1 fe621944 2020-11-10 stsp /* Split source by line breaks, and calculate a simplistic checksum. */
  2 fe621944 2020-11-10 stsp /*
  3 fe621944 2020-11-10 stsp  * Copyright (c) 2020 Neels Hofmeyr <neels@hofmeyr.de>
  4 fe621944 2020-11-10 stsp  *
  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.
  8 fe621944 2020-11-10 stsp  *
  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.
 16 fe621944 2020-11-10 stsp  */
 17 fe621944 2020-11-10 stsp 
 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>
 25 fe621944 2020-11-10 stsp 
 26 fe621944 2020-11-10 stsp #include <arraylist.h>
 27 fe621944 2020-11-10 stsp #include <diff_main.h>
 28 fe621944 2020-11-10 stsp 
 29 fe621944 2020-11-10 stsp #include "diff_internal.h"
 30 fe621944 2020-11-10 stsp #include "diff_debug.h"
 31 fe621944 2020-11-10 stsp 
 32 dea26038 2020-11-18 stsp unsigned int
 33 dea26038 2020-11-18 stsp diff_atom_hash_update(unsigned int hash, unsigned char atom_byte)
 34 dea26038 2020-11-18 stsp {
 35 dea26038 2020-11-18 stsp 	return hash * 23 + atom_byte;
 36 dea26038 2020-11-18 stsp }
 37 dea26038 2020-11-18 stsp 
 38 fe621944 2020-11-10 stsp static int
 39 fe621944 2020-11-10 stsp diff_data_atomize_text_lines_fd(struct diff_data *d)
 40 fe621944 2020-11-10 stsp {
 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 fe621944 2020-11-10 stsp 
 47 fe621944 2020-11-10 stsp 	while (array_size_estimate >>= 1)
 48 fe621944 2020-11-10 stsp 		pow2++;
 49 fe621944 2020-11-10 stsp 
 50 fe621944 2020-11-10 stsp 	ARRAYLIST_INIT(d->atoms, 1 << pow2);
 51 fe621944 2020-11-10 stsp 
 52 fe621944 2020-11-10 stsp 	if (fseek(d->root->f, 0L, SEEK_SET) == -1)
 53 fe621944 2020-11-10 stsp 		return errno;
 54 fe621944 2020-11-10 stsp 
 55 fe621944 2020-11-10 stsp 	while (pos < end) {
 56 fe621944 2020-11-10 stsp 		off_t line_end = pos;
 57 fe621944 2020-11-10 stsp 		unsigned int hash = 0;
 58 fe621944 2020-11-10 stsp 		unsigned char buf[512];
 59 fe621944 2020-11-10 stsp 		size_t r, i;
 60 fe621944 2020-11-10 stsp 		struct diff_atom *atom;
 61 fe621944 2020-11-10 stsp 		int eol = 0;
 62 fe621944 2020-11-10 stsp 
 63 fe621944 2020-11-10 stsp 		while (eol == 0 && line_end < end) {
 64 fe621944 2020-11-10 stsp 			r = fread(buf, sizeof(char), sizeof(buf), d->root->f);
 65 fe621944 2020-11-10 stsp 			if (r == 0 && ferror(d->root->f))
 66 fe621944 2020-11-10 stsp 				return errno;
 67 fe621944 2020-11-10 stsp 			i = 0;
 68 fe621944 2020-11-10 stsp 			while (eol == 0 && i < r) {
 69 fe621944 2020-11-10 stsp 				if (buf[i] != '\r' && buf[i] != '\n') {
 70 fe621944 2020-11-10 stsp 					if (!ignore_whitespace
 71 fe621944 2020-11-10 stsp 					    || !isspace(buf[i]))
 72 dea26038 2020-11-18 stsp 						hash = diff_atom_hash_update(
 73 dea26038 2020-11-18 stsp 						    hash, buf[i]);
 74 fe621944 2020-11-10 stsp 					line_end++;
 75 fe621944 2020-11-10 stsp 				} else
 76 fe621944 2020-11-10 stsp 					eol = buf[i];
 77 fe621944 2020-11-10 stsp 				i++;
 78 fe621944 2020-11-10 stsp 			}
 79 fe621944 2020-11-10 stsp 		}
 80 fe621944 2020-11-10 stsp 
 81 fe621944 2020-11-10 stsp 		/* When not at the end of data, the line ending char ('\r' or
 82 fe621944 2020-11-10 stsp 		 * '\n') must follow */
 83 fe621944 2020-11-10 stsp 		if (line_end < end)
 84 fe621944 2020-11-10 stsp 			line_end++;
 85 fe621944 2020-11-10 stsp 		/* If that was an '\r', also pull in any following '\n' */
 86 fe621944 2020-11-10 stsp 		if (line_end < end && eol == '\r') {
 87 fe621944 2020-11-10 stsp 			if (fseeko(d->root->f, line_end, SEEK_SET) == -1)
 88 fe621944 2020-11-10 stsp 				return errno;
 89 fe621944 2020-11-10 stsp 			r = fread(buf, sizeof(char), sizeof(buf), d->root->f);
 90 fe621944 2020-11-10 stsp 			if (r == 0 && ferror(d->root->f))
 91 fe621944 2020-11-10 stsp 				return errno;
 92 fe621944 2020-11-10 stsp 			if (r == 1 && buf[0] == '\n' )
 93 fe621944 2020-11-10 stsp 				line_end++;
 94 fe621944 2020-11-10 stsp 		}
 95 fe621944 2020-11-10 stsp 
 96 fe621944 2020-11-10 stsp 		/* Record the found line as diff atom */
 97 fe621944 2020-11-10 stsp 		ARRAYLIST_ADD(atom, d->atoms);
 98 fe621944 2020-11-10 stsp 		if (!atom)
 99 fe621944 2020-11-10 stsp 			return ENOMEM;
100 fe621944 2020-11-10 stsp 
101 fe621944 2020-11-10 stsp 		*atom = (struct diff_atom){
102 fe621944 2020-11-10 stsp 			.root = d,
103 fe621944 2020-11-10 stsp 			.pos = pos,
104 fe621944 2020-11-10 stsp 			.at = NULL,	/* atom data is not memory-mapped */
105 fe621944 2020-11-10 stsp 			.len = line_end - pos,
106 fe621944 2020-11-10 stsp 			.hash = hash,
107 fe621944 2020-11-10 stsp 		};
108 fe621944 2020-11-10 stsp 
109 fe621944 2020-11-10 stsp 		/* Starting point for next line: */
110 fe621944 2020-11-10 stsp 		pos = line_end;
111 fe621944 2020-11-10 stsp 		if (fseeko(d->root->f, pos, SEEK_SET) == -1)
112 fe621944 2020-11-10 stsp 			return errno;
113 fe621944 2020-11-10 stsp 	}
114 fe621944 2020-11-10 stsp 
115 fe621944 2020-11-10 stsp 	return DIFF_RC_OK;
116 fe621944 2020-11-10 stsp }
117 fe621944 2020-11-10 stsp 
118 fe621944 2020-11-10 stsp static int
119 fe621944 2020-11-10 stsp diff_data_atomize_text_lines_mmap(struct diff_data *d)
120 fe621944 2020-11-10 stsp {
121 fe621944 2020-11-10 stsp 	const uint8_t *pos = d->data;
122 fe621944 2020-11-10 stsp 	const uint8_t *end = pos + d->len;
123 fe621944 2020-11-10 stsp 	bool ignore_whitespace = (d->diff_flags & DIFF_FLAG_IGNORE_WHITESPACE);
124 fe621944 2020-11-10 stsp 
125 fe621944 2020-11-10 stsp 	unsigned int array_size_estimate = d->len / 50;
126 fe621944 2020-11-10 stsp 	unsigned int pow2 = 1;
127 fe621944 2020-11-10 stsp 	while (array_size_estimate >>= 1)
128 fe621944 2020-11-10 stsp 		pow2++;
129 fe621944 2020-11-10 stsp 
130 fe621944 2020-11-10 stsp 	ARRAYLIST_INIT(d->atoms, 1 << pow2);
131 fe621944 2020-11-10 stsp 
132 fe621944 2020-11-10 stsp 	while (pos < end) {
133 fe621944 2020-11-10 stsp 		const uint8_t *line_end = pos;
134 fe621944 2020-11-10 stsp 		unsigned int hash = 0;
135 fe621944 2020-11-10 stsp 
136 fe621944 2020-11-10 stsp 		while (line_end < end && *line_end != '\r' && *line_end != '\n') {
137 fe621944 2020-11-10 stsp 			if (!ignore_whitespace
138 fe621944 2020-11-10 stsp 			    || !isspace(*line_end))
139 fe621944 2020-11-10 stsp 				hash = hash * 23 + *line_end;
140 fe621944 2020-11-10 stsp 			line_end++;
141 fe621944 2020-11-10 stsp 		}
142 fe621944 2020-11-10 stsp 
143 fe621944 2020-11-10 stsp 		/* When not at the end of data, the line ending char ('\r' or
144 fe621944 2020-11-10 stsp 		 * '\n') must follow */
145 fe621944 2020-11-10 stsp 		if (line_end < end)
146 fe621944 2020-11-10 stsp 			line_end++;
147 fe621944 2020-11-10 stsp 		/* If that was an '\r', also pull in any following '\n' */
148 fe621944 2020-11-10 stsp 		if (line_end < end - 1 && line_end[0] == '\r' &&
149 fe621944 2020-11-10 stsp 		    line_end[1] == '\n')
150 fe621944 2020-11-10 stsp 			line_end++;
151 fe621944 2020-11-10 stsp 
152 fe621944 2020-11-10 stsp 		/* Record the found line as diff atom */
153 fe621944 2020-11-10 stsp 		struct diff_atom *atom;
154 fe621944 2020-11-10 stsp 		ARRAYLIST_ADD(atom, d->atoms);
155 fe621944 2020-11-10 stsp 		if (!atom)
156 fe621944 2020-11-10 stsp 			return ENOMEM;
157 fe621944 2020-11-10 stsp 
158 fe621944 2020-11-10 stsp 		*atom = (struct diff_atom){
159 fe621944 2020-11-10 stsp 			.root = d,
160 fe621944 2020-11-10 stsp 			.pos = (off_t)(pos - d->data),
161 fe621944 2020-11-10 stsp 			.at = pos,
162 fe621944 2020-11-10 stsp 			.len = line_end - pos,
163 fe621944 2020-11-10 stsp 			.hash = hash,
164 fe621944 2020-11-10 stsp 		};
165 fe621944 2020-11-10 stsp 
166 fe621944 2020-11-10 stsp 		/* Starting point for next line: */
167 fe621944 2020-11-10 stsp 		pos = line_end;
168 fe621944 2020-11-10 stsp 	}
169 fe621944 2020-11-10 stsp 
170 fe621944 2020-11-10 stsp 	return DIFF_RC_OK;
171 fe621944 2020-11-10 stsp }
172 fe621944 2020-11-10 stsp 
173 fe621944 2020-11-10 stsp static int
174 fe621944 2020-11-10 stsp diff_data_atomize_text_lines(struct diff_data *d)
175 fe621944 2020-11-10 stsp {
176 fe621944 2020-11-10 stsp 	if (d->data == NULL)
177 fe621944 2020-11-10 stsp 		return diff_data_atomize_text_lines_fd(d);
178 fe621944 2020-11-10 stsp 	else
179 fe621944 2020-11-10 stsp 		return diff_data_atomize_text_lines_mmap(d);
180 fe621944 2020-11-10 stsp }
181 fe621944 2020-11-10 stsp 
182 fe621944 2020-11-10 stsp int
183 fe621944 2020-11-10 stsp diff_atomize_text_by_line(void *func_data, struct diff_data *d)
184 fe621944 2020-11-10 stsp {
185 fe621944 2020-11-10 stsp 	return diff_data_atomize_text_lines(d);
186 fe621944 2020-11-10 stsp }