fast-import: tighten parsing of datarefs
The syntax for the use of mark references in fast-import demands either a SP (space) or LF (end-of-line) after a mark reference. Fast-import does not complain when garbage appears after a mark reference in some cases. Factor out parsing of mark references and complain if errant characters are found. Also be a little more careful when parsing "inline" and SHA1s, complaining if extra characters appear or if the form of the dataref is unrecognized. Buggy input can cause fast-import to produce the wrong output, silently, without error. This makes it difficult to track down buggy generators of fast-import streams. An example is seen in the last line of this commit command: commit refs/heads/S2 committer Name <name@example.com> 1112912893 -0400 data <<COMMIT commit message COMMIT from :1M 100644 :103 hello.c It is missing a newline and should be: [...] from :1 M 100644 :103 hello.c What fast-import does is to produce a commit with the same contents for hello.c as in refs/heads/S2^. What the buggy program was expecting was the contents of blob :103. While the resulting commit graph looked correct, the contents in some commits were wrong. Signed-off-by: Pete Wyckoff <pw@padd.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Pete Wyckoff committed
Apr 7, 2012 at 18:59 UTC
06454cb9a3bd2a34299779b147e388ff0f31c9f7
2 files changed
+364
-33
fast-import.c
+77
-33
@@ -2207,6 +2207,59 @@ static uintmax_t change_note_fanout(struct tree_entry *root,
2207
return do_change_note_fanout(root, root, hex_sha1, 0, path, 0, fanout);
2208
}
2209
2210
+/*
2211
+ * Given a pointer into a string, parse a mark reference:
2212
+ *
2213
+ * idnum ::= ':' bigint;
2214
+ *
2215
+ * Return the first character after the value in *endptr.
2216
+ *
2217
+ * Complain if the following character is not what is expected,
2218
+ * either a space or end of the string.
2219
+ */
2220
+static uintmax_t parse_mark_ref(const char *p, char **endptr)
2221
+{
2222
+ uintmax_t mark;
2223
+
2224
+ assert(*p == ':');
2225
+ p++;
2226
+ mark = strtoumax(p, endptr, 10);
2227
+ if (*endptr == p)
2228
+ die("No value after ':' in mark: %s", command_buf.buf);
2229
+ return mark;
2230
+}
2231
+
2232
+/*
2233
+ * Parse the mark reference, and complain if this is not the end of
2234
+ * the string.
2235
+ */
2236
+static uintmax_t parse_mark_ref_eol(const char *p)
2237
+{
2238
+ char *end;
2239
+ uintmax_t mark;
2240
+
2241
+ mark = parse_mark_ref(p, &end);
2242
+ if (*end != '\0')
2243
+ die("Garbage after mark: %s", command_buf.buf);
2244
+ return mark;
2245
+}
2246
+
2247
+/*
2248
+ * Parse the mark reference, demanding a trailing space. Return a
2249
+ * pointer to the space.
2250
+ */
2251
+static uintmax_t parse_mark_ref_space(const char **p)
2252
+{
2253
+ uintmax_t mark;
2254
+ char *end;
2255
+
2256
+ mark = parse_mark_ref(*p, &end);
2257
+ if (*end != ' ')
2258
+ die("Missing space after mark: %s", command_buf.buf);
2259
+ *p = end;
2260
+ return mark;
2261
+}
2262
+
2263
static void file_change_m(struct branch *b)
2264
{
2265
const char *p = command_buf.buf + 2;
@@ -2235,21 +2288,21 @@ static void file_change_m(struct branch *b)
2288
}
2289
2290
if (*p == ':') {
2238
- char *x;
2239
- oe = find_mark(strtoumax(p + 1, &x, 10));
2291
+ oe = find_mark(parse_mark_ref_space(&p));
2292
hashcpy(sha1, oe->idx.sha1);
2241
- p = x;
2242
- } else if (!prefixcmp(p, "inline")) {
2293
+ } else if (!prefixcmp(p, "inline ")) {
2294
inline_data = 1;
2244
- p += 6;
2295
+ p += strlen("inline"); /* advance to space */
2296
} else {
2297
if (get_sha1_hex(p, sha1))
2247
- die("Invalid SHA1: %s", command_buf.buf);
2298
+ die("Invalid dataref: %s", command_buf.buf);
2299
oe = find_object(sha1);
2300
p += 40;
2301
+ if (*p != ' ')
2302
+ die("Missing space after SHA1: %s", command_buf.buf);
2303
}
2251
- if (*p++ != ' ')
2252
- die("Missing space after SHA1: %s", command_buf.buf);
2304
+ assert(*p == ' ');
2305
+ p++; /* skip space */
2306
2307
strbuf_reset(&uq);
2308
if (!unquote_c_style(&uq, p, &endp)) {
@@ -2407,21 +2460,21 @@ static void note_change_n(struct branch *b, unsigned char *old_fanout)
2460
/* Now parse the notemodify command. */
2461
/* <dataref> or 'inline' */
2462
if (*p == ':') {
2410
- char *x;
2411
- oe = find_mark(strtoumax(p + 1, &x, 10));
2463
+ oe = find_mark(parse_mark_ref_space(&p));
2464
hashcpy(sha1, oe->idx.sha1);
2413
- p = x;
2414
- } else if (!prefixcmp(p, "inline")) {
2465
+ } else if (!prefixcmp(p, "inline ")) {
2466
inline_data = 1;
2416
- p += 6;
2467
+ p += strlen("inline"); /* advance to space */
2468
} else {
2469
if (get_sha1_hex(p, sha1))
2419
- die("Invalid SHA1: %s", command_buf.buf);
2470
+ die("Invalid dataref: %s", command_buf.buf);
2471
oe = find_object(sha1);
2472
p += 40;
2473
+ if (*p != ' ')
2474
+ die("Missing space after SHA1: %s", command_buf.buf);
2475
}
2423
- if (*p++ != ' ')
2424
- die("Missing space after SHA1: %s", command_buf.buf);
2476
+ assert(*p == ' ');
2477
+ p++; /* skip space */
2478
2479
/* <committish> */
2480
s = lookup_branch(p);
@@ -2430,7 +2483,7 @@ static void note_change_n(struct branch *b, unsigned char *old_fanout)
2483
die("Can't add a note on empty branch.");
2484
hashcpy(commit_sha1, s->sha1);
2485
} else if (*p == ':') {
2433
- uintmax_t commit_mark = strtoumax(p + 1, NULL, 10);
2486
+ uintmax_t commit_mark = parse_mark_ref_eol(p);
2487
struct object_entry *commit_oe = find_mark(commit_mark);
2488
if (commit_oe->type != OBJ_COMMIT)
2489
die("Mark :%" PRIuMAX " not a commit", commit_mark);
@@ -2537,7 +2590,7 @@ static int parse_from(struct branch *b)
2590
hashcpy(b->branch_tree.versions[0].sha1, t);
2591
hashcpy(b->branch_tree.versions[1].sha1, t);
2592
} else if (*from == ':') {
2540
- uintmax_t idnum = strtoumax(from + 1, NULL, 10);
2593
+ uintmax_t idnum = parse_mark_ref_eol(from);
2594
struct object_entry *oe = find_mark(idnum);
2595
if (oe->type != OBJ_COMMIT)
2596
die("Mark :%" PRIuMAX " not a commit", idnum);
@@ -2572,7 +2625,7 @@ static struct hash_list *parse_merge(unsigned int *count)
2625
if (s)
2626
hashcpy(n->sha1, s->sha1);
2627
else if (*from == ':') {
2575
- uintmax_t idnum = strtoumax(from + 1, NULL, 10);
2628
+ uintmax_t idnum = parse_mark_ref_eol(from);
2629
struct object_entry *oe = find_mark(idnum);
2630
if (oe->type != OBJ_COMMIT)
2631
die("Mark :%" PRIuMAX " not a commit", idnum);
@@ -2735,7 +2788,7 @@ static void parse_new_tag(void)
2788
type = OBJ_COMMIT;
2789
} else if (*from == ':') {
2790
struct object_entry *oe;
2738
- from_mark = strtoumax(from + 1, NULL, 10);
2791
+ from_mark = parse_mark_ref_eol(from);
2792
oe = find_mark(from_mark);
2793
type = oe->type;
2794
hashcpy(sha1, oe->idx.sha1);
@@ -2867,18 +2920,13 @@ static void parse_cat_blob(void)
2920
/* cat-blob SP <object> LF */
2921
p = command_buf.buf + strlen("cat-blob ");
2922
if (*p == ':') {
2870
- char *x;
2871
- oe = find_mark(strtoumax(p + 1, &x, 10));
2872
- if (x == p + 1)
2873
- die("Invalid mark: %s", command_buf.buf);
2923
+ oe = find_mark(parse_mark_ref_eol(p));
2924
if (!oe)
2925
die("Unknown mark: %s", command_buf.buf);
2876
- if (*x)
2877
- die("Garbage after mark: %s", command_buf.buf);
2926
hashcpy(sha1, oe->idx.sha1);
2927
} else {
2928
if (get_sha1_hex(p, sha1))
2881
- die("Invalid SHA1: %s", command_buf.buf);
2929
+ die("Invalid dataref: %s", command_buf.buf);
2930
if (p[40])
2931
die("Garbage after SHA1: %s", command_buf.buf);
2932
oe = find_object(sha1);
@@ -2944,17 +2992,13 @@ static struct object_entry *parse_treeish_dataref(const char **p)
2992
struct object_entry *e;
2993
2994
if (**p == ':') { /* <mark> */
2947
- char *endptr;
2948
- e = find_mark(strtoumax(*p + 1, &endptr, 10));
2949
- if (endptr == *p + 1)
2950
- die("Invalid mark: %s", command_buf.buf);
2995
+ e = find_mark(parse_mark_ref_space(p));
2996
if (!e)
2997
die("Unknown mark: %s", command_buf.buf);
2953
- *p = endptr;
2998
hashcpy(sha1, e->idx.sha1);
2999
} else { /* <sha1> */
3000
if (get_sha1_hex(*p, sha1))
2957
- die("Invalid SHA1: %s", command_buf.buf);
3001
+ die("Invalid dataref: %s", command_buf.buf);
3002
e = find_object(sha1);
3003
*p += 40;
3004
}
t/t9300-fast-import.sh
+287
@@ -2635,4 +2635,291 @@ test_expect_success \
2635
'n=$(grep $a verify | wc -l) &&
2636
test 1 = $n'
2637
2638
+###
2639
+### series S
2640
+###
2641
+#
2642
+# Make sure missing spaces and EOLs after mark references
2643
+# cause errors.
2644
+#
2645
+# Setup:
2646
+#
2647
+# 1--2--4
2648
+# \ /
2649
+# -3-
2650
+#
2651
+# commit marks: 301, 302, 303, 304
2652
+# blob marks: 403, 404, resp.
2653
+# note mark: 202
2654
+#
2655
+# The error message when a space is missing not at the
2656
+# end of the line is:
2657
+#
2658
+# Missing space after ..
2659
+#
2660
+# or when extra characters come after the mark at the end
2661
+# of the line:
2662
+#
2663
+# Garbage after ..
2664
+#
2665
+# or when the dataref is neither "inline " or a known SHA1,
2666
+#
2667
+# Invalid dataref ..
2668
+#
2669
+test_tick
2670
+
2671
+cat >input <<INPUT_END
2672
+commit refs/heads/S
2673
+mark :301
2674
+committer $GIT_COMMITTER_NAME <$GIT_COMMITTER_EMAIL> $GIT_COMMITTER_DATE
2675
+data <<COMMIT
2676
+commit 1
2677
+COMMIT
2678
+M 100644 inline hello.c
2679
+data <<BLOB
2680
+blob 1
2681
+BLOB
2682
+
2683
+commit refs/heads/S
2684
+mark :302
2685
+committer $GIT_COMMITTER_NAME <$GIT_COMMITTER_EMAIL> $GIT_COMMITTER_DATE
2686
+data <<COMMIT
2687
+commit 2
2688
+COMMIT
2689
+from :301
2690
+M 100644 inline hello.c
2691
+data <<BLOB
2692
+blob 2
2693
+BLOB
2694
+
2695
+blob
2696
+mark :403
2697
+data <<BLOB
2698
+blob 3
2699
+BLOB
2700
+
2701
+blob
2702
+mark :202
2703
+data <<BLOB
2704
+note 2
2705
+BLOB
2706
+INPUT_END
2707
+
2708
+test_expect_success 'S: initialize for S tests' '
2709
+ git fast-import --export-marks=marks <input
2710
+'
2711
+
2712
+#
2713
+# filemodify, three datarefs
2714
+#
2715
+test_expect_success 'S: filemodify with garbage after mark must fail' '
2716
+ test_must_fail git fast-import --import-marks=marks <<-EOF 2>err &&
2717
+ commit refs/heads/S
2718
+ committer $GIT_COMMITTER_NAME <$GIT_COMMITTER_EMAIL> $GIT_COMMITTER_DATE
2719
+ data <<COMMIT
2720
+ commit N
2721
+ COMMIT
2722
+ M 100644 :403x hello.c
2723
+ EOF
2724
+ cat err &&
2725
+ test_i18ngrep "space after mark" err
2726
+'
2727
+
2728
+# inline is misspelled; fast-import thinks it is some unknown dataref
2729
+test_expect_success 'S: filemodify with garbage after inline must fail' '
2730
+ test_must_fail git fast-import --import-marks=marks <<-EOF 2>err &&
2731
+ commit refs/heads/S
2732
+ committer $GIT_COMMITTER_NAME <$GIT_COMMITTER_EMAIL> $GIT_COMMITTER_DATE
2733
+ data <<COMMIT
2734
+ commit N
2735
+ COMMIT
2736
+ M 100644 inlineX hello.c
2737
+ data <<BLOB
2738
+ inline
2739
+ BLOB
2740
+ EOF
2741
+ cat err &&
2742
+ test_i18ngrep "nvalid dataref" err
2743
+'
2744
+
2745
+test_expect_success 'S: filemodify with garbage after sha1 must fail' '
2746
+ sha1=$(grep :403 marks | cut -d\ -f2) &&
2747
+ test_must_fail git fast-import --import-marks=marks <<-EOF 2>err &&
2748
+ commit refs/heads/S
2749
+ committer $GIT_COMMITTER_NAME <$GIT_COMMITTER_EMAIL> $GIT_COMMITTER_DATE
2750
+ data <<COMMIT
2751
+ commit N
2752
+ COMMIT
2753
+ M 100644 ${sha1}x hello.c
2754
+ EOF
2755
+ cat err &&
2756
+ test_i18ngrep "space after SHA1" err
2757
+'
2758
+
2759
+#
2760
+# notemodify, three ways to say dataref
2761
+#
2762
+test_expect_success 'S: notemodify with garabge after mark dataref must fail' '
2763
+ test_must_fail git fast-import --import-marks=marks <<-EOF 2>err &&
2764
+ commit refs/heads/S
2765
+ committer $GIT_COMMITTER_NAME <$GIT_COMMITTER_EMAIL> $GIT_COMMITTER_DATE
2766
+ data <<COMMIT
2767
+ commit S note dataref markref
2768
+ COMMIT
2769
+ N :202x :302
2770
+ EOF
2771
+ cat err &&
2772
+ test_i18ngrep "space after mark" err
2773
+'
2774
+
2775
+test_expect_success 'S: notemodify with garbage after inline dataref must fail' '
2776
+ test_must_fail git fast-import --import-marks=marks <<-EOF 2>err &&
2777
+ commit refs/heads/S
2778
+ committer $GIT_COMMITTER_NAME <$GIT_COMMITTER_EMAIL> $GIT_COMMITTER_DATE
2779
+ data <<COMMIT
2780
+ commit S note dataref inline
2781
+ COMMIT
2782
+ N inlineX :302
2783
+ data <<BLOB
2784
+ note blob
2785
+ BLOB
2786
+ EOF
2787
+ cat err &&
2788
+ test_i18ngrep "nvalid dataref" err
2789
+'
2790
+
2791
+test_expect_success 'S: notemodify with garbage after sha1 dataref must fail' '
2792
+ sha1=$(grep :202 marks | cut -d\ -f2) &&
2793
+ test_must_fail git fast-import --import-marks=marks <<-EOF 2>err &&
2794
+ commit refs/heads/S
2795
+ committer $GIT_COMMITTER_NAME <$GIT_COMMITTER_EMAIL> $GIT_COMMITTER_DATE
2796
+ data <<COMMIT
2797
+ commit S note dataref sha1
2798
+ COMMIT
2799
+ N ${sha1}x :302
2800
+ EOF
2801
+ cat err &&
2802
+ test_i18ngrep "space after SHA1" err
2803
+'
2804
+
2805
+#
2806
+# notemodify, mark in committish
2807
+#
2808
+test_expect_success 'S: notemodify with garbarge after mark committish must fail' '
2809
+ test_must_fail git fast-import --import-marks=marks <<-EOF 2>err &&
2810
+ commit refs/heads/Snotes
2811
+ committer $GIT_COMMITTER_NAME <$GIT_COMMITTER_EMAIL> $GIT_COMMITTER_DATE
2812
+ data <<COMMIT
2813
+ commit S note committish
2814
+ COMMIT
2815
+ N :202 :302x
2816
+ EOF
2817
+ cat err &&
2818
+ test_i18ngrep "after mark" err
2819
+'
2820
+
2821
+#
2822
+# from
2823
+#
2824
+test_expect_success 'S: from with garbage after mark must fail' '
2825
+ # no &&
2826
+ git fast-import --import-marks=marks --export-marks=marks <<-EOF 2>err
2827
+ commit refs/heads/S2
2828
+ mark :303
2829
+ committer $GIT_COMMITTER_NAME <$GIT_COMMITTER_EMAIL> $GIT_COMMITTER_DATE
2830
+ data <<COMMIT
2831
+ commit 3
2832
+ COMMIT
2833
+ from :301x
2834
+ M 100644 :403 hello.c
2835
+ EOF
2836
+
2837
+ ret=$? &&
2838
+ echo returned $ret &&
2839
+ test $ret -ne 0 && # failed, but it created the commit
2840
+
2841
+ # go create the commit, need it for merge test
2842
+ git fast-import --import-marks=marks --export-marks=marks <<-EOF &&
2843
+ commit refs/heads/S2
2844
+ mark :303
2845
+ committer $GIT_COMMITTER_NAME <$GIT_COMMITTER_EMAIL> $GIT_COMMITTER_DATE
2846
+ data <<COMMIT
2847
+ commit 3
2848
+ COMMIT
2849
+ from :301
2850
+ M 100644 :403 hello.c
2851
+ EOF
2852
+
2853
+ # now evaluate the error
2854
+ cat err &&
2855
+ test_i18ngrep "after mark" err
2856
+'
2857
+
2858
+
2859
+#
2860
+# merge
2861
+#
2862
+test_expect_success 'S: merge with garbage after mark must fail' '
2863
+ test_must_fail git fast-import --import-marks=marks <<-EOF 2>err &&
2864
+ commit refs/heads/S
2865
+ mark :304
2866
+ committer $GIT_COMMITTER_NAME <$GIT_COMMITTER_EMAIL> $GIT_COMMITTER_DATE
2867
+ data <<COMMIT
2868
+ merge 4
2869
+ COMMIT
2870
+ from :302
2871
+ merge :303x
2872
+ M 100644 :403 hello.c
2873
+ EOF
2874
+ cat err &&
2875
+ test_i18ngrep "after mark" err
2876
+'
2877
+
2878
+#
2879
+# tag, from markref
2880
+#
2881
+test_expect_success 'S: tag with garbage after mark must fail' '
2882
+ test_must_fail git fast-import --import-marks=marks <<-EOF 2>err &&
2883
+ tag refs/tags/Stag
2884
+ from :302x
2885
+ tagger $GIT_COMMITTER_NAME <$GIT_COMMITTER_EMAIL> $GIT_COMMITTER_DATE
2886
+ data <<TAG
2887
+ tag S
2888
+ TAG
2889
+ EOF
2890
+ cat err &&
2891
+ test_i18ngrep "after mark" err
2892
+'
2893
+
2894
+#
2895
+# cat-blob markref
2896
+#
2897
+test_expect_success 'S: cat-blob with garbage after mark must fail' '
2898
+ test_must_fail git fast-import --import-marks=marks <<-EOF 2>err &&
2899
+ cat-blob :403x
2900
+ EOF
2901
+ cat err &&
2902
+ test_i18ngrep "after mark" err
2903
+'
2904
+
2905
+#
2906
+# ls markref
2907
+#
2908
+test_expect_success 'S: ls with garbage after mark must fail' '
2909
+ test_must_fail git fast-import --import-marks=marks <<-EOF 2>err &&
2910
+ ls :302x hello.c
2911
+ EOF
2912
+ cat err &&
2913
+ test_i18ngrep "space after mark" err
2914
+'
2915
+
2916
+test_expect_success 'S: ls with garbage after sha1 must fail' '
2917
+ sha1=$(grep :302 marks | cut -d\ -f2) &&
2918
+ test_must_fail git fast-import --import-marks=marks <<-EOF 2>err &&
2919
+ ls ${sha1}x hello.c
2920
+ EOF
2921
+ cat err &&
2922
+ test_i18ngrep "space after tree-ish" err
2923
+'
2924
+
2925
test_done