pdftotext.cpp
1//========================================================================
2//
3// pdftotext.cpp
4//
5// Copyright 1997-2002 Glyph & Cog, LLC
6//
7//========================================================================
8
9#include <aconf.h>
10#include <stdio.h>
11#include <stdlib.h>
12#include <stddef.h>
13#include <string.h>
14#include "parseargs.h"
15#include "GString.h"
16#include "gmem.h"
17#include "GlobalParams.h"
18#include "Object.h"
19#include "Stream.h"
20#include "Array.h"
21#include "Dict.h"
22#include "XRef.h"
23#include "Catalog.h"
24#include "Page.h"
25#include "PDFDoc.h"
26#include "TextOutputDev.h"
27#include "CharTypes.h"
28#include "UnicodeMap.h"
29#include "Error.h"
30#include "config.h"
31
32static void printInfoString(FILE *f, Dict *infoDict, char *key,
33 char *text1, char *text2, UnicodeMap *uMap);
34static void printInfoDate(FILE *f, Dict *infoDict, char *key, char *fmt);
35
36static int firstPage = 1;
37static int lastPage = 0;
38static GBool rawOrder = gFalse;
39static GBool htmlMeta = gFalse;
40static char textEncName[128] = "";
41static char textEOL[16] = "";
42static char ownerPassword[33] = "";
43static char userPassword[33] = "";
44static GBool quiet = gFalse;
45static char cfgFileName[256] = "";
46static GBool printVersion = gFalse;
47static GBool printHelp = gFalse;
48
49static ArgDesc argDesc[] = {
50 {"-f", argInt, &firstPage, 0,
51 "first page to convert"},
52 {"-l", argInt, &lastPage, 0,
53 "last page to convert"},
54 {"-raw", argFlag, &rawOrder, 0,
55 "keep strings in content stream order"},
56 {"-htmlmeta", argFlag, &htmlMeta, 0,
57 "generate a simple HTML file, including the meta information"},
58 {"-enc", argString, textEncName, sizeof(textEncName),
59 "output text encoding name"},
60 {"-eol", argString, textEOL, sizeof(textEOL),
61 "output end-of-line convention (unix, dos, or mac)"},
62 {"-opw", argString, ownerPassword, sizeof(ownerPassword),
63 "owner password (for encrypted files)"},
64 {"-upw", argString, userPassword, sizeof(userPassword),
65 "user password (for encrypted files)"},
66 {"-q", argFlag, &quiet, 0,
67 "don't print any messages or errors"},
68 {"-cfg", argString, cfgFileName, sizeof(cfgFileName),
69 "configuration file to use in place of .xpdfrc"},
70 {"-v", argFlag, &printVersion, 0,
71 "print copyright and version info"},
72 {"-h", argFlag, &printHelp, 0,
73 "print usage information"},
74 {"-help", argFlag, &printHelp, 0,
75 "print usage information"},
76 {"--help", argFlag, &printHelp, 0,
77 "print usage information"},
78 {"-?", argFlag, &printHelp, 0,
79 "print usage information"},
80 {NULL}
81};
82
83int main(int argc, char *argv[]) {
84 PDFDoc *doc;
85 GString *fileName;
86 GString *textFileName;
87 GString *ownerPW, *userPW;
88 TextOutputDev *textOut;
89 FILE *f;
90 UnicodeMap *uMap;
91 Object info;
92 GBool ok;
93 char *p;
94 int exitCode;
95
96 exitCode = 99;
97
98 // parse args
99 ok = parseArgs(argDesc, &argc, argv);
100 if (!ok || argc < 2 || argc > 3 || printVersion || printHelp) {
101 fprintf(stderr, "pdftotext version %s\n", xpdfVersion);
102 fprintf(stderr, "%s\n", xpdfCopyright);
103 if (!printVersion) {
104 printUsage("pdftotext", "<PDF-file> [<text-file>]", argDesc);
105 }
106 goto err0;
107 }
108 fileName = new GString(argv[1]);
109
110 // read config file
111 globalParams = new GlobalParams(cfgFileName);
112 if (textEncName[0]) {
113 globalParams->setTextEncoding(textEncName);
114 }
115 if (textEOL[0]) {
116 if (!globalParams->setTextEOL(textEOL)) {
117 fprintf(stderr, "Bad '-eol' value on command line\n");
118 }
119 }
120 if (quiet) {
121 globalParams->setErrQuiet(quiet);
122 }
123
124 // get mapping to output encoding
125 if (!(uMap = globalParams->getTextEncoding())) {
126 error(-1, "Couldn't get text encoding");
127 delete fileName;
128 goto err1;
129 }
130
131 // open PDF file
132 if (ownerPassword[0]) {
133 ownerPW = new GString(ownerPassword);
134 } else {
135 ownerPW = NULL;
136 }
137 if (userPassword[0]) {
138 userPW = new GString(userPassword);
139 } else {
140 userPW = NULL;
141 }
142 doc = new PDFDoc(fileName, ownerPW, userPW);
143 if (userPW) {
144 delete userPW;
145 }
146 if (ownerPW) {
147 delete ownerPW;
148 }
149 if (!doc->isOk()) {
150 exitCode = 1;
151 goto err2;
152 }
153
154 // check for copy permission
155 if (!doc->okToCopy()) {
156 error(-1, "Copying of text from this document is not allowed.");
157 exitCode = 3;
158 goto err2;
159 }
160
161 // construct text file name
162 if (argc == 3) {
163 textFileName = new GString(argv[2]);
164 } else {
165 p = fileName->getCString() + fileName->getLength() - 4;
166 if (!strcmp(p, ".pdf") || !strcmp(p, ".PDF")) {
167 textFileName = new GString(fileName->getCString(),
168 fileName->getLength() - 4);
169 } else {
170 textFileName = fileName->copy();
171 }
172 textFileName->append(htmlMeta ? ".html" : ".txt");
173 }
174
175 // get page range
176 if (firstPage < 1) {
177 firstPage = 1;
178 }
179 if (lastPage < 1 || lastPage > doc->getNumPages()) {
180 lastPage = doc->getNumPages();
181 }
182
183 // write HTML header
184 if (htmlMeta) {
185 if (!textFileName->cmp("-")) {
186 f = stdout;
187 } else {
188 if (!(f = fopen(textFileName->getCString(), "wb"))) {
189 error(-1, "Couldn't open text file '%s'", textFileName->getCString());
190 exitCode = 2;
191 goto err3;
192 }
193 }
194 fputs("<html>\n", f);
195 fputs("<head>\n", f);
196 doc->getDocInfo(&info);
197 if (info.isDict()) {
198 printInfoString(f, info.getDict(), "Title", "<title>", "</title>\n",
199 uMap);
200 printInfoString(f, info.getDict(), "Subject",
201 "<meta name=\"Subject\" content=\"", "\">\n", uMap);
202 printInfoString(f, info.getDict(), "Keywords",
203 "<meta name=\"Keywords\" content=\"", "\">\n", uMap);
204 printInfoString(f, info.getDict(), "Author",
205 "<meta name=\"Author\" content=\"", "\">\n", uMap);
206 printInfoString(f, info.getDict(), "Creator",
207 "<meta name=\"Creator\" content=\"", "\">\n", uMap);
208 printInfoString(f, info.getDict(), "Producer",
209 "<meta name=\"Producer\" content=\"", "\">\n", uMap);
210 printInfoDate(f, info.getDict(), "CreationDate",
211 "<meta name=\"CreationDate\" content=\"\">\n");
212 printInfoDate(f, info.getDict(), "LastModifiedDate",
213 "<meta name=\"ModDate\" content=\"\">\n");
214 }
215 info.free();
216 fputs("</head>\n", f);
217 fputs("<body>\n", f);
218 fputs("<pre>\n", f);
219 if (f != stdout) {
220 fclose(f);
221 }
222 }
223
224 // write text file
225 textOut = new TextOutputDev(textFileName->getCString(), rawOrder, htmlMeta);
226 if (textOut->isOk()) {
227 doc->displayPages(textOut, firstPage, lastPage, 72, 0, gFalse);
228 } else {
229 delete textOut;
230 exitCode = 2;
231 goto err3;
232 }
233 delete textOut;
234
235 // write end of HTML file
236 if (htmlMeta) {
237 if (!textFileName->cmp("-")) {
238 f = stdout;
239 } else {
240 if (!(f = fopen(textFileName->getCString(), "ab"))) {
241 error(-1, "Couldn't open text file '%s'", textFileName->getCString());
242 exitCode = 2;
243 goto err3;
244 }
245 }
246 fputs("</pre>\n", f);
247 fputs("</body>\n", f);
248 fputs("</html>\n", f);
249 if (f != stdout) {
250 fclose(f);
251 }
252 }
253
254 exitCode = 0;
255
256 // clean up
257 err3:
258 delete textFileName;
259 err2:
260 delete doc;
261 uMap->decRefCnt();
262 err1:
263 delete globalParams;
264 err0:
265
266 // check for memory leaks
267 Object::memCheck(stderr);
268 gMemReport(stderr);
269
270 return exitCode;
271}
272
273static void printInfoString(FILE *f, Dict *infoDict, char *key,
274 char *text1, char *text2, UnicodeMap *uMap) {
275 Object obj;
276 GString *s1;
277 GBool isUnicode;
278 Unicode u;
279 char buf[8];
280 int i, n;
281
282 if (infoDict->lookup(key, &obj)->isString()) {
283 fputs(text1, f);
284 s1 = obj.getString();
285 if ((s1->getChar(0) & 0xff) == 0xfe &&
286 (s1->getChar(1) & 0xff) == 0xff) {
287 isUnicode = gTrue;
288 i = 2;
289 } else {
290 isUnicode = gFalse;
291 i = 0;
292 }
293 while (i < obj.getString()->getLength()) {
294 if (isUnicode) {
295 u = ((s1->getChar(i) & 0xff) << 8) |
296 (s1->getChar(i+1) & 0xff);
297 i += 2;
298 } else {
299 u = s1->getChar(i) & 0xff;
300 ++i;
301 }
302 n = uMap->mapUnicode(u, buf, sizeof(buf));
303 fwrite(buf, 1, n, f);
304 }
305 fputs(text2, f);
306 }
307 obj.free();
308}
309
310static void printInfoDate(FILE *f, Dict *infoDict, char *key, char *fmt) {
311 Object obj;
312 char *s;
313
314 if (infoDict->lookup(key, &obj)->isString()) {
315 s = obj.getString()->getCString();
316 if (s[0] == 'D' && s[1] == ':') {
317 s += 2;
318 }
319 fprintf(f, fmt, s);
320 }
321 obj.free();
322}