Merge branch 'pe/cleanup' into next
[gitweb.git] / diff.c
diff --git a/diff.c b/diff.c
index bb2a43b5b0725a5ef1896fcdd62e4d3e3ceccc04..e496905bad9853f912e1bbcc081c544524f8cb77 100644 (file)
--- a/diff.c
+++ b/diff.c
@@ -8,13 +8,62 @@
 #include "quote.h"
 #include "diff.h"
 #include "diffcore.h"
+#include "xdiff/xdiff.h"
 
-static const char *diff_opts = "-pu";
-static unsigned char null_sha1[20] = { 0, };
-
-static int reverse_diff;
 static int use_size_cache;
 
+int diff_rename_limit_default = -1;
+
+int git_diff_config(const char *var, const char *value)
+{
+       if (!strcmp(var, "diff.renamelimit")) {
+               diff_rename_limit_default = git_config_int(var, value);
+               return 0;
+       }
+
+       return git_default_config(var, value);
+}
+
+static char *quote_one(const char *str)
+{
+       int needlen;
+       char *xp;
+
+       if (!str)
+               return NULL;
+       needlen = quote_c_style(str, NULL, NULL, 0);
+       if (!needlen)
+               return strdup(str);
+       xp = xmalloc(needlen + 1);
+       quote_c_style(str, xp, NULL, 0);
+       return xp;
+}
+
+static char *quote_two(const char *one, const char *two)
+{
+       int need_one = quote_c_style(one, NULL, NULL, 1);
+       int need_two = quote_c_style(two, NULL, NULL, 1);
+       char *xp;
+
+       if (need_one + need_two) {
+               if (!need_one) need_one = strlen(one);
+               if (!need_two) need_one = strlen(two);
+
+               xp = xmalloc(need_one + need_two + 3);
+               xp[0] = '"';
+               quote_c_style(one, xp + 1, NULL, 1);
+               quote_c_style(two, xp + need_one + 1, NULL, 1);
+               strcpy(xp + need_one + need_two + 1, "\"");
+               return xp;
+       }
+       need_one = strlen(one);
+       need_two = strlen(two);
+       xp = xmalloc(need_one + need_two + 1);
+       strcpy(xp, one);
+       strcpy(xp + need_one, two);
+       return xp;
+}
+
 static const char *external_diff(void)
 {
        static const char *external_diff_cmd = NULL;
@@ -22,21 +71,7 @@ static const char *external_diff(void)
 
        if (done_preparing)
                return external_diff_cmd;
-
-       /*
-        * Default values above are meant to match the
-        * Linux kernel development style.  Examples of
-        * alternative styles you can specify via environment
-        * variables are:
-        *
-        * GIT_DIFF_OPTS="-c";
-        */
-       if (gitenv("GIT_EXTERNAL_DIFF"))
-               external_diff_cmd = gitenv("GIT_EXTERNAL_DIFF");
-
-       /* In case external diff fails... */
-       diff_opts = gitenv("GIT_DIFF_OPTS") ? : diff_opts;
-
+       external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
        done_preparing = 1;
        return external_diff_cmd;
 }
@@ -50,13 +85,12 @@ static struct diff_tempfile {
        char tmp_path[TEMPFILE_PATH_LEN];
 } diff_temp[2];
 
-static int count_lines(const char *filename)
+static int count_lines(const char *data, int size)
 {
-       FILE *in;
        int count, ch, completely_empty = 1, nl_just_seen = 0;
-       in = fopen(filename, "r");
        count = 0;
-       while ((ch = fgetc(in)) != EOF)
+       while (0 < size--) {
+               ch = *data++;
                if (ch == '\n') {
                        count++;
                        nl_just_seen = 1;
@@ -66,7 +100,7 @@ static int count_lines(const char *filename)
                        nl_just_seen = 0;
                        completely_empty = 0;
                }
-       fclose(in);
+       }
        if (completely_empty)
                return 0;
        if (!nl_just_seen)
@@ -89,12 +123,11 @@ static void print_line_count(int count)
        }
 }
 
-static void copy_file(int prefix, const char *filename)
+static void copy_file(int prefix, const char *data, int size)
 {
-       FILE *in;
        int ch, nl_just_seen = 1;
-       in = fopen(filename, "r");
-       while ((ch = fgetc(in)) != EOF) {
+       while (0 < size--) {
+               ch = *data++;
                if (nl_just_seen)
                        putchar(prefix);
                putchar(ch);
@@ -103,124 +136,163 @@ static void copy_file(int prefix, const char *filename)
                else
                        nl_just_seen = 0;
        }
-       fclose(in);
        if (!nl_just_seen)
                printf("\n\\ No newline at end of file\n");
 }
 
 static void emit_rewrite_diff(const char *name_a,
                              const char *name_b,
-                             struct diff_tempfile *temp)
+                             struct diff_filespec *one, 
+                             struct diff_filespec *two)
 {
        /* Use temp[i].name as input, name_a and name_b as labels */
        int lc_a, lc_b;
-       lc_a = count_lines(temp[0].name);
-       lc_b = count_lines(temp[1].name);
+       lc_a = count_lines(one->data, one->size);
+       lc_b = count_lines(two->data, two->size);
        printf("--- %s\n+++ %s\n@@ -", name_a, name_b);
        print_line_count(lc_a);
        printf(" +");
        print_line_count(lc_b);
        printf(" @@\n");
        if (lc_a)
-               copy_file('-', temp[0].name);
+               copy_file('-', one->data, one->size);
        if (lc_b)
-               copy_file('+', temp[1].name);
+               copy_file('+', two->data, two->size);
 }
 
-static void builtin_diff(const char *name_a,
-                        const char *name_b,
-                        struct diff_tempfile *temp,
-                        const char *xfrm_msg,
-                        int complete_rewrite)
+static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
 {
-       int i, next_at, cmd_size;
-       const char *const diff_cmd = "diff -L%s%s -L%s%s";
-       const char *const diff_arg  = "%s %s||:"; /* "||:" is to return 0 */
-       const char *input_name_sq[2];
-       const char *path0[2];
-       const char *path1[2];
-       const char *name_sq[2];
-       char *cmd;
-
-       name_sq[0] = sq_quote(name_a);
-       name_sq[1] = sq_quote(name_b);
-
-       /* diff_cmd and diff_arg have 6 %s in total which makes
-        * the sum of these strings 12 bytes larger than required.
-        * we use 2 spaces around diff-opts, and we need to count
-        * terminating NUL, so we subtract 9 here.
-        */
-       cmd_size = (strlen(diff_cmd) + strlen(diff_opts) +
-                       strlen(diff_arg) - 9);
-       for (i = 0; i < 2; i++) {
-               input_name_sq[i] = sq_quote(temp[i].name);
-               if (!strcmp(temp[i].name, "/dev/null")) {
-                       path0[i] = "/dev/null";
-                       path1[i] = "";
-               } else {
-                       path0[i] = i ? "b/" : "a/";
-                       path1[i] = name_sq[i];
-               }
-               cmd_size += (strlen(path0[i]) + strlen(path1[i]) +
-                            strlen(input_name_sq[i]));
+       if (!DIFF_FILE_VALID(one)) {
+               mf->ptr = ""; /* does not matter */
+               mf->size = 0;
+               return 0;
        }
+       else if (diff_populate_filespec(one, 0))
+               return -1;
+       mf->ptr = one->data;
+       mf->size = one->size;
+       return 0;
+}
 
-       cmd = xmalloc(cmd_size);
+struct emit_callback {
+       const char **label_path;
+};
 
-       next_at = 0;
-       next_at += snprintf(cmd+next_at, cmd_size-next_at,
-                           diff_cmd,
-                           path0[0], path1[0], path0[1], path1[1]);
-       next_at += snprintf(cmd+next_at, cmd_size-next_at,
-                           " %s ", diff_opts);
-       next_at += snprintf(cmd+next_at, cmd_size-next_at,
-                           diff_arg, input_name_sq[0], input_name_sq[1]);
+static int fn_out(void *priv, mmbuffer_t *mb, int nbuf)
+{
+       int i;
+       struct emit_callback *ecbdata = priv;
 
-       printf("diff --git a/%s b/%s\n", name_a, name_b);
-       if (!path1[0][0]) {
-               printf("new file mode %s\n", temp[1].mode);
+       if (ecbdata->label_path[0]) {
+               printf("--- %s\n", ecbdata->label_path[0]);
+               printf("+++ %s\n", ecbdata->label_path[1]);
+               ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
+       }
+       for (i = 0; i < nbuf; i++)
+               if (!fwrite(mb[i].ptr, mb[i].size, 1, stdout))
+                       return -1;
+       return 0;
+}
+
+#define FIRST_FEW_BYTES 8000
+static int mmfile_is_binary(mmfile_t *mf)
+{
+       long sz = mf->size;
+       if (FIRST_FEW_BYTES < sz)
+               sz = FIRST_FEW_BYTES;
+       if (memchr(mf->ptr, 0, sz))
+               return 1;
+       return 0;
+}
+
+static void builtin_diff(const char *name_a,
+                        const char *name_b,
+                        struct diff_filespec *one,
+                        struct diff_filespec *two,
+                        const char *xfrm_msg,
+                        int complete_rewrite)
+{
+       mmfile_t mf1, mf2;
+       const char *lbl[2];
+       char *a_one, *b_two;
+
+       a_one = quote_two("a/", name_a);
+       b_two = quote_two("b/", name_b);
+       lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
+       lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
+       printf("diff --git %s %s\n", a_one, b_two);
+       if (lbl[0][0] == '/') {
+               /* /dev/null */
+               printf("new file mode %06o\n", two->mode);
                if (xfrm_msg && xfrm_msg[0])
                        puts(xfrm_msg);
        }
-       else if (!path1[1][0]) {
-               printf("deleted file mode %s\n", temp[0].mode);
+       else if (lbl[1][0] == '/') {
+               printf("deleted file mode %06o\n", one->mode);
                if (xfrm_msg && xfrm_msg[0])
                        puts(xfrm_msg);
        }
        else {
-               if (strcmp(temp[0].mode, temp[1].mode)) {
-                       printf("old mode %s\n", temp[0].mode);
-                       printf("new mode %s\n", temp[1].mode);
+               if (one->mode != two->mode) {
+                       printf("old mode %06o\n", one->mode);
+                       printf("new mode %06o\n", two->mode);
                }
                if (xfrm_msg && xfrm_msg[0])
                        puts(xfrm_msg);
-               if (strncmp(temp[0].mode, temp[1].mode, 3))
-                       /* we do not run diff between different kind
-                        * of objects.
-                        */
-                       exit(0);
+               /*
+                * we do not run diff between different kind
+                * of objects.
+                */
+               if ((one->mode ^ two->mode) & S_IFMT)
+                       goto free_ab_and_return;
                if (complete_rewrite) {
-                       fflush(NULL);
-                       emit_rewrite_diff(name_a, name_b, temp);
-                       exit(0);
+                       emit_rewrite_diff(name_a, name_b, one, two);
+                       goto free_ab_and_return;
                }
        }
-       fflush(NULL);
-       execlp("/bin/sh","sh", "-c", cmd, NULL);
+
+       if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
+               die("unable to read files to diff");
+
+       if (mmfile_is_binary(&mf1) || mmfile_is_binary(&mf2))
+               printf("Binary files %s and %s differ\n", lbl[0], lbl[1]);
+       else {
+               /* Crazy xdl interfaces.. */
+               const char *diffopts = getenv("GIT_DIFF_OPTS");
+               xpparam_t xpp;
+               xdemitconf_t xecfg;
+               xdemitcb_t ecb;
+               struct emit_callback ecbdata;
+
+               ecbdata.label_path = lbl;
+               xpp.flags = XDF_NEED_MINIMAL;
+               xecfg.ctxlen = 3;
+               xecfg.flags = XDL_EMIT_FUNCNAMES;
+               if (!diffopts)
+                       ;
+               else if (!strncmp(diffopts, "--unified=", 10))
+                       xecfg.ctxlen = strtoul(diffopts + 10, NULL, 10);
+               else if (!strncmp(diffopts, "-u", 2))
+                       xecfg.ctxlen = strtoul(diffopts + 2, NULL, 10);
+               ecb.outf = fn_out;
+               ecb.priv = &ecbdata;
+               xdl_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
+       }
+
+ free_ab_and_return:
+       free(a_one);
+       free(b_two);
+       return;
 }
 
 struct diff_filespec *alloc_filespec(const char *path)
 {
        int namelen = strlen(path);
        struct diff_filespec *spec = xmalloc(sizeof(*spec) + namelen + 1);
+
+       memset(spec, 0, sizeof(*spec));
        spec->path = (char *)(spec + 1);
-       strcpy(spec->path, path);
-       spec->should_free = spec->should_munmap = 0;
-       spec->xfrm_flags = 0;
-       spec->size = 0;
-       spec->data = NULL;
-       spec->mode = 0;
-       memset(spec->sha1, 0, 20);
+       memcpy(spec->path, path, namelen+1);
        return spec;
 }
 
@@ -228,7 +300,7 @@ void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
                   unsigned short mode)
 {
        if (mode) {
-               spec->mode = DIFF_FILE_CANON_MODE(mode);
+               spec->mode = canon_mode(mode);
                memcpy(spec->sha1, sha1, 20);
                spec->sha1_valid = !!memcmp(sha1, null_sha1, 20);
        }
@@ -267,7 +339,7 @@ static int work_tree_matches(const char *name, const unsigned char *sha1)
        ce = active_cache[pos];
        if ((lstat(name, &st) < 0) ||
            !S_ISREG(st.st_mode) || /* careful! */
-           ce_match_stat(ce, &st) ||
+           ce_match_stat(ce, &st, 0) ||
            memcmp(sha1, ce->sha1, 20))
                return 0;
        /* we return 1 only when we can stat, it is a regular file,
@@ -405,19 +477,22 @@ int diff_populate_filespec(struct diff_filespec *s, int size_only)
        return 0;
 }
 
-void diff_free_filespec(struct diff_filespec *s)
+void diff_free_filespec_data(struct diff_filespec *s)
 {
        if (s->should_free)
                free(s->data);
        else if (s->should_munmap)
                munmap(s->data, s->size);
-       free(s);
+       s->should_free = s->should_munmap = 0;
+       s->data = NULL;
+       free(s->cnt_data);
+       s->cnt_data = NULL;
 }
 
 static void prep_temp_blob(struct diff_tempfile *temp,
                           void *blob,
                           unsigned long size,
-                          unsigned char *sha1,
+                          const unsigned char *sha1,
                           int mode)
 {
        int fd;
@@ -459,9 +534,9 @@ static void prepare_temp_file(const char *name,
                }
                if (S_ISLNK(st.st_mode)) {
                        int ret;
-                       char *buf, buf_[1024];
-                       buf = ((sizeof(buf_) < st.st_size) ?
-                              xmalloc(st.st_size) : buf_);
+                       char buf[PATH_MAX + 1]; /* ought to be SYMLINK_MAX */
+                       if (sizeof(buf) <= st.st_size)
+                               die("symlink too long: %s", name);
                        ret = readlink(name, buf, st.st_size);
                        if (ret < 0)
                                die("readlink(%s)", name);
@@ -510,6 +585,42 @@ static void remove_tempfile(void)
 static void remove_tempfile_on_signal(int signo)
 {
        remove_tempfile();
+       signal(SIGINT, SIG_DFL);
+       raise(signo);
+}
+
+static int spawn_prog(const char *pgm, const char **arg)
+{
+       pid_t pid;
+       int status;
+
+       fflush(NULL);
+       pid = fork();
+       if (pid < 0)
+               die("unable to fork");
+       if (!pid) {
+               execvp(pgm, (char *const*) arg);
+               exit(255);
+       }
+
+       while (waitpid(pid, &status, 0) < 0) {
+               if (errno == EINTR)
+                       continue;
+               return -1;
+       }
+
+       /* Earlier we did not check the exit status because
+        * diff exits non-zero if files are different, and
+        * we are not interested in knowing that.  It was a
+        * mistake which made it harder to quit a diff-*
+        * session that uses the git-apply-patch-script as
+        * the GIT_EXTERNAL_DIFF.  A custom GIT_EXTERNAL_DIFF
+        * should also exit non-zero only when it wants to
+        * abort the entire diff-* session.
+        */
+       if (WIFEXITED(status) && !WEXITSTATUS(status))
+               return 0;
+       return -1;
 }
 
 /* An external diff command takes:
@@ -526,14 +637,17 @@ static void run_external_diff(const char *pgm,
                              const char *xfrm_msg,
                              int complete_rewrite)
 {
+       const char *spawn_arg[10];
        struct diff_tempfile *temp = diff_temp;
-       pid_t pid;
-       int status;
+       int retval;
        static int atexit_asked = 0;
+       const char *othername;
+       const char **arg = &spawn_arg[0];
 
+       othername = (other? other : name);
        if (one && two) {
                prepare_temp_file(name, &temp[0], one);
-               prepare_temp_file(other ? : name, &temp[1], two);
+               prepare_temp_file(othername, &temp[1], two);
                if (! atexit_asked &&
                    (temp[0].name == temp[0].tmp_path ||
                     temp[1].name == temp[1].tmp_path)) {
@@ -543,114 +657,146 @@ static void run_external_diff(const char *pgm,
                signal(SIGINT, remove_tempfile_on_signal);
        }
 
-       fflush(NULL);
-       pid = fork();
-       if (pid < 0)
-               die("unable to fork");
-       if (!pid) {
-               if (pgm) {
-                       if (one && two) {
-                               const char *exec_arg[10];
-                               const char **arg = &exec_arg[0];
-                               *arg++ = pgm;
-                               *arg++ = name;
-                               *arg++ = temp[0].name;
-                               *arg++ = temp[0].hex;
-                               *arg++ = temp[0].mode;
-                               *arg++ = temp[1].name;
-                               *arg++ = temp[1].hex;
-                               *arg++ = temp[1].mode;
-                               if (other) {
-                                       *arg++ = other;
-                                       *arg++ = xfrm_msg;
-                               }
-                               *arg = NULL;
-                               execvp(pgm, (char *const*) exec_arg);
-                       }
-                       else
-                               execlp(pgm, pgm, name, NULL);
+       if (one && two) {
+               *arg++ = pgm;
+               *arg++ = name;
+               *arg++ = temp[0].name;
+               *arg++ = temp[0].hex;
+               *arg++ = temp[0].mode;
+               *arg++ = temp[1].name;
+               *arg++ = temp[1].hex;
+               *arg++ = temp[1].mode;
+               if (other) {
+                       *arg++ = other;
+                       *arg++ = xfrm_msg;
                }
-               /*
-                * otherwise we use the built-in one.
-                */
-               if (one && two)
-                       builtin_diff(name, other ? : name, temp, xfrm_msg,
-                                    complete_rewrite);
-               else
-                       printf("* Unmerged path %s\n", name);
-               exit(0);
+       } else {
+               *arg++ = pgm;
+               *arg++ = name;
        }
-       if (waitpid(pid, &status, 0) < 0 ||
-           !WIFEXITED(status) || WEXITSTATUS(status)) {
-               /* Earlier we did not check the exit status because
-                * diff exits non-zero if files are different, and
-                * we are not interested in knowing that.  It was a
-                * mistake which made it harder to quit a diff-*
-                * session that uses the git-apply-patch-script as
-                * the GIT_EXTERNAL_DIFF.  A custom GIT_EXTERNAL_DIFF
-                * should also exit non-zero only when it wants to
-                * abort the entire diff-* session.
-                */
-               remove_tempfile();
+       *arg = NULL;
+       retval = spawn_prog(pgm, spawn_arg);
+       remove_tempfile();
+       if (retval) {
                fprintf(stderr, "external diff died, stopping at %s.\n", name);
                exit(1);
        }
-       remove_tempfile();
 }
 
-static void run_diff(struct diff_filepair *p)
+static void run_diff_cmd(const char *pgm,
+                        const char *name,
+                        const char *other,
+                        struct diff_filespec *one,
+                        struct diff_filespec *two,
+                        const char *xfrm_msg,
+                        int complete_rewrite)
+{
+       if (pgm) {
+               run_external_diff(pgm, name, other, one, two, xfrm_msg,
+                                 complete_rewrite);
+               return;
+       }
+       if (one && two)
+               builtin_diff(name, other ? other : name,
+                            one, two, xfrm_msg, complete_rewrite);
+       else
+               printf("* Unmerged path %s\n", name);
+}
+
+static void diff_fill_sha1_info(struct diff_filespec *one)
+{
+       if (DIFF_FILE_VALID(one)) {
+               if (!one->sha1_valid) {
+                       struct stat st;
+                       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);
+               }
+       }
+       else
+               memset(one->sha1, 0, 20);
+}
+
+static void run_diff(struct diff_filepair *p, struct diff_options *o)
 {
        const char *pgm = external_diff();
-       char msg_[PATH_MAX*2+200], *xfrm_msg;
+       char msg[PATH_MAX*2+300], *xfrm_msg;
        struct diff_filespec *one;
        struct diff_filespec *two;
        const char *name;
        const char *other;
+       char *name_munged, *other_munged;
        int complete_rewrite = 0;
+       int len;
 
        if (DIFF_PAIR_UNMERGED(p)) {
                /* unmerged */
-               run_external_diff(pgm, p->one->path, NULL, NULL, NULL, NULL,
-                                 0);
+               run_diff_cmd(pgm, p->one->path, NULL, NULL, NULL, NULL, 0);
                return;
        }
 
        name = p->one->path;
        other = (strcmp(name, p->two->path) ? p->two->path : NULL);
+       name_munged = quote_one(name);
+       other_munged = quote_one(other);
        one = p->one; two = p->two;
+
+       diff_fill_sha1_info(one);
+       diff_fill_sha1_info(two);
+
+       len = 0;
        switch (p->status) {
        case DIFF_STATUS_COPIED:
-               sprintf(msg_,
-                       "similarity index %d%%\n"
-                       "copy from %s\n"
-                       "copy to %s",
-                       (int)(0.5 + p->score * 100.0/MAX_SCORE),
-                       name, other);
-               xfrm_msg = msg_;
+               len += snprintf(msg + len, sizeof(msg) - len,
+                               "similarity index %d%%\n"
+                               "copy from %s\n"
+                               "copy to %s\n",
+                               (int)(0.5 + p->score * 100.0/MAX_SCORE),
+                               name_munged, other_munged);
                break;
        case DIFF_STATUS_RENAMED:
-               sprintf(msg_,
-                       "similarity index %d%%\n"
-                       "rename from %s\n"
-                       "rename to %s",
-                       (int)(0.5 + p->score * 100.0/MAX_SCORE),
-                       name, other);
-               xfrm_msg = msg_;
+               len += snprintf(msg + len, sizeof(msg) - len,
+                               "similarity index %d%%\n"
+                               "rename from %s\n"
+                               "rename to %s\n",
+                               (int)(0.5 + p->score * 100.0/MAX_SCORE),
+                               name_munged, other_munged);
                break;
        case DIFF_STATUS_MODIFIED:
                if (p->score) {
-                       sprintf(msg_,
-                               "dissimilarity index %d%%",
-                               (int)(0.5 + p->score * 100.0/MAX_SCORE));
-                       xfrm_msg = msg_;
+                       len += snprintf(msg + len, sizeof(msg) - len,
+                                       "dissimilarity index %d%%\n",
+                                       (int)(0.5 + p->score *
+                                             100.0/MAX_SCORE));
                        complete_rewrite = 1;
                        break;
                }
                /* fallthru */
        default:
-               xfrm_msg = NULL;
+               /* nothing */
+               ;
        }
 
+       if (memcmp(one->sha1, two->sha1, 20)) {
+               char one_sha1[41];
+               int abbrev = o->full_index ? 40 : DEFAULT_ABBREV;
+               memcpy(one_sha1, sha1_to_hex(one->sha1), 41);
+
+               len += snprintf(msg + len, sizeof(msg) - len,
+                               "index %.*s..%.*s",
+                               abbrev, one_sha1, abbrev,
+                               sha1_to_hex(two->sha1));
+               if (one->mode == two->mode)
+                       len += snprintf(msg + len, sizeof(msg) - len,
+                                       " %06o", one->mode);
+               len += snprintf(msg + len, sizeof(msg) - len, "\n");
+       }
+
+       if (len)
+               msg[--len] = 0;
+       xfrm_msg = len ? msg : NULL;
+
        if (!pgm &&
            DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
            (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
@@ -658,57 +804,152 @@ static void run_diff(struct diff_filepair *p)
                 * needs to be split into deletion and creation.
                 */
                struct diff_filespec *null = alloc_filespec(two->path);
-               run_external_diff(NULL, name, other, one, null, xfrm_msg, 0);
+               run_diff_cmd(NULL, name, other, one, null, xfrm_msg, 0);
                free(null);
                null = alloc_filespec(one->path);
-               run_external_diff(NULL, name, other, null, two, xfrm_msg, 0);
+               run_diff_cmd(NULL, name, other, null, two, xfrm_msg, 0);
                free(null);
        }
        else
-               run_external_diff(pgm, name, other, one, two, xfrm_msg,
-                                 complete_rewrite);
+               run_diff_cmd(pgm, name, other, one, two, xfrm_msg,
+                            complete_rewrite);
+
+       free(name_munged);
+       free(other_munged);
+}
+
+void diff_setup(struct diff_options *options)
+{
+       memset(options, 0, sizeof(*options));
+       options->output_format = DIFF_FORMAT_RAW;
+       options->line_termination = '\n';
+       options->break_opt = -1;
+       options->rename_limit = -1;
+
+       options->change = diff_change;
+       options->add_remove = diff_addremove;
 }
 
-void diff_setup(int flags)
+int diff_setup_done(struct diff_options *options)
 {
-       if (flags & DIFF_SETUP_REVERSE)
-               reverse_diff = 1;
-       if (flags & DIFF_SETUP_USE_CACHE) {
+       if ((options->find_copies_harder &&
+            options->detect_rename != DIFF_DETECT_COPY) ||
+           (0 <= options->rename_limit && !options->detect_rename))
+               return -1;
+       if (options->detect_rename && options->rename_limit < 0)
+               options->rename_limit = diff_rename_limit_default;
+       if (options->setup & DIFF_SETUP_USE_CACHE) {
                if (!active_cache)
                        /* read-cache does not die even when it fails
                         * so it is safe for us to do this here.  Also
                         * it does not smudge active_cache or active_nr
                         * when it fails, so we do not have to worry about
-                        * cleaning it up oufselves either.
+                        * cleaning it up ourselves either.
                         */
                        read_cache();
        }
-       if (flags & DIFF_SETUP_USE_SIZE_CACHE)
+       if (options->setup & DIFF_SETUP_USE_SIZE_CACHE)
                use_size_cache = 1;
-       
+       if (options->abbrev <= 0 || 40 < options->abbrev)
+               options->abbrev = 40; /* full */
+
+       return 0;
+}
+
+int diff_opt_parse(struct diff_options *options, const char **av, int ac)
+{
+       const char *arg = av[0];
+       if (!strcmp(arg, "-p") || !strcmp(arg, "-u"))
+               options->output_format = DIFF_FORMAT_PATCH;
+       else if (!strcmp(arg, "-z"))
+               options->line_termination = 0;
+       else if (!strncmp(arg, "-l", 2))
+               options->rename_limit = strtoul(arg+2, NULL, 10);
+       else if (!strcmp(arg, "--full-index"))
+               options->full_index = 1;
+       else if (!strcmp(arg, "--name-only"))
+               options->output_format = DIFF_FORMAT_NAME;
+       else if (!strcmp(arg, "--name-status"))
+               options->output_format = DIFF_FORMAT_NAME_STATUS;
+       else if (!strcmp(arg, "-R"))
+               options->reverse_diff = 1;
+       else if (!strncmp(arg, "-S", 2))
+               options->pickaxe = arg + 2;
+       else if (!strcmp(arg, "-s"))
+               options->output_format = DIFF_FORMAT_NO_OUTPUT;
+       else if (!strncmp(arg, "-O", 2))
+               options->orderfile = arg + 2;
+       else if (!strncmp(arg, "--diff-filter=", 14))
+               options->filter = arg + 14;
+       else if (!strcmp(arg, "--pickaxe-all"))
+               options->pickaxe_opts = DIFF_PICKAXE_ALL;
+       else if (!strncmp(arg, "-B", 2)) {
+               if ((options->break_opt =
+                    diff_scoreopt_parse(arg)) == -1)
+                       return -1;
+       }
+       else if (!strncmp(arg, "-M", 2)) {
+               if ((options->rename_score =
+                    diff_scoreopt_parse(arg)) == -1)
+                       return -1;
+               options->detect_rename = DIFF_DETECT_RENAME;
+       }
+       else if (!strncmp(arg, "-C", 2)) {
+               if ((options->rename_score =
+                    diff_scoreopt_parse(arg)) == -1)
+                       return -1;
+               options->detect_rename = DIFF_DETECT_COPY;
+       }
+       else if (!strcmp(arg, "--find-copies-harder"))
+               options->find_copies_harder = 1;
+       else if (!strcmp(arg, "--abbrev"))
+               options->abbrev = DEFAULT_ABBREV;
+       else if (!strncmp(arg, "--abbrev=", 9)) {
+               options->abbrev = strtoul(arg + 9, NULL, 10);
+               if (options->abbrev < MINIMUM_ABBREV)
+                       options->abbrev = MINIMUM_ABBREV;
+               else if (40 < options->abbrev)
+                       options->abbrev = 40;
+       }
+       else
+               return 0;
+       return 1;
 }
 
 static int parse_num(const char **cp_p)
 {
-       int num, scale, ch, cnt;
+       unsigned long num, scale;
+       int ch, dot;
        const char *cp = *cp_p;
 
-       cnt = num = 0;
+       num = 0;
        scale = 1;
-       while ('0' <= (ch = *cp) && ch <= '9') {
-               if (cnt++ < 5) {
-                       /* We simply ignore more than 5 digits precision. */
-                       scale *= 10;
-                       num = num * 10 + ch - '0';
+       dot = 0;
+       for(;;) {
+               ch = *cp;
+               if ( !dot && ch == '.' ) {
+                       scale = 1;
+                       dot = 1;
+               } else if ( ch == '%' ) {
+                       scale = dot ? scale*100 : 100;
+                       cp++;   /* % is always at the end */
+                       break;
+               } else if ( ch >= '0' && ch <= '9' ) {
+                       if ( scale < 100000 ) {
+                               scale *= 10;
+                               num = (num*10) + (ch-'0');
+                       }
+               } else {
+                       break;
                }
-               *cp++;
+               cp++;
        }
        *cp_p = cp;
 
        /* user says num divided by scale and we say internally that
         * is MAX_SCORE * num / scale.
         */
-       return (MAX_SCORE * num / scale);
+       return (num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale);
 }
 
 int diff_scoreopt_parse(const char *opt)
@@ -762,33 +1003,61 @@ struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
        dp->status = 0;
        dp->source_stays = 0;
        dp->broken_pair = 0;
-       diff_q(queue, dp);
+       if (queue)
+               diff_q(queue, dp);
        return dp;
 }
 
 void diff_free_filepair(struct diff_filepair *p)
 {
-       diff_free_filespec(p->one);
-       diff_free_filespec(p->two);
+       diff_free_filespec_data(p->one);
+       diff_free_filespec_data(p->two);
+       free(p->one);
+       free(p->two);
        free(p);
 }
 
+/* This is different from find_unique_abbrev() in that
+ * it stuffs the result with dots for alignment.
+ */
+const char *diff_unique_abbrev(const unsigned char *sha1, int len)
+{
+       int abblen;
+       const char *abbrev;
+       if (len == 40)
+               return sha1_to_hex(sha1);
+
+       abbrev = find_unique_abbrev(sha1, len);
+       if (!abbrev)
+               return sha1_to_hex(sha1);
+       abblen = strlen(abbrev);
+       if (abblen < 37) {
+               static char hex[41];
+               if (len < abblen && abblen <= len + 2)
+                       sprintf(hex, "%s%.*s", abbrev, len+3-abblen, "..");
+               else
+                       sprintf(hex, "%s...", abbrev);
+               return hex;
+       }
+       return sha1_to_hex(sha1);
+}
+
 static void diff_flush_raw(struct diff_filepair *p,
                           int line_termination,
-                          int inter_name_termination)
+                          int inter_name_termination,
+                          struct diff_options *options)
 {
        int two_paths;
        char status[10];
+       int abbrev = options->abbrev;
+       const char *path_one, *path_two;
+       int output_format = options->output_format;
 
+       path_one = p->one->path;
+       path_two = p->two->path;
        if (line_termination) {
-               const char *const err =
-                       "path %s cannot be expressed without -z";
-               if (strchr(p->one->path, line_termination) ||
-                   strchr(p->one->path, inter_name_termination))
-                       die(err, p->one->path);
-               if (strchr(p->two->path, line_termination) ||
-                   strchr(p->two->path, inter_name_termination))
-                       die(err, p->two->path);
+               path_one = quote_one(path_one);
+               path_two = quote_one(path_two);
        }
 
        if (p->score)
@@ -811,22 +1080,36 @@ static void diff_flush_raw(struct diff_filepair *p,
                two_paths = 0;
                break;
        }
-       printf(":%06o %06o %s ",
-              p->one->mode, p->two->mode, sha1_to_hex(p->one->sha1));
-       printf("%s %s%c%s",
-              sha1_to_hex(p->two->sha1),
-              status,
-              inter_name_termination,
-              p->one->path);
+       if (output_format != DIFF_FORMAT_NAME_STATUS) {
+               printf(":%06o %06o %s ",
+                      p->one->mode, p->two->mode,
+                      diff_unique_abbrev(p->one->sha1, abbrev));
+               printf("%s ",
+                      diff_unique_abbrev(p->two->sha1, abbrev));
+       }
+       printf("%s%c%s", status, inter_name_termination, path_one);
        if (two_paths)
-               printf("%c%s", inter_name_termination, p->two->path);
+               printf("%c%s", inter_name_termination, path_two);
        putchar(line_termination);
+       if (path_one != p->one->path)
+               free((void*)path_one);
+       if (path_two != p->two->path)
+               free((void*)path_two);
 }
 
 static void diff_flush_name(struct diff_filepair *p,
+                           int inter_name_termination,
                            int line_termination)
 {
-       printf("%s%c", p->two->path, line_termination);
+       char *path = p->two->path;
+
+       if (line_termination)
+               path = quote_one(p->two->path);
+       else
+               path = p->two->path;
+       printf("%s%c", path, line_termination);
+       if (p->two->path != path)
+               free(path);
 }
 
 int diff_unmodified_pair(struct diff_filepair *p)
@@ -863,7 +1146,7 @@ int diff_unmodified_pair(struct diff_filepair *p)
        return 0;
 }
 
-static void diff_flush_patch(struct diff_filepair *p)
+static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
 {
        if (diff_unmodified_pair(p))
                return;
@@ -872,7 +1155,7 @@ static void diff_flush_patch(struct diff_filepair *p)
            (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
                return; /* no tree diffs in patch format */ 
 
-       run_diff(p);
+       run_diff(p, o);
 }
 
 int diff_queue_is_empty(void)
@@ -889,13 +1172,13 @@ int diff_queue_is_empty(void)
 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
 {
        fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
-               x, one ? : "",
+               x, one ? one : "",
                s->path,
                DIFF_FILE_VALID(s) ? "valid" : "invalid",
                s->mode,
                s->sha1_valid ? sha1_to_hex(s->sha1) : "");
        fprintf(stderr, "queue[%d] %s size %lu flags %d\n",
-               x, one ? : "",
+               x, one ? one : "",
                s->size, s->xfrm_flags);
 }
 
@@ -904,7 +1187,7 @@ void diff_debug_filepair(const struct diff_filepair *p, int i)
        diff_debug_filespec(p->one, i, "one");
        diff_debug_filespec(p->two, i, "two");
        fprintf(stderr, "score %d, status %c stays %d broken %d\n",
-               p->score, p->status ? : '?',
+               p->score, p->status ? p->status : '?',
                p->source_stays, p->broken_pair);
 }
 
@@ -952,7 +1235,9 @@ static void diff_resolve_rename_copy(void)
                        }
                        /* See if there is some other filepair that
                         * copies from the same source as us.  If so
-                        * we are a copy.  Otherwise we are a rename.
+                        * we are a copy.  Otherwise we are either a
+                        * copy if the path stays, or a rename if it
+                        * does not, but we already handled "stays" case.
                         */
                        for (j = i + 1; j < q->nr; j++) {
                                pp = q->queue[j];
@@ -982,37 +1267,42 @@ static void diff_resolve_rename_copy(void)
        diff_debug_queue("resolve-rename-copy done", q);
 }
 
-void diff_flush(int diff_output_style, int line_termination)
+void diff_flush(struct diff_options *options)
 {
        struct diff_queue_struct *q = &diff_queued_diff;
        int i;
        int inter_name_termination = '\t';
+       int diff_output_format = options->output_format;
+       int line_termination = options->line_termination;
 
        if (!line_termination)
                inter_name_termination = 0;
 
        for (i = 0; i < q->nr; i++) {
                struct diff_filepair *p = q->queue[i];
-               if ((diff_output_style == DIFF_FORMAT_NO_OUTPUT) ||
+               if ((diff_output_format == DIFF_FORMAT_NO_OUTPUT) ||
                    (p->status == DIFF_STATUS_UNKNOWN))
                        continue;
                if (p->status == 0)
                        die("internal error in diff-resolve-rename-copy");
-               switch (diff_output_style) {
+               switch (diff_output_format) {
                case DIFF_FORMAT_PATCH:
-                       diff_flush_patch(p);
+                       diff_flush_patch(p, options);
                        break;
                case DIFF_FORMAT_RAW:
+               case DIFF_FORMAT_NAME_STATUS:
                        diff_flush_raw(p, line_termination,
-                                      inter_name_termination);
+                                      inter_name_termination,
+                                      options);
                        break;
                case DIFF_FORMAT_NAME:
-                       diff_flush_name(p, line_termination);
+                       diff_flush_name(p,
+                                       inter_name_termination,
+                                       line_termination);
                        break;
                }
-       }
-       for (i = 0; i < q->nr; i++)
                diff_free_filepair(q->queue[i]);
+       }
        free(q->queue);
        q->queue = NULL;
        q->nr = q->alloc = 0;
@@ -1074,45 +1364,36 @@ static void diffcore_apply_filter(const char *filter)
        *q = outq;
 }
 
-void diffcore_std(const char **paths,
-                 int detect_rename, int rename_score,
-                 const char *pickaxe, int pickaxe_opts,
-                 int break_opt,
-                 const char *orderfile,
-                 const char *filter)
+void diffcore_std(struct diff_options *options)
 {
-       if (paths && paths[0])
-               diffcore_pathspec(paths);
-       if (break_opt != -1)
-               diffcore_break(break_opt);
-       if (detect_rename)
-               diffcore_rename(detect_rename, rename_score);
-       if (break_opt != -1)
+       if (options->paths && options->paths[0])
+               diffcore_pathspec(options->paths);
+       if (options->break_opt != -1)
+               diffcore_break(options->break_opt);
+       if (options->detect_rename)
+               diffcore_rename(options);
+       if (options->break_opt != -1)
                diffcore_merge_broken();
-       if (pickaxe)
-               diffcore_pickaxe(pickaxe, pickaxe_opts);
-       if (orderfile)
-               diffcore_order(orderfile);
+       if (options->pickaxe)
+               diffcore_pickaxe(options->pickaxe, options->pickaxe_opts);
+       if (options->orderfile)
+               diffcore_order(options->orderfile);
        diff_resolve_rename_copy();
-       diffcore_apply_filter(filter);
+       diffcore_apply_filter(options->filter);
 }
 
 
-void diffcore_std_no_resolve(const char **paths,
-                            const char *pickaxe, int pickaxe_opts,
-                            const char *orderfile,
-                            const char *filter)
+void diffcore_std_no_resolve(struct diff_options *options)
 {
-       if (paths && paths[0])
-               diffcore_pathspec(paths);
-       if (pickaxe)
-               diffcore_pickaxe(pickaxe, pickaxe_opts);
-       if (orderfile)
-               diffcore_order(orderfile);
-       diffcore_apply_filter(filter);
+       if (options->pickaxe)
+               diffcore_pickaxe(options->pickaxe, options->pickaxe_opts);
+       if (options->orderfile)
+               diffcore_order(options->orderfile);
+       diffcore_apply_filter(options->filter);
 }
 
-void diff_addremove(int addremove, unsigned mode,
+void diff_addremove(struct diff_options *options,
+                   int addremove, unsigned mode,
                    const unsigned char *sha1,
                    const char *base, const char *path)
 {
@@ -1131,7 +1412,7 @@ void diff_addremove(int addremove, unsigned mode,
         * Before the final output happens, they are pruned after
         * merged into rename/copy pairs as appropriate.
         */
-       if (reverse_diff)
+       if (options->reverse_diff)
                addremove = (addremove == '+' ? '-' :
                             addremove == '-' ? '+' : addremove);
 
@@ -1148,30 +1429,8 @@ void diff_addremove(int addremove, unsigned mode,
        diff_queue(&diff_queued_diff, one, two);
 }
 
-void diff_helper_input(unsigned old_mode,
-                      unsigned new_mode,
-                      const unsigned char *old_sha1,
-                      const unsigned char *new_sha1,
-                      const char *old_path,
-                      int status,
-                      int score,
-                      const char *new_path)
-{
-       struct diff_filespec *one, *two;
-       struct diff_filepair *dp;
-
-       one = alloc_filespec(old_path);
-       two = alloc_filespec(new_path);
-       if (old_mode)
-               fill_filespec(one, old_sha1, old_mode);
-       if (new_mode)
-               fill_filespec(two, new_sha1, new_mode);
-       dp = diff_queue(&diff_queued_diff, one, two);
-       dp->score = score * MAX_SCORE / 100;
-       dp->status = status;
-}
-
-void diff_change(unsigned old_mode, unsigned new_mode,
+void diff_change(struct diff_options *options,
+                unsigned old_mode, unsigned new_mode,
                 const unsigned char *old_sha1,
                 const unsigned char *new_sha1,
                 const char *base, const char *path) 
@@ -1179,7 +1438,7 @@ void diff_change(unsigned old_mode, unsigned new_mode,
        char concatpath[PATH_MAX];
        struct diff_filespec *one, *two;
 
-       if (reverse_diff) {
+       if (options->reverse_diff) {
                unsigned tmp;
                const unsigned char *tmp_c;
                tmp = old_mode; old_mode = new_mode; new_mode = tmp;
@@ -1195,7 +1454,8 @@ void diff_change(unsigned old_mode, unsigned new_mode,
        diff_queue(&diff_queued_diff, one, two);
 }
 
-void diff_unmerge(const char *path)
+void diff_unmerge(struct diff_options *options,
+                 const char *path)
 {
        struct diff_filespec *one, *two;
        one = alloc_filespec(path);