centraldir.c revision 90764cfc76d65ae67755a375eb8ef429e44b9822 
1#include "private.h"
2#include <stdio.h>
3#include <string.h>
4#include <stdlib.h>
5
6enum {
7    // finding the directory
8    CD_SIGNATURE = 0x06054b50,
9    EOCD_LEN     = 22,        // EndOfCentralDir len, excl. comment
10    MAX_COMMENT_LEN = 65535,
11    MAX_EOCD_SEARCH = MAX_COMMENT_LEN + EOCD_LEN,
12
13    // central directory entries
14    ENTRY_SIGNATURE = 0x02014b50,
15    ENTRY_LEN = 46,          // CentralDirEnt len, excl. var fields
16
17    // local file header
18    LFH_SIZE = 30,
19};
20
21unsigned int
22read_le_int(const unsigned char* buf)
23{
24    return buf[0] | (buf[1] << 8) | (buf[2] << 16) | (buf[3] << 24);
25}
26
27unsigned int
28read_le_short(const unsigned char* buf)
29{
30    return buf[0] | (buf[1] << 8);
31}
32
33static int
34read_central_dir_values(Zipfile* file, const unsigned char* buf, int len)
35{
36    if (len < EOCD_LEN) {
37        // looks like ZIP file got truncated
38        fprintf(stderr, " Zip EOCD: expected >= %d bytes, found %d\n",
39                EOCD_LEN, len);
40        return -1;
41    }
42
43    file->disknum = read_le_short(&buf[0x04]);
44    file->diskWithCentralDir = read_le_short(&buf[0x06]);
45    file->entryCount = read_le_short(&buf[0x08]);
46    file->totalEntryCount = read_le_short(&buf[0x0a]);
47    file->centralDirSize = read_le_int(&buf[0x0c]);
48    file->centralDirOffest = read_le_int(&buf[0x10]);
49    file->commentLen = read_le_short(&buf[0x14]);
50
51    if (file->commentLen > 0) {
52        if (EOCD_LEN + file->commentLen > len) {
53            fprintf(stderr, "EOCD(%d) + comment(%d) exceeds len (%d)\n",
54                    EOCD_LEN, file->commentLen, len);
55            return -1;
56        }
57        file->comment = buf + EOCD_LEN;
58    }
59
60    return 0;
61}
62
63static int
64read_central_directory_entry(Zipfile* file, Zipentry* entry,
65                const unsigned char** buf, ssize_t* len)
66{
67    const unsigned char* p;
68
69    unsigned short  versionMadeBy;
70    unsigned short  versionToExtract;
71    unsigned short  gpBitFlag;
72    unsigned short  compressionMethod;
73    unsigned short  lastModFileTime;
74    unsigned short  lastModFileDate;
75    unsigned long   crc32;
76    unsigned short  extraFieldLength;
77    unsigned short  fileCommentLength;
78    unsigned short  diskNumberStart;
79    unsigned short  internalAttrs;
80    unsigned long   externalAttrs;
81    unsigned long   localHeaderRelOffset;
82    const unsigned char*  extraField;
83    const unsigned char*  fileComment;
84    unsigned int dataOffset;
85    unsigned short lfhExtraFieldSize;
86
87
88    p = *buf;
89
90    if (*len < ENTRY_LEN) {
91        fprintf(stderr, "cde entry not large enough\n");
92        return -1;
93    }
94
95    if (read_le_int(&p[0x00]) != ENTRY_SIGNATURE) {
96        fprintf(stderr, "Whoops: didn't find expected signature\n");
97        return -1;
98    }
99
100    versionMadeBy = read_le_short(&p[0x04]);
101    versionToExtract = read_le_short(&p[0x06]);
102    gpBitFlag = read_le_short(&p[0x08]);
103    entry->compressionMethod = read_le_short(&p[0x0a]);
104    lastModFileTime = read_le_short(&p[0x0c]);
105    lastModFileDate = read_le_short(&p[0x0e]);
106    crc32 = read_le_int(&p[0x10]);
107    entry->compressedSize = read_le_int(&p[0x14]);
108    entry->uncompressedSize = read_le_int(&p[0x18]);
109    entry->fileNameLength = read_le_short(&p[0x1c]);
110    extraFieldLength = read_le_short(&p[0x1e]);
111    fileCommentLength = read_le_short(&p[0x20]);
112    diskNumberStart = read_le_short(&p[0x22]);
113    internalAttrs = read_le_short(&p[0x24]);
114    externalAttrs = read_le_int(&p[0x26]);
115    localHeaderRelOffset = read_le_int(&p[0x2a]);
116
117    p += ENTRY_LEN;
118
119    // filename
120    if (entry->fileNameLength != 0) {
121        entry->fileName = p;
122    } else {
123        entry->fileName = NULL;
124    }
125    p += entry->fileNameLength;
126
127    // extra field
128    if (extraFieldLength != 0) {
129        extraField = p;
130    } else {
131        extraField = NULL;
132    }
133    p += extraFieldLength;
134
135    // comment, if any
136    if (fileCommentLength != 0) {
137        fileComment = p;
138    } else {
139        fileComment = NULL;
140    }
141    p += fileCommentLength;
142
143    *buf = p;
144
145    // the size of the extraField in the central dir is how much data there is,
146    // but the one in the local file header also contains some padding.
147    p = file->buf + localHeaderRelOffset;
148    extraFieldLength = read_le_short(&p[0x1c]);
149
150    dataOffset = localHeaderRelOffset + LFH_SIZE
151        + entry->fileNameLength + extraFieldLength;
152    entry->data = file->buf + dataOffset;
153#if 0
154    printf("file->buf=%p entry->data=%p dataOffset=%x localHeaderRelOffset=%d "
155           "entry->fileNameLength=%d extraFieldLength=%d\n",
156           file->buf, entry->data, dataOffset, localHeaderRelOffset,
157           entry->fileNameLength, extraFieldLength);
158#endif
159    return 0;
160}
161
162/*
163 * Find the central directory and read the contents.
164 *
165 * The fun thing about ZIP archives is that they may or may not be
166 * readable from start to end.  In some cases, notably for archives
167 * that were written to stdout, the only length information is in the
168 * central directory at the end of the file.
169 *
170 * Of course, the central directory can be followed by a variable-length
171 * comment field, so we have to scan through it backwards.  The comment
172 * is at most 64K, plus we have 18 bytes for the end-of-central-dir stuff
173 * itself, plus apparently sometimes people throw random junk on the end
174 * just for the fun of it.
175 *
176 * This is all a little wobbly.  If the wrong value ends up in the EOCD
177 * area, we're hosed.  This appears to be the way that everbody handles
178 * it though, so we're in pretty good company if this fails.
179 */
180int
181read_central_dir(Zipfile *file)
182{
183    int err;
184
185    const unsigned char* buf = file->buf;
186    ssize_t bufsize = file->bufsize;
187    const unsigned char* eocd;
188    const unsigned char* p;
189    const unsigned char* start;
190    ssize_t len;
191    int i;
192
193    // too small to be a ZIP archive?
194    if (bufsize < EOCD_LEN) {
195        fprintf(stderr, "Length is %d -- too small\n", bufsize);
196        goto bail;
197    }
198
199    // find the end-of-central-dir magic
200    if (bufsize > MAX_EOCD_SEARCH) {
201        start = buf + bufsize - MAX_EOCD_SEARCH;
202    } else {
203        start = buf;
204    }
205    p = buf + bufsize - 4;
206    while (p >= start) {
207        if (*p == 0x50 && read_le_int(p) == CD_SIGNATURE) {
208            eocd = p;
209            break;
210        }
211        p--;
212    }
213    if (p < start) {
214        fprintf(stderr, "EOCD not found, not Zip\n");
215        goto bail;
216    }
217
218    // extract eocd values
219    err = read_central_dir_values(file, eocd, (buf+bufsize)-eocd);
220    if (err != 0) {
221        goto bail;
222    }
223
224    if (file->disknum != 0
225          || file->diskWithCentralDir != 0
226          || file->entryCount != file->totalEntryCount) {
227        fprintf(stderr, "Archive spanning not supported\n");
228        goto bail;
229    }
230
231    // Loop through and read the central dir entries.
232    p = buf + file->centralDirOffest;
233    len = (buf+bufsize)-p;
234    for (i=0; i < file->totalEntryCount; i++) {
235        Zipentry* entry = malloc(sizeof(Zipentry));
236        memset(entry, 0, sizeof(Zipentry));
237
238        err = read_central_directory_entry(file, entry, &p, &len);
239        if (err != 0) {
240            fprintf(stderr, "read_central_directory_entry failed\n");
241            free(entry);
242            goto bail;
243        }
244
245        // add it to our list
246        entry->next = file->entries;
247        file->entries = entry;
248    }
249
250    return 0;
251bail:
252    return -1;
253}
254