2 * Auxiliary functions for processing PDF files
4 * (c) 2018 Martin Mares <mj@ucw.cz>
15 #include "pdf-tools.h"
17 #include <qpdf/QUtil.hh>
18 #include <qpdf/Pl_Concatenate.hh>
25 void debug(const char *msg, ...)
31 fprintf(stderr, "%*s", debug_indent, "");
32 vfprintf(stderr, msg, args);
37 void warn(const char *msg, ...)
41 fprintf(stderr, "WARNING: ");
42 vfprintf(stderr, msg, args);
47 void die(const char *msg, ...)
51 fprintf(stderr, "ERROR: ");
52 vfprintf(stderr, msg, args);
58 void bad(const char *msg, ...)
63 vsnprintf(buf, sizeof(buf), msg, args);
66 printf("error: %s\n", buf);
70 /*** Transformation matrices ***/
72 // Construct string representation of a transformation matrix
73 string pdf_matrix::to_string() {
75 for (int i=0; i<6; i++) {
79 snprintf(buf, sizeof(buf), "%.3f", m[i]);
85 /*** Bounding boxes ***/
87 QPDFObjectHandle BBox::to_array()
89 QPDFObjectHandle a = QPDFObjectHandle::newArray();
90 a.appendItem(QPDFObjectHandle::newReal(x_min, 1));
91 a.appendItem(QPDFObjectHandle::newReal(y_min, 1));
92 a.appendItem(QPDFObjectHandle::newReal(x_max, 1));
93 a.appendItem(QPDFObjectHandle::newReal(y_max, 1));
97 string BBox::to_rect()
100 snprintf(buf, sizeof(buf), "%.1f %.1f %.1f %.1f", x_min, y_min, width(), height());
104 bool BBox::parse(QPDFObjectHandle h)
106 if (!h.isArray() || h.getArrayNItems() != 4)
109 for (int i=0; i<4; i++) {
110 QPDFObjectHandle item = h.getArrayItem(i);
111 if (!item.isNumber())
113 x[i] = item.getNumericValue();
122 void BBox::transform(pdf_matrix &m)
124 m.apply(&x_min, &y_min);
125 m.apply(&x_max, &y_max);
132 /*** Unicode strings ***/
134 // Construct PDF representation of a UTF-8 string
135 QPDFObjectHandle unicode_string(string s)
137 // If it is ASCII only, use the string directly
138 bool ascii_only = true;
140 if (c < 0x20 || c > 0x7e)
143 return QPDFObjectHandle::newString(s);
145 // Use iconv to convert the string to big-endian UTF-16
146 iconv_t conv = iconv_open("UTF-16BE", "UTF-8");
147 if (conv == (iconv_t) -1)
148 die("Cannot initialize iconv: %m");
150 char *in_ptr = (char *) s.c_str(); // Work around bad API of iconv()
151 size_t in_len = strlen(in_ptr);
152 size_t out_len = 2*in_len + 2; // Worst case (including the BOM)
153 char out_buf[out_len];
154 char *out_ptr = out_buf;
155 size_t res = iconv(conv, &in_ptr, &in_len, &out_ptr, &out_len);
156 if (res == (size_t) -1)
157 die("iconv failed: %m");
159 die("iconv stopped before the end of input");
163 // Package UTF-16 in a PDF string
167 for (char *p = out_buf; p < out_ptr; p++)
169 return QPDFObjectHandle::newString(out);
172 /*** Conversion of pages to XObjects ***/
174 static BBox get_trim_box(QPDFObjectHandle page)
176 static const char * const boxes[] = { "/TrimBox", "/CropBox", "/MediaBox", NULL };
177 for (int i=0; boxes[i]; i++)
178 if (page.hasKey(boxes[i]))
179 return BBox(page.getKey(boxes[i]));
180 warn("Page has no trimbox, falling back to A4");
181 return BBox(0, 0, A4_WIDTH, A4_HEIGHT);
184 /* Conversion of pages to XObjects is inspired by CUPS's pdftopdf filter. */
185 class CombineFromContents_Provider : public QPDFObjectHandle::StreamDataProvider {
187 vector<QPDFObjectHandle> contents;
189 CombineFromContents_Provider(const vector<QPDFObjectHandle> &contents) : contents(contents) { }
190 void provideStreamData(int objid UNUSED, int generation UNUSED, Pipeline* pipeline) {
191 Pl_Concatenate concat("concat", pipeline);
192 for (int i=0; i < (int)contents.size(); i++)
193 contents[i].pipeStreamData(&concat, true, false, false);
194 concat.manualFinish();
198 QPDFObjectHandle page_to_xobject(QPDF *out, QPDFObjectHandle page)
200 page.assertPageObject();
202 QPDFObjectHandle xo_stream = QPDFObjectHandle::newStream(out);
203 QPDFObjectHandle xo_dict = xo_stream.getDict();
205 xo_dict.replaceKey("/Type", QPDFObjectHandle::newName("/XObject"));
206 xo_dict.replaceKey("/Subtype", QPDFObjectHandle::newName("/Form"));
207 xo_dict.replaceKey("/FormType", QPDFObjectHandle::newInteger(1));
209 BBox box = get_trim_box(page);
210 xo_dict.replaceKey("/BBox", box.to_array());
212 xo_dict.replaceKey("/Resources", page.getKey("/Resources"));
213 if (page.hasKey("/Group"))
214 xo_dict.replaceKey("/Group", page.getKey("/Group"));
216 if (page.hasKey("/UserUnit")) {
217 double u = page.getKey("/UserUnit").getNumericValue();
218 QPDFObjectHandle m = QPDFObjectHandle::newArray();
219 m.appendItem(QPDFObjectHandle::newReal(u, 3));
220 m.appendItem(QPDFObjectHandle::newReal(0, 0));
221 m.appendItem(QPDFObjectHandle::newReal(0, 0));
222 m.appendItem(QPDFObjectHandle::newReal(u, 3));
223 m.appendItem(QPDFObjectHandle::newReal(0, 0));
224 m.appendItem(QPDFObjectHandle::newReal(0, 0));
225 xo_dict.replaceKey("/Matrix", m);
228 vector<QPDFObjectHandle> contents = page.getPageContents();
229 auto ph = PointerHolder<QPDFObjectHandle::StreamDataProvider>(new CombineFromContents_Provider(contents));
230 xo_stream.replaceStreamData(ph, QPDFObjectHandle::newNull(), QPDFObjectHandle::newNull());