Show More
@@ -1,213 +1,263 b'' | |||
|
1 | 1 | /* |
|
2 | 2 | bdiff.c - efficient binary diff extension for Mercurial |
|
3 | 3 | |
|
4 | 4 | Copyright 2005, 2006 Matt Mackall <mpm@selenic.com> |
|
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 | Based roughly on Python difflib |
|
10 | 10 | */ |
|
11 | 11 | |
|
12 | 12 | #define PY_SSIZE_T_CLEAN |
|
13 | 13 | #include <Python.h> |
|
14 | 14 | #include <limits.h> |
|
15 | 15 | #include <stdlib.h> |
|
16 | 16 | #include <string.h> |
|
17 | 17 | |
|
18 | 18 | #include "bdiff.h" |
|
19 | 19 | #include "bitmanipulation.h" |
|
20 | 20 | #include "util.h" |
|
21 | 21 | |
|
22 | 22 | static PyObject *blocks(PyObject *self, PyObject *args) |
|
23 | 23 | { |
|
24 | 24 | PyObject *sa, *sb, *rl = NULL, *m; |
|
25 | 25 | struct bdiff_line *a, *b; |
|
26 | 26 | struct bdiff_hunk l, *h; |
|
27 | 27 | int an, bn, count, pos = 0; |
|
28 | 28 | |
|
29 | 29 | l.next = NULL; |
|
30 | 30 | |
|
31 | 31 | if (!PyArg_ParseTuple(args, "SS:bdiff", &sa, &sb)) |
|
32 | 32 | return NULL; |
|
33 | 33 | |
|
34 | 34 | an = bdiff_splitlines(PyBytes_AsString(sa), PyBytes_Size(sa), &a); |
|
35 | 35 | bn = bdiff_splitlines(PyBytes_AsString(sb), PyBytes_Size(sb), &b); |
|
36 | 36 | |
|
37 | 37 | if (!a || !b) |
|
38 | 38 | goto nomem; |
|
39 | 39 | |
|
40 | 40 | count = bdiff_diff(a, an, b, bn, &l); |
|
41 | 41 | if (count < 0) |
|
42 | 42 | goto nomem; |
|
43 | 43 | |
|
44 | 44 | rl = PyList_New(count); |
|
45 | 45 | if (!rl) |
|
46 | 46 | goto nomem; |
|
47 | 47 | |
|
48 | 48 | for (h = l.next; h; h = h->next) { |
|
49 | 49 | m = Py_BuildValue("iiii", h->a1, h->a2, h->b1, h->b2); |
|
50 | 50 | PyList_SetItem(rl, pos, m); |
|
51 | 51 | pos++; |
|
52 | 52 | } |
|
53 | 53 | |
|
54 | 54 | nomem: |
|
55 | 55 | free(a); |
|
56 | 56 | free(b); |
|
57 | 57 | bdiff_freehunks(l.next); |
|
58 | 58 | return rl ? rl : PyErr_NoMemory(); |
|
59 | 59 | } |
|
60 | 60 | |
|
61 | 61 | static PyObject *bdiff(PyObject *self, PyObject *args) |
|
62 | 62 | { |
|
63 | 63 | char *sa, *sb, *rb, *ia, *ib; |
|
64 | 64 | PyObject *result = NULL; |
|
65 | 65 | struct bdiff_line *al, *bl; |
|
66 | 66 | struct bdiff_hunk l, *h; |
|
67 | 67 | int an, bn, count; |
|
68 | 68 | Py_ssize_t len = 0, la, lb, li = 0, lcommon = 0, lmax; |
|
69 | 69 | PyThreadState *_save; |
|
70 | 70 | |
|
71 | 71 | l.next = NULL; |
|
72 | 72 | |
|
73 | 73 | if (!PyArg_ParseTuple(args, "s#s#:bdiff", &sa, &la, &sb, &lb)) |
|
74 | 74 | return NULL; |
|
75 | 75 | |
|
76 | 76 | if (la > UINT_MAX || lb > UINT_MAX) { |
|
77 | 77 | PyErr_SetString(PyExc_ValueError, "bdiff inputs too large"); |
|
78 | 78 | return NULL; |
|
79 | 79 | } |
|
80 | 80 | |
|
81 | 81 | _save = PyEval_SaveThread(); |
|
82 | 82 | |
|
83 | 83 | lmax = la > lb ? lb : la; |
|
84 | 84 | for (ia = sa, ib = sb; li < lmax && *ia == *ib; ++li, ++ia, ++ib) |
|
85 | 85 | if (*ia == '\n') |
|
86 | 86 | lcommon = li + 1; |
|
87 | 87 | /* we can almost add: if (li == lmax) lcommon = li; */ |
|
88 | 88 | |
|
89 | 89 | an = bdiff_splitlines(sa + lcommon, la - lcommon, &al); |
|
90 | 90 | bn = bdiff_splitlines(sb + lcommon, lb - lcommon, &bl); |
|
91 | 91 | if (!al || !bl) |
|
92 | 92 | goto nomem; |
|
93 | 93 | |
|
94 | 94 | count = bdiff_diff(al, an, bl, bn, &l); |
|
95 | 95 | if (count < 0) |
|
96 | 96 | goto nomem; |
|
97 | 97 | |
|
98 | 98 | /* calculate length of output */ |
|
99 | 99 | la = lb = 0; |
|
100 | 100 | for (h = l.next; h; h = h->next) { |
|
101 | 101 | if (h->a1 != la || h->b1 != lb) |
|
102 | 102 | len += 12 + bl[h->b1].l - bl[lb].l; |
|
103 | 103 | la = h->a2; |
|
104 | 104 | lb = h->b2; |
|
105 | 105 | } |
|
106 | 106 | PyEval_RestoreThread(_save); |
|
107 | 107 | _save = NULL; |
|
108 | 108 | |
|
109 | 109 | result = PyBytes_FromStringAndSize(NULL, len); |
|
110 | 110 | |
|
111 | 111 | if (!result) |
|
112 | 112 | goto nomem; |
|
113 | 113 | |
|
114 | 114 | /* build binary patch */ |
|
115 | 115 | rb = PyBytes_AsString(result); |
|
116 | 116 | la = lb = 0; |
|
117 | 117 | |
|
118 | 118 | for (h = l.next; h; h = h->next) { |
|
119 | 119 | if (h->a1 != la || h->b1 != lb) { |
|
120 | 120 | len = bl[h->b1].l - bl[lb].l; |
|
121 | 121 | putbe32((uint32_t)(al[la].l + lcommon - al->l), rb); |
|
122 | 122 | putbe32((uint32_t)(al[h->a1].l + lcommon - al->l), |
|
123 | 123 | rb + 4); |
|
124 | 124 | putbe32((uint32_t)len, rb + 8); |
|
125 | 125 | memcpy(rb + 12, bl[lb].l, len); |
|
126 | 126 | rb += 12 + len; |
|
127 | 127 | } |
|
128 | 128 | la = h->a2; |
|
129 | 129 | lb = h->b2; |
|
130 | 130 | } |
|
131 | 131 | |
|
132 | 132 | nomem: |
|
133 | 133 | if (_save) |
|
134 | 134 | PyEval_RestoreThread(_save); |
|
135 | 135 | free(al); |
|
136 | 136 | free(bl); |
|
137 | 137 | bdiff_freehunks(l.next); |
|
138 | 138 | return result ? result : PyErr_NoMemory(); |
|
139 | 139 | } |
|
140 | 140 | |
|
141 | 141 | /* |
|
142 | 142 | * If allws != 0, remove all whitespace (' ', \t and \r). Otherwise, |
|
143 | 143 | * reduce whitespace sequences to a single space and trim remaining whitespace |
|
144 | 144 | * from end of lines. |
|
145 | 145 | */ |
|
146 | 146 | static PyObject *fixws(PyObject *self, PyObject *args) |
|
147 | 147 | { |
|
148 | 148 | PyObject *s, *result = NULL; |
|
149 | 149 | char allws, c; |
|
150 | 150 | const char *r; |
|
151 | 151 | Py_ssize_t i, rlen, wlen = 0; |
|
152 | 152 | char *w; |
|
153 | 153 | |
|
154 | 154 | if (!PyArg_ParseTuple(args, "Sb:fixws", &s, &allws)) |
|
155 | 155 | return NULL; |
|
156 | 156 | r = PyBytes_AsString(s); |
|
157 | 157 | rlen = PyBytes_Size(s); |
|
158 | 158 | |
|
159 | 159 | w = (char *)PyMem_Malloc(rlen ? rlen : 1); |
|
160 | 160 | if (!w) |
|
161 | 161 | goto nomem; |
|
162 | 162 | |
|
163 | 163 | for (i = 0; i != rlen; i++) { |
|
164 | 164 | c = r[i]; |
|
165 | 165 | if (c == ' ' || c == '\t' || c == '\r') { |
|
166 | 166 | if (!allws && (wlen == 0 || w[wlen - 1] != ' ')) |
|
167 | 167 | w[wlen++] = ' '; |
|
168 | 168 | } else if (c == '\n' && !allws && wlen > 0 && |
|
169 | 169 | w[wlen - 1] == ' ') { |
|
170 | 170 | w[wlen - 1] = '\n'; |
|
171 | 171 | } else { |
|
172 | 172 | w[wlen++] = c; |
|
173 | 173 | } |
|
174 | 174 | } |
|
175 | 175 | |
|
176 | 176 | result = PyBytes_FromStringAndSize(w, wlen); |
|
177 | 177 | |
|
178 | 178 | nomem: |
|
179 | 179 | PyMem_Free(w); |
|
180 | 180 | return result ? result : PyErr_NoMemory(); |
|
181 | 181 | } |
|
182 | 182 | |
|
183 | static bool sliceintolist(PyObject *list, Py_ssize_t destidx, | |
|
184 | const char *source, Py_ssize_t len) | |
|
185 | { | |
|
186 | PyObject *sliced = PyBytes_FromStringAndSize(source, len); | |
|
187 | if (sliced == NULL) | |
|
188 | return false; | |
|
189 | PyList_SET_ITEM(list, destidx, sliced); | |
|
190 | return true; | |
|
191 | } | |
|
192 | ||
|
193 | static PyObject *splitnewlines(PyObject *self, PyObject *args) | |
|
194 | { | |
|
195 | const char *text; | |
|
196 | Py_ssize_t nelts = 0, size, i, start = 0; | |
|
197 | PyObject *result = NULL; | |
|
198 | ||
|
199 | if (!PyArg_ParseTuple(args, "s#", &text, &size)) { | |
|
200 | goto abort; | |
|
201 | } | |
|
202 | if (!size) { | |
|
203 | return PyList_New(0); | |
|
204 | } | |
|
205 | /* This loops to size-1 because if the last byte is a newline, | |
|
206 | * we don't want to perform a split there. */ | |
|
207 | for (i = 0; i < size - 1; ++i) { | |
|
208 | if (text[i] == '\n') { | |
|
209 | ++nelts; | |
|
210 | } | |
|
211 | } | |
|
212 | if ((result = PyList_New(nelts + 1)) == NULL) | |
|
213 | goto abort; | |
|
214 | nelts = 0; | |
|
215 | for (i = 0; i < size - 1; ++i) { | |
|
216 | if (text[i] == '\n') { | |
|
217 | if (!sliceintolist(result, nelts++, text + start, | |
|
218 | i - start + 1)) | |
|
219 | goto abort; | |
|
220 | start = i + 1; | |
|
221 | } | |
|
222 | } | |
|
223 | if (!sliceintolist(result, nelts++, text + start, size - start)) | |
|
224 | goto abort; | |
|
225 | return result; | |
|
226 | abort: | |
|
227 | Py_XDECREF(result); | |
|
228 | return NULL; | |
|
229 | } | |
|
230 | ||
|
183 | 231 | static char mdiff_doc[] = "Efficient binary diff."; |
|
184 | 232 | |
|
185 | 233 | static PyMethodDef methods[] = { |
|
186 | 234 | {"bdiff", bdiff, METH_VARARGS, "calculate a binary diff\n"}, |
|
187 | 235 | {"blocks", blocks, METH_VARARGS, "find a list of matching lines\n"}, |
|
188 | 236 | {"fixws", fixws, METH_VARARGS, "normalize diff whitespaces\n"}, |
|
237 | {"splitnewlines", splitnewlines, METH_VARARGS, | |
|
238 | "like str.splitlines, but only split on newlines\n"}, | |
|
189 | 239 | {NULL, NULL}, |
|
190 | 240 | }; |
|
191 | 241 | |
|
192 |
static const int version = |
|
|
242 | static const int version = 2; | |
|
193 | 243 | |
|
194 | 244 | #ifdef IS_PY3K |
|
195 | 245 | static struct PyModuleDef bdiff_module = { |
|
196 | 246 | PyModuleDef_HEAD_INIT, "bdiff", mdiff_doc, -1, methods, |
|
197 | 247 | }; |
|
198 | 248 | |
|
199 | 249 | PyMODINIT_FUNC PyInit_bdiff(void) |
|
200 | 250 | { |
|
201 | 251 | PyObject *m; |
|
202 | 252 | m = PyModule_Create(&bdiff_module); |
|
203 | 253 | PyModule_AddIntConstant(m, "version", version); |
|
204 | 254 | return m; |
|
205 | 255 | } |
|
206 | 256 | #else |
|
207 | 257 | PyMODINIT_FUNC initbdiff(void) |
|
208 | 258 | { |
|
209 | 259 | PyObject *m; |
|
210 | 260 | m = Py_InitModule3("bdiff", methods, mdiff_doc); |
|
211 | 261 | PyModule_AddIntConstant(m, "version", version); |
|
212 | 262 | } |
|
213 | 263 | #endif |
@@ -1,521 +1,512 b'' | |||
|
1 | 1 | # mdiff.py - diff and patch routines for mercurial |
|
2 | 2 | # |
|
3 | 3 | # Copyright 2005, 2006 Matt Mackall <mpm@selenic.com> |
|
4 | 4 | # |
|
5 | 5 | # This software may be used and distributed according to the terms of the |
|
6 | 6 | # GNU General Public License version 2 or any later version. |
|
7 | 7 | |
|
8 | 8 | from __future__ import absolute_import |
|
9 | 9 | |
|
10 | 10 | import re |
|
11 | 11 | import struct |
|
12 | 12 | import zlib |
|
13 | 13 | |
|
14 | 14 | from .i18n import _ |
|
15 | 15 | from . import ( |
|
16 | 16 | error, |
|
17 | 17 | policy, |
|
18 | 18 | pycompat, |
|
19 | 19 | util, |
|
20 | 20 | ) |
|
21 | 21 | |
|
22 | 22 | _missing_newline_marker = "\\ No newline at end of file\n" |
|
23 | 23 | |
|
24 | 24 | bdiff = policy.importmod(r'bdiff') |
|
25 | 25 | mpatch = policy.importmod(r'mpatch') |
|
26 | 26 | |
|
27 | 27 | blocks = bdiff.blocks |
|
28 | 28 | fixws = bdiff.fixws |
|
29 | 29 | patches = mpatch.patches |
|
30 | 30 | patchedsize = mpatch.patchedsize |
|
31 | 31 | textdiff = bdiff.bdiff |
|
32 | ||
|
33 | def splitnewlines(text): | |
|
34 | '''like str.splitlines, but only split on newlines.''' | |
|
35 | lines = [l + '\n' for l in text.split('\n')] | |
|
36 | if lines: | |
|
37 | if lines[-1] == '\n': | |
|
38 | lines.pop() | |
|
39 | else: | |
|
40 | lines[-1] = lines[-1][:-1] | |
|
41 | return lines | |
|
32 | splitnewlines = bdiff.splitnewlines | |
|
42 | 33 | |
|
43 | 34 | class diffopts(object): |
|
44 | 35 | '''context is the number of context lines |
|
45 | 36 | text treats all files as text |
|
46 | 37 | showfunc enables diff -p output |
|
47 | 38 | git enables the git extended patch format |
|
48 | 39 | nodates removes dates from diff headers |
|
49 | 40 | nobinary ignores binary files |
|
50 | 41 | noprefix disables the 'a/' and 'b/' prefixes (ignored in plain mode) |
|
51 | 42 | ignorews ignores all whitespace changes in the diff |
|
52 | 43 | ignorewsamount ignores changes in the amount of whitespace |
|
53 | 44 | ignoreblanklines ignores changes whose lines are all blank |
|
54 | 45 | upgrade generates git diffs to avoid data loss |
|
55 | 46 | ''' |
|
56 | 47 | |
|
57 | 48 | defaults = { |
|
58 | 49 | 'context': 3, |
|
59 | 50 | 'text': False, |
|
60 | 51 | 'showfunc': False, |
|
61 | 52 | 'git': False, |
|
62 | 53 | 'nodates': False, |
|
63 | 54 | 'nobinary': False, |
|
64 | 55 | 'noprefix': False, |
|
65 | 56 | 'index': 0, |
|
66 | 57 | 'ignorews': False, |
|
67 | 58 | 'ignorewsamount': False, |
|
68 | 59 | 'ignorewseol': False, |
|
69 | 60 | 'ignoreblanklines': False, |
|
70 | 61 | 'upgrade': False, |
|
71 | 62 | 'showsimilarity': False, |
|
72 | 63 | 'worddiff': False, |
|
73 | 64 | } |
|
74 | 65 | |
|
75 | 66 | def __init__(self, **opts): |
|
76 | 67 | opts = pycompat.byteskwargs(opts) |
|
77 | 68 | for k in self.defaults.keys(): |
|
78 | 69 | v = opts.get(k) |
|
79 | 70 | if v is None: |
|
80 | 71 | v = self.defaults[k] |
|
81 | 72 | setattr(self, k, v) |
|
82 | 73 | |
|
83 | 74 | try: |
|
84 | 75 | self.context = int(self.context) |
|
85 | 76 | except ValueError: |
|
86 | 77 | raise error.Abort(_('diff context lines count must be ' |
|
87 | 78 | 'an integer, not %r') % self.context) |
|
88 | 79 | |
|
89 | 80 | def copy(self, **kwargs): |
|
90 | 81 | opts = dict((k, getattr(self, k)) for k in self.defaults) |
|
91 | 82 | opts = pycompat.strkwargs(opts) |
|
92 | 83 | opts.update(kwargs) |
|
93 | 84 | return diffopts(**opts) |
|
94 | 85 | |
|
95 | 86 | defaultopts = diffopts() |
|
96 | 87 | |
|
97 | 88 | def wsclean(opts, text, blank=True): |
|
98 | 89 | if opts.ignorews: |
|
99 | 90 | text = bdiff.fixws(text, 1) |
|
100 | 91 | elif opts.ignorewsamount: |
|
101 | 92 | text = bdiff.fixws(text, 0) |
|
102 | 93 | if blank and opts.ignoreblanklines: |
|
103 | 94 | text = re.sub('\n+', '\n', text).strip('\n') |
|
104 | 95 | if opts.ignorewseol: |
|
105 | 96 | text = re.sub(br'[ \t\r\f]+\n', r'\n', text) |
|
106 | 97 | return text |
|
107 | 98 | |
|
108 | 99 | def splitblock(base1, lines1, base2, lines2, opts): |
|
109 | 100 | # The input lines matches except for interwoven blank lines. We |
|
110 | 101 | # transform it into a sequence of matching blocks and blank blocks. |
|
111 | 102 | lines1 = [(wsclean(opts, l) and 1 or 0) for l in lines1] |
|
112 | 103 | lines2 = [(wsclean(opts, l) and 1 or 0) for l in lines2] |
|
113 | 104 | s1, e1 = 0, len(lines1) |
|
114 | 105 | s2, e2 = 0, len(lines2) |
|
115 | 106 | while s1 < e1 or s2 < e2: |
|
116 | 107 | i1, i2, btype = s1, s2, '=' |
|
117 | 108 | if (i1 >= e1 or lines1[i1] == 0 |
|
118 | 109 | or i2 >= e2 or lines2[i2] == 0): |
|
119 | 110 | # Consume the block of blank lines |
|
120 | 111 | btype = '~' |
|
121 | 112 | while i1 < e1 and lines1[i1] == 0: |
|
122 | 113 | i1 += 1 |
|
123 | 114 | while i2 < e2 and lines2[i2] == 0: |
|
124 | 115 | i2 += 1 |
|
125 | 116 | else: |
|
126 | 117 | # Consume the matching lines |
|
127 | 118 | while i1 < e1 and lines1[i1] == 1 and lines2[i2] == 1: |
|
128 | 119 | i1 += 1 |
|
129 | 120 | i2 += 1 |
|
130 | 121 | yield [base1 + s1, base1 + i1, base2 + s2, base2 + i2], btype |
|
131 | 122 | s1 = i1 |
|
132 | 123 | s2 = i2 |
|
133 | 124 | |
|
134 | 125 | def hunkinrange(hunk, linerange): |
|
135 | 126 | """Return True if `hunk` defined as (start, length) is in `linerange` |
|
136 | 127 | defined as (lowerbound, upperbound). |
|
137 | 128 | |
|
138 | 129 | >>> hunkinrange((5, 10), (2, 7)) |
|
139 | 130 | True |
|
140 | 131 | >>> hunkinrange((5, 10), (6, 12)) |
|
141 | 132 | True |
|
142 | 133 | >>> hunkinrange((5, 10), (13, 17)) |
|
143 | 134 | True |
|
144 | 135 | >>> hunkinrange((5, 10), (3, 17)) |
|
145 | 136 | True |
|
146 | 137 | >>> hunkinrange((5, 10), (1, 3)) |
|
147 | 138 | False |
|
148 | 139 | >>> hunkinrange((5, 10), (18, 20)) |
|
149 | 140 | False |
|
150 | 141 | >>> hunkinrange((5, 10), (1, 5)) |
|
151 | 142 | False |
|
152 | 143 | >>> hunkinrange((5, 10), (15, 27)) |
|
153 | 144 | False |
|
154 | 145 | """ |
|
155 | 146 | start, length = hunk |
|
156 | 147 | lowerbound, upperbound = linerange |
|
157 | 148 | return lowerbound < start + length and start < upperbound |
|
158 | 149 | |
|
159 | 150 | def blocksinrange(blocks, rangeb): |
|
160 | 151 | """filter `blocks` like (a1, a2, b1, b2) from items outside line range |
|
161 | 152 | `rangeb` from ``(b1, b2)`` point of view. |
|
162 | 153 | |
|
163 | 154 | Return `filteredblocks, rangea` where: |
|
164 | 155 | |
|
165 | 156 | * `filteredblocks` is list of ``block = (a1, a2, b1, b2), stype`` items of |
|
166 | 157 | `blocks` that are inside `rangeb` from ``(b1, b2)`` point of view; a |
|
167 | 158 | block ``(b1, b2)`` being inside `rangeb` if |
|
168 | 159 | ``rangeb[0] < b2 and b1 < rangeb[1]``; |
|
169 | 160 | * `rangea` is the line range w.r.t. to ``(a1, a2)`` parts of `blocks`. |
|
170 | 161 | """ |
|
171 | 162 | lbb, ubb = rangeb |
|
172 | 163 | lba, uba = None, None |
|
173 | 164 | filteredblocks = [] |
|
174 | 165 | for block in blocks: |
|
175 | 166 | (a1, a2, b1, b2), stype = block |
|
176 | 167 | if lbb >= b1 and ubb <= b2 and stype == '=': |
|
177 | 168 | # rangeb is within a single "=" hunk, restrict back linerange1 |
|
178 | 169 | # by offsetting rangeb |
|
179 | 170 | lba = lbb - b1 + a1 |
|
180 | 171 | uba = ubb - b1 + a1 |
|
181 | 172 | else: |
|
182 | 173 | if b1 <= lbb < b2: |
|
183 | 174 | if stype == '=': |
|
184 | 175 | lba = a2 - (b2 - lbb) |
|
185 | 176 | else: |
|
186 | 177 | lba = a1 |
|
187 | 178 | if b1 < ubb <= b2: |
|
188 | 179 | if stype == '=': |
|
189 | 180 | uba = a1 + (ubb - b1) |
|
190 | 181 | else: |
|
191 | 182 | uba = a2 |
|
192 | 183 | if hunkinrange((b1, (b2 - b1)), rangeb): |
|
193 | 184 | filteredblocks.append(block) |
|
194 | 185 | if lba is None or uba is None or uba < lba: |
|
195 | 186 | raise error.Abort(_('line range exceeds file size')) |
|
196 | 187 | return filteredblocks, (lba, uba) |
|
197 | 188 | |
|
198 | 189 | def allblocks(text1, text2, opts=None, lines1=None, lines2=None): |
|
199 | 190 | """Return (block, type) tuples, where block is an mdiff.blocks |
|
200 | 191 | line entry. type is '=' for blocks matching exactly one another |
|
201 | 192 | (bdiff blocks), '!' for non-matching blocks and '~' for blocks |
|
202 | 193 | matching only after having filtered blank lines. |
|
203 | 194 | line1 and line2 are text1 and text2 split with splitnewlines() if |
|
204 | 195 | they are already available. |
|
205 | 196 | """ |
|
206 | 197 | if opts is None: |
|
207 | 198 | opts = defaultopts |
|
208 | 199 | if opts.ignorews or opts.ignorewsamount or opts.ignorewseol: |
|
209 | 200 | text1 = wsclean(opts, text1, False) |
|
210 | 201 | text2 = wsclean(opts, text2, False) |
|
211 | 202 | diff = bdiff.blocks(text1, text2) |
|
212 | 203 | for i, s1 in enumerate(diff): |
|
213 | 204 | # The first match is special. |
|
214 | 205 | # we've either found a match starting at line 0 or a match later |
|
215 | 206 | # in the file. If it starts later, old and new below will both be |
|
216 | 207 | # empty and we'll continue to the next match. |
|
217 | 208 | if i > 0: |
|
218 | 209 | s = diff[i - 1] |
|
219 | 210 | else: |
|
220 | 211 | s = [0, 0, 0, 0] |
|
221 | 212 | s = [s[1], s1[0], s[3], s1[2]] |
|
222 | 213 | |
|
223 | 214 | # bdiff sometimes gives huge matches past eof, this check eats them, |
|
224 | 215 | # and deals with the special first match case described above |
|
225 | 216 | if s[0] != s[1] or s[2] != s[3]: |
|
226 | 217 | type = '!' |
|
227 | 218 | if opts.ignoreblanklines: |
|
228 | 219 | if lines1 is None: |
|
229 | 220 | lines1 = splitnewlines(text1) |
|
230 | 221 | if lines2 is None: |
|
231 | 222 | lines2 = splitnewlines(text2) |
|
232 | 223 | old = wsclean(opts, "".join(lines1[s[0]:s[1]])) |
|
233 | 224 | new = wsclean(opts, "".join(lines2[s[2]:s[3]])) |
|
234 | 225 | if old == new: |
|
235 | 226 | type = '~' |
|
236 | 227 | yield s, type |
|
237 | 228 | yield s1, '=' |
|
238 | 229 | |
|
239 | 230 | def unidiff(a, ad, b, bd, fn1, fn2, binary, opts=defaultopts): |
|
240 | 231 | """Return a unified diff as a (headers, hunks) tuple. |
|
241 | 232 | |
|
242 | 233 | If the diff is not null, `headers` is a list with unified diff header |
|
243 | 234 | lines "--- <original>" and "+++ <new>" and `hunks` is a generator yielding |
|
244 | 235 | (hunkrange, hunklines) coming from _unidiff(). |
|
245 | 236 | Otherwise, `headers` and `hunks` are empty. |
|
246 | 237 | |
|
247 | 238 | Set binary=True if either a or b should be taken as a binary file. |
|
248 | 239 | """ |
|
249 | 240 | def datetag(date, fn=None): |
|
250 | 241 | if not opts.git and not opts.nodates: |
|
251 | 242 | return '\t%s' % date |
|
252 | 243 | if fn and ' ' in fn: |
|
253 | 244 | return '\t' |
|
254 | 245 | return '' |
|
255 | 246 | |
|
256 | 247 | sentinel = [], () |
|
257 | 248 | if not a and not b: |
|
258 | 249 | return sentinel |
|
259 | 250 | |
|
260 | 251 | if opts.noprefix: |
|
261 | 252 | aprefix = bprefix = '' |
|
262 | 253 | else: |
|
263 | 254 | aprefix = 'a/' |
|
264 | 255 | bprefix = 'b/' |
|
265 | 256 | |
|
266 | 257 | epoch = util.datestr((0, 0)) |
|
267 | 258 | |
|
268 | 259 | fn1 = util.pconvert(fn1) |
|
269 | 260 | fn2 = util.pconvert(fn2) |
|
270 | 261 | |
|
271 | 262 | if binary: |
|
272 | 263 | if a and b and len(a) == len(b) and a == b: |
|
273 | 264 | return sentinel |
|
274 | 265 | headerlines = [] |
|
275 | 266 | hunks = (None, ['Binary file %s has changed\n' % fn1]), |
|
276 | 267 | elif not a: |
|
277 | 268 | without_newline = not b.endswith('\n') |
|
278 | 269 | b = splitnewlines(b) |
|
279 | 270 | if a is None: |
|
280 | 271 | l1 = '--- /dev/null%s' % datetag(epoch) |
|
281 | 272 | else: |
|
282 | 273 | l1 = "--- %s%s%s" % (aprefix, fn1, datetag(ad, fn1)) |
|
283 | 274 | l2 = "+++ %s%s" % (bprefix + fn2, datetag(bd, fn2)) |
|
284 | 275 | headerlines = [l1, l2] |
|
285 | 276 | size = len(b) |
|
286 | 277 | hunkrange = (0, 0, 1, size) |
|
287 | 278 | hunklines = ["@@ -0,0 +1,%d @@\n" % size] + ["+" + e for e in b] |
|
288 | 279 | if without_newline: |
|
289 | 280 | hunklines[-1] += '\n' |
|
290 | 281 | hunklines.append(_missing_newline_marker) |
|
291 | 282 | hunks = (hunkrange, hunklines), |
|
292 | 283 | elif not b: |
|
293 | 284 | without_newline = not a.endswith('\n') |
|
294 | 285 | a = splitnewlines(a) |
|
295 | 286 | l1 = "--- %s%s%s" % (aprefix, fn1, datetag(ad, fn1)) |
|
296 | 287 | if b is None: |
|
297 | 288 | l2 = '+++ /dev/null%s' % datetag(epoch) |
|
298 | 289 | else: |
|
299 | 290 | l2 = "+++ %s%s%s" % (bprefix, fn2, datetag(bd, fn2)) |
|
300 | 291 | headerlines = [l1, l2] |
|
301 | 292 | size = len(a) |
|
302 | 293 | hunkrange = (1, size, 0, 0) |
|
303 | 294 | hunklines = ["@@ -1,%d +0,0 @@\n" % size] + ["-" + e for e in a] |
|
304 | 295 | if without_newline: |
|
305 | 296 | hunklines[-1] += '\n' |
|
306 | 297 | hunklines.append(_missing_newline_marker) |
|
307 | 298 | hunks = (hunkrange, hunklines), |
|
308 | 299 | else: |
|
309 | 300 | hunks = _unidiff(a, b, opts=opts) |
|
310 | 301 | if not next(hunks): |
|
311 | 302 | return sentinel |
|
312 | 303 | |
|
313 | 304 | headerlines = [ |
|
314 | 305 | "--- %s%s%s" % (aprefix, fn1, datetag(ad, fn1)), |
|
315 | 306 | "+++ %s%s%s" % (bprefix, fn2, datetag(bd, fn2)), |
|
316 | 307 | ] |
|
317 | 308 | |
|
318 | 309 | return headerlines, hunks |
|
319 | 310 | |
|
320 | 311 | def _unidiff(t1, t2, opts=defaultopts): |
|
321 | 312 | """Yield hunks of a headerless unified diff from t1 and t2 texts. |
|
322 | 313 | |
|
323 | 314 | Each hunk consists of a (hunkrange, hunklines) tuple where `hunkrange` is a |
|
324 | 315 | tuple (s1, l1, s2, l2) representing the range information of the hunk to |
|
325 | 316 | form the '@@ -s1,l1 +s2,l2 @@' header and `hunklines` is a list of lines |
|
326 | 317 | of the hunk combining said header followed by line additions and |
|
327 | 318 | deletions. |
|
328 | 319 | |
|
329 | 320 | The hunks are prefixed with a bool. |
|
330 | 321 | """ |
|
331 | 322 | l1 = splitnewlines(t1) |
|
332 | 323 | l2 = splitnewlines(t2) |
|
333 | 324 | def contextend(l, len): |
|
334 | 325 | ret = l + opts.context |
|
335 | 326 | if ret > len: |
|
336 | 327 | ret = len |
|
337 | 328 | return ret |
|
338 | 329 | |
|
339 | 330 | def contextstart(l): |
|
340 | 331 | ret = l - opts.context |
|
341 | 332 | if ret < 0: |
|
342 | 333 | return 0 |
|
343 | 334 | return ret |
|
344 | 335 | |
|
345 | 336 | lastfunc = [0, ''] |
|
346 | 337 | def yieldhunk(hunk): |
|
347 | 338 | (astart, a2, bstart, b2, delta) = hunk |
|
348 | 339 | aend = contextend(a2, len(l1)) |
|
349 | 340 | alen = aend - astart |
|
350 | 341 | blen = b2 - bstart + aend - a2 |
|
351 | 342 | |
|
352 | 343 | func = "" |
|
353 | 344 | if opts.showfunc: |
|
354 | 345 | lastpos, func = lastfunc |
|
355 | 346 | # walk backwards from the start of the context up to the start of |
|
356 | 347 | # the previous hunk context until we find a line starting with an |
|
357 | 348 | # alphanumeric char. |
|
358 | 349 | for i in xrange(astart - 1, lastpos - 1, -1): |
|
359 | 350 | if l1[i][0:1].isalnum(): |
|
360 | 351 | func = ' ' + l1[i].rstrip()[:40] |
|
361 | 352 | lastfunc[1] = func |
|
362 | 353 | break |
|
363 | 354 | # by recording this hunk's starting point as the next place to |
|
364 | 355 | # start looking for function lines, we avoid reading any line in |
|
365 | 356 | # the file more than once. |
|
366 | 357 | lastfunc[0] = astart |
|
367 | 358 | |
|
368 | 359 | # zero-length hunk ranges report their start line as one less |
|
369 | 360 | if alen: |
|
370 | 361 | astart += 1 |
|
371 | 362 | if blen: |
|
372 | 363 | bstart += 1 |
|
373 | 364 | |
|
374 | 365 | hunkrange = astart, alen, bstart, blen |
|
375 | 366 | hunklines = ( |
|
376 | 367 | ["@@ -%d,%d +%d,%d @@%s\n" % (hunkrange + (func,))] |
|
377 | 368 | + delta |
|
378 | 369 | + [' ' + l1[x] for x in xrange(a2, aend)] |
|
379 | 370 | ) |
|
380 | 371 | # If either file ends without a newline and the last line of |
|
381 | 372 | # that file is part of a hunk, a marker is printed. If the |
|
382 | 373 | # last line of both files is identical and neither ends in |
|
383 | 374 | # a newline, print only one marker. That's the only case in |
|
384 | 375 | # which the hunk can end in a shared line without a newline. |
|
385 | 376 | skip = False |
|
386 | 377 | if not t1.endswith('\n') and astart + alen == len(l1) + 1: |
|
387 | 378 | for i in xrange(len(hunklines) - 1, -1, -1): |
|
388 | 379 | if hunklines[i].startswith(('-', ' ')): |
|
389 | 380 | if hunklines[i].startswith(' '): |
|
390 | 381 | skip = True |
|
391 | 382 | hunklines[i] += '\n' |
|
392 | 383 | hunklines.insert(i + 1, _missing_newline_marker) |
|
393 | 384 | break |
|
394 | 385 | if not skip and not t2.endswith('\n') and bstart + blen == len(l2) + 1: |
|
395 | 386 | for i in xrange(len(hunklines) - 1, -1, -1): |
|
396 | 387 | if hunklines[i].startswith('+'): |
|
397 | 388 | hunklines[i] += '\n' |
|
398 | 389 | hunklines.insert(i + 1, _missing_newline_marker) |
|
399 | 390 | break |
|
400 | 391 | yield hunkrange, hunklines |
|
401 | 392 | |
|
402 | 393 | # bdiff.blocks gives us the matching sequences in the files. The loop |
|
403 | 394 | # below finds the spaces between those matching sequences and translates |
|
404 | 395 | # them into diff output. |
|
405 | 396 | # |
|
406 | 397 | hunk = None |
|
407 | 398 | ignoredlines = 0 |
|
408 | 399 | has_hunks = False |
|
409 | 400 | for s, stype in allblocks(t1, t2, opts, l1, l2): |
|
410 | 401 | a1, a2, b1, b2 = s |
|
411 | 402 | if stype != '!': |
|
412 | 403 | if stype == '~': |
|
413 | 404 | # The diff context lines are based on t1 content. When |
|
414 | 405 | # blank lines are ignored, the new lines offsets must |
|
415 | 406 | # be adjusted as if equivalent blocks ('~') had the |
|
416 | 407 | # same sizes on both sides. |
|
417 | 408 | ignoredlines += (b2 - b1) - (a2 - a1) |
|
418 | 409 | continue |
|
419 | 410 | delta = [] |
|
420 | 411 | old = l1[a1:a2] |
|
421 | 412 | new = l2[b1:b2] |
|
422 | 413 | |
|
423 | 414 | b1 -= ignoredlines |
|
424 | 415 | b2 -= ignoredlines |
|
425 | 416 | astart = contextstart(a1) |
|
426 | 417 | bstart = contextstart(b1) |
|
427 | 418 | prev = None |
|
428 | 419 | if hunk: |
|
429 | 420 | # join with the previous hunk if it falls inside the context |
|
430 | 421 | if astart < hunk[1] + opts.context + 1: |
|
431 | 422 | prev = hunk |
|
432 | 423 | astart = hunk[1] |
|
433 | 424 | bstart = hunk[3] |
|
434 | 425 | else: |
|
435 | 426 | if not has_hunks: |
|
436 | 427 | has_hunks = True |
|
437 | 428 | yield True |
|
438 | 429 | for x in yieldhunk(hunk): |
|
439 | 430 | yield x |
|
440 | 431 | if prev: |
|
441 | 432 | # we've joined the previous hunk, record the new ending points. |
|
442 | 433 | hunk[1] = a2 |
|
443 | 434 | hunk[3] = b2 |
|
444 | 435 | delta = hunk[4] |
|
445 | 436 | else: |
|
446 | 437 | # create a new hunk |
|
447 | 438 | hunk = [astart, a2, bstart, b2, delta] |
|
448 | 439 | |
|
449 | 440 | delta[len(delta):] = [' ' + x for x in l1[astart:a1]] |
|
450 | 441 | delta[len(delta):] = ['-' + x for x in old] |
|
451 | 442 | delta[len(delta):] = ['+' + x for x in new] |
|
452 | 443 | |
|
453 | 444 | if hunk: |
|
454 | 445 | if not has_hunks: |
|
455 | 446 | has_hunks = True |
|
456 | 447 | yield True |
|
457 | 448 | for x in yieldhunk(hunk): |
|
458 | 449 | yield x |
|
459 | 450 | elif not has_hunks: |
|
460 | 451 | yield False |
|
461 | 452 | |
|
462 | 453 | def b85diff(to, tn): |
|
463 | 454 | '''print base85-encoded binary diff''' |
|
464 | 455 | def fmtline(line): |
|
465 | 456 | l = len(line) |
|
466 | 457 | if l <= 26: |
|
467 | 458 | l = pycompat.bytechr(ord('A') + l - 1) |
|
468 | 459 | else: |
|
469 | 460 | l = pycompat.bytechr(l - 26 + ord('a') - 1) |
|
470 | 461 | return '%c%s\n' % (l, util.b85encode(line, True)) |
|
471 | 462 | |
|
472 | 463 | def chunk(text, csize=52): |
|
473 | 464 | l = len(text) |
|
474 | 465 | i = 0 |
|
475 | 466 | while i < l: |
|
476 | 467 | yield text[i:i + csize] |
|
477 | 468 | i += csize |
|
478 | 469 | |
|
479 | 470 | if to is None: |
|
480 | 471 | to = '' |
|
481 | 472 | if tn is None: |
|
482 | 473 | tn = '' |
|
483 | 474 | |
|
484 | 475 | if to == tn: |
|
485 | 476 | return '' |
|
486 | 477 | |
|
487 | 478 | # TODO: deltas |
|
488 | 479 | ret = [] |
|
489 | 480 | ret.append('GIT binary patch\n') |
|
490 | 481 | ret.append('literal %d\n' % len(tn)) |
|
491 | 482 | for l in chunk(zlib.compress(tn)): |
|
492 | 483 | ret.append(fmtline(l)) |
|
493 | 484 | ret.append('\n') |
|
494 | 485 | |
|
495 | 486 | return ''.join(ret) |
|
496 | 487 | |
|
497 | 488 | def patchtext(bin): |
|
498 | 489 | pos = 0 |
|
499 | 490 | t = [] |
|
500 | 491 | while pos < len(bin): |
|
501 | 492 | p1, p2, l = struct.unpack(">lll", bin[pos:pos + 12]) |
|
502 | 493 | pos += 12 |
|
503 | 494 | t.append(bin[pos:pos + l]) |
|
504 | 495 | pos += l |
|
505 | 496 | return "".join(t) |
|
506 | 497 | |
|
507 | 498 | def patch(a, bin): |
|
508 | 499 | if len(a) == 0: |
|
509 | 500 | # skip over trivial delta header |
|
510 | 501 | return util.buffer(bin, 12) |
|
511 | 502 | return mpatch.patches(a, [bin]) |
|
512 | 503 | |
|
513 | 504 | # similar to difflib.SequenceMatcher.get_matching_blocks |
|
514 | 505 | def get_matching_blocks(a, b): |
|
515 | 506 | return [(d[0], d[2], d[1] - d[0]) for d in bdiff.blocks(a, b)] |
|
516 | 507 | |
|
517 | 508 | def trivialdiffheader(length): |
|
518 | 509 | return struct.pack(">lll", 0, 0, length) if length else '' |
|
519 | 510 | |
|
520 | 511 | def replacediffheader(oldlen, newlen): |
|
521 | 512 | return struct.pack(">lll", 0, oldlen, newlen) |
@@ -1,116 +1,116 b'' | |||
|
1 | 1 | # policy.py - module policy logic for Mercurial. |
|
2 | 2 | # |
|
3 | 3 | # Copyright 2015 Gregory Szorc <gregory.szorc@gmail.com> |
|
4 | 4 | # |
|
5 | 5 | # This software may be used and distributed according to the terms of the |
|
6 | 6 | # GNU General Public License version 2 or any later version. |
|
7 | 7 | |
|
8 | 8 | from __future__ import absolute_import |
|
9 | 9 | |
|
10 | 10 | import os |
|
11 | 11 | import sys |
|
12 | 12 | |
|
13 | 13 | # Rules for how modules can be loaded. Values are: |
|
14 | 14 | # |
|
15 | 15 | # c - require C extensions |
|
16 | 16 | # allow - allow pure Python implementation when C loading fails |
|
17 | 17 | # cffi - required cffi versions (implemented within pure module) |
|
18 | 18 | # cffi-allow - allow pure Python implementation if cffi version is missing |
|
19 | 19 | # py - only load pure Python modules |
|
20 | 20 | # |
|
21 | 21 | # By default, fall back to the pure modules so the in-place build can |
|
22 | 22 | # run without recompiling the C extensions. This will be overridden by |
|
23 | 23 | # __modulepolicy__ generated by setup.py. |
|
24 | 24 | policy = b'allow' |
|
25 | 25 | _packageprefs = { |
|
26 | 26 | # policy: (versioned package, pure package) |
|
27 | 27 | b'c': (r'cext', None), |
|
28 | 28 | b'allow': (r'cext', r'pure'), |
|
29 | 29 | b'cffi': (r'cffi', None), |
|
30 | 30 | b'cffi-allow': (r'cffi', r'pure'), |
|
31 | 31 | b'py': (None, r'pure'), |
|
32 | 32 | } |
|
33 | 33 | |
|
34 | 34 | try: |
|
35 | 35 | from . import __modulepolicy__ |
|
36 | 36 | policy = __modulepolicy__.modulepolicy |
|
37 | 37 | except ImportError: |
|
38 | 38 | pass |
|
39 | 39 | |
|
40 | 40 | # PyPy doesn't load C extensions. |
|
41 | 41 | # |
|
42 | 42 | # The canonical way to do this is to test platform.python_implementation(). |
|
43 | 43 | # But we don't import platform and don't bloat for it here. |
|
44 | 44 | if r'__pypy__' in sys.builtin_module_names: |
|
45 | 45 | policy = b'cffi' |
|
46 | 46 | |
|
47 | 47 | # Our C extensions aren't yet compatible with Python 3. So use pure Python |
|
48 | 48 | # on Python 3 for now. |
|
49 | 49 | if sys.version_info[0] >= 3: |
|
50 | 50 | policy = b'py' |
|
51 | 51 | |
|
52 | 52 | # Environment variable can always force settings. |
|
53 | 53 | if sys.version_info[0] >= 3: |
|
54 | 54 | if r'HGMODULEPOLICY' in os.environ: |
|
55 | 55 | policy = os.environ[r'HGMODULEPOLICY'].encode(r'utf-8') |
|
56 | 56 | else: |
|
57 | 57 | policy = os.environ.get(r'HGMODULEPOLICY', policy) |
|
58 | 58 | |
|
59 | 59 | def _importfrom(pkgname, modname): |
|
60 | 60 | # from .<pkgname> import <modname> (where . is looked through this module) |
|
61 | 61 | fakelocals = {} |
|
62 | 62 | pkg = __import__(pkgname, globals(), fakelocals, [modname], level=1) |
|
63 | 63 | try: |
|
64 | 64 | fakelocals[modname] = mod = getattr(pkg, modname) |
|
65 | 65 | except AttributeError: |
|
66 | 66 | raise ImportError(r'cannot import name %s' % modname) |
|
67 | 67 | # force import; fakelocals[modname] may be replaced with the real module |
|
68 | 68 | getattr(mod, r'__doc__', None) |
|
69 | 69 | return fakelocals[modname] |
|
70 | 70 | |
|
71 | 71 | # keep in sync with "version" in C modules |
|
72 | 72 | _cextversions = { |
|
73 | 73 | (r'cext', r'base85'): 1, |
|
74 |
(r'cext', r'bdiff'): |
|
|
74 | (r'cext', r'bdiff'): 2, | |
|
75 | 75 | (r'cext', r'diffhelpers'): 1, |
|
76 | 76 | (r'cext', r'mpatch'): 1, |
|
77 | 77 | (r'cext', r'osutil'): 3, |
|
78 | 78 | (r'cext', r'parsers'): 4, |
|
79 | 79 | } |
|
80 | 80 | |
|
81 | 81 | # map import request to other package or module |
|
82 | 82 | _modredirects = { |
|
83 | 83 | (r'cext', r'charencode'): (r'cext', r'parsers'), |
|
84 | 84 | (r'cffi', r'base85'): (r'pure', r'base85'), |
|
85 | 85 | (r'cffi', r'charencode'): (r'pure', r'charencode'), |
|
86 | 86 | (r'cffi', r'diffhelpers'): (r'pure', r'diffhelpers'), |
|
87 | 87 | (r'cffi', r'parsers'): (r'pure', r'parsers'), |
|
88 | 88 | } |
|
89 | 89 | |
|
90 | 90 | def _checkmod(pkgname, modname, mod): |
|
91 | 91 | expected = _cextversions.get((pkgname, modname)) |
|
92 | 92 | actual = getattr(mod, r'version', None) |
|
93 | 93 | if actual != expected: |
|
94 | 94 | raise ImportError(r'cannot import module %s.%s ' |
|
95 | 95 | r'(expected version: %d, actual: %r)' |
|
96 | 96 | % (pkgname, modname, expected, actual)) |
|
97 | 97 | |
|
98 | 98 | def importmod(modname): |
|
99 | 99 | """Import module according to policy and check API version""" |
|
100 | 100 | try: |
|
101 | 101 | verpkg, purepkg = _packageprefs[policy] |
|
102 | 102 | except KeyError: |
|
103 | 103 | raise ImportError(r'invalid HGMODULEPOLICY %r' % policy) |
|
104 | 104 | assert verpkg or purepkg |
|
105 | 105 | if verpkg: |
|
106 | 106 | pn, mn = _modredirects.get((verpkg, modname), (verpkg, modname)) |
|
107 | 107 | try: |
|
108 | 108 | mod = _importfrom(pn, mn) |
|
109 | 109 | if pn == verpkg: |
|
110 | 110 | _checkmod(pn, mn, mod) |
|
111 | 111 | return mod |
|
112 | 112 | except ImportError: |
|
113 | 113 | if not purepkg: |
|
114 | 114 | raise |
|
115 | 115 | pn, mn = _modredirects.get((purepkg, modname), (purepkg, modname)) |
|
116 | 116 | return _importfrom(pn, mn) |
@@ -1,92 +1,102 b'' | |||
|
1 | 1 | # bdiff.py - Python implementation of bdiff.c |
|
2 | 2 | # |
|
3 | 3 | # Copyright 2009 Matt Mackall <mpm@selenic.com> and others |
|
4 | 4 | # |
|
5 | 5 | # This software may be used and distributed according to the terms of the |
|
6 | 6 | # GNU General Public License version 2 or any later version. |
|
7 | 7 | |
|
8 | 8 | from __future__ import absolute_import |
|
9 | 9 | |
|
10 | 10 | import difflib |
|
11 | 11 | import re |
|
12 | 12 | import struct |
|
13 | 13 | |
|
14 | 14 | def splitnewlines(text): |
|
15 | 15 | '''like str.splitlines, but only split on newlines.''' |
|
16 | 16 | lines = [l + '\n' for l in text.split('\n')] |
|
17 | 17 | if lines: |
|
18 | 18 | if lines[-1] == '\n': |
|
19 | 19 | lines.pop() |
|
20 | 20 | else: |
|
21 | 21 | lines[-1] = lines[-1][:-1] |
|
22 | 22 | return lines |
|
23 | 23 | |
|
24 | 24 | def _normalizeblocks(a, b, blocks): |
|
25 | 25 | prev = None |
|
26 | 26 | r = [] |
|
27 | 27 | for curr in blocks: |
|
28 | 28 | if prev is None: |
|
29 | 29 | prev = curr |
|
30 | 30 | continue |
|
31 | 31 | shift = 0 |
|
32 | 32 | |
|
33 | 33 | a1, b1, l1 = prev |
|
34 | 34 | a1end = a1 + l1 |
|
35 | 35 | b1end = b1 + l1 |
|
36 | 36 | |
|
37 | 37 | a2, b2, l2 = curr |
|
38 | 38 | a2end = a2 + l2 |
|
39 | 39 | b2end = b2 + l2 |
|
40 | 40 | if a1end == a2: |
|
41 | 41 | while (a1end + shift < a2end and |
|
42 | 42 | a[a1end + shift] == b[b1end + shift]): |
|
43 | 43 | shift += 1 |
|
44 | 44 | elif b1end == b2: |
|
45 | 45 | while (b1end + shift < b2end and |
|
46 | 46 | a[a1end + shift] == b[b1end + shift]): |
|
47 | 47 | shift += 1 |
|
48 | 48 | r.append((a1, b1, l1 + shift)) |
|
49 | 49 | prev = a2 + shift, b2 + shift, l2 - shift |
|
50 | 50 | r.append(prev) |
|
51 | 51 | return r |
|
52 | 52 | |
|
53 | 53 | def bdiff(a, b): |
|
54 | 54 | a = bytes(a).splitlines(True) |
|
55 | 55 | b = bytes(b).splitlines(True) |
|
56 | 56 | |
|
57 | 57 | if not a: |
|
58 | 58 | s = "".join(b) |
|
59 | 59 | return s and (struct.pack(">lll", 0, 0, len(s)) + s) |
|
60 | 60 | |
|
61 | 61 | bin = [] |
|
62 | 62 | p = [0] |
|
63 | 63 | for i in a: |
|
64 | 64 | p.append(p[-1] + len(i)) |
|
65 | 65 | |
|
66 | 66 | d = difflib.SequenceMatcher(None, a, b).get_matching_blocks() |
|
67 | 67 | d = _normalizeblocks(a, b, d) |
|
68 | 68 | la = 0 |
|
69 | 69 | lb = 0 |
|
70 | 70 | for am, bm, size in d: |
|
71 | 71 | s = "".join(b[lb:bm]) |
|
72 | 72 | if am > la or s: |
|
73 | 73 | bin.append(struct.pack(">lll", p[la], p[am], len(s)) + s) |
|
74 | 74 | la = am + size |
|
75 | 75 | lb = bm + size |
|
76 | 76 | |
|
77 | 77 | return "".join(bin) |
|
78 | 78 | |
|
79 | 79 | def blocks(a, b): |
|
80 | 80 | an = splitnewlines(a) |
|
81 | 81 | bn = splitnewlines(b) |
|
82 | 82 | d = difflib.SequenceMatcher(None, an, bn).get_matching_blocks() |
|
83 | 83 | d = _normalizeblocks(an, bn, d) |
|
84 | 84 | return [(i, i + n, j, j + n) for (i, j, n) in d] |
|
85 | 85 | |
|
86 | 86 | def fixws(text, allws): |
|
87 | 87 | if allws: |
|
88 | 88 | text = re.sub('[ \t\r]+', '', text) |
|
89 | 89 | else: |
|
90 | 90 | text = re.sub('[ \t\r]+', ' ', text) |
|
91 | 91 | text = text.replace(' \n', '\n') |
|
92 | 92 | return text |
|
93 | ||
|
94 | def splitnewlines(text): | |
|
95 | '''like str.splitlines, but only split on newlines.''' | |
|
96 | lines = [l + '\n' for l in text.split('\n')] | |
|
97 | if lines: | |
|
98 | if lines[-1] == '\n': | |
|
99 | lines.pop() | |
|
100 | else: | |
|
101 | lines[-1] = lines[-1][:-1] | |
|
102 | return lines |
General Comments 0
You need to be logged in to leave comments.
Login now