Mercurial > hgrepos > Python2 > PyMuPDF

annotate mupdf-source/thirdparty/gumbo-parser/examples/clean_text.cc @ 29:f76e6575dca9 v1.26.4+1

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
+++++ v1.26.4+1
author Franz Glasner <fzglas.hg@dom66.de>
date Fri, 19 Sep 2025 19:59:23 +0200
parents b50eed0cc0ef
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
2
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
1 // Copyright 2013 Google Inc. All Rights Reserved.
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
2 //
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
3 // Licensed under the Apache License, Version 2.0 (the "License");
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
4 // you may not use this file except in compliance with the License.
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
5 // You may obtain a copy of the License at
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
6 //
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
7 // http://www.apache.org/licenses/LICENSE-2.0
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
8 //
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
9 // Unless required by applicable law or agreed to in writing, software
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
10 // distributed under the License is distributed on an "AS IS" BASIS,
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
11 // WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
12 // See the License for the specific language governing permissions and
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
13 // limitations under the License.
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
14 //
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
15 // Author: jdtang@google.com (Jonathan Tang)
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
16 //
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
17 // Gets the cleantext of a page.
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
18
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
19 #include <fstream>
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
20 #include <iostream>
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
21 #include <stdlib.h>
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
22 #include <string>
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
23
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
24 #include "gumbo.h"
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
25
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
26 static std::string cleantext(GumboNode* node) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
27 if (node->type == GUMBO_NODE_TEXT) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
28 return std::string(node->v.text.text);
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
29 } else if (node->type == GUMBO_NODE_ELEMENT &&
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
30 node->v.element.tag != GUMBO_TAG_SCRIPT &&
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
31 node->v.element.tag != GUMBO_TAG_STYLE) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
32 std::string contents = "";
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
33 GumboVector* children = &node->v.element.children;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
34 for (unsigned int i = 0; i < children->length; ++i) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
35 const std::string text = cleantext((GumboNode*) children->data[i]);
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
36 if (i != 0 && !text.empty()) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
37 contents.append(" ");
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
38 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
39 contents.append(text);
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
40 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
41 return contents;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
42 } else {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
43 return "";
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
44 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
45 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
46
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
47 int main(int argc, char** argv) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
48 if (argc != 2) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
49 std::cout << "Usage: clean_text <html filename>\n";
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
50 exit(EXIT_FAILURE);
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
51 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
52 const char* filename = argv[1];
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
53
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
54 std::ifstream in(filename, std::ios::in | std::ios::binary);
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
55 if (!in) {
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
56 std::cout << "File " << filename << " not found!\n";
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
57 exit(EXIT_FAILURE);
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
58 }
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
59
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
60 std::string contents;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
61 in.seekg(0, std::ios::end);
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
62 contents.resize(in.tellg());
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
63 in.seekg(0, std::ios::beg);
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
64 in.read(&contents[0], contents.size());
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
65 in.close();
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
66
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
67 GumboOutput* output = gumbo_parse(contents.c_str());
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
68 std::cout << cleantext(output->root) << std::endl;
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
69 gumbo_destroy_output(&kGumboDefaultOptions, output);
b50eed0cc0ef ADD: MuPDF v1.26.7: the MuPDF source as downloaded by a default build of PyMuPDF 1.26.4.
Franz Glasner <fzglas.hg@dom66.de>
parents:
diff changeset
70 }