Merge branch 'jc/maint-doc-em-dash'
[git] / userdiff.c
1 #include "cache.h"
2 #include "userdiff.h"
3 #include "cache.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 }, word_regex }
12 static struct userdiff_driver builtin_drivers[] = {
13 PATTERNS("html", "^[ \t]*(<[Hh][1-6][ \t].*>.*)$",
14          "[^<>= \t]+|[^[:space:]]|[\x80-\xff]+"),
15 PATTERNS("java",
16          "!^[ \t]*(catch|do|for|if|instanceof|new|return|switch|throw|while)\n"
17          "^[ \t]*(([A-Za-z_][A-Za-z_0-9]*[ \t]+)+[A-Za-z_][A-Za-z_0-9]*[ \t]*\\([^;]*)$",
18          /* -- */
19          "[a-zA-Z_][a-zA-Z0-9_]*"
20          "|[-+0-9.e]+[fFlL]?|0[xXbB]?[0-9a-fA-F]+[lL]?"
21          "|[-+*/<>%&^|=!]="
22          "|--|\\+\\+|<<=?|>>>?=?|&&|\\|\\|"
23          "|[^[:space:]]|[\x80-\xff]+"),
24 PATTERNS("objc",
25          /* Negate C statements that can look like functions */
26          "!^[ \t]*(do|for|if|else|return|switch|while)\n"
27          /* Objective-C methods */
28          "^[ \t]*([-+][ \t]*\\([ \t]*[A-Za-z_][A-Za-z_0-9* \t]*\\)[ \t]*[A-Za-z_].*)$\n"
29          /* C functions */
30          "^[ \t]*(([A-Za-z_][A-Za-z_0-9]*[ \t]+)+[A-Za-z_][A-Za-z_0-9]*[ \t]*\\([^;]*)$\n"
31          /* Objective-C class/protocol definitions */
32          "^(@(implementation|interface|protocol)[ \t].*)$",
33          /* -- */
34          "[a-zA-Z_][a-zA-Z0-9_]*"
35          "|[-+0-9.e]+[fFlL]?|0[xXbB]?[0-9a-fA-F]+[lL]?"
36          "|[-+*/<>%&^|=!]=|--|\\+\\+|<<=?|>>=?|&&|\\|\\||::|->"
37          "|[^[:space:]]|[\x80-\xff]+"),
38 PATTERNS("pascal",
39          "^((procedure|function|constructor|destructor|interface|"
40                 "implementation|initialization|finalization)[ \t]*.*)$"
41          "\n"
42          "^(.*=[ \t]*(class|record).*)$",
43          /* -- */
44          "[a-zA-Z_][a-zA-Z0-9_]*"
45          "|[-+0-9.e]+|0[xXbB]?[0-9a-fA-F]+"
46          "|<>|<=|>=|:=|\\.\\."
47          "|[^[:space:]]|[\x80-\xff]+"),
48 PATTERNS("php",
49          "^[\t ]*(((public|protected|private|static)[\t ]+)*function.*)$\n"
50          "^[\t ]*(class.*)$",
51          /* -- */
52          "[a-zA-Z_][a-zA-Z0-9_]*"
53          "|[-+0-9.e]+|0[xXbB]?[0-9a-fA-F]+"
54          "|[-+*/<>%&^|=!.]=|--|\\+\\+|<<=?|>>=?|===|&&|\\|\\||::|->"
55          "|[^[:space:]]|[\x80-\xff]+"),
56 PATTERNS("python", "^[ \t]*((class|def)[ \t].*)$",
57          /* -- */
58          "[a-zA-Z_][a-zA-Z0-9_]*"
59          "|[-+0-9.e]+[jJlL]?|0[xX]?[0-9a-fA-F]+[lL]?"
60          "|[-+*/<>%&^|=!]=|//=?|<<=?|>>=?|\\*\\*=?"
61          "|[^[:space:]|[\x80-\xff]+"),
62          /* -- */
63 PATTERNS("ruby", "^[ \t]*((class|module|def)[ \t].*)$",
64          /* -- */
65          "(@|@@|\\$)?[a-zA-Z_][a-zA-Z0-9_]*"
66          "|[-+0-9.e]+|0[xXbB]?[0-9a-fA-F]+|\\?(\\\\C-)?(\\\\M-)?."
67          "|//=?|[-+*/<>%&^|=!]=|<<=?|>>=?|===|\\.{1,3}|::|[!=]~"
68          "|[^[:space:]|[\x80-\xff]+"),
69 PATTERNS("bibtex", "(@[a-zA-Z]{1,}[ \t]*\\{{0,1}[ \t]*[^ \t\"@',\\#}{~%]*).*$",
70          "[={}\"]|[^={}\" \t]+"),
71 PATTERNS("tex", "^(\\\\((sub)*section|chapter|part)\\*{0,1}\\{.*)$",
72          "\\\\[a-zA-Z@]+|\\\\.|[a-zA-Z0-9\x80-\xff]+|[^[:space:]]"),
73 PATTERNS("cpp",
74          /* Jump targets or access declarations */
75          "!^[ \t]*[A-Za-z_][A-Za-z_0-9]*:.*$\n"
76          /* C/++ functions/methods at top level */
77          "^([A-Za-z_][A-Za-z_0-9]*([ \t]+[A-Za-z_][A-Za-z_0-9]*([ \t]*::[ \t]*[^[:space:]]+)?){1,}[ \t]*\\([^;]*)$\n"
78          /* compound type at top level */
79          "^((struct|class|enum)[^;]*)$",
80          /* -- */
81          "[a-zA-Z_][a-zA-Z0-9_]*"
82          "|[-+0-9.e]+[fFlL]?|0[xXbB]?[0-9a-fA-F]+[lL]?"
83          "|[-+*/<>%&^|=!]=|--|\\+\\+|<<=?|>>=?|&&|\\|\\||::|->"
84          "|[^[:space:]]|[\x80-\xff]+"),
85 PATTERNS("csharp",
86          /* Keywords */
87          "!^[ \t]*(do|while|for|if|else|instanceof|new|return|switch|case|throw|catch|using)\n"
88          /* Methods and constructors */
89          "^[ \t]*(((static|public|internal|private|protected|new|virtual|sealed|override|unsafe)[ \t]+)*[][<>@.~_[:alnum:]]+[ \t]+[<>@._[:alnum:]]+[ \t]*\\(.*\\))[ \t]*$\n"
90          /* Properties */
91          "^[ \t]*(((static|public|internal|private|protected|new|virtual|sealed|override|unsafe)[ \t]+)*[][<>@.~_[:alnum:]]+[ \t]+[@._[:alnum:]]+)[ \t]*$\n"
92          /* Type definitions */
93          "^[ \t]*(((static|public|internal|private|protected|new|unsafe|sealed|abstract|partial)[ \t]+)*(class|enum|interface|struct)[ \t]+.*)$\n"
94          /* Namespace */
95          "^[ \t]*(namespace[ \t]+.*)$",
96          /* -- */
97          "[a-zA-Z_][a-zA-Z0-9_]*"
98          "|[-+0-9.e]+[fFlL]?|0[xXbB]?[0-9a-fA-F]+[lL]?"
99          "|[-+*/<>%&^|=!]=|--|\\+\\+|<<=?|>>=?|&&|\\|\\||::|->"
100          "|[^[:space:]]|[\x80-\xff]+"),
101 { "default", NULL, -1, { NULL, 0 } },
102 };
103 #undef PATTERNS
104
105 static struct userdiff_driver driver_true = {
106         "diff=true",
107         NULL,
108         0,
109         { NULL, 0 }
110 };
111
112 static struct userdiff_driver driver_false = {
113         "!diff",
114         NULL,
115         1,
116         { NULL, 0 }
117 };
118
119 static struct userdiff_driver *userdiff_find_by_namelen(const char *k, int len)
120 {
121         int i;
122         for (i = 0; i < ndrivers; i++) {
123                 struct userdiff_driver *drv = drivers + i;
124                 if (!strncmp(drv->name, k, len) && !drv->name[len])
125                         return drv;
126         }
127         for (i = 0; i < ARRAY_SIZE(builtin_drivers); i++) {
128                 struct userdiff_driver *drv = builtin_drivers + i;
129                 if (!strncmp(drv->name, k, len) && !drv->name[len])
130                         return drv;
131         }
132         return NULL;
133 }
134
135 static struct userdiff_driver *parse_driver(const char *var,
136                 const char *value, const char *type)
137 {
138         struct userdiff_driver *drv;
139         const char *dot;
140         const char *name;
141         int namelen;
142
143         if (prefixcmp(var, "diff."))
144                 return NULL;
145         dot = strrchr(var, '.');
146         if (dot == var + 4)
147                 return NULL;
148         if (strcmp(type, dot+1))
149                 return NULL;
150
151         name = var + 5;
152         namelen = dot - name;
153         drv = userdiff_find_by_namelen(name, namelen);
154         if (!drv) {
155                 ALLOC_GROW(drivers, ndrivers+1, drivers_alloc);
156                 drv = &drivers[ndrivers++];
157                 memset(drv, 0, sizeof(*drv));
158                 drv->name = xmemdupz(name, namelen);
159                 drv->binary = -1;
160         }
161         return drv;
162 }
163
164 static int parse_funcname(struct userdiff_funcname *f, const char *k,
165                 const char *v, int cflags)
166 {
167         if (git_config_string(&f->pattern, k, v) < 0)
168                 return -1;
169         f->cflags = cflags;
170         return 1;
171 }
172
173 static int parse_string(const char **d, const char *k, const char *v)
174 {
175         if (git_config_string(d, k, v) < 0)
176                 return -1;
177         return 1;
178 }
179
180 static int parse_tristate(int *b, const char *k, const char *v)
181 {
182         if (v && !strcasecmp(v, "auto"))
183                 *b = -1;
184         else
185                 *b = git_config_bool(k, v);
186         return 1;
187 }
188
189 static int parse_bool(int *b, const char *k, const char *v)
190 {
191         *b = git_config_bool(k, v);
192         return 1;
193 }
194
195 int userdiff_config(const char *k, const char *v)
196 {
197         struct userdiff_driver *drv;
198
199         if ((drv = parse_driver(k, v, "funcname")))
200                 return parse_funcname(&drv->funcname, k, v, 0);
201         if ((drv = parse_driver(k, v, "xfuncname")))
202                 return parse_funcname(&drv->funcname, k, v, REG_EXTENDED);
203         if ((drv = parse_driver(k, v, "binary")))
204                 return parse_tristate(&drv->binary, k, v);
205         if ((drv = parse_driver(k, v, "command")))
206                 return parse_string(&drv->external, k, v);
207         if ((drv = parse_driver(k, v, "textconv")))
208                 return parse_string(&drv->textconv, k, v);
209         if ((drv = parse_driver(k, v, "cachetextconv")))
210                 return parse_bool(&drv->textconv_want_cache, k, v);
211         if ((drv = parse_driver(k, v, "wordregex")))
212                 return parse_string(&drv->word_regex, k, v);
213
214         return 0;
215 }
216
217 struct userdiff_driver *userdiff_find_by_name(const char *name) {
218         int len = strlen(name);
219         return userdiff_find_by_namelen(name, len);
220 }
221
222 struct userdiff_driver *userdiff_find_by_path(const char *path)
223 {
224         static struct git_attr *attr;
225         struct git_attr_check check;
226
227         if (!attr)
228                 attr = git_attr("diff");
229         check.attr = attr;
230
231         if (!path)
232                 return NULL;
233         if (git_checkattr(path, 1, &check))
234                 return NULL;
235
236         if (ATTR_TRUE(check.value))
237                 return &driver_true;
238         if (ATTR_FALSE(check.value))
239                 return &driver_false;
240         if (ATTR_UNSET(check.value))
241                 return NULL;
242         return userdiff_find_by_name(check.value);
243 }