1 fe621944 2020-11-10 stsp /* Split source by line breaks, and calculate a simplistic checksum. */
3 fe621944 2020-11-10 stsp * Copyright (c) 2020 Neels Hofmeyr <neels@hofmeyr.de>
5 fe621944 2020-11-10 stsp * Permission to use, copy, modify, and distribute this software for any
6 fe621944 2020-11-10 stsp * purpose with or without fee is hereby granted, provided that the above
7 fe621944 2020-11-10 stsp * copyright notice and this permission notice appear in all copies.
9 fe621944 2020-11-10 stsp * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
10 fe621944 2020-11-10 stsp * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
11 fe621944 2020-11-10 stsp * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
12 fe621944 2020-11-10 stsp * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
13 fe621944 2020-11-10 stsp * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
14 fe621944 2020-11-10 stsp * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
15 fe621944 2020-11-10 stsp * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
18 fe621944 2020-11-10 stsp #include <errno.h>
19 fe621944 2020-11-10 stsp #include <stdbool.h>
20 f3c44083 2020-11-14 naddy #include <stdint.h>
21 fe621944 2020-11-10 stsp #include <stdio.h>
22 fe621944 2020-11-10 stsp #include <stdlib.h>
23 fe621944 2020-11-10 stsp #include <unistd.h>
24 fe621944 2020-11-10 stsp #include <ctype.h>
26 fe621944 2020-11-10 stsp #include <arraylist.h>
27 fe621944 2020-11-10 stsp #include <diff_main.h>
29 fe621944 2020-11-10 stsp #include "diff_internal.h"
30 fe621944 2020-11-10 stsp #include "diff_debug.h"
32 f8c2e76a 2022-08-03 stsp unsigned int
33 dea26038 2020-11-18 stsp diff_atom_hash_update(unsigned int hash, unsigned char atom_byte)
35 dea26038 2020-11-18 stsp return hash * 23 + atom_byte;
39 fe621944 2020-11-10 stsp diff_data_atomize_text_lines_fd(struct diff_data *d)
41 fe621944 2020-11-10 stsp off_t pos = 0;
42 fe621944 2020-11-10 stsp const off_t end = pos + d->len;
43 fe621944 2020-11-10 stsp unsigned int array_size_estimate = d->len / 50;
44 fe621944 2020-11-10 stsp unsigned int pow2 = 1;
45 fe621944 2020-11-10 stsp bool ignore_whitespace = (d->diff_flags & DIFF_FLAG_IGNORE_WHITESPACE);
46 b67f3bcb 2020-11-21 stsp bool embedded_nul = false;
48 fe621944 2020-11-10 stsp while (array_size_estimate >>= 1)
51 fe621944 2020-11-10 stsp ARRAYLIST_INIT(d->atoms, 1 << pow2);
53 fe621944 2020-11-10 stsp if (fseek(d->root->f, 0L, SEEK_SET) == -1)
54 fe621944 2020-11-10 stsp return errno;
56 fe621944 2020-11-10 stsp while (pos < end) {
57 fe621944 2020-11-10 stsp off_t line_end = pos;
58 fe621944 2020-11-10 stsp unsigned int hash = 0;
59 fe621944 2020-11-10 stsp unsigned char buf[512];
60 fe621944 2020-11-10 stsp size_t r, i;
61 fe621944 2020-11-10 stsp struct diff_atom *atom;
62 fe621944 2020-11-10 stsp int eol = 0;
64 fe621944 2020-11-10 stsp while (eol == 0 && line_end < end) {
65 fe621944 2020-11-10 stsp r = fread(buf, sizeof(char), sizeof(buf), d->root->f);
66 fe621944 2020-11-10 stsp if (r == 0 && ferror(d->root->f))
69 fe621944 2020-11-10 stsp while (eol == 0 && i < r) {
70 fe621944 2020-11-10 stsp if (buf[i] != '\r' && buf[i] != '\n') {
71 fe621944 2020-11-10 stsp if (!ignore_whitespace
72 cee4532d 2022-11-17 op || !isspace((unsigned char)buf[i]))
73 dea26038 2020-11-18 stsp hash = diff_atom_hash_update(
74 dea26038 2020-11-18 stsp hash, buf[i]);
75 b67f3bcb 2020-11-21 stsp if (buf[i] == '\0')
76 b67f3bcb 2020-11-21 stsp embedded_nul = true;
79 fe621944 2020-11-10 stsp eol = buf[i];
84 fe621944 2020-11-10 stsp /* When not at the end of data, the line ending char ('\r' or
85 fe621944 2020-11-10 stsp * '\n') must follow */
86 fe621944 2020-11-10 stsp if (line_end < end)
88 fe621944 2020-11-10 stsp /* If that was an '\r', also pull in any following '\n' */
89 fe621944 2020-11-10 stsp if (line_end < end && eol == '\r') {
90 fe621944 2020-11-10 stsp if (fseeko(d->root->f, line_end, SEEK_SET) == -1)
91 fe621944 2020-11-10 stsp return errno;
92 fe621944 2020-11-10 stsp r = fread(buf, sizeof(char), sizeof(buf), d->root->f);
93 fe621944 2020-11-10 stsp if (r == 0 && ferror(d->root->f))
95 8d504b53 2022-07-26 op if (r > 0 && buf[0] == '\n')
99 fe621944 2020-11-10 stsp /* Record the found line as diff atom */
100 fe621944 2020-11-10 stsp ARRAYLIST_ADD(atom, d->atoms);
102 fe621944 2020-11-10 stsp return ENOMEM;
104 fe621944 2020-11-10 stsp *atom = (struct diff_atom){
106 fe621944 2020-11-10 stsp .pos = pos,
107 fe621944 2020-11-10 stsp .at = NULL, /* atom data is not memory-mapped */
108 fe621944 2020-11-10 stsp .len = line_end - pos,
109 fe621944 2020-11-10 stsp .hash = hash,
112 fe621944 2020-11-10 stsp /* Starting point for next line: */
113 fe621944 2020-11-10 stsp pos = line_end;
114 fe621944 2020-11-10 stsp if (fseeko(d->root->f, pos, SEEK_SET) == -1)
115 fe621944 2020-11-10 stsp return errno;
118 b67f3bcb 2020-11-21 stsp /* File are considered binary if they contain embedded '\0' bytes. */
119 b67f3bcb 2020-11-21 stsp if (embedded_nul)
120 b67f3bcb 2020-11-21 stsp d->atomizer_flags |= DIFF_ATOMIZER_FOUND_BINARY_DATA;
122 fe621944 2020-11-10 stsp return DIFF_RC_OK;
126 fe621944 2020-11-10 stsp diff_data_atomize_text_lines_mmap(struct diff_data *d)
128 fe621944 2020-11-10 stsp const uint8_t *pos = d->data;
129 fe621944 2020-11-10 stsp const uint8_t *end = pos + d->len;
130 fe621944 2020-11-10 stsp bool ignore_whitespace = (d->diff_flags & DIFF_FLAG_IGNORE_WHITESPACE);
131 b67f3bcb 2020-11-21 stsp bool embedded_nul = false;
132 fe621944 2020-11-10 stsp unsigned int array_size_estimate = d->len / 50;
133 fe621944 2020-11-10 stsp unsigned int pow2 = 1;
134 fe621944 2020-11-10 stsp while (array_size_estimate >>= 1)
137 fe621944 2020-11-10 stsp ARRAYLIST_INIT(d->atoms, 1 << pow2);
139 fe621944 2020-11-10 stsp while (pos < end) {
140 fe621944 2020-11-10 stsp const uint8_t *line_end = pos;
141 fe621944 2020-11-10 stsp unsigned int hash = 0;
143 fe621944 2020-11-10 stsp while (line_end < end && *line_end != '\r' && *line_end != '\n') {
144 fe621944 2020-11-10 stsp if (!ignore_whitespace
145 cee4532d 2022-11-17 op || !isspace((unsigned char)*line_end))
146 1c5fbba3 2022-08-03 op hash = diff_atom_hash_update(hash, *line_end);
147 b67f3bcb 2020-11-21 stsp if (*line_end == '\0')
148 b67f3bcb 2020-11-21 stsp embedded_nul = true;
149 fe621944 2020-11-10 stsp line_end++;
152 fe621944 2020-11-10 stsp /* When not at the end of data, the line ending char ('\r' or
153 fe621944 2020-11-10 stsp * '\n') must follow */
154 8d504b53 2022-07-26 op if (line_end < end && *line_end == '\r')
155 fe621944 2020-11-10 stsp line_end++;
156 8d504b53 2022-07-26 op if (line_end < end && *line_end == '\n')
157 fe621944 2020-11-10 stsp line_end++;
159 fe621944 2020-11-10 stsp /* Record the found line as diff atom */
160 fe621944 2020-11-10 stsp struct diff_atom *atom;
161 fe621944 2020-11-10 stsp ARRAYLIST_ADD(atom, d->atoms);
163 fe621944 2020-11-10 stsp return ENOMEM;
165 fe621944 2020-11-10 stsp *atom = (struct diff_atom){
167 fe621944 2020-11-10 stsp .pos = (off_t)(pos - d->data),
169 fe621944 2020-11-10 stsp .len = line_end - pos,
170 fe621944 2020-11-10 stsp .hash = hash,
173 fe621944 2020-11-10 stsp /* Starting point for next line: */
174 fe621944 2020-11-10 stsp pos = line_end;
177 b67f3bcb 2020-11-21 stsp /* File are considered binary if they contain embedded '\0' bytes. */
178 b67f3bcb 2020-11-21 stsp if (embedded_nul)
179 b67f3bcb 2020-11-21 stsp d->atomizer_flags |= DIFF_ATOMIZER_FOUND_BINARY_DATA;
181 fe621944 2020-11-10 stsp return DIFF_RC_OK;
185 fe621944 2020-11-10 stsp diff_data_atomize_text_lines(struct diff_data *d)
187 fe621944 2020-11-10 stsp if (d->data == NULL)
188 fe621944 2020-11-10 stsp return diff_data_atomize_text_lines_fd(d);
190 fe621944 2020-11-10 stsp return diff_data_atomize_text_lines_mmap(d);
194 fe621944 2020-11-10 stsp diff_atomize_text_by_line(void *func_data, struct diff_data *d)
196 fe621944 2020-11-10 stsp return diff_data_atomize_text_lines(d);