Merge branch 'feature/custom_css'

docs/man.md

@@ -71,6 +71,18 @@ reMarkable options:
       If the target directory does not exist it will be created. If not 
       specified, the root directory will be used.
 
+Output customization:
+
+--css=FILENAME
+      Path to a CSS file with custom styling for the HTML output. This option 
+      is ignored for any of the other providers. The code for the HTML 
+      provider contains the default CSS style, which can be used as a starting 
+      point.
+
+--font-urls=FILENAME
+      Path to a file with font urls (one per line) for the HTML output. This 
+      will generally be used in combination with the ``--css`` option.
+
 System settings:
 
 You'll only need to specify these options if the programs are not available on 

paper2remarkable/log.py

@@ -9,8 +9,8 @@
 """
 
 # NOTE: I know about the logging module, but this was easier because one of the
-# dependencies was using that and it became complicated. This one is obviously
-# not thread-safe and is very simple.
+# dependencies was using that and it interfered with our logging. The logger
+# class below is obviously not thread-safe and is very simple.
 
 import datetime
 import sys

paper2remarkable/providers/_base.py

@@ -49,6 +49,8 @@ def __init__(
         pdftk_path="pdftk",
         qpdf_path="qpdf",
         gs_path="gs",
+        css_path=None,
+        font_urls_path=None,
         cookiejar=None,
     ):
         self.upload = upload
@@ -60,9 +62,12 @@ def __init__(
         self.pdftk_path = pdftk_path
         self.qpdf_path = qpdf_path
         self.gs_path = gs_path
-        self.informer = Informer()
+        self.css_path = css_path
+        self.font_urls_path = font_urls_path
         self.cookiejar = cookiejar
 
+        self.informer = Informer()
+
         self.pdftool = check_pdftool(self.pdftk_path, self.qpdf_path)
 
         # wait time to not hit the server too frequently

paper2remarkable/providers/html.py

@@ -13,13 +13,13 @@
 
 import html2text
 import markdown
+import os
 import re
 import readability
 import titlecase
 import unidecode
 import urllib
 import weasyprint
-import weasyprint.fonts
 
 from ._base import Provider
 from ._info import Informer
@@ -34,7 +34,6 @@
 logger = Logger()
 
 CSS = """
-@import url('https://fonts.googleapis.com/css?family=EB+Garamond|Noto+Serif|Inconsolata&display=swap');
 @page { size: 702px 936px; margin: 1in; }
 a { color: black; }
 img { display: block; margin: 0 auto; text-align: center; max-width: 70%; max-height: 300px; }
@@ -48,6 +47,13 @@
 code { font-family: 'Inconsolata'; font-size: .7rem; background: #efefef; }
 """
 
+# NOTE: For some reason, Weasyprint no longer accepts the @import statement in
+# the CSS to load the fonts. This may have to do with recent changes they've
+# introduced. Providing the font urls separately does seem to work.
+FONT_URLS = [
+    "https://fonts.googleapis.com/css2?family=EB+Garamond&family=Noto+Serif&family=Inconsolata"
+]
+
 
 def url_fetcher(url):
     if url.startswith("//"):
@@ -168,6 +174,30 @@ def preprocess_html(self, pdf_url, title, article):
         html_article = md.convert(article)
         return html_article
 
+    def get_css(self):
+        if self.css_path is None:
+            return CSS
+        if not os.path.exists(self.css_path):
+            logger.warning(
+                f"CSS file {self.css_path} doesn't exist, using default style."
+            )
+            return CSS
+        with open(self.css_path, "r") as fp:
+            css = fp.read()
+        return css
+
+    def get_font_urls(self):
+        if self.font_urls_path is None:
+            return FONT_URLS
+        if not os.path.exists(self.font_urls_path):
+            logger.warning(
+                f"Font urls file {self.font_urls_path} doesn't exist, using default."
+            )
+            return FONT_URLS
+        with open(self.font_urls_path, "r") as fp:
+            font_urls = [l.strip() for l in fp.read().split("\n")]
+        return font_urls
+
     def retrieve_pdf(self, pdf_url, filename):
         """Turn the HTML article in a clean pdf file
 
@@ -193,11 +223,11 @@ def retrieve_pdf(self, pdf_url, filename):
             with open("./paper.html", "w") as fp:
                 fp.write(html_article)
 
-        font_config = weasyprint.fonts.FontConfiguration()
         html = weasyprint.HTML(string=html_article, url_fetcher=url_fetcher)
-        css = weasyprint.CSS(string=CSS, font_config=font_config)
-
-        html.write_pdf(filename, stylesheets=[css], font_config=font_config)
+        css = self.get_css()
+        font_urls = self.get_font_urls()
+        style = weasyprint.CSS(string=css)
+        html.write_pdf(filename, stylesheets=[style] + font_urls)
 
     def validate(src):
         # first check if it is a valid url

paper2remarkable/ui.py

@@ -107,6 +107,14 @@ def parse_args():
         help="path to rmapi executable (default: rmapi)",
         default="rmapi",
     )
+    parser.add_argument(
+        "--css", help="path to custom CSS file for HTML output", default=None
+    )
+    parser.add_argument(
+        "--font-urls",
+        help="path to custom font urls file for HTML output",
+        default=None,
+    )
     parser.add_argument(
         "input",
         help="One or more URLs to a paper or paths to local PDF files",
@@ -229,6 +237,8 @@ def main():
             pdftk_path=args.pdftk,
             qpdf_path=args.qpdf,
             gs_path=args.gs,
+            css_path=args.css,
+            font_urls_path=args.font_urls,
             cookiejar=cookiejar,
         )
         prov.run(new_input, filename=filename)

tests/test_html.py

@@ -7,6 +7,9 @@
 
 """
 
+import os
+import pdfplumber
+import tempfile
 import unittest
 
 from paper2remarkable.providers.html import HTML
@@ -24,6 +27,38 @@ def test_experimental_fix_lazy_loading(self):
         expected_image = "https://www.seriouseats.com/images/2015/01/20150118-tea-max-falkowitz-3.jpg"
         self.assertIn(expected_image, html_article)
 
+    def test_custom_css(self):
+        test_css = """
+        @page { size: 702px 936px; margin: 1in; }
+        img { display: block; margin: 0 auto; text-align: center; max-width: 70%; max-height: 300px; }
+        h1,h2,h3 { font-family: 'Montserrat'; }
+        p, li { font-size: 12pt; line-height: 2; font-family: 'Montserrat'; text-align: left; }
+        """
+
+        test_font_urls = [
+            "https://fonts.googleapis.com/css2?family=Montserrat&display=swap"
+        ]
+
+        tmpfd, tempfname_css = tempfile.mkstemp(prefix="p2r_", suffix=".css")
+        with os.fdopen(tmpfd, "w") as fp:
+            fp.write(test_css)
+
+        tmpfd, tempfname_urls = tempfile.mkstemp(prefix="p2r_", suffix=".txt")
+        with os.fdopen(tmpfd, "w") as fp:
+            fp.write("\n".join(test_font_urls))
+
+        url = "https://hbr.org/2019/11/getting-your-team-to-do-more-than-meet-deadlines"
+        prov = HTML(
+            upload=False, css_path=tempfname_css, font_urls_path=tempfname_urls
+        )
+        filename = prov.run(url)
+        with pdfplumber.open(filename) as pdf:
+            self.assertEqual(8, len(pdf.pages))
+
+        os.unlink(tempfname_css)
+        os.unlink(tempfname_urls)
+        os.unlink(filename)
+
 
 if __name__ == "__main__":
     unittest.main()