Third run of making variable and function names more readable, integer scans and ip parsers

3 files changed, 114 insertions, 93 deletions

diff --git a/tests/t_ip.c b/tests/t_ip.c
index 2fe3a72..8d71679 100644
--- a/tests/t_ip.c
+++ b/tests/t_ip.c
@@ -95,27 +95,27 @@ static void case_scan_forms(void) {
   char ip[16];
 
   test_case("2.2 form 1, every group spelled out");
-  CHECK_INT(ip_scan("1111:2222:3333:4444:5555:6666:7777:8888", ip), 39);
+  CHECK_INT(ip_parse("1111:2222:3333:4444:5555:6666:7777:8888", ip), 39);
   CHECK_MEM(ip, "\x11\x11\x22\x22\x33\x33\x44\x44\x55\x55\x66\x66\x77\x77\x88\x88", 16);
 
   test_case("2.2 form 2, zero groups abbreviated");
-  CHECK_INT(ip_scan("2001:db8::1", ip), 11);
+  CHECK_INT(ip_parse("2001:db8::1", ip), 11);
   CHECK_MEM(ip, "\x20\x01\x0d\xb8\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x01", 16);
-  CHECK_INT(ip_scan("::1", ip), 3);
+  CHECK_INT(ip_parse("::1", ip), 3);
   CHECK_MEM(ip, "\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x01", 16);
-  CHECK_INT(ip_scan("::", ip), 2);
+  CHECK_INT(ip_parse("::", ip), 2);
   CHECK_MEM(ip, "\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00", 16);
-  CHECK_INT(ip_scan("1::", ip), 3);
+  CHECK_INT(ip_parse("1::", ip), 3);
   CHECK_MEM(ip, "\x00\x01\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00", 16);
 
   test_case("2.2 form 3, a dotted quad in the last two groups");
-  CHECK_INT(ip_scan("::ffff:1.2.3.4", ip), 14);
+  CHECK_INT(ip_parse("::ffff:1.2.3.4", ip), 14);
   CHECK_MEM(ip, "\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\xff\xff\x01\x02\x03\x04", 16);
-  CHECK_INT(ip_scan("1:2:3:4:5:6:1.2.3.4", ip), 19);
+  CHECK_INT(ip_parse("1:2:3:4:5:6:1.2.3.4", ip), 19);
   CHECK_MEM(ip, "\x00\x01\x00\x02\x00\x03\x00\x04\x00\x05\x00\x06\x01\x02\x03\x04", 16);
 
   test_case("upper case input is accepted");
-  CHECK_INT(ip_scan("2001:DB8::1", ip), 11);
+  CHECK_INT(ip_parse("2001:DB8::1", ip), 11);
   CHECK_MEM(ip, "\x20\x01\x0d\xb8\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x01", 16);
 }
 
@@ -125,7 +125,7 @@ static void case_scan_accepts_bare_ipv4(void) {
   /* Every address opentracker holds is sixteen bytes, so an operator writing
      a plain IPv4 address in the config has to land as the v4 mapped form */
   test_case("a bare IPv4 address is stored v4 mapped");
-  CHECK_INT(ip_scan("10.0.0.1", ip), 8);
+  CHECK_INT(ip_parse("10.0.0.1", ip), 8);
   CHECK_MEM(ip, "\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\xff\xff\x0a\x00\x00\x01", 16);
   CHECK(ip_is_v4mapped(ip));
 }
@@ -136,39 +136,39 @@ static void case_scan_reports_length(void) {
   /* The config parser continues from where the address ended, which is how
      "address:port" and "address/bits" are read */
   test_case("scanning stops at the character after the address");
-  CHECK_INT(ip_scan("10.0.0.1:6969", ip), 8);
-  CHECK_INT(ip_scan("2001:db8::1/64", ip), 11);
-  CHECK_INT(ip_scan("::1]", ip), 3);
-  CHECK_INT(ip_scan("1:2:3:4:5:6:7:8 trailing", ip), 15);
+  CHECK_INT(ip_parse("10.0.0.1:6969", ip), 8);
+  CHECK_INT(ip_parse("2001:db8::1/64", ip), 11);
+  CHECK_INT(ip_parse("::1]", ip), 3);
+  CHECK_INT(ip_parse("1:2:3:4:5:6:7:8 trailing", ip), 15);
 }
 
 static void case_scan_rejects(void) {
   char ip[16];
 
   test_case("an empty or junk string is refused");
-  CHECK_INT(ip_scan("", ip), 0);
-  CHECK_INT(ip_scan("x", ip), 0);
-  CHECK_INT(ip_scan(":", ip), 0);
-  CHECK_INT(ip_scan(":1:2", ip), 0);
+  CHECK_INT(ip_parse("", ip), 0);
+  CHECK_INT(ip_parse("x", ip), 0);
+  CHECK_INT(ip_parse(":", ip), 0);
+  CHECK_INT(ip_parse(":1:2", ip), 0);
 
   test_case("too few groups without :: are refused");
-  CHECK_INT(ip_scan("1:2:3:4:5:6:7", ip), 0);
-  CHECK_INT(ip_scan("2001:db8:1", ip), 0);
+  CHECK_INT(ip_parse("1:2:3:4:5:6:7", ip), 0);
+  CHECK_INT(ip_parse("2001:db8:1", ip), 0);
 
   /* libowfat accepted a prefix of each of these and carried on. Taking only
      part of a malformed address is worse than refusing it, especially where
      the text comes from a config file or an X-Forwarded-For header. */
   test_case("a second :: is refused rather than truncated");
-  CHECK_INT(ip_scan("1::2::3", ip), 0);
-  CHECK_INT(ip_scan("::1::", ip), 0);
-  CHECK_INT(ip_scan("::::", ip), 0);
+  CHECK_INT(ip_parse("1::2::3", ip), 0);
+  CHECK_INT(ip_parse("::1::", ip), 0);
+  CHECK_INT(ip_parse("::::", ip), 0);
 
   test_case("more than eight groups is refused");
-  CHECK_INT(ip_scan("1:2:3:4:5:6:7:8:9", ip), 0);
+  CHECK_INT(ip_parse("1:2:3:4:5:6:7:8:9", ip), 0);
 
   test_case("a group of more than four digits is refused, not truncated");
-  CHECK_INT(ip_scan("12345::1", ip), 0);
-  CHECK_INT(ip_scan("fffff::1", ip), 0);
+  CHECK_INT(ip_parse("12345::1", ip), 0);
+  CHECK_INT(ip_parse("fffff::1", ip), 0);
 }
 
 static void case_round_trip(void) {
@@ -187,7 +187,7 @@ static void case_round_trip(void) {
 
   test_case("every formatted address scans back to the same bytes");
   for (i = 0; samples[i]; ++i) {
-    size_t n = ip_scan(samples[i], ip);
+    size_t n = ip_parse(samples[i], ip);
     if (n != strlen(samples[i])) {
       test_failf(__FILE__, __LINE__, "\"%s\" consumed %zu of %zu", samples[i], n, strlen(samples[i]));
       return;
@@ -197,7 +197,7 @@ static void case_round_trip(void) {
       test_failf(__FILE__, __LINE__, "\"%s\" formatted back as \"%s\"", samples[i], text);
       return;
     }
-    if (!ip_scan(text, again) || memcmp(ip, again, 16)) {
+    if (!ip_parse(text, again) || memcmp(ip, again, 16)) {
       test_failf(__FILE__, __LINE__, "\"%s\" did not survive a second pass", samples[i]);
       return;
     }
diff --git a/tests/t_scan.c b/tests/t_scan.c
index 89f1590..aab4413 100644
--- a/tests/t_scan.c
+++ b/tests/t_scan.c
@@ -15,13 +15,13 @@ static void case_plain_numbers(void) {
   uint64_t value;
 
   test_case("a decimal number is read and its length reported");
-  CHECK_INT(scan_udec("0", &value, 65535), 1);
+  CHECK_INT(scan_uint("0", &value, 65535), 1);
   CHECK_INT(value, 0);
-  CHECK_INT(scan_udec("7", &value, 65535), 1);
+  CHECK_INT(scan_uint("7", &value, 65535), 1);
   CHECK_INT(value, 7);
-  CHECK_INT(scan_udec("6969", &value, 65535), 4);
+  CHECK_INT(scan_uint("6969", &value, 65535), 4);
   CHECK_INT(value, 6969);
-  CHECK_INT(scan_udec("65535", &value, 65535), 5);
+  CHECK_INT(scan_uint("65535", &value, 65535), 5);
   CHECK_INT(value, 65535);
 }
 
@@ -29,11 +29,11 @@ static void case_stops_at_non_digit(void) {
   uint64_t value;
 
   test_case("scanning stops at the first character that is not a digit");
-  CHECK_INT(scan_udec("6969:rest", &value, 65535), 4);
+  CHECK_INT(scan_uint("6969:rest", &value, 65535), 4);
   CHECK_INT(value, 6969);
-  CHECK_INT(scan_udec("128/8", &value, 65535), 3);
+  CHECK_INT(scan_uint("128/8", &value, 65535), 3);
   CHECK_INT(value, 128);
-  CHECK_INT(scan_udec("42\n", &value, 65535), 2);
+  CHECK_INT(scan_uint("42\n", &value, 65535), 2);
   CHECK_INT(value, 42);
 }
 
@@ -41,9 +41,9 @@ static void case_rejects_whitespace(void) {
   uint64_t value = 0xdead;
 
   test_case("leading whitespace is not skipped");
-  CHECK_INT(scan_udec(" 1", &value, 65535), 0);
-  CHECK_INT(scan_udec("\t1", &value, 65535), 0);
-  CHECK_INT(scan_udec("\n1", &value, 65535), 0);
+  CHECK_INT(scan_uint(" 1", &value, 65535), 0);
+  CHECK_INT(scan_uint("\t1", &value, 65535), 0);
+  CHECK_INT(scan_uint("\n1", &value, 65535), 0);
   CHECK_INT(value, 0xdead);
 }
 
@@ -51,9 +51,9 @@ static void case_rejects_sign(void) {
   uint64_t value = 0xdead;
 
   test_case("a sign is not accepted");
-  CHECK_INT(scan_udec("+1", &value, 65535), 0);
-  CHECK_INT(scan_udec("-1", &value, 65535), 0);
-  CHECK_INT(scan_udec("-0", &value, 65535), 0);
+  CHECK_INT(scan_uint("+1", &value, 65535), 0);
+  CHECK_INT(scan_uint("-1", &value, 65535), 0);
+  CHECK_INT(scan_uint("-0", &value, 65535), 0);
   CHECK_INT(value, 0xdead);
 }
 
@@ -61,9 +61,9 @@ static void case_rejects_empty_and_junk(void) {
   uint64_t value = 0xdead;
 
   test_case("a string with no leading digit yields nothing");
-  CHECK_INT(scan_udec("", &value, 65535), 0);
-  CHECK_INT(scan_udec("x", &value, 65535), 0);
-  CHECK_INT(scan_udec(":80", &value, 65535), 0);
+  CHECK_INT(scan_uint("", &value, 65535), 0);
+  CHECK_INT(scan_uint("x", &value, 65535), 0);
+  CHECK_INT(scan_uint(":80", &value, 65535), 0);
   CHECK_INT(value, 0xdead);
 }
 
@@ -71,9 +71,9 @@ static void case_no_alternative_bases(void) {
   uint64_t value;
 
   test_case("hexadecimal and octal notations are read as plain decimal");
-  CHECK_INT(scan_udec("0x10", &value, 65535), 1);
+  CHECK_INT(scan_uint("0x10", &value, 65535), 1);
   CHECK_INT(value, 0);
-  CHECK_INT(scan_udec("010", &value, 65535), 3);
+  CHECK_INT(scan_uint("010", &value, 65535), 3);
   CHECK_INT(value, 10);
 }
 
@@ -81,13 +81,13 @@ static void case_refuses_out_of_range(void) {
   uint64_t value = 0xdead;
 
   test_case("a value above max is refused, not truncated");
-  CHECK_INT(scan_udec("65536", &value, 65535), 0);
-  CHECK_INT(scan_udec("99999", &value, 65535), 0);
-  CHECK_INT(scan_udec("129", &value, 128), 0);
+  CHECK_INT(scan_uint("65536", &value, 65535), 0);
+  CHECK_INT(scan_uint("99999", &value, 65535), 0);
+  CHECK_INT(scan_uint("129", &value, 128), 0);
   CHECK_INT(value, 0xdead);
 
   test_case("the boundary itself is accepted");
-  CHECK_INT(scan_udec("128", &value, 128), 3);
+  CHECK_INT(scan_uint("128", &value, 128), 3);
   CHECK_INT(value, 128);
 }
 
@@ -95,21 +95,21 @@ static void case_sixty_four_bit_range(void) {
   uint64_t value;
 
   test_case("the full unsigned 64 bit range is available");
-  CHECK_INT(scan_udec("18446744073709551615", &value, UINT64_MAX), 20);
+  CHECK_INT(scan_uint("18446744073709551615", &value, UINT64_MAX), 20);
   CHECK(value == UINT64_MAX);
 
   test_case("a number too large for 64 bits is refused");
-  CHECK_INT(scan_udec("18446744073709551616", &value, UINT64_MAX), 0);
-  CHECK_INT(scan_udec("99999999999999999999999999", &value, UINT64_MAX), 0);
+  CHECK_INT(scan_uint("18446744073709551616", &value, UINT64_MAX), 0);
+  CHECK_INT(scan_uint("99999999999999999999999999", &value, UINT64_MAX), 0);
 }
 
 static void case_leading_zeros_allowed(void) {
   uint64_t value;
 
   test_case("leading zeros are accepted outside bencode");
-  CHECK_INT(scan_udec("007", &value, 65535), 3);
+  CHECK_INT(scan_uint("007", &value, 65535), 3);
   CHECK_INT(value, 7);
-  CHECK_INT(scan_udec("0000", &value, 65535), 4);
+  CHECK_INT(scan_uint("0000", &value, 65535), 4);
   CHECK_INT(value, 0);
 }
 
@@ -117,9 +117,9 @@ static void case_bencode_rejects_leading_zeros(void) {
   uint64_t value = 0xdead;
 
   test_case("bencode refuses a leading zero on anything but zero");
-  CHECK_INT(scan_udec_bencode("007", &value, 65535), 0);
-  CHECK_INT(scan_udec_bencode("0000", &value, 65535), 0);
-  CHECK_INT(scan_udec_bencode("01", &value, 65535), 0);
+  CHECK_INT(scan_uint_bencode("007", &value, 65535), 0);
+  CHECK_INT(scan_uint_bencode("0000", &value, 65535), 0);
+  CHECK_INT(scan_uint_bencode("01", &value, 65535), 0);
   CHECK_INT(value, 0xdead);
 }
 
@@ -127,11 +127,11 @@ static void case_bencode_accepts_canonical(void) {
   uint64_t value;
 
   test_case("bencode accepts a canonically written number");
-  CHECK_INT(scan_udec_bencode("0", &value, 65535), 1);
+  CHECK_INT(scan_uint_bencode("0", &value, 65535), 1);
   CHECK_INT(value, 0);
-  CHECK_INT(scan_udec_bencode("0:", &value, 65535), 1);
+  CHECK_INT(scan_uint_bencode("0:", &value, 65535), 1);
   CHECK_INT(value, 0);
-  CHECK_INT(scan_udec_bencode("160:", &value, 65535), 3);
+  CHECK_INT(scan_uint_bencode("160:", &value, 65535), 3);
   CHECK_INT(value, 160);
 }
 
@@ -139,14 +139,64 @@ static void case_bencode_inherits_the_rest(void) {
   uint64_t value = 0xdead;
 
   test_case("bencode is no laxer than the plain scanner");
-  CHECK_INT(scan_udec_bencode(" 1", &value, 65535), 0);
-  CHECK_INT(scan_udec_bencode("-1", &value, 65535), 0);
-  CHECK_INT(scan_udec_bencode("+1", &value, 65535), 0);
-  CHECK_INT(scan_udec_bencode("65536", &value, 65535), 0);
+  CHECK_INT(scan_uint_bencode(" 1", &value, 65535), 0);
+  CHECK_INT(scan_uint_bencode("-1", &value, 65535), 0);
+  CHECK_INT(scan_uint_bencode("+1", &value, 65535), 0);
+  CHECK_INT(scan_uint_bencode("65536", &value, 65535), 0);
   CHECK_INT(value, 0xdead);
 }
 
+/* ------------------------------------------- fixed width signed integers */
+
+/* A value lifted out of a query string arrives as a pointer and a length
+   with nothing terminating it, and the whole of it has to be the number. */
+
+static void case_int_whole_field(void) {
+  int64_t value = 0;
+
+  test_case("the whole field is read, sign and all");
+  CHECK_INT(scan_int("123", 3, &value), 0);
+  CHECK_INT(value, 123);
+  CHECK_INT(scan_int("-45", 3, &value), 0);
+  CHECK_INT(value, -45);
+
+  test_case("and only as far as the length given");
+  CHECK_INT(scan_int("12345", 3, &value), 0);
+  CHECK_INT(value, 123);
+}
+
+static void case_int_rejects_the_rest(void) {
+  int64_t value = 4711;
+
+  test_case("anything that is not all digits is refused");
+  CHECK_INT(scan_int("12a", 3, &value), -1);
+  CHECK_INT(scan_int("", 0, &value), -1);
+  CHECK_INT(scan_int("-", 1, &value), -1);
+  CHECK_INT(scan_int(" 12", 3, &value), -1);
+  CHECK_INT(scan_int("+12", 3, &value), -1);
+
+  test_case("and a refused field leaves the number alone");
+  CHECK_INT(value, 4711);
+}
+
+static void case_int_overflow(void) {
+  int64_t value = 0;
+
+  test_case("a field too long to fit is refused rather than wrapped");
+  CHECK_INT(scan_int("9223372036854775807", 19, &value), 0);
+  CHECK_INT(value, INT64_MAX);
+  CHECK_INT(scan_int("9223372036854775808", 19, &value), -1);
+  CHECK_INT(scan_int("99999999999999999999", 20, &value), -1);
+
+  test_case("including the negative side");
+  CHECK_INT(scan_int("-9223372036854775807", 20, &value), 0);
+  CHECK_INT(value, -INT64_MAX);
+}
+
 void test_scan(void) {
+  case_int_whole_field();
+  case_int_rejects_the_rest();
+  case_int_overflow();
   case_plain_numbers();
   case_stops_at_non_digit();
   case_rejects_whitespace();
diff --git a/tests/t_urlencoded.c b/tests/t_urlencoded.c
index 1239e6b..6e0ee31 100644
--- a/tests/t_urlencoded.c
+++ b/tests/t_urlencoded.c
@@ -228,33 +228,6 @@ static void case_find_keywords(void) {
   CHECK_INT(urlencoded_match_keyword(keywords, &p, SCAN_SEARCHPATH_PARAM), -3);
 }
 
-static void case_fixed_int(void) {
-  char positive[] = "123";
-  char negative[] = "-45";
-  char trailing[] = "12a";
-  int  value;
-
-  test_case("scan_fixed_int returns the count of unparsed bytes");
-  CHECK_INT(scan_fixed_int(positive, 3, &value), 0);
-  CHECK_INT(value, 123);
-
-  CHECK_INT(scan_fixed_int(negative, 3, &value), 0);
-  CHECK_INT(value, -45);
-
-  CHECK_INT(scan_fixed_int(trailing, 3, &value), 1);
-  CHECK_INT(value, 12);
-}
-
-static void case_fixed_int_bounded(void) {
-  char digits[] = "12345";
-  int  value;
-
-  test_case("scan_fixed_int honours the length it is given");
-  CHECK_INT(scan_fixed_int(digits, 3, &value), 0);
-  CHECK_INT(value, 123);
-}
-
-
 /* The encoding direction. Which bytes survive unescaped is fixed by what
    existing parsers of the urlencoded full scrape expect, so these cases pin
    the set rather than describe a preference. */
@@ -361,8 +334,6 @@ void test_urlencoded(void) {
   case_raw_high_byte();
   case_skipvalue();
   case_find_keywords();
-  case_fixed_int();
-  case_fixed_int_bounded();
   case_fmt_passes_safe_bytes();
   case_fmt_escapes_the_rest();
   case_fmt_hex_is_lower_case();