src/scanner.c

Wed, 22 May 2013 10:26:22 +0200

author
Mike Becker <universe@uap-core.de>
date
Wed, 22 May 2013 10:26:22 +0200
changeset 42
0402b9b41b0a
parent 41
c2e73e175341
child 43
104e75d18ede
permissions
-rw-r--r--

file list is now sorted

10
ecf787666f44 refactored sources
Mike Becker <universe@uap-core.de>
parents: 8
diff changeset
1 /*
34
fa9bda32de17 moved src files to src subdirectory and added licence text
Mike Becker <universe@uap-core.de>
parents: 30
diff changeset
2 * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS HEADER.
36
a7ff583e153f updated copyright year + added make install + removed project files
Mike Becker <universe@uap-core.de>
parents: 34
diff changeset
3 * Copyright 2013 Mike Becker. All rights reserved.
34
fa9bda32de17 moved src files to src subdirectory and added licence text
Mike Becker <universe@uap-core.de>
parents: 30
diff changeset
4 *
fa9bda32de17 moved src files to src subdirectory and added licence text
Mike Becker <universe@uap-core.de>
parents: 30
diff changeset
5 * Redistribution and use in source and binary forms, with or without
fa9bda32de17 moved src files to src subdirectory and added licence text
Mike Becker <universe@uap-core.de>
parents: 30
diff changeset
6 * modification, are permitted provided that the following conditions are met:
fa9bda32de17 moved src files to src subdirectory and added licence text
Mike Becker <universe@uap-core.de>
parents: 30
diff changeset
7 *
fa9bda32de17 moved src files to src subdirectory and added licence text
Mike Becker <universe@uap-core.de>
parents: 30
diff changeset
8 * 1. Redistributions of source code must retain the above copyright
fa9bda32de17 moved src files to src subdirectory and added licence text
Mike Becker <universe@uap-core.de>
parents: 30
diff changeset
9 * notice, this list of conditions and the following disclaimer.
fa9bda32de17 moved src files to src subdirectory and added licence text
Mike Becker <universe@uap-core.de>
parents: 30
diff changeset
10 *
fa9bda32de17 moved src files to src subdirectory and added licence text
Mike Becker <universe@uap-core.de>
parents: 30
diff changeset
11 * 2. Redistributions in binary form must reproduce the above copyright
fa9bda32de17 moved src files to src subdirectory and added licence text
Mike Becker <universe@uap-core.de>
parents: 30
diff changeset
12 * notice, this list of conditions and the following disclaimer in the
fa9bda32de17 moved src files to src subdirectory and added licence text
Mike Becker <universe@uap-core.de>
parents: 30
diff changeset
13 * documentation and/or other materials provided with the distribution.
fa9bda32de17 moved src files to src subdirectory and added licence text
Mike Becker <universe@uap-core.de>
parents: 30
diff changeset
14 *
fa9bda32de17 moved src files to src subdirectory and added licence text
Mike Becker <universe@uap-core.de>
parents: 30
diff changeset
15 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
fa9bda32de17 moved src files to src subdirectory and added licence text
Mike Becker <universe@uap-core.de>
parents: 30
diff changeset
16 * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
fa9bda32de17 moved src files to src subdirectory and added licence text
Mike Becker <universe@uap-core.de>
parents: 30
diff changeset
17 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
fa9bda32de17 moved src files to src subdirectory and added licence text
Mike Becker <universe@uap-core.de>
parents: 30
diff changeset
18 * DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
fa9bda32de17 moved src files to src subdirectory and added licence text
Mike Becker <universe@uap-core.de>
parents: 30
diff changeset
19 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
fa9bda32de17 moved src files to src subdirectory and added licence text
Mike Becker <universe@uap-core.de>
parents: 30
diff changeset
20 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
fa9bda32de17 moved src files to src subdirectory and added licence text
Mike Becker <universe@uap-core.de>
parents: 30
diff changeset
21 * SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
fa9bda32de17 moved src files to src subdirectory and added licence text
Mike Becker <universe@uap-core.de>
parents: 30
diff changeset
22 * CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
fa9bda32de17 moved src files to src subdirectory and added licence text
Mike Becker <universe@uap-core.de>
parents: 30
diff changeset
23 * OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
fa9bda32de17 moved src files to src subdirectory and added licence text
Mike Becker <universe@uap-core.de>
parents: 30
diff changeset
24 * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
fa9bda32de17 moved src files to src subdirectory and added licence text
Mike Becker <universe@uap-core.de>
parents: 30
diff changeset
25 *
20
43725438ac50 Changed author comments + added signatures for upcomming bfile heuristics
Mike Becker <universe@uap-core.de>
parents: 18
diff changeset
26 * scanner.c
10
ecf787666f44 refactored sources
Mike Becker <universe@uap-core.de>
parents: 8
diff changeset
27 *
ecf787666f44 refactored sources
Mike Becker <universe@uap-core.de>
parents: 8
diff changeset
28 * Created on: 23.05.2011
20
43725438ac50 Changed author comments + added signatures for upcomming bfile heuristics
Mike Becker <universe@uap-core.de>
parents: 18
diff changeset
29 * Author: Mike
10
ecf787666f44 refactored sources
Mike Becker <universe@uap-core.de>
parents: 8
diff changeset
30 */
3
510d6b198dde Moved some functions to functions.c
Mike Becker <universe@uap-core.de>
parents: 1
diff changeset
31
8
28319b20968c encapsulated common operations
Mike Becker <universe@uap-core.de>
parents: 6
diff changeset
32
10
ecf787666f44 refactored sources
Mike Becker <universe@uap-core.de>
parents: 8
diff changeset
33 #include "scanner.h"
ecf787666f44 refactored sources
Mike Becker <universe@uap-core.de>
parents: 8
diff changeset
34 #include "suffix_fnc.h"
21
91e0890464b0 implemented bfile heuristics option + TODO: implement algorithm
Mike Becker <universe@uap-core.de>
parents: 20
diff changeset
35 #include "bfile_heuristics.h"
27
95a958e3de88 added regexp_parser struct and compile function
Mike Becker <universe@uap-core.de>
parents: 25
diff changeset
36 #include "regex_parser.h"
23
778388400f7b encapsulated scanner arguments + enabled optimizer + empty file is no bfile
Mike Becker <universe@uap-core.de>
parents: 22
diff changeset
37 #include <sys/stat.h>
3
510d6b198dde Moved some functions to functions.c
Mike Becker <universe@uap-core.de>
parents: 1
diff changeset
38
41
c2e73e175341 replaced direct scanning of directories with a two-step approach (first: create filename list, second: scan)
Mike Becker <universe@uap-core.de>
parents: 40
diff changeset
39 typedef struct filelist filelist_t;
23
778388400f7b encapsulated scanner arguments + enabled optimizer + empty file is no bfile
Mike Becker <universe@uap-core.de>
parents: 22
diff changeset
40
41
c2e73e175341 replaced direct scanning of directories with a two-step approach (first: create filename list, second: scan)
Mike Becker <universe@uap-core.de>
parents: 40
diff changeset
41 struct filelist {
c2e73e175341 replaced direct scanning of directories with a two-step approach (first: create filename list, second: scan)
Mike Becker <universe@uap-core.de>
parents: 40
diff changeset
42 char *displayname;
c2e73e175341 replaced direct scanning of directories with a two-step approach (first: create filename list, second: scan)
Mike Becker <universe@uap-core.de>
parents: 40
diff changeset
43 int displayname_len;
c2e73e175341 replaced direct scanning of directories with a two-step approach (first: create filename list, second: scan)
Mike Becker <universe@uap-core.de>
parents: 40
diff changeset
44 char *filename;
c2e73e175341 replaced direct scanning of directories with a two-step approach (first: create filename list, second: scan)
Mike Becker <universe@uap-core.de>
parents: 40
diff changeset
45 int st_mode;
c2e73e175341 replaced direct scanning of directories with a two-step approach (first: create filename list, second: scan)
Mike Becker <universe@uap-core.de>
parents: 40
diff changeset
46 filelist_t *next;
c2e73e175341 replaced direct scanning of directories with a two-step approach (first: create filename list, second: scan)
Mike Becker <universe@uap-core.de>
parents: 40
diff changeset
47 };
c2e73e175341 replaced direct scanning of directories with a two-step approach (first: create filename list, second: scan)
Mike Becker <universe@uap-core.de>
parents: 40
diff changeset
48
c2e73e175341 replaced direct scanning of directories with a two-step approach (first: create filename list, second: scan)
Mike Becker <universe@uap-core.de>
parents: 40
diff changeset
49 filelist_t *buildFileList(scanner_t scanner, settings_t* settings,
c2e73e175341 replaced direct scanning of directories with a two-step approach (first: create filename list, second: scan)
Mike Becker <universe@uap-core.de>
parents: 40
diff changeset
50 filelist_t* list) {
c2e73e175341 replaced direct scanning of directories with a two-step approach (first: create filename list, second: scan)
Mike Becker <universe@uap-core.de>
parents: 40
diff changeset
51
23
778388400f7b encapsulated scanner arguments + enabled optimizer + empty file is no bfile
Mike Becker <universe@uap-core.de>
parents: 22
diff changeset
52 DIR *dirf;
3
510d6b198dde Moved some functions to functions.c
Mike Becker <universe@uap-core.de>
parents: 1
diff changeset
53 struct dirent *entry;
23
778388400f7b encapsulated scanner arguments + enabled optimizer + empty file is no bfile
Mike Becker <universe@uap-core.de>
parents: 22
diff changeset
54 struct stat statbuf;
41
c2e73e175341 replaced direct scanning of directories with a two-step approach (first: create filename list, second: scan)
Mike Becker <universe@uap-core.de>
parents: 40
diff changeset
55
23
778388400f7b encapsulated scanner arguments + enabled optimizer + empty file is no bfile
Mike Becker <universe@uap-core.de>
parents: 22
diff changeset
56 if ((dirf = opendir(scanner.dir)) == NULL) {
34
fa9bda32de17 moved src files to src subdirectory and added licence text
Mike Becker <universe@uap-core.de>
parents: 30
diff changeset
57 printf("%s", scanner.dir);
23
778388400f7b encapsulated scanner arguments + enabled optimizer + empty file is no bfile
Mike Becker <universe@uap-core.de>
parents: 22
diff changeset
58 perror(" Directory access failed");
778388400f7b encapsulated scanner arguments + enabled optimizer + empty file is no bfile
Mike Becker <universe@uap-core.de>
parents: 22
diff changeset
59 return 0;
778388400f7b encapsulated scanner arguments + enabled optimizer + empty file is no bfile
Mike Becker <universe@uap-core.de>
parents: 22
diff changeset
60 }
778388400f7b encapsulated scanner arguments + enabled optimizer + empty file is no bfile
Mike Becker <universe@uap-core.de>
parents: 22
diff changeset
61
778388400f7b encapsulated scanner arguments + enabled optimizer + empty file is no bfile
Mike Becker <universe@uap-core.de>
parents: 22
diff changeset
62 while ((entry = readdir(dirf)) != NULL) {
3
510d6b198dde Moved some functions to functions.c
Mike Becker <universe@uap-core.de>
parents: 1
diff changeset
63 if (strcmp(entry->d_name, ".") != 0 && strcmp(entry->d_name, "..") != 0) {
41
c2e73e175341 replaced direct scanning of directories with a two-step approach (first: create filename list, second: scan)
Mike Becker <universe@uap-core.de>
parents: 40
diff changeset
64
c2e73e175341 replaced direct scanning of directories with a two-step approach (first: create filename list, second: scan)
Mike Becker <universe@uap-core.de>
parents: 40
diff changeset
65 /* Create new filelist entry */
c2e73e175341 replaced direct scanning of directories with a two-step approach (first: create filename list, second: scan)
Mike Becker <universe@uap-core.de>
parents: 40
diff changeset
66 filelist_t *newentry = (filelist_t*) malloc(sizeof(filelist_t));
42
0402b9b41b0a file list is now sorted
Mike Becker <universe@uap-core.de>
parents: 41
diff changeset
67 newentry->next = NULL;
41
c2e73e175341 replaced direct scanning of directories with a two-step approach (first: create filename list, second: scan)
Mike Becker <universe@uap-core.de>
parents: 40
diff changeset
68
42
0402b9b41b0a file list is now sorted
Mike Becker <universe@uap-core.de>
parents: 41
diff changeset
69 newentry->displayname_len = strlen(entry->d_name);
0402b9b41b0a file list is now sorted
Mike Becker <universe@uap-core.de>
parents: 41
diff changeset
70 newentry->displayname = (char*) malloc(newentry->displayname_len+1);
0402b9b41b0a file list is now sorted
Mike Becker <universe@uap-core.de>
parents: 41
diff changeset
71 memcpy(newentry->displayname, entry->d_name, newentry->displayname_len);
0402b9b41b0a file list is now sorted
Mike Becker <universe@uap-core.de>
parents: 41
diff changeset
72 newentry->displayname[newentry->displayname_len] = 0;
41
c2e73e175341 replaced direct scanning of directories with a two-step approach (first: create filename list, second: scan)
Mike Becker <universe@uap-core.de>
parents: 40
diff changeset
73
42
0402b9b41b0a file list is now sorted
Mike Becker <universe@uap-core.de>
parents: 41
diff changeset
74 newentry->st_mode = 0;
41
c2e73e175341 replaced direct scanning of directories with a two-step approach (first: create filename list, second: scan)
Mike Becker <universe@uap-core.de>
parents: 40
diff changeset
75
40
5938a9b74e8e improved printing the results and fixed bug where the filename was displayed as an empty string
Mike Becker <universe@uap-core.de>
parents: 36
diff changeset
76 /* Construct absolute pathname string */
41
c2e73e175341 replaced direct scanning of directories with a two-step approach (first: create filename list, second: scan)
Mike Becker <universe@uap-core.de>
parents: 40
diff changeset
77 size_t dirnamelen = strlen(scanner.dir);
42
0402b9b41b0a file list is now sorted
Mike Becker <universe@uap-core.de>
parents: 41
diff changeset
78 char *filename = (char*) malloc(2+dirnamelen+newentry->displayname_len);
41
c2e73e175341 replaced direct scanning of directories with a two-step approach (first: create filename list, second: scan)
Mike Becker <universe@uap-core.de>
parents: 40
diff changeset
79 memcpy(filename, scanner.dir, dirnamelen);
c2e73e175341 replaced direct scanning of directories with a two-step approach (first: create filename list, second: scan)
Mike Becker <universe@uap-core.de>
parents: 40
diff changeset
80 filename[dirnamelen] = settings->fileSeparator;
42
0402b9b41b0a file list is now sorted
Mike Becker <universe@uap-core.de>
parents: 41
diff changeset
81 memcpy(filename+dirnamelen+1, entry->d_name, newentry->displayname_len);
0402b9b41b0a file list is now sorted
Mike Becker <universe@uap-core.de>
parents: 41
diff changeset
82 filename[1+dirnamelen+newentry->displayname_len] = 0;
0402b9b41b0a file list is now sorted
Mike Becker <universe@uap-core.de>
parents: 41
diff changeset
83 newentry->filename = filename;
14
ee9333c91dda some minor fixes + makefile now injects revisionnumber into cline.h
Mike Becker <universe@uap-core.de>
parents: 10
diff changeset
84
22
4508da679ffb completed binary file heuristics
Mike Becker <universe@uap-core.de>
parents: 21
diff changeset
85 /* Check for subdirectory */
23
778388400f7b encapsulated scanner arguments + enabled optimizer + empty file is no bfile
Mike Becker <universe@uap-core.de>
parents: 22
diff changeset
86 if (stat(filename, &statbuf) == 0) {
42
0402b9b41b0a file list is now sorted
Mike Becker <universe@uap-core.de>
parents: 41
diff changeset
87 newentry->st_mode = statbuf.st_mode;
23
778388400f7b encapsulated scanner arguments + enabled optimizer + empty file is no bfile
Mike Becker <universe@uap-core.de>
parents: 22
diff changeset
88 } else {
778388400f7b encapsulated scanner arguments + enabled optimizer + empty file is no bfile
Mike Becker <universe@uap-core.de>
parents: 22
diff changeset
89 perror(" Error in stat call");
3
510d6b198dde Moved some functions to functions.c
Mike Becker <universe@uap-core.de>
parents: 1
diff changeset
90 continue;
510d6b198dde Moved some functions to functions.c
Mike Becker <universe@uap-core.de>
parents: 1
diff changeset
91 }
42
0402b9b41b0a file list is now sorted
Mike Becker <universe@uap-core.de>
parents: 41
diff changeset
92
0402b9b41b0a file list is now sorted
Mike Becker <universe@uap-core.de>
parents: 41
diff changeset
93 if (list) {
0402b9b41b0a file list is now sorted
Mike Becker <universe@uap-core.de>
parents: 41
diff changeset
94 // create fake root to have a pointer on the true root
0402b9b41b0a file list is now sorted
Mike Becker <universe@uap-core.de>
parents: 41
diff changeset
95 filelist_t root;
0402b9b41b0a file list is now sorted
Mike Becker <universe@uap-core.de>
parents: 41
diff changeset
96 root.next = list;
0402b9b41b0a file list is now sorted
Mike Becker <universe@uap-core.de>
parents: 41
diff changeset
97 filelist_t *parent = &root;
0402b9b41b0a file list is now sorted
Mike Becker <universe@uap-core.de>
parents: 41
diff changeset
98 while (parent->next &&
0402b9b41b0a file list is now sorted
Mike Becker <universe@uap-core.de>
parents: 41
diff changeset
99 (strcasecmp(parent->next->displayname, newentry->displayname) < 0 ||
0402b9b41b0a file list is now sorted
Mike Becker <universe@uap-core.de>
parents: 41
diff changeset
100 (!S_ISDIR(newentry->st_mode) && S_ISDIR(parent->next->st_mode))
0402b9b41b0a file list is now sorted
Mike Becker <universe@uap-core.de>
parents: 41
diff changeset
101 ) &&
0402b9b41b0a file list is now sorted
Mike Becker <universe@uap-core.de>
parents: 41
diff changeset
102 (!S_ISDIR(newentry->st_mode) || S_ISDIR(parent->next->st_mode))
0402b9b41b0a file list is now sorted
Mike Becker <universe@uap-core.de>
parents: 41
diff changeset
103 ) {
0402b9b41b0a file list is now sorted
Mike Becker <universe@uap-core.de>
parents: 41
diff changeset
104 parent = parent->next;
0402b9b41b0a file list is now sorted
Mike Becker <universe@uap-core.de>
parents: 41
diff changeset
105 }
0402b9b41b0a file list is now sorted
Mike Becker <universe@uap-core.de>
parents: 41
diff changeset
106 newentry->next = parent->next;
0402b9b41b0a file list is now sorted
Mike Becker <universe@uap-core.de>
parents: 41
diff changeset
107 parent->next = newentry;
0402b9b41b0a file list is now sorted
Mike Becker <universe@uap-core.de>
parents: 41
diff changeset
108 list = root.next;
0402b9b41b0a file list is now sorted
Mike Becker <universe@uap-core.de>
parents: 41
diff changeset
109 } else {
0402b9b41b0a file list is now sorted
Mike Becker <universe@uap-core.de>
parents: 41
diff changeset
110 list = newentry;
0402b9b41b0a file list is now sorted
Mike Becker <universe@uap-core.de>
parents: 41
diff changeset
111 }
41
c2e73e175341 replaced direct scanning of directories with a two-step approach (first: create filename list, second: scan)
Mike Becker <universe@uap-core.de>
parents: 40
diff changeset
112 }
c2e73e175341 replaced direct scanning of directories with a two-step approach (first: create filename list, second: scan)
Mike Becker <universe@uap-core.de>
parents: 40
diff changeset
113 }
c2e73e175341 replaced direct scanning of directories with a two-step approach (first: create filename list, second: scan)
Mike Becker <universe@uap-core.de>
parents: 40
diff changeset
114
c2e73e175341 replaced direct scanning of directories with a two-step approach (first: create filename list, second: scan)
Mike Becker <universe@uap-core.de>
parents: 40
diff changeset
115 closedir(dirf);
c2e73e175341 replaced direct scanning of directories with a two-step approach (first: create filename list, second: scan)
Mike Becker <universe@uap-core.de>
parents: 40
diff changeset
116
c2e73e175341 replaced direct scanning of directories with a two-step approach (first: create filename list, second: scan)
Mike Becker <universe@uap-core.de>
parents: 40
diff changeset
117 return list;
c2e73e175341 replaced direct scanning of directories with a two-step approach (first: create filename list, second: scan)
Mike Becker <universe@uap-core.de>
parents: 40
diff changeset
118 }
3
510d6b198dde Moved some functions to functions.c
Mike Becker <universe@uap-core.de>
parents: 1
diff changeset
119
41
c2e73e175341 replaced direct scanning of directories with a two-step approach (first: create filename list, second: scan)
Mike Becker <universe@uap-core.de>
parents: 40
diff changeset
120 int scanDirectory(scanner_t scanner, settings_t* settings) {
c2e73e175341 replaced direct scanning of directories with a two-step approach (first: create filename list, second: scan)
Mike Becker <universe@uap-core.de>
parents: 40
diff changeset
121
c2e73e175341 replaced direct scanning of directories with a two-step approach (first: create filename list, second: scan)
Mike Becker <universe@uap-core.de>
parents: 40
diff changeset
122 int lines, a;
c2e73e175341 replaced direct scanning of directories with a two-step approach (first: create filename list, second: scan)
Mike Becker <universe@uap-core.de>
parents: 40
diff changeset
123 int lineSum = 0;
c2e73e175341 replaced direct scanning of directories with a two-step approach (first: create filename list, second: scan)
Mike Becker <universe@uap-core.de>
parents: 40
diff changeset
124 bool bfile;
c2e73e175341 replaced direct scanning of directories with a two-step approach (first: create filename list, second: scan)
Mike Becker <universe@uap-core.de>
parents: 40
diff changeset
125
c2e73e175341 replaced direct scanning of directories with a two-step approach (first: create filename list, second: scan)
Mike Becker <universe@uap-core.de>
parents: 40
diff changeset
126 filelist_t *filelist = buildFileList(scanner, settings, NULL);
c2e73e175341 replaced direct scanning of directories with a two-step approach (first: create filename list, second: scan)
Mike Becker <universe@uap-core.de>
parents: 40
diff changeset
127
c2e73e175341 replaced direct scanning of directories with a two-step approach (first: create filename list, second: scan)
Mike Becker <universe@uap-core.de>
parents: 40
diff changeset
128 while (filelist != NULL) {
c2e73e175341 replaced direct scanning of directories with a two-step approach (first: create filename list, second: scan)
Mike Becker <universe@uap-core.de>
parents: 40
diff changeset
129
c2e73e175341 replaced direct scanning of directories with a two-step approach (first: create filename list, second: scan)
Mike Becker <universe@uap-core.de>
parents: 40
diff changeset
130 /* Scan subdirectories */
42
0402b9b41b0a file list is now sorted
Mike Becker <universe@uap-core.de>
parents: 41
diff changeset
131 if (!S_ISREG(filelist->st_mode)) {
41
c2e73e175341 replaced direct scanning of directories with a two-step approach (first: create filename list, second: scan)
Mike Becker <universe@uap-core.de>
parents: 40
diff changeset
132 printf("%*s\n", filelist->displayname_len+scanner.spaces,
c2e73e175341 replaced direct scanning of directories with a two-step approach (first: create filename list, second: scan)
Mike Becker <universe@uap-core.de>
parents: 40
diff changeset
133 filelist->displayname);
42
0402b9b41b0a file list is now sorted
Mike Becker <universe@uap-core.de>
parents: 41
diff changeset
134 if (settings->recursive && S_ISDIR(filelist->st_mode)) {
41
c2e73e175341 replaced direct scanning of directories with a two-step approach (first: create filename list, second: scan)
Mike Becker <universe@uap-core.de>
parents: 40
diff changeset
135 scanDirectory((scanner_t) {filelist->filename, scanner.spaces+1},
c2e73e175341 replaced direct scanning of directories with a two-step approach (first: create filename list, second: scan)
Mike Becker <universe@uap-core.de>
parents: 40
diff changeset
136 settings);
c2e73e175341 replaced direct scanning of directories with a two-step approach (first: create filename list, second: scan)
Mike Becker <universe@uap-core.de>
parents: 40
diff changeset
137 }
c2e73e175341 replaced direct scanning of directories with a two-step approach (first: create filename list, second: scan)
Mike Becker <universe@uap-core.de>
parents: 40
diff changeset
138 } else {
30
d642fdb6745e allow parallel use of -s and -S
Mike Becker <universe@uap-core.de>
parents: 28
diff changeset
139 if ((settings->includeSuffixes->count == 0
41
c2e73e175341 replaced direct scanning of directories with a two-step approach (first: create filename list, second: scan)
Mike Becker <universe@uap-core.de>
parents: 40
diff changeset
140 || testSuffix(filelist->displayname, settings->includeSuffixes))
c2e73e175341 replaced direct scanning of directories with a two-step approach (first: create filename list, second: scan)
Mike Becker <universe@uap-core.de>
parents: 40
diff changeset
141 && !testSuffix(filelist->displayname, settings->excludeSuffixes)) {
c2e73e175341 replaced direct scanning of directories with a two-step approach (first: create filename list, second: scan)
Mike Becker <universe@uap-core.de>
parents: 40
diff changeset
142
25
802c5382f499 Added line buffer (and warning message - there is no regexp parser, though)
Mike Becker <universe@uap-core.de>
parents: 23
diff changeset
143 /* Count lines */
802c5382f499 Added line buffer (and warning message - there is no regexp parser, though)
Mike Becker <universe@uap-core.de>
parents: 23
diff changeset
144 lines = 0;
802c5382f499 Added line buffer (and warning message - there is no regexp parser, though)
Mike Becker <universe@uap-core.de>
parents: 23
diff changeset
145 bfile = false;
802c5382f499 Added line buffer (and warning message - there is no regexp parser, though)
Mike Becker <universe@uap-core.de>
parents: 23
diff changeset
146 bfile_reset(settings->bfileHeuristics);
27
95a958e3de88 added regexp_parser struct and compile function
Mike Becker <universe@uap-core.de>
parents: 25
diff changeset
147 char line_buffer[REGEX_MAX_LINELENGTH];
25
802c5382f499 Added line buffer (and warning message - there is no regexp parser, though)
Mike Becker <universe@uap-core.de>
parents: 23
diff changeset
148 int line_buffer_offset = 0;
802c5382f499 Added line buffer (and warning message - there is no regexp parser, though)
Mike Becker <universe@uap-core.de>
parents: 23
diff changeset
149
41
c2e73e175341 replaced direct scanning of directories with a two-step approach (first: create filename list, second: scan)
Mike Becker <universe@uap-core.de>
parents: 40
diff changeset
150 FILE *file = fopen(filelist->filename, "r");
3
510d6b198dde Moved some functions to functions.c
Mike Becker <universe@uap-core.de>
parents: 1
diff changeset
151 if (file == NULL) {
41
c2e73e175341 replaced direct scanning of directories with a two-step approach (first: create filename list, second: scan)
Mike Becker <universe@uap-core.de>
parents: 40
diff changeset
152 printf("%*s", filelist->displayname_len+scanner.spaces,
c2e73e175341 replaced direct scanning of directories with a two-step approach (first: create filename list, second: scan)
Mike Becker <universe@uap-core.de>
parents: 40
diff changeset
153 filelist->displayname);
3
510d6b198dde Moved some functions to functions.c
Mike Becker <universe@uap-core.de>
parents: 1
diff changeset
154 perror(" File acces failed");
41
c2e73e175341 replaced direct scanning of directories with a two-step approach (first: create filename list, second: scan)
Mike Becker <universe@uap-core.de>
parents: 40
diff changeset
155 } else {
c2e73e175341 replaced direct scanning of directories with a two-step approach (first: create filename list, second: scan)
Mike Becker <universe@uap-core.de>
parents: 40
diff changeset
156 do {
c2e73e175341 replaced direct scanning of directories with a two-step approach (first: create filename list, second: scan)
Mike Becker <universe@uap-core.de>
parents: 40
diff changeset
157 a = fgetc(file);
c2e73e175341 replaced direct scanning of directories with a two-step approach (first: create filename list, second: scan)
Mike Becker <universe@uap-core.de>
parents: 40
diff changeset
158
c2e73e175341 replaced direct scanning of directories with a two-step approach (first: create filename list, second: scan)
Mike Becker <universe@uap-core.de>
parents: 40
diff changeset
159 bfile = bfile_check(settings->bfileHeuristics, a);
3
510d6b198dde Moved some functions to functions.c
Mike Becker <universe@uap-core.de>
parents: 1
diff changeset
160
41
c2e73e175341 replaced direct scanning of directories with a two-step approach (first: create filename list, second: scan)
Mike Becker <universe@uap-core.de>
parents: 40
diff changeset
161 if (a == 10 || a == EOF) {
c2e73e175341 replaced direct scanning of directories with a two-step approach (first: create filename list, second: scan)
Mike Becker <universe@uap-core.de>
parents: 40
diff changeset
162 line_buffer[line_buffer_offset] = 0;
c2e73e175341 replaced direct scanning of directories with a two-step approach (first: create filename list, second: scan)
Mike Becker <universe@uap-core.de>
parents: 40
diff changeset
163 if (regex_parser_do(settings->regex, line_buffer) == 0) {
c2e73e175341 replaced direct scanning of directories with a two-step approach (first: create filename list, second: scan)
Mike Becker <universe@uap-core.de>
parents: 40
diff changeset
164 /* Only subtract lines when matching has finished */
c2e73e175341 replaced direct scanning of directories with a two-step approach (first: create filename list, second: scan)
Mike Becker <universe@uap-core.de>
parents: 40
diff changeset
165 if (!regex_parser_matching(settings->regex)) {
c2e73e175341 replaced direct scanning of directories with a two-step approach (first: create filename list, second: scan)
Mike Becker <universe@uap-core.de>
parents: 40
diff changeset
166 lines -= settings->regex->matched_lines;
c2e73e175341 replaced direct scanning of directories with a two-step approach (first: create filename list, second: scan)
Mike Becker <universe@uap-core.de>
parents: 40
diff changeset
167 }
c2e73e175341 replaced direct scanning of directories with a two-step approach (first: create filename list, second: scan)
Mike Becker <universe@uap-core.de>
parents: 40
diff changeset
168 }
21
91e0890464b0 implemented bfile heuristics option + TODO: implement algorithm
Mike Becker <universe@uap-core.de>
parents: 20
diff changeset
169
41
c2e73e175341 replaced direct scanning of directories with a two-step approach (first: create filename list, second: scan)
Mike Becker <universe@uap-core.de>
parents: 40
diff changeset
170 line_buffer_offset = 0;
c2e73e175341 replaced direct scanning of directories with a two-step approach (first: create filename list, second: scan)
Mike Becker <universe@uap-core.de>
parents: 40
diff changeset
171 lines++;
c2e73e175341 replaced direct scanning of directories with a two-step approach (first: create filename list, second: scan)
Mike Becker <universe@uap-core.de>
parents: 40
diff changeset
172 } else {
c2e73e175341 replaced direct scanning of directories with a two-step approach (first: create filename list, second: scan)
Mike Becker <universe@uap-core.de>
parents: 40
diff changeset
173 if (line_buffer_offset < REGEX_MAX_LINELENGTH) {
c2e73e175341 replaced direct scanning of directories with a two-step approach (first: create filename list, second: scan)
Mike Becker <universe@uap-core.de>
parents: 40
diff changeset
174 line_buffer[line_buffer_offset] = a;
c2e73e175341 replaced direct scanning of directories with a two-step approach (first: create filename list, second: scan)
Mike Becker <universe@uap-core.de>
parents: 40
diff changeset
175 line_buffer_offset++;
c2e73e175341 replaced direct scanning of directories with a two-step approach (first: create filename list, second: scan)
Mike Becker <universe@uap-core.de>
parents: 40
diff changeset
176 } else {
c2e73e175341 replaced direct scanning of directories with a two-step approach (first: create filename list, second: scan)
Mike Becker <universe@uap-core.de>
parents: 40
diff changeset
177 line_buffer[line_buffer_offset-1] = 0;
c2e73e175341 replaced direct scanning of directories with a two-step approach (first: create filename list, second: scan)
Mike Becker <universe@uap-core.de>
parents: 40
diff changeset
178 settings->confusing_lnlen = true;
28
72a98cbcb9f1 added regex parser
Mike Becker <universe@uap-core.de>
parents: 27
diff changeset
179 }
72a98cbcb9f1 added regex parser
Mike Becker <universe@uap-core.de>
parents: 27
diff changeset
180 }
41
c2e73e175341 replaced direct scanning of directories with a two-step approach (first: create filename list, second: scan)
Mike Becker <universe@uap-core.de>
parents: 40
diff changeset
181 } while (!bfile && a != EOF);
c2e73e175341 replaced direct scanning of directories with a two-step approach (first: create filename list, second: scan)
Mike Becker <universe@uap-core.de>
parents: 40
diff changeset
182 fclose(file);
25
802c5382f499 Added line buffer (and warning message - there is no regexp parser, though)
Mike Becker <universe@uap-core.de>
parents: 23
diff changeset
183
41
c2e73e175341 replaced direct scanning of directories with a two-step approach (first: create filename list, second: scan)
Mike Becker <universe@uap-core.de>
parents: 40
diff changeset
184 /* Print and sum line count */
c2e73e175341 replaced direct scanning of directories with a two-step approach (first: create filename list, second: scan)
Mike Becker <universe@uap-core.de>
parents: 40
diff changeset
185 if (bfile) {
c2e73e175341 replaced direct scanning of directories with a two-step approach (first: create filename list, second: scan)
Mike Becker <universe@uap-core.de>
parents: 40
diff changeset
186 if (!settings->matchesOnly) {
c2e73e175341 replaced direct scanning of directories with a two-step approach (first: create filename list, second: scan)
Mike Becker <universe@uap-core.de>
parents: 40
diff changeset
187 printf("%*s%*s%19s\n", filelist->displayname_len+scanner.spaces,
c2e73e175341 replaced direct scanning of directories with a two-step approach (first: create filename list, second: scan)
Mike Becker <universe@uap-core.de>
parents: 40
diff changeset
188 filelist->displayname,
c2e73e175341 replaced direct scanning of directories with a two-step approach (first: create filename list, second: scan)
Mike Becker <universe@uap-core.de>
parents: 40
diff changeset
189 60-filelist->displayname_len-scanner.spaces, "", "binary");
c2e73e175341 replaced direct scanning of directories with a two-step approach (first: create filename list, second: scan)
Mike Becker <universe@uap-core.de>
parents: 40
diff changeset
190 }
25
802c5382f499 Added line buffer (and warning message - there is no regexp parser, though)
Mike Becker <universe@uap-core.de>
parents: 23
diff changeset
191 } else {
41
c2e73e175341 replaced direct scanning of directories with a two-step approach (first: create filename list, second: scan)
Mike Becker <universe@uap-core.de>
parents: 40
diff changeset
192 lineSum += lines;
c2e73e175341 replaced direct scanning of directories with a two-step approach (first: create filename list, second: scan)
Mike Becker <universe@uap-core.de>
parents: 40
diff changeset
193 printf("%*s%*s%13d lines\n",
c2e73e175341 replaced direct scanning of directories with a two-step approach (first: create filename list, second: scan)
Mike Becker <universe@uap-core.de>
parents: 40
diff changeset
194 filelist->displayname_len+scanner.spaces, filelist->displayname,
c2e73e175341 replaced direct scanning of directories with a two-step approach (first: create filename list, second: scan)
Mike Becker <universe@uap-core.de>
parents: 40
diff changeset
195 60-filelist->displayname_len-scanner.spaces, "", lines);
3
510d6b198dde Moved some functions to functions.c
Mike Becker <universe@uap-core.de>
parents: 1
diff changeset
196 }
21
91e0890464b0 implemented bfile heuristics option + TODO: implement algorithm
Mike Becker <universe@uap-core.de>
parents: 20
diff changeset
197 }
16
bc9a0fefd892 fixed makefile to run safely on compile errors + added -V option to cline
Mike Becker <universe@uap-core.de>
parents: 14
diff changeset
198 } else {
3
510d6b198dde Moved some functions to functions.c
Mike Becker <universe@uap-core.de>
parents: 1
diff changeset
199 if (!settings->matchesOnly) {
22
4508da679ffb completed binary file heuristics
Mike Becker <universe@uap-core.de>
parents: 21
diff changeset
200 /* Print hint */
40
5938a9b74e8e improved printing the results and fixed bug where the filename was displayed as an empty string
Mike Becker <universe@uap-core.de>
parents: 36
diff changeset
201 printf("%*s%*s%19s\n",
41
c2e73e175341 replaced direct scanning of directories with a two-step approach (first: create filename list, second: scan)
Mike Becker <universe@uap-core.de>
parents: 40
diff changeset
202 filelist->displayname_len+scanner.spaces, filelist->displayname,
c2e73e175341 replaced direct scanning of directories with a two-step approach (first: create filename list, second: scan)
Mike Becker <universe@uap-core.de>
parents: 40
diff changeset
203 60-filelist->displayname_len-scanner.spaces, "", "no match");
3
510d6b198dde Moved some functions to functions.c
Mike Becker <universe@uap-core.de>
parents: 1
diff changeset
204 }
510d6b198dde Moved some functions to functions.c
Mike Becker <universe@uap-core.de>
parents: 1
diff changeset
205 }
510d6b198dde Moved some functions to functions.c
Mike Becker <universe@uap-core.de>
parents: 1
diff changeset
206 }
41
c2e73e175341 replaced direct scanning of directories with a two-step approach (first: create filename list, second: scan)
Mike Becker <universe@uap-core.de>
parents: 40
diff changeset
207
c2e73e175341 replaced direct scanning of directories with a two-step approach (first: create filename list, second: scan)
Mike Becker <universe@uap-core.de>
parents: 40
diff changeset
208 free(filelist->filename);
c2e73e175341 replaced direct scanning of directories with a two-step approach (first: create filename list, second: scan)
Mike Becker <universe@uap-core.de>
parents: 40
diff changeset
209 free(filelist->displayname);
c2e73e175341 replaced direct scanning of directories with a two-step approach (first: create filename list, second: scan)
Mike Becker <universe@uap-core.de>
parents: 40
diff changeset
210 filelist_t *freethis = filelist;
c2e73e175341 replaced direct scanning of directories with a two-step approach (first: create filename list, second: scan)
Mike Becker <universe@uap-core.de>
parents: 40
diff changeset
211 filelist = filelist->next;
c2e73e175341 replaced direct scanning of directories with a two-step approach (first: create filename list, second: scan)
Mike Becker <universe@uap-core.de>
parents: 40
diff changeset
212 free(freethis);
3
510d6b198dde Moved some functions to functions.c
Mike Becker <universe@uap-core.de>
parents: 1
diff changeset
213 }
23
778388400f7b encapsulated scanner arguments + enabled optimizer + empty file is no bfile
Mike Becker <universe@uap-core.de>
parents: 22
diff changeset
214
3
510d6b198dde Moved some functions to functions.c
Mike Becker <universe@uap-core.de>
parents: 1
diff changeset
215 return lineSum;
510d6b198dde Moved some functions to functions.c
Mike Becker <universe@uap-core.de>
parents: 1
diff changeset
216 }

mercurial