http-backend: respect CONTENT_LENGTH for receive-pack
[git] / userdiff.c
1 #include "cache.h"
2 #include "config.h"
3 #include "userdiff.h"
4 #include "attr.h"
5
6 static struct userdiff_driver *drivers;
7 static int ndrivers;
8 static int drivers_alloc;
9
10 #define PATTERNS(name, pattern, word_regex)                     \
11         { name, NULL, -1, { pattern, REG_EXTENDED },            \
12           word_regex "|[^[:space:]]|[\xc0-\xff][\x80-\xbf]+" }
13 #define IPATTERN(name, pattern, word_regex)                     \
14         { name, NULL, -1, { pattern, REG_EXTENDED | REG_ICASE }, \
15           word_regex "|[^[:space:]]|[\xc0-\xff][\x80-\xbf]+" }
16 static struct userdiff_driver builtin_drivers[] = {
17 IPATTERN("ada",
18          "!^(.*[ \t])?(is[ \t]+new|renames|is[ \t]+separate)([ \t].*)?$\n"
19          "!^[ \t]*with[ \t].*$\n"
20          "^[ \t]*((procedure|function)[ \t]+.*)$\n"
21          "^[ \t]*((package|protected|task)[ \t]+.*)$",
22          /* -- */
23          "[a-zA-Z][a-zA-Z0-9_]*"
24          "|[-+]?[0-9][0-9#_.aAbBcCdDeEfF]*([eE][+-]?[0-9_]+)?"
25          "|=>|\\.\\.|\\*\\*|:=|/=|>=|<=|<<|>>|<>"),
26 IPATTERN("fortran",
27          "!^([C*]|[ \t]*!)\n"
28          "!^[ \t]*MODULE[ \t]+PROCEDURE[ \t]\n"
29          "^[ \t]*((END[ \t]+)?(PROGRAM|MODULE|BLOCK[ \t]+DATA"
30                 "|([^'\" \t]+[ \t]+)*(SUBROUTINE|FUNCTION))[ \t]+[A-Z].*)$",
31          /* -- */
32          "[a-zA-Z][a-zA-Z0-9_]*"
33          "|\\.([Ee][Qq]|[Nn][Ee]|[Gg][TtEe]|[Ll][TtEe]|[Tt][Rr][Uu][Ee]|[Ff][Aa][Ll][Ss][Ee]|[Aa][Nn][Dd]|[Oo][Rr]|[Nn]?[Ee][Qq][Vv]|[Nn][Oo][Tt])\\."
34          /* numbers and format statements like 2E14.4, or ES12.6, 9X.
35           * Don't worry about format statements without leading digits since
36           * they would have been matched above as a variable anyway. */
37          "|[-+]?[0-9.]+([AaIiDdEeFfLlTtXx][Ss]?[-+]?[0-9.]*)?(_[a-zA-Z0-9][a-zA-Z0-9_]*)?"
38          "|//|\\*\\*|::|[/<>=]="),
39 IPATTERN("fountain", "^((\\.[^.]|(int|ext|est|int\\.?/ext|i/e)[. ]).*)$",
40          "[^ \t-]+"),
41 PATTERNS("golang",
42          /* Functions */
43          "^[ \t]*(func[ \t]*.*(\\{[ \t]*)?)\n"
44          /* Structs and interfaces */
45          "^[ \t]*(type[ \t].*(struct|interface)[ \t]*(\\{[ \t]*)?)",
46          /* -- */
47          "[a-zA-Z_][a-zA-Z0-9_]*"
48          "|[-+0-9.eE]+i?|0[xX]?[0-9a-fA-F]+i?"
49          "|[-+*/<>%&^|=!:]=|--|\\+\\+|<<=?|>>=?|&\\^=?|&&|\\|\\||<-|\\.{3}"),
50 PATTERNS("html", "^[ \t]*(<[Hh][1-6]([ \t].*)?>.*)$",
51          "[^<>= \t]+"),
52 PATTERNS("java",
53          "!^[ \t]*(catch|do|for|if|instanceof|new|return|switch|throw|while)\n"
54          "^[ \t]*(([A-Za-z_][A-Za-z_0-9]*[ \t]+)+[A-Za-z_][A-Za-z_0-9]*[ \t]*\\([^;]*)$",
55          /* -- */
56          "[a-zA-Z_][a-zA-Z0-9_]*"
57          "|[-+0-9.e]+[fFlL]?|0[xXbB]?[0-9a-fA-F]+[lL]?"
58          "|[-+*/<>%&^|=!]="
59          "|--|\\+\\+|<<=?|>>>?=?|&&|\\|\\|"),
60 PATTERNS("matlab",
61          "^[[:space:]]*((classdef|function)[[:space:]].*)$|^%%[[:space:]].*$",
62          "[a-zA-Z_][a-zA-Z0-9_]*|[-+0-9.e]+|[=~<>]=|\\.[*/\\^']|\\|\\||&&"),
63 PATTERNS("objc",
64          /* Negate C statements that can look like functions */
65          "!^[ \t]*(do|for|if|else|return|switch|while)\n"
66          /* Objective-C methods */
67          "^[ \t]*([-+][ \t]*\\([ \t]*[A-Za-z_][A-Za-z_0-9* \t]*\\)[ \t]*[A-Za-z_].*)$\n"
68          /* C functions */
69          "^[ \t]*(([A-Za-z_][A-Za-z_0-9]*[ \t]+)+[A-Za-z_][A-Za-z_0-9]*[ \t]*\\([^;]*)$\n"
70          /* Objective-C class/protocol definitions */
71          "^(@(implementation|interface|protocol)[ \t].*)$",
72          /* -- */
73          "[a-zA-Z_][a-zA-Z0-9_]*"
74          "|[-+0-9.e]+[fFlL]?|0[xXbB]?[0-9a-fA-F]+[lL]?"
75          "|[-+*/<>%&^|=!]=|--|\\+\\+|<<=?|>>=?|&&|\\|\\||::|->"),
76 PATTERNS("pascal",
77          "^(((class[ \t]+)?(procedure|function)|constructor|destructor|interface|"
78                 "implementation|initialization|finalization)[ \t]*.*)$"
79          "\n"
80          "^(.*=[ \t]*(class|record).*)$",
81          /* -- */
82          "[a-zA-Z_][a-zA-Z0-9_]*"
83          "|[-+0-9.e]+|0[xXbB]?[0-9a-fA-F]+"
84          "|<>|<=|>=|:=|\\.\\."),
85 PATTERNS("perl",
86          "^package .*\n"
87          "^sub [[:alnum:]_':]+[ \t]*"
88                 "(\\([^)]*\\)[ \t]*)?" /* prototype */
89                 /*
90                  * Attributes.  A regex can't count nested parentheses,
91                  * so just slurp up whatever we see, taking care not
92                  * to accept lines like "sub foo; # defined elsewhere".
93                  *
94                  * An attribute could contain a semicolon, but at that
95                  * point it seems reasonable enough to give up.
96                  */
97                 "(:[^;#]*)?"
98                 "(\\{[ \t]*)?" /* brace can come here or on the next line */
99                 "(#.*)?$\n" /* comment */
100          "^(BEGIN|END|INIT|CHECK|UNITCHECK|AUTOLOAD|DESTROY)[ \t]*"
101                 "(\\{[ \t]*)?" /* brace can come here or on the next line */
102                 "(#.*)?$\n"
103          "^=head[0-9] .*",      /* POD */
104          /* -- */
105          "[[:alpha:]_'][[:alnum:]_']*"
106          "|0[xb]?[0-9a-fA-F_]*"
107          /* taking care not to interpret 3..5 as (3.)(.5) */
108          "|[0-9a-fA-F_]+(\\.[0-9a-fA-F_]+)?([eE][-+]?[0-9_]+)?"
109          "|=>|-[rwxoRWXOezsfdlpSugkbctTBMAC>]|~~|::"
110          "|&&=|\\|\\|=|//=|\\*\\*="
111          "|&&|\\|\\||//|\\+\\+|--|\\*\\*|\\.\\.\\.?"
112          "|[-+*/%.^&<>=!|]="
113          "|=~|!~"
114          "|<<|<>|<=>|>>"),
115 PATTERNS("php",
116          "^[\t ]*(((public|protected|private|static)[\t ]+)*function.*)$\n"
117          "^[\t ]*(class.*)$",
118          /* -- */
119          "[a-zA-Z_][a-zA-Z0-9_]*"
120          "|[-+0-9.e]+|0[xXbB]?[0-9a-fA-F]+"
121          "|[-+*/<>%&^|=!.]=|--|\\+\\+|<<=?|>>=?|===|&&|\\|\\||::|->"),
122 PATTERNS("python", "^[ \t]*((class|def)[ \t].*)$",
123          /* -- */
124          "[a-zA-Z_][a-zA-Z0-9_]*"
125          "|[-+0-9.e]+[jJlL]?|0[xX]?[0-9a-fA-F]+[lL]?"
126          "|[-+*/<>%&^|=!]=|//=?|<<=?|>>=?|\\*\\*=?"),
127          /* -- */
128 PATTERNS("ruby", "^[ \t]*((class|module|def)[ \t].*)$",
129          /* -- */
130          "(@|@@|\\$)?[a-zA-Z_][a-zA-Z0-9_]*"
131          "|[-+0-9.e]+|0[xXbB]?[0-9a-fA-F]+|\\?(\\\\C-)?(\\\\M-)?."
132          "|//=?|[-+*/<>%&^|=!]=|<<=?|>>=?|===|\\.{1,3}|::|[!=]~"),
133 PATTERNS("bibtex", "(@[a-zA-Z]{1,}[ \t]*\\{{0,1}[ \t]*[^ \t\"@',\\#}{~%]*).*$",
134          "[={}\"]|[^={}\" \t]+"),
135 PATTERNS("tex", "^(\\\\((sub)*section|chapter|part)\\*{0,1}\\{.*)$",
136          "\\\\[a-zA-Z@]+|\\\\.|[a-zA-Z0-9\x80-\xff]+"),
137 PATTERNS("cpp",
138          /* Jump targets or access declarations */
139          "!^[ \t]*[A-Za-z_][A-Za-z_0-9]*:[[:space:]]*($|/[/*])\n"
140          /* functions/methods, variables, and compounds at top level */
141          "^((::[[:space:]]*)?[A-Za-z_].*)$",
142          /* -- */
143          "[a-zA-Z_][a-zA-Z0-9_]*"
144          "|[-+0-9.e]+[fFlL]?|0[xXbB]?[0-9a-fA-F]+[lLuU]*"
145          "|[-+*/<>%&^|=!]=|--|\\+\\+|<<=?|>>=?|&&|\\|\\||::|->\\*?|\\.\\*"),
146 PATTERNS("csharp",
147          /* Keywords */
148          "!^[ \t]*(do|while|for|if|else|instanceof|new|return|switch|case|throw|catch|using)\n"
149          /* Methods and constructors */
150          "^[ \t]*(((static|public|internal|private|protected|new|virtual|sealed|override|unsafe|async)[ \t]+)*[][<>@.~_[:alnum:]]+[ \t]+[<>@._[:alnum:]]+[ \t]*\\(.*\\))[ \t]*$\n"
151          /* Properties */
152          "^[ \t]*(((static|public|internal|private|protected|new|virtual|sealed|override|unsafe)[ \t]+)*[][<>@.~_[:alnum:]]+[ \t]+[@._[:alnum:]]+)[ \t]*$\n"
153          /* Type definitions */
154          "^[ \t]*(((static|public|internal|private|protected|new|unsafe|sealed|abstract|partial)[ \t]+)*(class|enum|interface|struct)[ \t]+.*)$\n"
155          /* Namespace */
156          "^[ \t]*(namespace[ \t]+.*)$",
157          /* -- */
158          "[a-zA-Z_][a-zA-Z0-9_]*"
159          "|[-+0-9.e]+[fFlL]?|0[xXbB]?[0-9a-fA-F]+[lL]?"
160          "|[-+*/<>%&^|=!]=|--|\\+\\+|<<=?|>>=?|&&|\\|\\||::|->"),
161 IPATTERN("css",
162          "![:;][[:space:]]*$\n"
163          "^[_a-z0-9].*$",
164          /* -- */
165          /*
166           * This regex comes from W3C CSS specs. Should theoretically also
167           * allow ISO 10646 characters U+00A0 and higher,
168           * but they are not handled in this regex.
169           */
170          "-?[_a-zA-Z][-_a-zA-Z0-9]*" /* identifiers */
171          "|-?[0-9]+|\\#[0-9a-fA-F]+" /* numbers */
172 ),
173 { "default", NULL, -1, { NULL, 0 } },
174 };
175 #undef PATTERNS
176 #undef IPATTERN
177
178 static struct userdiff_driver driver_true = {
179         "diff=true",
180         NULL,
181         0,
182         { NULL, 0 }
183 };
184
185 static struct userdiff_driver driver_false = {
186         "!diff",
187         NULL,
188         1,
189         { NULL, 0 }
190 };
191
192 static struct userdiff_driver *userdiff_find_by_namelen(const char *k, int len)
193 {
194         int i;
195         for (i = 0; i < ndrivers; i++) {
196                 struct userdiff_driver *drv = drivers + i;
197                 if (!strncmp(drv->name, k, len) && !drv->name[len])
198                         return drv;
199         }
200         for (i = 0; i < ARRAY_SIZE(builtin_drivers); i++) {
201                 struct userdiff_driver *drv = builtin_drivers + i;
202                 if (!strncmp(drv->name, k, len) && !drv->name[len])
203                         return drv;
204         }
205         return NULL;
206 }
207
208 static int parse_funcname(struct userdiff_funcname *f, const char *k,
209                 const char *v, int cflags)
210 {
211         if (git_config_string(&f->pattern, k, v) < 0)
212                 return -1;
213         f->cflags = cflags;
214         return 0;
215 }
216
217 static int parse_tristate(int *b, const char *k, const char *v)
218 {
219         if (v && !strcasecmp(v, "auto"))
220                 *b = -1;
221         else
222                 *b = git_config_bool(k, v);
223         return 0;
224 }
225
226 static int parse_bool(int *b, const char *k, const char *v)
227 {
228         *b = git_config_bool(k, v);
229         return 0;
230 }
231
232 int userdiff_config(const char *k, const char *v)
233 {
234         struct userdiff_driver *drv;
235         const char *name, *type;
236         int namelen;
237
238         if (parse_config_key(k, "diff", &name, &namelen, &type) || !name)
239                 return 0;
240
241         drv = userdiff_find_by_namelen(name, namelen);
242         if (!drv) {
243                 ALLOC_GROW(drivers, ndrivers+1, drivers_alloc);
244                 drv = &drivers[ndrivers++];
245                 memset(drv, 0, sizeof(*drv));
246                 drv->name = xmemdupz(name, namelen);
247                 drv->binary = -1;
248         }
249
250         if (!strcmp(type, "funcname"))
251                 return parse_funcname(&drv->funcname, k, v, 0);
252         if (!strcmp(type, "xfuncname"))
253                 return parse_funcname(&drv->funcname, k, v, REG_EXTENDED);
254         if (!strcmp(type, "binary"))
255                 return parse_tristate(&drv->binary, k, v);
256         if (!strcmp(type, "command"))
257                 return git_config_string(&drv->external, k, v);
258         if (!strcmp(type, "textconv"))
259                 return git_config_string(&drv->textconv, k, v);
260         if (!strcmp(type, "cachetextconv"))
261                 return parse_bool(&drv->textconv_want_cache, k, v);
262         if (!strcmp(type, "wordregex"))
263                 return git_config_string(&drv->word_regex, k, v);
264
265         return 0;
266 }
267
268 struct userdiff_driver *userdiff_find_by_name(const char *name) {
269         int len = strlen(name);
270         return userdiff_find_by_namelen(name, len);
271 }
272
273 struct userdiff_driver *userdiff_find_by_path(const char *path)
274 {
275         static struct attr_check *check;
276
277         if (!check)
278                 check = attr_check_initl("diff", NULL);
279         if (!path)
280                 return NULL;
281         if (git_check_attr(path, check))
282                 return NULL;
283
284         if (ATTR_TRUE(check->items[0].value))
285                 return &driver_true;
286         if (ATTR_FALSE(check->items[0].value))
287                 return &driver_false;
288         if (ATTR_UNSET(check->items[0].value))
289                 return NULL;
290         return userdiff_find_by_name(check->items[0].value);
291 }
292
293 struct userdiff_driver *userdiff_get_textconv(struct userdiff_driver *driver)
294 {
295         if (!driver->textconv)
296                 return NULL;
297
298         if (driver->textconv_want_cache && !driver->textconv_cache) {
299                 struct notes_cache *c = xmalloc(sizeof(*c));
300                 struct strbuf name = STRBUF_INIT;
301
302                 strbuf_addf(&name, "textconv/%s", driver->name);
303                 notes_cache_init(c, name.buf, driver->textconv);
304                 driver->textconv_cache = c;
305                 strbuf_release(&name);
306         }
307
308         return driver;
309 }