1 //========================================================================
2 //
3 // pdfseparate.cc
4 //
5 // This file is licensed under the GPLv2 or later
6 //
7 // Copyright (C) 2011, 2012, 2015 Thomas Freitag <Thomas.Freitag@alfa.de>
8 // Copyright (C) 2012-2014, 2017, 2018, 2021 Albert Astals Cid <aacid@kde.org>
9 // Copyright (C) 2013, 2016 Pino Toscano <pino@kde.org>
10 // Copyright (C) 2013 Daniel Kahn Gillmor <dkg@fifthhorseman.net>
11 // Copyright (C) 2013 Suzuki Toshiya <mpsuzuki@hiroshima-u.ac.jp>
12 // Copyright (C) 2017 Léonard Michelet <leonard.michelet@smile.fr>
13 // Copyright (C) 2017 Adrian Johnson <ajohnson@redneon.com>
14 // Copyright (C) 2018 Adam Reichold <adam.reichold@t-online.de>
15 // Copyright (C) 2019 Oliver Sander <oliver.sander@tu-dresden.de>
16 //
17 //========================================================================
18 #include "config.h"
19 #include <poppler-config.h>
20 #include <cstdio>
21 #include <cstdlib>
22 #include <cstddef>
23 #include <cstring>
24 #include "parseargs.h"
25 #include "goo/GooString.h"
26 #include "PDFDoc.h"
27 #include "ErrorCodes.h"
28 #include "GlobalParams.h"
29 #include "Win32Console.h"
30 #include <cctype>
31
32 static int firstPage = 0;
33 static int lastPage = 0;
34 static bool printVersion = false;
35 static bool printHelp = false;
36
37 static const ArgDesc argDesc[] = { { "-f", argInt, &firstPage, 0, "first page to extract" },
38 { "-l", argInt, &lastPage, 0, "last page to extract" },
39 { "-v", argFlag, &printVersion, 0, "print copyright and version info" },
40 { "-h", argFlag, &printHelp, 0, "print usage information" },
41 { "-help", argFlag, &printHelp, 0, "print usage information" },
42 { "--help", argFlag, &printHelp, 0, "print usage information" },
43 { "-?", argFlag, &printHelp, 0, "print usage information" },
44 {} };
45
extractPages(const char * srcFileName,const char * destFileName)46 static bool extractPages(const char *srcFileName, const char *destFileName)
47 {
48 char pathName[4096];
49 GooString *gfileName = new GooString(srcFileName);
50 PDFDoc *doc = new PDFDoc(gfileName, nullptr, nullptr, nullptr);
51
52 if (!doc->isOk()) {
53 error(errSyntaxError, -1, "Could not extract page(s) from damaged file ('{0:s}')", srcFileName);
54 delete doc;
55 return false;
56 }
57
58 // destFileName can have multiple %% and one %d
59 // We use auxDestFileName to replace all the valid % appearances
60 // by 'A' (random char that is not %), if at the end of replacing
61 // any of the valid appearances there is still any % around, the
62 // pattern is wrong
63 if (firstPage == 0 && lastPage == 0) {
64 firstPage = 1;
65 lastPage = doc->getNumPages();
66 }
67 if (lastPage == 0)
68 lastPage = doc->getNumPages();
69 if (firstPage == 0)
70 firstPage = 1;
71 if (lastPage < firstPage) {
72 error(errCommandLine, -1, "Wrong page range given: the first page ({0:d}) can not be after the last page ({1:d}).", firstPage, lastPage);
73 delete doc;
74 return false;
75 }
76 bool foundmatch = false;
77 char *auxDestFileName = strdup(destFileName);
78 char *p = strstr(auxDestFileName, "%d");
79 if (p != nullptr) {
80 foundmatch = true;
81 *p = 'A';
82 } else {
83 char pattern[6];
84 for (int i = 2; i < 10; i++) {
85 sprintf(pattern, "%%0%dd", i);
86 p = strstr(auxDestFileName, pattern);
87 if (p != nullptr) {
88 foundmatch = true;
89 *p = 'A';
90 break;
91 }
92 }
93 }
94 if (!foundmatch && firstPage != lastPage) {
95 error(errSyntaxError, -1, "'{0:s}' must contain '%d' (or any variant respecting printf format) if more than one page should be extracted, in order to print the page number", destFileName);
96 free(auxDestFileName);
97 delete doc;
98 return false;
99 }
100
101 // at this point auxDestFileName can only contain %%
102 p = strstr(auxDestFileName, "%%");
103 while (p != nullptr) {
104 *p = 'A';
105 *(p + 1) = 'A';
106 p = strstr(p, "%%");
107 }
108
109 // at this point any other % is wrong
110 p = strstr(auxDestFileName, "%");
111 if (p != nullptr) {
112 error(errSyntaxError, -1, "'{0:s}' can only contain one '%d' pattern", destFileName);
113 free(auxDestFileName);
114 delete doc;
115 return false;
116 }
117 free(auxDestFileName);
118
119 for (int pageNo = firstPage; pageNo <= lastPage; pageNo++) {
120 snprintf(pathName, sizeof(pathName) - 1, destFileName, pageNo);
121 GooString *gpageName = new GooString(pathName);
122 PDFDoc *pagedoc = new PDFDoc(new GooString(srcFileName), nullptr, nullptr, nullptr);
123 int errCode = pagedoc->savePageAs(gpageName, pageNo);
124 if (errCode != errNone) {
125 delete gpageName;
126 delete doc;
127 delete pagedoc;
128 return false;
129 }
130 delete pagedoc;
131 delete gpageName;
132 }
133 delete doc;
134 return true;
135 }
136
137 static constexpr int kOtherError = 99;
138
main(int argc,char * argv[])139 int main(int argc, char *argv[])
140 {
141 // parse args
142 Win32Console win32console(&argc, &argv);
143 const bool parseOK = parseArgs(argDesc, &argc, argv);
144 if (!parseOK || argc != 3 || printVersion || printHelp) {
145 fprintf(stderr, "pdfseparate version %s\n", PACKAGE_VERSION);
146 fprintf(stderr, "%s\n", popplerCopyright);
147 fprintf(stderr, "%s\n", xpdfCopyright);
148 if (!printVersion) {
149 printUsage("pdfseparate", "<PDF-sourcefile> <PDF-pattern-destfile>", argDesc);
150 }
151 if (printVersion || printHelp) {
152 return 0;
153 } else {
154 return kOtherError;
155 }
156 }
157 globalParams = std::make_unique<GlobalParams>();
158 const bool extractOK = extractPages(argv[1], argv[2]);
159 return extractOK ? 0 : kOtherError;
160 }
161