hdoc-tools 0.19.8 → 0.21.0

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
package/hdoc-build-pdf.js CHANGED
@@ -1,173 +1,219 @@
1
- (function () {
2
- 'use strict';
3
-
4
- const axios = require('axios'),
5
- cheerio = require('cheerio'),
6
- fs = require('fs-extra'),
7
- mime = require('mime-types'),
8
- path = require('path'),
9
- hdoc = require(path.join(__dirname, 'hdoc-module.js'));
10
-
11
- let hb_logo = '',
12
- footer = '',
13
- header = '';
14
-
15
- const get_footer = function (template_path) {
16
- let footer_content = null;
17
- try {
18
- footer_content = fs.readFileSync(path.join(template_path, 'template-footer.html'), 'utf8');
19
- } catch (err) {
20
- console.error(`Error loading template: ${err}`);
21
- }
22
- return footer_content;
23
- };
24
-
25
- const get_header = function (template_path) {
26
- let header_content = null;
27
- try {
28
- header_content = fs.readFileSync(path.join(template_path, 'template-header.html'), 'utf8');
29
- } catch (err) {
30
- console.error(`Error loading template: ${err}`);
31
- }
32
- return header_content;
33
- };
34
-
35
- exports.process_images = async function (file_path, html_source, verbose) {
36
- const book_work_root = file_path.path.replace(file_path.relativePath, '');
37
- if (verbose) console.log('Parsing img tags from HTML source');
38
-
39
- // Use cheerio to parse html
40
- const $ = cheerio.load(html_source);
41
-
42
- // Get iFrames from HTML, to replace with a tags
43
- let iframes = [];
44
- const iframe_html = $('iframe').map(function () {
45
- const response = {
46
- html: $.html(this),
47
- src: $(this).attr('src'),
48
- title: $(this).attr('title') ? $(this).attr('title') : 'No Link Title Provided'
49
- };
50
- return response;
51
- }).get();
52
- iframes.push(...iframe_html);
53
- for (let i = 0; i < iframes.length; i++) {
54
- const link = `<p><a href="${iframes[i].src}">${iframes[i].title}</a></p>`;
55
- const regex = new RegExp(`<iframe.*src="${iframes[i].src.replace('/', '\\/')}".*</iframe>`);
56
- html_source = html_source.replace(regex, link);
57
- }
58
-
59
- // Get image links from HTML, to embed into the pdf
60
- let imgs = [];
61
- const srcs = $('img').map(function (i) {
62
- return $(this).attr('src');
63
- }).get();
64
- imgs.push(...srcs);
65
- for (let i = 0; i < imgs.length; i++) {
66
- if (!hdoc.valid_url(imgs[i])) {
67
- // Internal link
68
- const image_path = path.join(book_work_root, imgs[i].replace('_books/', ''));
69
- try {
70
- const image_buffer = fs.readFileSync(image_path);
71
- const mime_type = mime.lookup(image_path);
72
- let image_b64 = image_buffer.toString("base64");
73
- image_b64 = `data:${mime_type};base64,${image_b64}`;
74
- html_source = html_source.replace(imgs[i], image_b64);
75
- } catch (err) {
76
- console.error('Error reading image from HTML source [', image_path, '] -', err);
77
- return null;
78
- }
79
- } else {
80
- // External Link
81
- try {
82
- const file_response = await axios.get(imgs[i]);
83
- if (file_response.status === 200) {
84
- const image_buffer = file_response.data;
85
- const mime_type = mime.lookup(imgs[i]);
86
- let image_b64 = image_buffer.toString("base64");
87
- image_b64 = `data:${mime_type};base64,${image_b64}`;
88
- html_source = html_source.replace(imgs[i], image_b64);
89
- } else {
90
- throw `Unexpected Status ${file_response.status}`;
91
- }
92
- } catch (err) {
93
- console.error(`Error downloading external source [${imgs[i]}] - ${err}`);
94
- }
95
- }
96
- }
97
-
98
- return html_source;
99
- };
100
-
101
- exports.generate_pdf = async function (browser, pdf_template_path, pdf_template_content, book_config, html_source, target_file, css_templates, verbose = false) {
102
- let pdf_size = 0;
103
-
104
- // Cache footer
105
- if (footer === '') footer = get_footer(pdf_template_path);
106
-
107
- // Read svg logo file into buffer, convert to B64 string
108
- if (hb_logo === '') {
109
- const hb_logo_path = path.join(pdf_template_path, 'images', 'hornbill-logo-full.svg');
110
- try {
111
- const hb_logo_file_buffer = fs.readFileSync(hb_logo_path);
112
- hb_logo = hb_logo_file_buffer.toString("base64");
113
- hb_logo = `data:image/svg+xml;base64,${hb_logo}`;
114
- } catch (err) {
115
- console.error('Error reading logo from template:', err);
116
- return pdf_size;
117
- }
118
- }
119
-
120
- // Cache header
121
- if (header === '') {
122
- header = get_header(pdf_template_path).replace('{{book_title}}', book_config.title).replace('{{hb_logo}}', hb_logo);
123
- }
124
-
125
- html_source = pdf_template_content.replace('{{book_title}}', book_config.title).replace('{{document_content}}', html_source);
126
-
127
- const page = await browser.newPage();
128
-
129
- // To reflect CSS used for screens instead of print
130
- await page.emulateMediaType('screen');
131
-
132
- // Set HTML content from HTML source
133
- await page.setContent(html_source, {
134
- waitUntil: 'domcontentloaded'
135
- });
136
- for (let i = 0; i < css_templates.length; i++) {
137
- try {
138
- await page.addStyleTag({
139
- content: css_templates[i]
140
- });
141
- } catch (e) {
142
- console.error(`Error applying template for [${target_file}]: ${e}`);
143
- }
144
- }
145
-
146
- try {
147
- const pdf_gen = await page.pdf({
148
- path: target_file,
149
- printBackground: true,
150
- format: 'A4',
151
- displayHeaderFooter: true,
152
- headerTemplate: header,
153
- footerTemplate: footer,
154
- margin: {
155
- top: "90px",
156
- right: "30px",
157
- bottom: "60px",
158
- left: "30px"
159
- },
160
- timeout: 0
161
- });
162
- let currdate = new Date;
163
- let datetime = currdate.toISOString();
164
- if (verbose) console.log(`[${datetime}] PDF generation success: ${target_file}`);
165
-
166
- pdf_size = pdf_gen.byteLength;
167
- } catch (err) {
168
- console.error(`Error generating PDF ${target_file} - ${err}`);
169
- }
170
- await page.close();
171
- return pdf_size;
172
- };
173
- })();
1
+ (() => {
2
+ const axios = require("axios");
3
+ const cheerio = require("cheerio");
4
+ const fs = require("fs-extra");
5
+ const mime = require("mime-types");
6
+ const path = require("node:path");
7
+ const hdoc = require(path.join(__dirname, "hdoc-module.js"));
8
+
9
+ let hb_logo = "";
10
+ let footer = "";
11
+ let header = "";
12
+
13
+ const get_footer = (template_path) => {
14
+ let footer_content = null;
15
+ try {
16
+ footer_content = fs.readFileSync(
17
+ path.join(template_path, "template-footer.html"),
18
+ "utf8",
19
+ );
20
+ } catch (err) {
21
+ console.error(`Error loading template: ${err}`);
22
+ }
23
+ return footer_content;
24
+ };
25
+
26
+ const get_header = (template_path) => {
27
+ let header_content = null;
28
+ try {
29
+ header_content = fs.readFileSync(
30
+ path.join(template_path, "template-header.html"),
31
+ "utf8",
32
+ );
33
+ } catch (err) {
34
+ console.error(`Error loading template: ${err}`);
35
+ }
36
+ return header_content;
37
+ };
38
+
39
+ exports.process_images = async (file_path, html_source, verbose) => {
40
+ const book_work_root = file_path.path.replace(file_path.relativePath, "");
41
+ if (verbose) console.log("Parsing img tags from HTML source");
42
+
43
+ let processed_html_source = html_source;
44
+ // Use cheerio to parse html
45
+ const $ = cheerio.load(processed_html_source);
46
+
47
+ // Get iFrames from HTML, to replace with a tags
48
+ const iframes = [];
49
+ const iframe_html = $("iframe")
50
+ .map(function () {
51
+ const response = {
52
+ html: $.html(this),
53
+ src: $(this).attr("src"),
54
+ title: $(this).attr("title")
55
+ ? $(this).attr("title")
56
+ : "No Link Title Provided",
57
+ };
58
+ return response;
59
+ })
60
+ .get();
61
+ iframes.push(...iframe_html);
62
+ for (let i = 0; i < iframes.length; i++) {
63
+ const link = `<p><a href="${iframes[i].src}">${iframes[i].title}</a></p>`;
64
+ const regex = new RegExp(
65
+ `<iframe.*src="${iframes[i].src.replace("/", "\\/")}".*</iframe>`,
66
+ );
67
+ processed_html_source = processed_html_source.replace(regex, link);
68
+ }
69
+
70
+ // Get image links from HTML, to embed into the pdf
71
+ const imgs = [];
72
+ const srcs = $("img")
73
+ .map(function (i) {
74
+ return $(this).attr("src");
75
+ })
76
+ .get();
77
+ imgs.push(...srcs);
78
+ for (let i = 0; i < imgs.length; i++) {
79
+ if (!hdoc.valid_url(imgs[i])) {
80
+ // Internal link
81
+ const image_path = path.join(
82
+ book_work_root,
83
+ imgs[i].replace("_books/", ""),
84
+ );
85
+ try {
86
+ const image_buffer = fs.readFileSync(image_path);
87
+ const mime_type = mime.lookup(image_path);
88
+ let image_b64 = image_buffer.toString("base64");
89
+ image_b64 = `data:${mime_type};base64,${image_b64}`;
90
+ processed_html_source = processed_html_source.replace(
91
+ imgs[i],
92
+ image_b64,
93
+ );
94
+ } catch (err) {
95
+ console.error(
96
+ "Error reading image from HTML source [",
97
+ image_path,
98
+ "] -",
99
+ err,
100
+ );
101
+ return null;
102
+ }
103
+ } else {
104
+ // External Link
105
+ try {
106
+ const file_response = await axios.get(imgs[i]);
107
+ if (file_response.status === 200) {
108
+ const image_buffer = file_response.data;
109
+ const mime_type = mime.lookup(imgs[i]);
110
+ let image_b64 = image_buffer.toString("base64");
111
+ image_b64 = `data:${mime_type};base64,${image_b64}`;
112
+ processed_html_source = processed_html_source.replace(
113
+ imgs[i],
114
+ image_b64,
115
+ );
116
+ } else {
117
+ throw `Unexpected Status ${file_response.status}`;
118
+ }
119
+ } catch (err) {
120
+ console.error(
121
+ `Error downloading external source [${imgs[i]}] - ${err}`,
122
+ );
123
+ }
124
+ }
125
+ }
126
+
127
+ return processed_html_source;
128
+ };
129
+
130
+ exports.generate_pdf = async (
131
+ browser,
132
+ pdf_template_path,
133
+ pdf_template_content,
134
+ book_config,
135
+ html_source,
136
+ target_file,
137
+ css_templates,
138
+ verbose = false,
139
+ ) => {
140
+ let pdf_size = 0;
141
+ // Cache footer
142
+ if (footer === "") footer = get_footer(pdf_template_path);
143
+
144
+ // Read svg logo file into buffer, convert to B64 string
145
+ if (hb_logo === "") {
146
+ const hb_logo_path = path.join(
147
+ pdf_template_path,
148
+ "images",
149
+ "hornbill-logo-full.svg",
150
+ );
151
+ try {
152
+ const hb_logo_file_buffer = fs.readFileSync(hb_logo_path);
153
+ hb_logo = hb_logo_file_buffer.toString("base64");
154
+ hb_logo = `data:image/svg+xml;base64,${hb_logo}`;
155
+ } catch (err) {
156
+ console.error("Error reading logo from template:", err);
157
+ return pdf_size;
158
+ }
159
+ }
160
+
161
+ // Cache header
162
+ if (header === "") {
163
+ header = get_header(pdf_template_path)
164
+ .replace("{{book_title}}", book_config.title)
165
+ .replace("{{hb_logo}}", hb_logo);
166
+ }
167
+
168
+ const processed_html_source = pdf_template_content
169
+ .replace("{{book_title}}", book_config.title)
170
+ .replace("{{document_content}}", html_source);
171
+
172
+ const page = await browser.newPage();
173
+
174
+ // To reflect CSS used for screens instead of print
175
+ await page.emulateMediaType("screen");
176
+
177
+ // Set HTML content from HTML source
178
+ await page.setContent(processed_html_source, {
179
+ waitUntil: "domcontentloaded",
180
+ });
181
+ for (let i = 0; i < css_templates.length; i++) {
182
+ try {
183
+ await page.addStyleTag({
184
+ content: css_templates[i],
185
+ });
186
+ } catch (e) {
187
+ console.error(`Error applying template for [${target_file}]: ${e}`);
188
+ }
189
+ }
190
+
191
+ try {
192
+ const pdf_gen = await page.pdf({
193
+ path: target_file,
194
+ printBackground: true,
195
+ format: "A4",
196
+ displayHeaderFooter: true,
197
+ headerTemplate: header,
198
+ footerTemplate: footer,
199
+ margin: {
200
+ top: "90px",
201
+ right: "30px",
202
+ bottom: "60px",
203
+ left: "30px",
204
+ },
205
+ timeout: 0,
206
+ });
207
+ const currdate = new Date();
208
+ const datetime = currdate.toISOString();
209
+ if (verbose)
210
+ console.log(`[${datetime}] PDF generation success: ${target_file}`);
211
+
212
+ pdf_size = pdf_gen.byteLength;
213
+ } catch (err) {
214
+ console.error(`Error generating PDF ${target_file} - ${err}`);
215
+ }
216
+ await page.close();
217
+ return pdf_size;
218
+ };
219
+ })();