use weasyprint to generate pdfs from html + css

NYCPlanning · Oct 2, 2024 · da37bc1 · da37bc1
1 parent f0a5edc
commit da37bc1
Show file tree

Hide file tree

Showing 3 changed files with 33 additions and 13 deletions.
diff --git a/dcpy/lifecycle/package/generate_metadata_assets.py b/dcpy/lifecycle/package/generate_metadata_assets.py
@@ -8,16 +8,21 @@
 DEFAULT_DATA_DICTIONARY_TEMPLATE_PATH = (
     RESOURCES_PATH / "data_dictionary_template.jinja"
 )
+DEFAULT_DATA_DICTIONARY_STYLESHEET_PATH = RESOURCES_PATH / "data_dictionary.css"
 
 
-def generate_pdf_from_html(output_html_path: Path, output_pdf_path: Path) -> Path:
+def generate_pdf_from_html(
+    output_html_path: Path,
+    output_pdf_path: Path,
+    stylesheet_path: Path,
+) -> Path:
     subprocess.run(
         [
-            "pandoc",
+            "weasyprint",
             output_html_path,
-            "-o",
             output_pdf_path,
-            "--pdf-engine=weasyprint",
+            "-s",
+            stylesheet_path,
         ],
         check=True,
     )

diff --git a/dcpy/lifecycle/package/resources/data_dictionary.css b/dcpy/lifecycle/package/resources/data_dictionary.css
@@ -0,0 +1,22 @@
+html {
+  color: #393939;
+  font-family: Fira Sans;
+  font-size: 11pt;
+  font-weight: 300;
+  line-height: 1.5;
+}
+h1 {
+  string-set: top-right-text content();
+}
+@page {
+  @top-left {
+    content: url("DCP_Logo.svg");
+  }
+  @top-right {
+    content: string(top-right-text);
+  }
+  @bottom-right-corner {
+    content: counter(page);
+    text-align: center;
+  }
+}
diff --git a/dcpy/test/lifecycle/package/test_generate_data_dictionary.py b/dcpy/test/lifecycle/package/test_generate_data_dictionary.py
@@ -17,15 +17,7 @@ class TestDataDictionary(TestCase):
     output_pdf_path = TEMP_DATA_PATH / "metadata.pdf"
     output_xlsx_path = TEMP_DATA_PATH / "my_data_dictionary.pdf"
 
-    def test_generate_html_from_yaml(self):
-        html_path = generate_metadata_assets.generate_html_from_yaml(
-            yaml_file_path=self.yaml_file_path,
-            output_html_path=self.output_html_path,
-            html_template_path=generate_metadata_assets.DEFAULT_DATA_DICTIONARY_TEMPLATE_PATH,
-        )
-        assert html_path.exists()
-
-    def test_generate_pdf_from_html(self):
+    def test_generate_pdf_from_yaml(self):
         html_path = generate_metadata_assets.generate_html_from_yaml(
             yaml_file_path=self.yaml_file_path,
             output_html_path=self.output_html_path,
@@ -34,6 +26,7 @@ def test_generate_pdf_from_html(self):
         pdf_path = generate_metadata_assets.generate_pdf_from_html(
             output_html_path=html_path,
             output_pdf_path=self.output_pdf_path,
+            stylesheet_path=generate_metadata_assets.DEFAULT_DATA_DICTIONARY_STYLESHEET_PATH,
         )
         assert pdf_path.exists()