vcs-svn: improve reporting of input errors
[gitweb.git] / vcs-svn / svndump.c
index 65bd335aa26f71b2df82ae490e354736a3a73992..15f822ea844ba64a12d19e02e12f244b2f99d4cb 100644 (file)
@@ -30,7 +30,9 @@
 /* Create memory pool for log messages */
 obj_pool_gen(log, char, 4096)
 
-static char* log_copy(uint32_t length, char *log)
+static struct line_buffer input = LINE_BUFFER_INIT;
+
+static char *log_copy(uint32_t length, const char *log)
 {
        char *buffer;
        log_free(log_pool.size);
@@ -40,7 +42,7 @@ static char* log_copy(uint32_t length, char *log)
 }
 
 static struct {
-       uint32_t action, propLength, textLength, srcRev, mark, type;
+       uint32_t action, propLength, textLength, srcRev, type;
        uint32_t src[REPO_MAX_PATH_DEPTH], dst[REPO_MAX_PATH_DEPTH];
        uint32_t text_delta, prop_delta;
 } node_ctx;
@@ -73,7 +75,6 @@ static void reset_node_ctx(char *fname)
        node_ctx.src[0] = ~0;
        node_ctx.srcRev = 0;
        pool_tok_seq(REPO_MAX_PATH_DEPTH, node_ctx.dst, "/", fname);
-       node_ctx.mark = 0;
        node_ctx.text_delta = 0;
        node_ctx.prop_delta = 0;
 }
@@ -116,84 +117,170 @@ static void init_keys(void)
        keys.prop_delta = pool_intern("Prop-delta");
 }
 
-static void read_props(void)
+static void handle_property(uint32_t key, const char *val, uint32_t len,
+                               uint32_t *type_set)
 {
-       uint32_t len;
-       uint32_t key = ~0;
-       char *val = NULL;
-       char *t;
-       while ((t = buffer_read_line()) && strcmp(t, "PROPS-END")) {
-               if (!strncmp(t, "K ", 2)) {
-                       len = atoi(&t[2]);
-                       key = pool_intern(buffer_read_string(len));
-                       buffer_read_line();
-               } else if (!strncmp(t, "V ", 2)) {
-                       len = atoi(&t[2]);
-                       val = buffer_read_string(len);
-                       if (key == keys.svn_log) {
-                               /* Value length excludes terminating nul. */
-                               rev_ctx.log = log_copy(len + 1, val);
-                       } else if (key == keys.svn_author) {
-                               rev_ctx.author = pool_intern(val);
-                       } else if (key == keys.svn_date) {
-                               if (parse_date_basic(val, &rev_ctx.timestamp, NULL))
-                                       fprintf(stderr, "Invalid timestamp: %s\n", val);
-                       } else if (key == keys.svn_executable) {
-                               node_ctx.type = REPO_MODE_EXE;
-                       } else if (key == keys.svn_special) {
-                               node_ctx.type = REPO_MODE_LNK;
-                       }
-                       key = ~0;
-                       buffer_read_line();
+       if (key == keys.svn_log) {
+               if (!val)
+                       die("invalid dump: unsets svn:log");
+               /* Value length excludes terminating nul. */
+               rev_ctx.log = log_copy(len + 1, val);
+       } else if (key == keys.svn_author) {
+               rev_ctx.author = pool_intern(val);
+       } else if (key == keys.svn_date) {
+               if (!val)
+                       die("invalid dump: unsets svn:date");
+               if (parse_date_basic(val, &rev_ctx.timestamp, NULL))
+                       warning("invalid timestamp: %s", val);
+       } else if (key == keys.svn_executable || key == keys.svn_special) {
+               if (*type_set) {
+                       if (!val)
+                               return;
+                       die("invalid dump: sets type twice");
+               }
+               if (!val) {
+                       node_ctx.type = REPO_MODE_BLB;
+                       return;
                }
+               *type_set = 1;
+               node_ctx.type = key == keys.svn_executable ?
+                               REPO_MODE_EXE :
+                               REPO_MODE_LNK;
        }
 }
 
-static void handle_node(void)
+static void die_short_read(void)
 {
-       uint32_t old_mode = 0;
+       if (buffer_ferror(&input))
+               die_errno("error reading dump file");
+       die("invalid dump: unexpected end of file");
+}
 
-       if (node_ctx.text_delta || node_ctx.prop_delta)
-               die("text and property deltas not supported");
+static void read_props(void)
+{
+       uint32_t key = ~0;
+       const char *t;
+       /*
+        * NEEDSWORK: to support simple mode changes like
+        *      K 11
+        *      svn:special
+        *      V 1
+        *      *
+        *      D 14
+        *      svn:executable
+        * we keep track of whether a mode has been set and reset to
+        * plain file only if not.  We should be keeping track of the
+        * symlink and executable bits separately instead.
+        */
+       uint32_t type_set = 0;
+       while ((t = buffer_read_line(&input)) && strcmp(t, "PROPS-END")) {
+               uint32_t len;
+               const char *val;
+               const char type = t[0];
+               int ch;
 
-       if (node_ctx.propLength != LENGTH_UNKNOWN && node_ctx.propLength)
-               read_props();
+               if (!type || t[1] != ' ')
+                       die("invalid property line: %s\n", t);
+               len = atoi(&t[2]);
+               val = buffer_read_string(&input, len);
+               if (!val || strlen(val) != len)
+                       die_short_read();
 
-       if (node_ctx.srcRev)
-               old_mode = repo_copy(node_ctx.srcRev, node_ctx.src, node_ctx.dst);
+               /* Discard trailing newline. */
+               ch = buffer_read_char(&input);
+               if (ch == EOF)
+                       die_short_read();
+               if (ch != '\n')
+                       die("invalid dump: expected newline after %s", val);
 
-       if (node_ctx.textLength != LENGTH_UNKNOWN &&
-           node_ctx.type != REPO_MODE_DIR)
-               node_ctx.mark = next_blob_mark();
+               switch (type) {
+               case 'K':
+                       key = pool_intern(val);
+                       continue;
+               case 'D':
+                       key = pool_intern(val);
+                       val = NULL;
+                       len = 0;
+                       /* fall through */
+               case 'V':
+                       handle_property(key, val, len, &type_set);
+                       key = ~0;
+                       continue;
+               default:
+                       die("invalid property line: %s\n", t);
+               }
+       }
+}
+
+static void handle_node(void)
+{
+       uint32_t mark = 0;
+       const uint32_t type = node_ctx.type;
+       const int have_props = node_ctx.propLength != LENGTH_UNKNOWN;
+       const int have_text = node_ctx.textLength != LENGTH_UNKNOWN;
 
+       if (node_ctx.text_delta)
+               die("text deltas not supported");
+       if (have_text)
+               mark = next_blob_mark();
        if (node_ctx.action == NODEACT_DELETE) {
+               if (have_text || have_props || node_ctx.srcRev)
+                       die("invalid dump: deletion node has "
+                               "copyfrom info, text, or properties");
+               return repo_delete(node_ctx.dst);
+       }
+       if (node_ctx.action == NODEACT_REPLACE) {
                repo_delete(node_ctx.dst);
-       } else if (node_ctx.action == NODEACT_CHANGE ||
-                          node_ctx.action == NODEACT_REPLACE) {
-               if (node_ctx.action == NODEACT_REPLACE &&
-                   node_ctx.type == REPO_MODE_DIR)
-                       repo_replace(node_ctx.dst, node_ctx.mark);
-               else if (node_ctx.propLength != LENGTH_UNKNOWN)
-                       repo_modify(node_ctx.dst, node_ctx.type, node_ctx.mark);
-               else if (node_ctx.textLength != LENGTH_UNKNOWN)
-                       old_mode = repo_replace(node_ctx.dst, node_ctx.mark);
+               node_ctx.action = NODEACT_ADD;
+       }
+       if (node_ctx.srcRev) {
+               repo_copy(node_ctx.srcRev, node_ctx.src, node_ctx.dst);
+               if (node_ctx.action == NODEACT_ADD)
+                       node_ctx.action = NODEACT_CHANGE;
+       }
+       if (have_text && type == REPO_MODE_DIR)
+               die("invalid dump: directories cannot have text attached");
+
+       /*
+        * Decide on the new content (mark) and mode (node_ctx.type).
+        */
+       if (node_ctx.action == NODEACT_CHANGE && !~*node_ctx.dst) {
+               if (type != REPO_MODE_DIR)
+                       die("invalid dump: root of tree is not a regular file");
+       } else if (node_ctx.action == NODEACT_CHANGE) {
+               uint32_t mode;
+               if (!have_text)
+                       mark = repo_read_path(node_ctx.dst);
+               mode = repo_read_mode(node_ctx.dst);
+               if (mode == REPO_MODE_DIR && type != REPO_MODE_DIR)
+                       die("invalid dump: cannot modify a directory into a file");
+               if (mode != REPO_MODE_DIR && type == REPO_MODE_DIR)
+                       die("invalid dump: cannot modify a file into a directory");
+               node_ctx.type = mode;
        } else if (node_ctx.action == NODEACT_ADD) {
-               if (node_ctx.srcRev && node_ctx.propLength != LENGTH_UNKNOWN)
-                       repo_modify(node_ctx.dst, node_ctx.type, node_ctx.mark);
-               else if (node_ctx.srcRev && node_ctx.textLength != LENGTH_UNKNOWN)
-                       old_mode = repo_replace(node_ctx.dst, node_ctx.mark);
-               else if ((node_ctx.type == REPO_MODE_DIR && !node_ctx.srcRev) ||
-                        node_ctx.textLength != LENGTH_UNKNOWN)
-                       repo_add(node_ctx.dst, node_ctx.type, node_ctx.mark);
+               if (!have_text && type != REPO_MODE_DIR)
+                       die("invalid dump: adds node without text");
+       } else {
+               die("invalid dump: Node-path block lacks Node-action");
        }
 
-       if (node_ctx.propLength == LENGTH_UNKNOWN && old_mode)
-               node_ctx.type = old_mode;
+       /*
+        * Adjust mode to reflect properties.
+        */
+       if (have_props) {
+               if (!node_ctx.prop_delta)
+                       node_ctx.type = type;
+               if (node_ctx.propLength)
+                       read_props();
+       }
 
-       if (node_ctx.mark)
-               fast_export_blob(node_ctx.type, node_ctx.mark, node_ctx.textLength);
-       else if (node_ctx.textLength != LENGTH_UNKNOWN)
-               buffer_skip_bytes(node_ctx.textLength);
+       /*
+        * Save the result.
+        */
+       repo_add(node_ctx.dst, node_ctx.type, mark);
+       if (have_text)
+               fast_export_blob(node_ctx.type, mark,
+                                node_ctx.textLength, &input);
 }
 
 static void handle_revision(void)
@@ -212,7 +299,7 @@ void svndump_read(const char *url)
        uint32_t key;
 
        reset_dump_ctx(pool_intern(url));
-       while ((t = buffer_read_line())) {
+       while ((t = buffer_read_line(&input))) {
                val = strstr(t, ": ");
                if (!val)
                        continue;
@@ -223,7 +310,7 @@ void svndump_read(const char *url)
                if (key == keys.svn_fs_dump_format_version) {
                        dump_ctx.version = atoi(val);
                        if (dump_ctx.version > 3)
-                               die("expected svn dump format version <= 3, found %d",
+                               die("expected svn dump format version <= 3, found %"PRIu32,
                                    dump_ctx.version);
                } else if (key == keys.uuid) {
                        dump_ctx.uuid = pool_intern(val);
@@ -273,7 +360,11 @@ void svndump_read(const char *url)
                        node_ctx.prop_delta = !strcmp(val, "true");
                } else if (key == keys.content_length) {
                        len = atoi(val);
-                       buffer_read_line();
+                       t = buffer_read_line(&input);
+                       if (!t)
+                               die_short_read();
+                       if (*t)
+                               die("invalid dump: expected blank line after content length header");
                        if (active_ctx == REV_CTX) {
                                read_props();
                        } else if (active_ctx == NODE_CTX) {
@@ -281,10 +372,13 @@ void svndump_read(const char *url)
                                active_ctx = REV_CTX;
                        } else {
                                fprintf(stderr, "Unexpected content length header: %"PRIu32"\n", len);
-                               buffer_skip_bytes(len);
+                               if (buffer_skip_bytes(&input, len) != len)
+                                       die_short_read();
                        }
                }
        }
+       if (buffer_ferror(&input))
+               die_short_read();
        if (active_ctx == NODE_CTX)
                handle_node();
        if (active_ctx != DUMP_CTX)
@@ -293,7 +387,7 @@ void svndump_read(const char *url)
 
 int svndump_init(const char *filename)
 {
-       if (buffer_init(filename))
+       if (buffer_init(&input, filename))
                return error("cannot open %s: %s", filename, strerror(errno));
        repo_init();
        reset_dump_ctx(~0);
@@ -310,7 +404,7 @@ void svndump_deinit(void)
        reset_dump_ctx(~0);
        reset_rev_ctx(0);
        reset_node_ctx(NULL);
-       if (buffer_deinit())
+       if (buffer_deinit(&input))
                fprintf(stderr, "Input error\n");
        if (ferror(stdout))
                fprintf(stderr, "Output error\n");
@@ -319,7 +413,7 @@ void svndump_deinit(void)
 void svndump_reset(void)
 {
        log_reset();
-       buffer_reset();
+       buffer_reset(&input);
        repo_reset();
        reset_dump_ctx(~0);
        reset_rev_ctx(0);