Show More
@@ -1,168 +1,169 b'' | |||
|
1 | 1 | /* |
|
2 | 2 | parsers.c - efficient content parsing |
|
3 | 3 | |
|
4 | 4 | Copyright 2008 Matt Mackall <mpm@selenic.com> and others |
|
5 | 5 | |
|
6 | 6 | This software may be used and distributed according to the terms of |
|
7 | 7 | the GNU General Public License, incorporated herein by reference. |
|
8 | 8 | */ |
|
9 | 9 | |
|
10 | 10 | #include <Python.h> |
|
11 | 11 | #include <ctype.h> |
|
12 | 12 | #include <string.h> |
|
13 | 13 | |
|
14 | 14 | static int hexdigit(char c) |
|
15 | 15 | { |
|
16 | 16 | if (c >= '0' && c <= '9') |
|
17 | 17 | return c - '0'; |
|
18 | 18 | |
|
19 | 19 | if (c >= 'A' && c <= 'F') |
|
20 | 20 | return c - 'A' + 10; |
|
21 | 21 | |
|
22 | 22 | if (c >= 'a' && c <= 'f') |
|
23 | 23 | return c - 'a' + 10; |
|
24 | 24 | |
|
25 | 25 | return -1; |
|
26 | 26 | } |
|
27 | 27 | |
|
28 | 28 | /* |
|
29 | 29 | * Turn a hex-encoded string into binary. |
|
30 | 30 | */ |
|
31 | 31 | static PyObject *unhexlify(const char *str, int len) |
|
32 | 32 | { |
|
33 | 33 | PyObject *ret = NULL; |
|
34 |
char *c |
|
|
34 | const char *c; | |
|
35 | char *d; | |
|
35 | 36 | |
|
36 | 37 | if (len % 2) { |
|
37 | 38 | PyErr_SetString(PyExc_ValueError, |
|
38 | 39 | "input is not even in length"); |
|
39 | 40 | goto bail; |
|
40 | 41 | } |
|
41 | 42 | |
|
42 | 43 | ret = PyString_FromStringAndSize(NULL, len / 2); |
|
43 | 44 | if (!ret) |
|
44 | 45 | goto bail; |
|
45 | 46 | |
|
46 | 47 | d = PyString_AsString(ret); |
|
47 | 48 | if (!d) |
|
48 | 49 | goto bail; |
|
49 | 50 | |
|
50 | 51 | for (c = str; c < str + len;) { |
|
51 | 52 | int hi = hexdigit(*c++); |
|
52 | 53 | int lo = hexdigit(*c++); |
|
53 | 54 | |
|
54 | 55 | if (hi == -1 || lo == -1) { |
|
55 | 56 | PyErr_SetString(PyExc_ValueError, |
|
56 | 57 | "input contains non-hex character"); |
|
57 | 58 | goto bail; |
|
58 | 59 | } |
|
59 | 60 | |
|
60 | 61 | *d++ = (hi << 4) | lo; |
|
61 | 62 | } |
|
62 | 63 | |
|
63 | 64 | goto done; |
|
64 | 65 | |
|
65 | 66 | bail: |
|
66 | 67 | Py_XDECREF(ret); |
|
67 | 68 | ret = NULL; |
|
68 | 69 | done: |
|
69 | 70 | return ret; |
|
70 | 71 | } |
|
71 | 72 | |
|
72 | 73 | /* |
|
73 | 74 | * This code assumes that a manifest is stitched together with newline |
|
74 | 75 | * ('\n') characters. |
|
75 | 76 | */ |
|
76 | 77 | static PyObject *parse_manifest(PyObject *self, PyObject *args) |
|
77 | 78 | { |
|
78 | 79 | PyObject *mfdict, *fdict; |
|
79 | 80 | char *str, *cur, *start, *zero; |
|
80 | 81 | int len; |
|
81 | 82 | |
|
82 | 83 | if (!PyArg_ParseTuple(args, "O!O!s#:parse_manifest", |
|
83 | 84 | &PyDict_Type, &mfdict, |
|
84 | 85 | &PyDict_Type, &fdict, |
|
85 | 86 | &str, &len)) |
|
86 | 87 | goto quit; |
|
87 | 88 | |
|
88 | 89 | for (start = cur = str, zero = NULL; cur < str + len; cur++) { |
|
89 | 90 | PyObject *file = NULL, *node = NULL; |
|
90 | 91 | PyObject *flags = NULL; |
|
91 | 92 | int nlen; |
|
92 | 93 | |
|
93 | 94 | if (!*cur) { |
|
94 | 95 | zero = cur; |
|
95 | 96 | continue; |
|
96 | 97 | } |
|
97 | 98 | else if (*cur != '\n') |
|
98 | 99 | continue; |
|
99 | 100 | |
|
100 | 101 | if (!zero) { |
|
101 | 102 | PyErr_SetString(PyExc_ValueError, |
|
102 | 103 | "manifest entry has no separator"); |
|
103 | 104 | goto quit; |
|
104 | 105 | } |
|
105 | 106 | |
|
106 | 107 | file = PyString_FromStringAndSize(start, zero - start); |
|
107 | 108 | if (!file) |
|
108 | 109 | goto bail; |
|
109 | 110 | |
|
110 | 111 | nlen = cur - zero - 1; |
|
111 | 112 | |
|
112 | 113 | node = unhexlify(zero + 1, nlen > 40 ? 40 : nlen); |
|
113 | 114 | if (!node) |
|
114 | 115 | goto bail; |
|
115 | 116 | |
|
116 | 117 | if (nlen > 40) { |
|
117 | 118 | PyObject *flags; |
|
118 | 119 | |
|
119 | 120 | flags = PyString_FromStringAndSize(zero + 41, |
|
120 | 121 | nlen - 40); |
|
121 | 122 | if (!flags) |
|
122 | 123 | goto bail; |
|
123 | 124 | |
|
124 | 125 | if (PyDict_SetItem(fdict, file, flags) == -1) |
|
125 | 126 | goto bail; |
|
126 | 127 | } |
|
127 | 128 | |
|
128 | 129 | if (PyDict_SetItem(mfdict, file, node) == -1) |
|
129 | 130 | goto bail; |
|
130 | 131 | |
|
131 | 132 | start = cur + 1; |
|
132 | 133 | zero = NULL; |
|
133 | 134 | |
|
134 | 135 | Py_XDECREF(flags); |
|
135 | 136 | Py_XDECREF(node); |
|
136 | 137 | Py_XDECREF(file); |
|
137 | 138 | continue; |
|
138 | 139 | bail: |
|
139 | 140 | Py_XDECREF(flags); |
|
140 | 141 | Py_XDECREF(node); |
|
141 | 142 | Py_XDECREF(file); |
|
142 | 143 | goto quit; |
|
143 | 144 | } |
|
144 | 145 | |
|
145 | 146 | if (len > 0 && *(cur - 1) != '\n') { |
|
146 | 147 | PyErr_SetString(PyExc_ValueError, |
|
147 | 148 | "manifest contains trailing garbage"); |
|
148 | 149 | goto quit; |
|
149 | 150 | } |
|
150 | 151 | |
|
151 | 152 | Py_INCREF(Py_None); |
|
152 | 153 | return Py_None; |
|
153 | 154 | |
|
154 | 155 | quit: |
|
155 | 156 | return NULL; |
|
156 | 157 | } |
|
157 | 158 | |
|
158 | 159 | static char parsers_doc[] = "Efficient content parsing."; |
|
159 | 160 | |
|
160 | 161 | static PyMethodDef methods[] = { |
|
161 | 162 | {"parse_manifest", parse_manifest, METH_VARARGS, "parse a manifest\n"}, |
|
162 | 163 | {NULL, NULL} |
|
163 | 164 | }; |
|
164 | 165 | |
|
165 | 166 | PyMODINIT_FUNC initparsers(void) |
|
166 | 167 | { |
|
167 | 168 | Py_InitModule3("parsers", methods, parsers_doc); |
|
168 | 169 | } |
General Comments 0
You need to be logged in to leave comments.
Login now