
        else if (!strcmp(arg, "--no-ext-diff"))
                DIFF_OPT_CLR(options, ALLOW_EXTERNAL);
+       else if (!strcmp(arg, "--textconv"))
+               DIFF_OPT_SET(options, ALLOW_TEXTCONV);
+       else if (!strcmp(arg, "--no-textconv"))
+               DIFF_OPT_CLR(options, ALLOW_TEXTCONV);
        else if (!strcmp(arg, "--ignore-submodules"))
                DIFF_OPT_SET(options, IGNORE_SUBMODULES);
 
@@ -2490,6 +2609,9 @@ int diff_opt_parse(struct diff_options *options, const char **av, int ac)
                options->b_prefix = arg + 13;
        else if (!strcmp(arg, "--no-prefix"))
                options->a_prefix = options->b_prefix = "";
+       else if (opt_arg(arg, '\0', "inter-hunk-context",
+                        &options->interhunkcontext))
+               ;
        else if (!prefixcmp(arg, "--output=")) {
                options->file = fopen(arg + strlen("--output="), "w");
                options->close_file = 1;