From 5d41d5775c7d33503d57aacbf45390608c59ec3e Mon Sep 17 00:00:00 2001
From: Vladimir Timofeenko <id@vtimofeenko.com>
Date: Sat, 24 Dec 2022 15:42:32 -0800
Subject: [PATCH] nix: move to weasypdf from wkhtmltopdf
Closes #1873
---
nix/checks/solr.nix | 2 --
nix/dev-vm/default.nix | 2 +-
nix/modules/joex.nix | 64 ++++++++++++++++++++++++++++++++++++++++--
3 files changed, 63 insertions(+), 5 deletions(-)
diff --git a/nix/checks/solr.nix b/nix/checks/solr.nix
index 10150a42..f2171eab 100644
--- a/nix/checks/solr.nix
+++ b/nix/checks/solr.nix
@@ -6,8 +6,6 @@
nixpkgs.config = {
permittedInsecurePackages = [
"solr-8.6.3"
- # NOTE: Qtwebkit is a dep for wkhtmltopdf, this line is needed until #201765 is fixed in nixpkgs
- "qtwebkit-5.212.0-alpha4"
];
};
diff --git a/nix/dev-vm/default.nix b/nix/dev-vm/default.nix
index 6f4e93f6..01e172a1 100644
--- a/nix/dev-vm/default.nix
+++ b/nix/dev-vm/default.nix
@@ -16,7 +16,7 @@
};
# Otherwise oomkiller kills docspell
- virtualisation.memorySize = 2048;
+ virtualisation.memorySize = 4096;
virtualisation.forwardPorts = [
# SSH
diff --git a/nix/modules/joex.nix b/nix/modules/joex.nix
index f31d25fa..7890d690 100644
--- a/nix/modules/joex.nix
+++ b/nix/modules/joex.nix
@@ -178,8 +178,24 @@ let
wkhtmlpdf = {
command = {
- program = "${pkgs.wkhtmltopdf}/bin/wkhtmltopdf";
- args = [ "-s" "A4" "--encoding" "UTF-8" "-" "{{outfile}}" ];
+ program = "";
+ args = [ "--encoding" "UTF-8" "-" "{{outfile}}" ];
+ timeout = "2 minutes";
+ };
+ working-dir = "/tmp/docspell-convert";
+ };
+
+ weasyprint = {
+ command = {
+ program = "${pkgs.python310Packages.weasyprint}/bin/weasyprint";
+ args = [
+ "--optimize-size"
+ "all"
+ "--encoding"
+ "{{encoding}}"
+ "-"
+ "{{outfile}}"
+ ];
timeout = "2 minutes";
};
working-dir = "/tmp/docspell-convert";
@@ -1207,6 +1223,11 @@ in
converted to a PDF file.
'';
};
+ html-converter = mkOption {
+ type = types.enum [ "wkhtmlpdf" "weasyprint" ];
+ default = "weasyprint";
+ description = "Which tool to use for converting html to pdfs";
+ };
wkhtmlpdf = mkOption {
type = types.submodule ({
options = {
@@ -1246,6 +1267,45 @@ in
wkhtmltopdf is used.
'';
};
+ weasyprint = mkOption {
+ type = types.submodule ({
+ options = {
+ working-dir = mkOption {
+ type = types.str;
+ default = defaults.convert.weasyprint.working-dir;
+ description = "Directory where the conversion processes can put their temp files";
+ };
+ command = mkOption {
+ type = types.submodule ({
+ options = {
+ program = mkOption {
+ type = types.str;
+ default = defaults.convert.weasyprint.command.program;
+ description = "The path to the executable.";
+ };
+ args = mkOption {
+ type = types.listOf types.str;
+ default = defaults.convert.weasyprint.command.args;
+ description = "The arguments to the program";
+ };
+ timeout = mkOption {
+ type = types.str;
+ default = defaults.convert.weasyprint.command.timeout;
+ description = "The timeout when executing the command";
+ };
+ };
+ });
+ default = defaults.convert.weasyprint.command;
+ description = "The system command";
+ };
+ };
+ });
+ default = defaults.convert.weasyprint;
+ description = ''
+ To convert HTML files into PDF files, the external tool
+ weasyprint is used.
+ '';
+ };
tesseract = mkOption {
type = types.submodule ({
options = {