Hermit/tests/test_diff.c

/* Built-in unified diff: GNU `diff -u` layout on unambiguous cases, and a
 * randomized round trip through our own patcher. */
#define _POSIX_C_SOURCE 200809L

#include "cst_agent.h"
#include "cst_tools.h"

#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <unistd.h>
#include "cst_test_util.h"

static int nfail;

#define EXPECT(cond, msg)                                                      \
  do {                                                                         \
    if (!(cond)) {                                                             \
      fprintf(stderr, "FAIL %s:%d: %s\n", __FILE__, __LINE__, msg);            \
      nfail++;                                                                 \
    }                                                                          \
  } while (0)

static void put(const char *p, const char *s) {
  FILE *fp = fopen(p, "wb");
  if (fp) {
    fputs(s, fp);
    fclose(fp);
  }
}

/* GNU diff output, or NULL when diff is not installed. */
static char *gnu_diff(const char *a, const char *b) {
#ifdef _WIN32
  (void)a;
  (void)b;
  return NULL; /* no GNU diff to compare against */
#else
  char pa[] = "/tmp/cstdiffaXXXXXX", pb[] = "/tmp/cstdiffbXXXXXX";
  int fa = mkstemp(pa), fb = mkstemp(pb);
  if (fa < 0 || fb < 0)
    return NULL;
  close(fa);
  close(fb);
  put(pa, a);
  put(pb, b);
  char cmd[256];
  snprintf(cmd, sizeof(cmd),
           "LC_ALL=C diff -u --label a/f --label b/f %s %s 2>/dev/null", pa,
           pb);
  FILE *pp = popen(cmd, "r");
  char *out = NULL;
  if (pp) {
    size_t cap = 4096, n = 0;
    out = malloc(cap);
    int ch;
    while (out && (ch = fgetc(pp)) != EOF) {
      if (n + 1 >= cap) {
        cap *= 2;
        out = realloc(out, cap);
      }
      if (out)
        out[n++] = (char)ch;
    }
    int st = pclose(pp);
    if (out)
      out[n] = 0;
    if (st == -1 || (WEXITSTATUS(st) != 0 && WEXITSTATUS(st) != 1)) {
      free(out);
      out = NULL;
    }
  }
  unlink(pa);
  unlink(pb);
  return out;
#endif
}

static int have_gnu_diff(void) {
#ifdef _WIN32
  return 0; /* gnu_diff() has no implementation there */
#else
  /* Android's toybox diff (and busybox's) print the same "@@ -1 +1 @@" for a
   * one-line change but group hunks and context their own way, so the layout
   * comparisons only make sense against real diffutils. */
  FILE *pp = popen("LC_ALL=C diff --version 2>/dev/null", "r");
  char line[128] = "";
  if (pp) {
    if (!fgets(line, sizeof(line), pp))
      line[0] = 0;
    pclose(pp);
  }
  if (!strstr(line, "GNU diffutils"))
    return 0;
  char *o = gnu_diff("x\n", "y\n");
  int ok = o && strstr(o, "@@ -1 +1 @@");
  free(o);
  return ok;
#endif
}

static char *mine(const char *a, const char *b) {
  return cst_diff_unified("f", a, strlen(a), b, strlen(b));
}

static void same_as_gnu(const char *name, const char *a, const char *b,
                        int gnu) {
  char *m = mine(a, b);
  EXPECT(m != NULL, name);
  if (gnu && m) {
    char *g = gnu_diff(a, b);
    if (g && strcmp(g, m) != 0) {
      fprintf(stderr, "FAIL %s: differs from diff -u\n--- ours\n%s--- gnu\n%s",
              name, m, g);
      nfail++;
    }
    free(g);
  }
  free(m);
}

/* Apply our diff with our patcher; must reproduce b. */
static int round_trip(const char *a, const char *b) {
  char *d = mine(a, b);
  if (!d)
    return 0;
  int ok;
  if (!d[0]) {
    ok = strcmp(a, b) == 0;
  } else {
    const char *h = strstr(d, "\n+++ ");
    h = h ? strchr(h + 1, '\n') : NULL;
    char *out = NULL;
    size_t out_len = 0;
    int nh = 0;
    char err[256];
    int rc = h ? cst_patch_apply_buffer(a, strlen(a), h + 1, strlen(h + 1),
                                        &out, &out_len, &nh, err, sizeof(err))
               : -1;
    ok = rc == 0 && out && out_len == strlen(b) && memcmp(out, b, out_len) == 0;
    if (!ok)
      fprintf(stderr, "round trip failed (%s)\n--- a\n%s--- b\n%s--- diff\n%s",
              rc ? err : "content", a, b, d);
    free(out);
  }
  free(d);
  return ok;
}

static char *random_text(unsigned *seed, int lines) {
  static const char *const words[] = {"alpha", "beta", "gamma", "delta",
                                      "{",     "}",    "",      "return 0;"};
  size_t cap = (size_t)lines * 16 + 2, n = 0;
  char *s = malloc(cap);
  for (int i = 0; i < lines; i++) {
    const char *w = words[test_rand(seed) % 8];
    n += (size_t)snprintf(s + n, cap - n, "%s%d\n", w, test_rand(seed) % 3);
  }
  s[n] = 0;
  if (n && test_rand(seed) % 5 == 0)
    s[n - 1] = 0; /* no newline at end of file */
  return s;
}

static char *mutate(unsigned *seed, const char *a) {
  size_t n = strlen(a), cap = n * 2 + 256, o = 0;
  char *b = malloc(cap);
  const char *p = a;
  while (*p) {
    const char *e = strchr(p, '\n');
    size_t len = e ? (size_t)(e - p) + 1 : strlen(p);
    int r = test_rand(seed) % 10;
    if (r == 0) {
      /* drop the line */
    } else if (r == 1) {
      o += (size_t)snprintf(b + o, cap - o, "inserted%d\n", test_rand(seed) % 9);
      memcpy(b + o, p, len);
      o += len;
    } else if (r == 2) {
      o += (size_t)snprintf(b + o, cap - o, "changed%d\n", test_rand(seed) % 9);
    } else {
      memcpy(b + o, p, len);
      o += len;
    }
    p += len;
  }
  b[o] = 0;
  return b;
}

int main(void) {
  char *m = mine("same\n", "same\n");
  EXPECT(m && m[0] == 0, "identical -> empty");
  free(m);

  int gnu = have_gnu_diff();
  if (!gnu)
    fprintf(stderr, "note: GNU diff not found; layout checks use round trip "
                    "only\n");

  same_as_gnu("modify", "a\nb\nc\nd\ne\n", "a\nb\nC\nd\ne\n", gnu);
  same_as_gnu("insert head", "a\nb\n", "z\na\nb\n", gnu);
  same_as_gnu("append", "a\nb\n", "a\nb\nc\n", gnu);
  same_as_gnu("delete", "a\nb\nc\n", "a\nc\n", gnu);
  same_as_gnu("new file", "", "x\ny\n", gnu);
  same_as_gnu("emptied", "x\ny\n", "", gnu);
  same_as_gnu("no eol added", "a\nb", "a\nb\n", gnu);
  same_as_gnu("no eol kept", "a\nb", "a\nB", gnu);
  same_as_gnu("two hunks",
              "1\n2\n3\n4\n5\n6\n7\n8\n9\n10\n11\n12\n13\n14\n15\n",
              "1\nTWO\n3\n4\n5\n6\n7\n8\n9\n10\n11\n12\n13\nFOURTEEN\n15\n",
              gnu);
  same_as_gnu("merged hunks", "1\n2\n3\n4\n5\n6\n7\n8\n9\n10\n",
              "1\nX\n3\n4\n5\n6\n7\nY\n9\n10\n", gnu);

  m = mine("a\n", "b\n");
  EXPECT(m && strncmp(m, "--- a/f\n+++ b/f\n@@ -1 +1 @@\n-a\n+b\n", 34) == 0,
         "header layout");
  free(m);

  unsigned seed = 12345;
  int ok = 0, total = 400;
  for (int i = 0; i < total; i++) {
    char *a = random_text(&seed, 1 + (int)(test_rand(&seed) % 60));
    char *b = mutate(&seed, a);
    ok += round_trip(a, b);
    free(a);
    free(b);
  }
  EXPECT(ok == total, "random round trips");

  /* very different large inputs hit the fallback and still round trip */
  {
    size_t n = 6000;
    char *a = malloc(n * 8), *b = malloc(n * 8);
    size_t oa = 0, ob = 0;
    for (size_t i = 0; i < n; i++) {
      oa += (size_t)sprintf(a + oa, "a%zu\n", i);
      ob += (size_t)sprintf(b + ob, "b%zu\n", i);
    }
    EXPECT(round_trip(a, b), "fallback round trip");
    free(a);
    free(b);
  }

  if (nfail) {
    fprintf(stderr, "test_diff: %d failure(s)\n", nfail);
    return 1;
  }
  printf("test_diff: OK (%d random round trips)\n", total);
  return 0;
}