summaryrefslogtreecommitdiff
path: root/tests/t_scan_urlencoded_query.c
diff options
context:
space:
mode:
authorDirk Engling <erdgeist@erdgeist.org>2026-10-24 01:17:50 +0200
committerDirk Engling <erdgeist@erdgeist.org>2026-10-24 01:17:50 +0200
commitf24ec48b27d4cb8dd0555840206dedbb45f1423c (patch)
tree4cf772a476be6b06f52b80abc05f952a33209711 /tests/t_scan_urlencoded_query.c
parentf33203e15a00449152283cab2b878c0886fcede5 (diff)
Add a unit test harness and a first set of tests
make check builds tests/runner from the modules that link without the io subsystem and runs it.
Diffstat (limited to 'tests/t_scan_urlencoded_query.c')
-rw-r--r--tests/t_scan_urlencoded_query.c274
1 files changed, 274 insertions, 0 deletions
diff --git a/tests/t_scan_urlencoded_query.c b/tests/t_scan_urlencoded_query.c
new file mode 100644
index 0000000..4c1ef23
--- /dev/null
+++ b/tests/t_scan_urlencoded_query.c
@@ -0,0 +1,274 @@
1/* This software was written by Dirk Engling <erdgeist@erdgeist.org>
2 It is considered beerware. Prost. Skol. Cheers or whatever.
3
4 $id$ */
5
6/* System */
7#include <string.h>
8
9/* Opentracker */
10#include "scan_urlencoded_query.h"
11#include "test.h"
12
13/* The scanner returns the number of bytes written to the destination, or
14 -1 for a parse error and -2 for a hard terminator encountered before any
15 output was produced. It advances the source pointer past the separator it
16 stopped on, except for hard terminators and the '&' ending a parameter,
17 which it leaves for the next call to see. */
18
19static void case_path(void) {
20 char query[] = "/announce?info_hash=A";
21 char dest[64];
22 char *p = query;
23 ssize_t n;
24
25 test_case("path stops at the question mark");
26 n = scan_urlencoded_query(&p, dest, SCAN_PATH);
27 CHECK_INT(n, 9);
28 dest[n] = 0;
29 CHECK_STR(dest, "/announce");
30 CHECK_INT(*p, 'i');
31}
32
33static void case_param_and_value(void) {
34 char query[] = "info_hash=abc&port=80";
35 char dest[64];
36 char *p = query;
37 ssize_t n;
38
39 test_case("parameter stops at '=', value stops at '&'");
40 n = scan_urlencoded_query(&p, dest, SCAN_SEARCHPATH_PARAM);
41 CHECK_INT(n, 9);
42 dest[n] = 0;
43 CHECK_STR(dest, "info_hash");
44
45 n = scan_urlencoded_query(&p, dest, SCAN_SEARCHPATH_VALUE);
46 CHECK_INT(n, 3);
47 dest[n] = 0;
48 CHECK_STR(dest, "abc");
49
50 n = scan_urlencoded_query(&p, dest, SCAN_SEARCHPATH_PARAM);
51 CHECK_INT(n, 4);
52 dest[n] = 0;
53 CHECK_STR(dest, "port");
54
55 n = scan_urlencoded_query(&p, dest, SCAN_SEARCHPATH_VALUE);
56 CHECK_INT(n, 2);
57 dest[n] = 0;
58 CHECK_STR(dest, "80");
59}
60
61static void case_percent_decoding(void) {
62 char query[] = "%41%42%43&x=1";
63 char dest[64];
64 char *p = query;
65 ssize_t n;
66
67 test_case("percent escapes decode");
68 n = scan_urlencoded_query(&p, dest, SCAN_SEARCHPATH_VALUE);
69 CHECK_INT(n, 3);
70 CHECK_MEM(dest, "ABC", 3);
71}
72
73static void case_binary_info_hash(void) {
74 /* The twenty byte info_hash is what the announce path really decodes */
75 char query[] = "%00%01%02%ff%fe%7f%80ABCDEFGHIJKLM&x=1";
76 unsigned char want[] = {0x00, 0x01, 0x02, 0xff, 0xfe, 0x7f, 0x80, 'A', 'B', 'C', 'D', 'E', 'F', 'G', 'H', 'I', 'J', 'K', 'L', 'M'};
77 char dest[64];
78 char *p = query;
79 ssize_t n;
80
81 test_case("twenty byte binary value survives decoding");
82 n = scan_urlencoded_query(&p, dest, SCAN_SEARCHPATH_VALUE);
83 CHECK_INT(n, 20);
84 CHECK_MEM(dest, want, 20);
85}
86
87static void case_in_place(void) {
88 char query[] = "%41%42&x=1";
89 char *p = query;
90 ssize_t n;
91
92 test_case("decoding in place over the source is safe");
93 n = scan_urlencoded_query(&p, query, SCAN_SEARCHPATH_VALUE);
94 CHECK_INT(n, 2);
95 CHECK_MEM(query, "AB", 2);
96}
97
98static void case_bad_escape(void) {
99 char query[] = "%zz";
100 char dest[64];
101 char *p = query;
102 ssize_t n;
103
104 test_case("a non hexadecimal escape is a parse error");
105 n = scan_urlencoded_query(&p, dest, SCAN_SEARCHPATH_VALUE);
106 CHECK_INT(n, -1);
107}
108
109static void case_truncated_escape(void) {
110 char query[] = "%4";
111 char dest[64];
112 char *p = query;
113 ssize_t n;
114
115 test_case("an escape cut short by the terminator is a parse error");
116 n = scan_urlencoded_query(&p, dest, SCAN_SEARCHPATH_VALUE);
117 CHECK_INT(n, -1);
118}
119
120static void case_terminator(void) {
121 char empty[] = "";
122 char space[] = " rest";
123 char dest[64];
124 char *p = empty;
125 ssize_t n;
126
127 test_case("a hard terminator before any output reports end of input");
128 n = scan_urlencoded_query(&p, dest, SCAN_SEARCHPATH_PARAM);
129 CHECK_INT(n, -2);
130
131 p = space;
132 n = scan_urlencoded_query(&p, dest, SCAN_SEARCHPATH_PARAM);
133 CHECK_INT(n, -2);
134}
135
136static void case_terminator_after_output(void) {
137 char query[] = "abc";
138 char dest[64];
139 char *p = query;
140 ssize_t n;
141
142 test_case("a hard terminator after output is left for the next call");
143 n = scan_urlencoded_query(&p, dest, SCAN_SEARCHPATH_VALUE);
144 CHECK_INT(n, 3);
145 CHECK_INT(*p, 0);
146 n = scan_urlencoded_query(&p, dest, SCAN_SEARCHPATH_VALUE);
147 CHECK_INT(n, -2);
148}
149
150/* The character matrix makes '?' terminate a path only, and '=' terminate a
151 path or a parameter only. Both are ordinary characters inside a value, so a
152 value is free to contain a nested query string. */
153
154static void case_separators_are_contextual(void) {
155 char question[] = "a?b&x=1";
156 char equals[] = "a=b&x=1";
157 char dest[64];
158 char *p = question;
159 ssize_t n;
160
161 test_case("a question mark is literal inside a value");
162 n = scan_urlencoded_query(&p, dest, SCAN_SEARCHPATH_VALUE);
163 CHECK_INT(n, 3);
164 dest[n] = 0;
165 CHECK_STR(dest, "a?b");
166
167 test_case("an equals sign is literal inside a value");
168 p = equals;
169 n = scan_urlencoded_query(&p, dest, SCAN_SEARCHPATH_VALUE);
170 CHECK_INT(n, 3);
171 dest[n] = 0;
172 CHECK_STR(dest, "a=b");
173}
174
175static void case_ampersand_in_path(void) {
176 char query[] = "/path&x";
177 char dest[64];
178 char *p = query;
179 ssize_t n;
180
181 test_case("an ampersand in a path is a parse error");
182 n = scan_urlencoded_query(&p, dest, SCAN_PATH);
183 CHECK_INT(n, -1);
184}
185
186static void case_raw_high_byte(void) {
187 char query[] = {'a', 'b', (char)0x80, 0};
188 char dest[64];
189 char *p = query;
190 ssize_t n;
191
192 test_case("an undecoded byte above ASCII is a parse error");
193 n = scan_urlencoded_query(&p, dest, SCAN_SEARCHPATH_VALUE);
194 CHECK_INT(n, -1);
195}
196
197static void case_skipvalue(void) {
198 char query[] = "abc&next=1";
199 char bare[] = "abc";
200 char *p = query;
201
202 test_case("skipvalue steps over the value and its separator");
203 scan_urlencoded_skipvalue(&p);
204 CHECK_STR(p, "next=1");
205
206 test_case("skipvalue leaves a hard terminator in place");
207 p = bare;
208 scan_urlencoded_skipvalue(&p);
209 CHECK_INT(*p, 0);
210}
211
212static void case_find_keywords(void) {
213 static const ot_keywords keywords[] = {{"port", 1}, {"left", 2}, {"compact", 3}, {NULL, -3}};
214 char match[] = "left=0";
215 char nomatch[] = "bogus=1";
216 char prefix[] = "por=1";
217 char *p = match;
218
219 test_case("a known keyword returns its value");
220 CHECK_INT(scan_find_keywords(keywords, &p, SCAN_SEARCHPATH_PARAM), 2);
221
222 test_case("an unknown keyword returns -3");
223 p = nomatch;
224 CHECK_INT(scan_find_keywords(keywords, &p, SCAN_SEARCHPATH_PARAM), -3);
225
226 test_case("a prefix of a keyword does not match it");
227 p = prefix;
228 CHECK_INT(scan_find_keywords(keywords, &p, SCAN_SEARCHPATH_PARAM), -3);
229}
230
231static void case_fixed_int(void) {
232 char positive[] = "123";
233 char negative[] = "-45";
234 char trailing[] = "12a";
235 int value;
236
237 test_case("scan_fixed_int returns the count of unparsed bytes");
238 CHECK_INT(scan_fixed_int(positive, 3, &value), 0);
239 CHECK_INT(value, 123);
240
241 CHECK_INT(scan_fixed_int(negative, 3, &value), 0);
242 CHECK_INT(value, -45);
243
244 CHECK_INT(scan_fixed_int(trailing, 3, &value), 1);
245 CHECK_INT(value, 12);
246}
247
248static void case_fixed_int_bounded(void) {
249 char digits[] = "12345";
250 int value;
251
252 test_case("scan_fixed_int honours the length it is given");
253 CHECK_INT(scan_fixed_int(digits, 3, &value), 0);
254 CHECK_INT(value, 123);
255}
256
257void test_scan_urlencoded_query(void) {
258 case_path();
259 case_param_and_value();
260 case_percent_decoding();
261 case_binary_info_hash();
262 case_in_place();
263 case_bad_escape();
264 case_truncated_escape();
265 case_terminator();
266 case_terminator_after_output();
267 case_separators_are_contextual();
268 case_ampersand_in_path();
269 case_raw_high_byte();
270 case_skipvalue();
271 case_find_keywords();
272 case_fixed_int();
273 case_fixed_int_bounded();
274}