summaryrefslogtreecommitdiff
path: root/xpdf/pdftotext.cc
diff options
context:
space:
mode:
Diffstat (limited to 'xpdf/pdftotext.cc')
-rw-r--r--xpdf/pdftotext.cc189
1 files changed, 56 insertions, 133 deletions
diff --git a/xpdf/pdftotext.cc b/xpdf/pdftotext.cc
index 5296ac4..758413e 100644
--- a/xpdf/pdftotext.cc
+++ b/xpdf/pdftotext.cc
@@ -2,7 +2,7 @@
//
// pdftotext.cc
//
-// Copyright 1997-2003 Glyph & Cog, LLC
+// Copyright 1997-2013 Glyph & Cog, LLC
//
//========================================================================
@@ -11,6 +11,10 @@
#include <stdlib.h>
#include <stddef.h>
#include <string.h>
+#ifdef DEBUG_FP_LINUX
+# include <fenv.h>
+# include <fpu_control.h>
+#endif
#include "parseargs.h"
#include "GString.h"
#include "gmem.h"
@@ -26,21 +30,19 @@
#include "TextOutputDev.h"
#include "CharTypes.h"
#include "UnicodeMap.h"
+#include "TextString.h"
#include "Error.h"
#include "config.h"
-static void printInfoString(FILE *f, Dict *infoDict, const char *key,
- const char *text1, const char *text2,
- UnicodeMap *uMap);
-static void printInfoDate(FILE *f, Dict *infoDict, const char *key,
- const char *fmt);
-
static int firstPage = 1;
static int lastPage = 0;
static GBool physLayout = gFalse;
-static double fixedPitch = 0;
+static GBool tableLayout = gFalse;
+static GBool linePrinter = gFalse;
static GBool rawOrder = gFalse;
-static GBool htmlMeta = gFalse;
+static double fixedPitch = 0;
+static double fixedLineSpacing = 0;
+static GBool clipText = gFalse;
static char textEncName[128] = "";
static char textEOL[16] = "";
static GBool noPageBreaks = gFalse;
@@ -58,12 +60,18 @@ static ArgDesc argDesc[] = {
"last page to convert"},
{"-layout", argFlag, &physLayout, 0,
"maintain original physical layout"},
- {"-fixed", argFP, &fixedPitch, 0,
- "assume fixed-pitch (or tabular) text"},
+ {"-table", argFlag, &tableLayout, 0,
+ "similar to -layout, but optimized for tables"},
+ {"-lineprinter", argFlag, &linePrinter, 0,
+ "use strict fixed-pitch/height layout"},
{"-raw", argFlag, &rawOrder, 0,
"keep strings in content stream order"},
- {"-htmlmeta", argFlag, &htmlMeta, 0,
- "generate a simple HTML file, including the meta information"},
+ {"-fixed", argFP, &fixedPitch, 0,
+ "assume fixed-pitch (or tabular) text"},
+ {"-linespacing", argFP, &fixedLineSpacing, 0,
+ "fixed line spacing for LinePrinter mode"},
+ {"-clip", argFlag, &clipText, 0,
+ "separate clipped text"},
{"-enc", argString, textEncName, sizeof(textEncName),
"output text encoding name"},
{"-eol", argString, textEOL, sizeof(textEOL),
@@ -96,14 +104,29 @@ int main(int argc, char *argv[]) {
GString *fileName;
GString *textFileName;
GString *ownerPW, *userPW;
+ TextOutputControl textOutControl;
TextOutputDev *textOut;
- FILE *f;
UnicodeMap *uMap;
Object info;
GBool ok;
char *p;
int exitCode;
+#ifdef DEBUG_FP_LINUX
+ // enable exceptions on floating point div-by-zero
+ feenableexcept(FE_DIVBYZERO);
+ // force 64-bit rounding: this avoids changes in output when minor
+ // code changes result in spills of x87 registers; it also avoids
+ // differences in output with valgrind's 64-bit floating point
+ // emulation (yes, this is a kludge; but it's pretty much
+ // unavoidable given the x87 instruction set; see gcc bug 323 for
+ // more info)
+ fpu_control_t cw;
+ _FPU_GETCW(cw);
+ cw = (cw & ~_FPU_EXTENDED) | _FPU_DOUBLE;
+ _FPU_SETCW(cw);
+#endif
+
exitCode = 99;
// parse args
@@ -117,9 +140,6 @@ int main(int argc, char *argv[]) {
goto err0;
}
fileName = new GString(argv[1]);
- if (fixedPitch) {
- physLayout = gTrue;
- }
// read config file
globalParams = new GlobalParams(cfgFileName);
@@ -187,7 +207,7 @@ int main(int argc, char *argv[]) {
} else {
textFileName = fileName->copy();
}
- textFileName->append(htmlMeta ? ".html" : ".txt");
+ textFileName->append(".txt");
}
// get page range
@@ -198,50 +218,25 @@ int main(int argc, char *argv[]) {
lastPage = doc->getNumPages();
}
- // write HTML header
- if (htmlMeta) {
- if (!textFileName->cmp("-")) {
- f = stdout;
- } else {
- if (!(f = fopen(textFileName->getCString(), "wb"))) {
- error(errIO, -1, "Couldn't open text file '{0:t}'", textFileName);
- exitCode = 2;
- goto err3;
- }
- }
- fputs("<html>\n", f);
- fputs("<head>\n", f);
- doc->getDocInfo(&info);
- if (info.isDict()) {
- printInfoString(f, info.getDict(), "Title", "<title>", "</title>\n",
- uMap);
- printInfoString(f, info.getDict(), "Subject",
- "<meta name=\"Subject\" content=\"", "\">\n", uMap);
- printInfoString(f, info.getDict(), "Keywords",
- "<meta name=\"Keywords\" content=\"", "\">\n", uMap);
- printInfoString(f, info.getDict(), "Author",
- "<meta name=\"Author\" content=\"", "\">\n", uMap);
- printInfoString(f, info.getDict(), "Creator",
- "<meta name=\"Creator\" content=\"", "\">\n", uMap);
- printInfoString(f, info.getDict(), "Producer",
- "<meta name=\"Producer\" content=\"", "\">\n", uMap);
- printInfoDate(f, info.getDict(), "CreationDate",
- "<meta name=\"CreationDate\" content=\"%s\">\n");
- printInfoDate(f, info.getDict(), "LastModifiedDate",
- "<meta name=\"ModDate\" content=\"%s\">\n");
- }
- info.free();
- fputs("</head>\n", f);
- fputs("<body>\n", f);
- fputs("<pre>\n", f);
- if (f != stdout) {
- fclose(f);
- }
- }
-
// write text file
- textOut = new TextOutputDev(textFileName->getCString(),
- physLayout, fixedPitch, rawOrder, htmlMeta);
+ if (tableLayout) {
+ textOutControl.mode = textOutTableLayout;
+ textOutControl.fixedPitch = fixedPitch;
+ } else if (physLayout) {
+ textOutControl.mode = textOutPhysLayout;
+ textOutControl.fixedPitch = fixedPitch;
+ } else if (linePrinter) {
+ textOutControl.mode = textOutLinePrinter;
+ textOutControl.fixedPitch = fixedPitch;
+ textOutControl.fixedLineSpacing = fixedLineSpacing;
+ } else if (rawOrder) {
+ textOutControl.mode = textOutRawOrder;
+ } else {
+ textOutControl.mode = textOutReadingOrder;
+ }
+ textOutControl.clipText = clipText;
+ textOut = new TextOutputDev(textFileName->getCString(), &textOutControl,
+ gFalse);
if (textOut->isOk()) {
doc->displayPages(textOut, firstPage, lastPage, 72, 72, 0,
gFalse, gTrue, gFalse);
@@ -252,25 +247,6 @@ int main(int argc, char *argv[]) {
}
delete textOut;
- // write end of HTML file
- if (htmlMeta) {
- if (!textFileName->cmp("-")) {
- f = stdout;
- } else {
- if (!(f = fopen(textFileName->getCString(), "ab"))) {
- error(errIO, -1, "Couldn't open text file '{0:t}'", textFileName);
- exitCode = 2;
- goto err3;
- }
- }
- fputs("</pre>\n", f);
- fputs("</body>\n", f);
- fputs("</html>\n", f);
- if (f != stdout) {
- fclose(f);
- }
- }
-
exitCode = 0;
// clean up
@@ -289,56 +265,3 @@ int main(int argc, char *argv[]) {
return exitCode;
}
-
-static void printInfoString(FILE *f, Dict *infoDict, const char *key,
- const char *text1, const char *text2,
- UnicodeMap *uMap) {
- Object obj;
- GString *s1;
- GBool isUnicode;
- Unicode u;
- char buf[8];
- int i, n;
-
- if (infoDict->lookup(key, &obj)->isString()) {
- fputs(text1, f);
- s1 = obj.getString();
- if ((s1->getChar(0) & 0xff) == 0xfe &&
- (s1->getChar(1) & 0xff) == 0xff) {
- isUnicode = gTrue;
- i = 2;
- } else {
- isUnicode = gFalse;
- i = 0;
- }
- while (i < obj.getString()->getLength()) {
- if (isUnicode) {
- u = ((s1->getChar(i) & 0xff) << 8) |
- (s1->getChar(i+1) & 0xff);
- i += 2;
- } else {
- u = s1->getChar(i) & 0xff;
- ++i;
- }
- n = uMap->mapUnicode(u, buf, sizeof(buf));
- fwrite(buf, 1, n, f);
- }
- fputs(text2, f);
- }
- obj.free();
-}
-
-static void printInfoDate(FILE *f, Dict *infoDict, const char *key,
- const char *fmt) {
- Object obj;
- char *s;
-
- if (infoDict->lookup(key, &obj)->isString()) {
- s = obj.getString()->getCString();
- if (s[0] == 'D' && s[1] == ':') {
- s += 2;
- }
- fprintf(f, fmt, s);
- }
- obj.free();
-}
generated by cgit v1.2.3 (git 2.39.1) at 2024-06-07 16:16:13 +0000