strbuf_read_file enhancement, and use it.
[gitweb.git] / tree-diff.c
index c8275823d0eb976e95b256f2bce5497f45d5da77..26bdbdd2bfea5eab99b9f1c38936efe56f1ab45a 100644 (file)
@@ -3,11 +3,11 @@
  */
 #include "cache.h"
 #include "diff.h"
+#include "diffcore.h"
 #include "tree.h"
 
-static char *malloc_base(const char *base, const char *path, int pathlen)
+static char *malloc_base(const char *base, int baselen, const char *path, int pathlen)
 {
-       int baselen = strlen(base);
        char *newbase = xmalloc(baselen + pathlen + 2);
        memcpy(newbase, base, baselen);
        memcpy(newbase + baselen, path, pathlen);
@@ -16,9 +16,9 @@ static char *malloc_base(const char *base, const char *path, int pathlen)
 }
 
 static void show_entry(struct diff_options *opt, const char *prefix, struct tree_desc *desc,
-                      const char *base);
+                      const char *base, int baselen);
 
-static int compare_tree_entry(struct tree_desc *t1, struct tree_desc *t2, const char *base, struct diff_options *opt)
+static int compare_tree_entry(struct tree_desc *t1, struct tree_desc *t2, const char *base, int baselen, struct diff_options *opt)
 {
        unsigned mode1, mode2;
        const char *path1, *path2;
@@ -28,15 +28,15 @@ static int compare_tree_entry(struct tree_desc *t1, struct tree_desc *t2, const
        sha1 = tree_entry_extract(t1, &path1, &mode1);
        sha2 = tree_entry_extract(t2, &path2, &mode2);
 
-       pathlen1 = strlen(path1);
-       pathlen2 = strlen(path2);
+       pathlen1 = tree_entry_len(path1, sha1);
+       pathlen2 = tree_entry_len(path2, sha2);
        cmp = base_name_compare(path1, pathlen1, mode1, path2, pathlen2, mode2);
        if (cmp < 0) {
-               show_entry(opt, "-", t1, base);
+               show_entry(opt, "-", t1, base, baselen);
                return -1;
        }
        if (cmp > 0) {
-               show_entry(opt, "+", t2, base);
+               show_entry(opt, "+", t2, base, baselen);
                return 1;
        }
        if (!opt->find_copies_harder && !hashcmp(sha1, sha2) && mode1 == mode2)
@@ -47,14 +47,14 @@ static int compare_tree_entry(struct tree_desc *t1, struct tree_desc *t2, const
         * file, we need to consider it a remove and an add.
         */
        if (S_ISDIR(mode1) != S_ISDIR(mode2)) {
-               show_entry(opt, "-", t1, base);
-               show_entry(opt, "+", t2, base);
+               show_entry(opt, "-", t1, base, baselen);
+               show_entry(opt, "+", t2, base, baselen);
                return 0;
        }
 
        if (opt->recursive && S_ISDIR(mode1)) {
                int retval;
-               char *newbase = malloc_base(base, path1, pathlen1);
+               char *newbase = malloc_base(base, baselen, path1, pathlen1);
                if (opt->tree_in_recursive)
                        opt->change(opt, mode1, mode2,
                                    sha1, sha2, base, path1);
@@ -67,32 +67,46 @@ static int compare_tree_entry(struct tree_desc *t1, struct tree_desc *t2, const
        return 0;
 }
 
-static int interesting(struct tree_desc *desc, const char *base, struct diff_options *opt)
+/*
+ * Is a tree entry interesting given the pathspec we have?
+ *
+ * Return:
+ *  - 2 for "yes, and all subsequent entries will be"
+ *  - 1 for yes
+ *  - zero for no
+ *  - negative for "no, and no subsequent entries will be either"
+ */
+static int tree_entry_interesting(struct tree_desc *desc, const char *base, int baselen, struct diff_options *opt)
 {
        const char *path;
+       const unsigned char *sha1;
        unsigned mode;
        int i;
-       int baselen, pathlen;
+       int pathlen;
+       int never_interesting = -1;
 
        if (!opt->nr_paths)
                return 1;
 
-       (void)tree_entry_extract(desc, &path, &mode);
+       sha1 = tree_entry_extract(desc, &path, &mode);
 
-       pathlen = strlen(path);
-       baselen = strlen(base);
+       pathlen = tree_entry_len(path, sha1);
 
-       for (i=0; i < opt->nr_paths; i++) {
+       for (i = 0; i < opt->nr_paths; i++) {
                const char *match = opt->paths[i];
                int matchlen = opt->pathlens[i];
+               int m = -1; /* signals that we haven't called strncmp() */
 
                if (baselen >= matchlen) {
                        /* If it doesn't match, move along... */
                        if (strncmp(base, match, matchlen))
                                continue;
 
-                       /* The base is a subdirectory of a path which was specified. */
-                       return 1;
+                       /*
+                        * The base is a subdirectory of a path which
+                        * was specified, so all of them are interesting.
+                        */
+                       return 2;
                }
 
                /* Does the base match? */
@@ -102,6 +116,37 @@ static int interesting(struct tree_desc *desc, const char *base, struct diff_opt
                match += baselen;
                matchlen -= baselen;
 
+               if (never_interesting) {
+                       /*
+                        * We have not seen any match that sorts later
+                        * than the current path.
+                        */
+
+                       /*
+                        * Does match sort strictly earlier than path
+                        * with their common parts?
+                        */
+                       m = strncmp(match, path,
+                                   (matchlen < pathlen) ? matchlen : pathlen);
+                       if (m < 0)
+                               continue;
+
+                       /*
+                        * If we come here even once, that means there is at
+                        * least one pathspec that would sort equal to or
+                        * later than the path we are currently looking at.
+                        * In other words, if we have never reached this point
+                        * after iterating all pathspecs, it means all
+                        * pathspecs are either outside of base, or inside the
+                        * base but sorts strictly earlier than the current
+                        * one.  In either case, they will never match the
+                        * subsequent entries.  In such a case, we initialized
+                        * the variable to -1 and that is what will be
+                        * returned, allowing the caller to terminate early.
+                        */
+                       never_interesting = 0;
+               }
+
                if (pathlen > matchlen)
                        continue;
 
@@ -112,27 +157,50 @@ static int interesting(struct tree_desc *desc, const char *base, struct diff_opt
                                continue;
                }
 
-               if (strncmp(path, match, pathlen))
-                       continue;
-
-               return 1;
+               if (m == -1)
+                       /*
+                        * we cheated and did not do strncmp(), so we do
+                        * that here.
+                        */
+                       m = strncmp(match, path, pathlen);
+
+               /*
+                * If common part matched earlier then it is a hit,
+                * because we rejected the case where path is not a
+                * leading directory and is shorter than match.
+                */
+               if (!m)
+                       return 1;
        }
-       return 0; /* No matches */
+       return never_interesting; /* No matches */
 }
 
 /* A whole sub-tree went away or appeared */
-static void show_tree(struct diff_options *opt, const char *prefix, struct tree_desc *desc, const char *base)
+static void show_tree(struct diff_options *opt, const char *prefix, struct tree_desc *desc, const char *base, int baselen)
 {
+       int all_interesting = 0;
        while (desc->size) {
-               if (interesting(desc, base, opt))
-                       show_entry(opt, prefix, desc, base);
+               int show;
+
+               if (all_interesting)
+                       show = 1;
+               else {
+                       show = tree_entry_interesting(desc, base, baselen,
+                                                     opt);
+                       if (show == 2)
+                               all_interesting = 1;
+               }
+               if (show < 0)
+                       break;
+               if (show)
+                       show_entry(opt, prefix, desc, base, baselen);
                update_tree_entry(desc);
        }
 }
 
 /* A file entry went away or appeared */
 static void show_entry(struct diff_options *opt, const char *prefix, struct tree_desc *desc,
-                      const char *base)
+                      const char *base, int baselen)
 {
        unsigned mode;
        const char *path;
@@ -140,16 +208,18 @@ static void show_entry(struct diff_options *opt, const char *prefix, struct tree
 
        if (opt->recursive && S_ISDIR(mode)) {
                enum object_type type;
-               char *newbase = malloc_base(base, path, strlen(path));
+               int pathlen = tree_entry_len(path, sha1);
+               char *newbase = malloc_base(base, baselen, path, pathlen);
                struct tree_desc inner;
                void *tree;
+               unsigned long size;
 
-               tree = read_sha1_file(sha1, &type, &inner.size);
+               tree = read_sha1_file(sha1, &type, &size);
                if (!tree || type != OBJ_TREE)
                        die("corrupt tree sha %s", sha1_to_hex(sha1));
 
-               inner.buf = tree;
-               show_tree(opt, prefix, &inner, newbase);
+               init_tree_desc(&inner, tree, size);
+               show_tree(opt, prefix, &inner, newbase, baselen + 1 + pathlen);
 
                free(tree);
                free(newbase);
@@ -158,28 +228,54 @@ static void show_entry(struct diff_options *opt, const char *prefix, struct tree
        }
 }
 
-int diff_tree(struct tree_desc *t1, struct tree_desc *t2, const char *base, struct diff_options *opt)
+static void skip_uninteresting(struct tree_desc *t, const char *base, int baselen, struct diff_options *opt)
 {
-       while (t1->size | t2->size) {
-               if (opt->nr_paths && t1->size && !interesting(t1, base, opt)) {
-                       update_tree_entry(t1);
-                       continue;
+       int all_interesting = 0;
+       while (t->size) {
+               int show;
+
+               if (all_interesting)
+                       show = 1;
+               else {
+                       show = tree_entry_interesting(t, base, baselen, opt);
+                       if (show == 2)
+                               all_interesting = 1;
                }
-               if (opt->nr_paths && t2->size && !interesting(t2, base, opt)) {
-                       update_tree_entry(t2);
+               if (!show) {
+                       update_tree_entry(t);
                        continue;
                }
+               /* Skip it all? */
+               if (show < 0)
+                       t->size = 0;
+               return;
+       }
+}
+
+int diff_tree(struct tree_desc *t1, struct tree_desc *t2, const char *base, struct diff_options *opt)
+{
+       int baselen = strlen(base);
+
+       for (;;) {
+               if (opt->quiet && opt->has_changes)
+                       break;
+               if (opt->nr_paths) {
+                       skip_uninteresting(t1, base, baselen, opt);
+                       skip_uninteresting(t2, base, baselen, opt);
+               }
                if (!t1->size) {
-                       show_entry(opt, "+", t2, base);
+                       if (!t2->size)
+                               break;
+                       show_entry(opt, "+", t2, base, baselen);
                        update_tree_entry(t2);
                        continue;
                }
                if (!t2->size) {
-                       show_entry(opt, "-", t1, base);
+                       show_entry(opt, "-", t1, base, baselen);
                        update_tree_entry(t1);
                        continue;
                }
-               switch (compare_tree_entry(t1, t2, base, opt)) {
+               switch (compare_tree_entry(t1, t2, base, baselen, opt)) {
                case -1:
                        update_tree_entry(t1);
                        continue;
@@ -195,21 +291,99 @@ int diff_tree(struct tree_desc *t1, struct tree_desc *t2, const char *base, stru
        return 0;
 }
 
+/*
+ * Does it look like the resulting diff might be due to a rename?
+ *  - single entry
+ *  - not a valid previous file
+ */
+static inline int diff_might_be_rename(void)
+{
+       return diff_queued_diff.nr == 1 &&
+               !DIFF_FILE_VALID(diff_queued_diff.queue[0]->one);
+}
+
+static void try_to_follow_renames(struct tree_desc *t1, struct tree_desc *t2, const char *base, struct diff_options *opt)
+{
+       struct diff_options diff_opts;
+       struct diff_queue_struct *q = &diff_queued_diff;
+       struct diff_filepair *choice;
+       const char *paths[1];
+       int i;
+
+       /* Remove the file creation entry from the diff queue, and remember it */
+       choice = q->queue[0];
+       q->nr = 0;
+
+       diff_setup(&diff_opts);
+       diff_opts.recursive = 1;
+       diff_opts.detect_rename = DIFF_DETECT_RENAME;
+       diff_opts.output_format = DIFF_FORMAT_NO_OUTPUT;
+       diff_opts.single_follow = opt->paths[0];
+       paths[0] = NULL;
+       diff_tree_setup_paths(paths, &diff_opts);
+       if (diff_setup_done(&diff_opts) < 0)
+               die("unable to set up diff options to follow renames");
+       diff_tree(t1, t2, base, &diff_opts);
+       diffcore_std(&diff_opts);
+
+       /* Go through the new set of filepairing, and see if we find a more interesting one */
+       for (i = 0; i < q->nr; i++) {
+               struct diff_filepair *p = q->queue[i];
+
+               /*
+                * Found a source? Not only do we use that for the new
+                * diff_queued_diff, we will also use that as the path in
+                * the future!
+                */
+               if ((p->status == 'R' || p->status == 'C') && !strcmp(p->two->path, opt->paths[0])) {
+                       /* Switch the file-pairs around */
+                       q->queue[i] = choice;
+                       choice = p;
+
+                       /* Update the path we use from now on.. */
+                       opt->paths[0] = xstrdup(p->one->path);
+                       diff_tree_setup_paths(opt->paths, opt);
+                       break;
+               }
+       }
+
+       /*
+        * Then, discard all the non-relevane file pairs...
+        */
+       for (i = 0; i < q->nr; i++) {
+               struct diff_filepair *p = q->queue[i];
+               diff_free_filepair(p);
+       }
+
+       /*
+        * .. and re-instate the one we want (which might be either the
+        * original one, or the rename/copy we found)
+        */
+       q->queue[0] = choice;
+       q->nr = 1;
+}
+
 int diff_tree_sha1(const unsigned char *old, const unsigned char *new, const char *base, struct diff_options *opt)
 {
        void *tree1, *tree2;
        struct tree_desc t1, t2;
+       unsigned long size1, size2;
        int retval;
 
-       tree1 = read_object_with_reference(old, tree_type, &t1.size, NULL);
+       tree1 = read_object_with_reference(old, tree_type, &size1, NULL);
        if (!tree1)
                die("unable to read source tree (%s)", sha1_to_hex(old));
-       tree2 = read_object_with_reference(new, tree_type, &t2.size, NULL);
+       tree2 = read_object_with_reference(new, tree_type, &size2, NULL);
        if (!tree2)
                die("unable to read destination tree (%s)", sha1_to_hex(new));
-       t1.buf = tree1;
-       t2.buf = tree2;
+       init_tree_desc(&t1, tree1, size1);
+       init_tree_desc(&t2, tree2, size2);
        retval = diff_tree(&t1, &t2, base, opt);
+       if (opt->follow_renames && diff_might_be_rename()) {
+               init_tree_desc(&t1, tree1, size1);
+               init_tree_desc(&t2, tree2, size2);
+               try_to_follow_renames(&t1, &t2, base, opt);
+       }
        free(tree1);
        free(tree2);
        return retval;
@@ -219,15 +393,15 @@ int diff_root_tree_sha1(const unsigned char *new, const char *base, struct diff_
 {
        int retval;
        void *tree;
+       unsigned long size;
        struct tree_desc empty, real;
 
-       tree = read_object_with_reference(new, tree_type, &real.size, NULL);
+       tree = read_object_with_reference(new, tree_type, &size, NULL);
        if (!tree)
                die("unable to read root tree (%s)", sha1_to_hex(new));
-       real.buf = tree;
+       init_tree_desc(&real, tree, size);
 
-       empty.size = 0;
-       empty.buf = "";
+       init_tree_desc(&empty, "", 0);
        retval = diff_tree(&empty, &real, base, opt);
        free(tree);
        return retval;