Initial commit

Author: eikek

.envrc

@@ -0,0 +1,11 @@
+#! /bin/sh
+
+# reload when these files change
+watch_file flake.nix
+watch_file flake.lock
+
+{
+  # shell gc root dir
+  mkdir -p "$(direnv_layout_dir)"
+  eval "$(nix print-dev-env --profile $(direnv_layout_dir)/flake-profile)"
+}

.gitignore

@@ -0,0 +1,4 @@
+.direnv/
+target/
+test/tmp/
+result

Dockerfile

@@ -0,0 +1,16 @@
+FROM debian:bookworm-slim
+
+RUN apt-get -y update && \
+    apt-get -y upgrade && \
+    apt-get install -y guile-2.2 guile-json qpdf poppler-utils
+
+RUN mkdir -p /opt/addon
+COPY src/addon.scm /opt/addon/
+RUN chmod 755 /opt/addon/addon.scm
+
+# test
+RUN qpdf --version && pdftotext -h
+
+# let it compile into the compile cache
+RUN /opt/addon/addon.scm || true
+ENTRYPOINT [ "/opt/addon/addon.scm" ]

LICENSE.txt

@@ -0,0 +1,56 @@
+# Rotate PDFs
+
+This is a simple addon for [Docspell](https://docspell.org) that
+allows to rotate attached pdfs files.
+
+
+## Prerequisites
+
+This addon supports these runners: `nix-flake`, `docker` and
+`trivial`.
+
+It is recommended to install [nix](https://nixos.org) on the machine
+running joex. This allows to use the `nix-flake` runner which can
+build the addon with all dependencies automatically.
+
+Otherwise, for the trivial runner, you need to install these tools
+manually: qpdf and pdftotext (might be provided by poppler-utils or
+xpdf)
+
+
+## Usage
+
+It expects arguments as a json file to know how to rotate. It is
+currently very basic, you can only set the degree to rotate and it
+applies it to all pdfs.
+
+``` json
+{ "degree": "90" }
+```
+
+Have a look at the
+[qpdf](https://qpdf.readthedocs.io/en/stable/cli.html#option-rotate)
+manual for possible values.
+
+
+## Testing
+
+Install [direnv](https://direnv.net/) and [nix](https://nixos.org) and
+allow the source root via `direnv allow`. This applies the `devShell`
+settings from `flake.nix`. Then build the addon:
+
+```
+nix build
+```
+
+Now you can run it:
+
+```
+./result/bin/rotate-pdf-addon
+```
+
+It will run on the test files provided in `test/` and put results in
+`test/tmp`.
+
+For quicker turnaround you can also run the source file itself. This
+works, because `devShell` puts all required binaries in path.

README.md

@@ -0,0 +1,36 @@
+meta:
+  name: "rotate-pdf-addon"
+  version: "0.1.0"
+  description: |
+    Allows to rotate pdfs.
+
+    Please specify the angle for rotation via a json document.
+    Example:
+
+    ```
+    { "degree": "90" }
+    ```
+
+    *Note that this is applied to ALL attachments!*
+
+    This is rotating "right". The possible values for `degree` can be
+    looked up in the [qpdf
+    manual](https://qpdf.readthedocs.io/en/stable/cli.html#option-rotate).
+
+triggers:
+  - existing-item
+
+runner:
+  nix:
+    enable: true
+
+  docker:
+    enable: true
+
+  trivial:
+    enable: true
+    exec: src/addon.scm
+
+options:
+  networking: false
+  collectOutput: true

docspell-addon.yml

@@ -0,0 +1,54 @@
+{
+  description = "A docspell addon for basic audio file support";
+
+  inputs = {
+    utils.url = "github:numtide/flake-utils";
+
+    # Nixpkgs / NixOS version to use.
+    nixpkgs.url = "nixpkgs/nixos-21.11";
+  };
+
+  outputs = { self, nixpkgs, utils }:
+    utils.lib.eachDefaultSystem (system:
+      let
+        pkgs = import nixpkgs {
+          inherit system;
+          overlays = [
+
+          ];
+        };
+        name = "rotate-pdf-addon";
+      in rec {
+        packages.${name} = pkgs.callPackage ./nix/addon.nix {
+          inherit name;
+        };
+
+        defaultPackage = packages.${name};
+
+        apps.${name} = utils.lib.mkApp {
+          inherit name;
+          drv = packages.${name};
+        };
+        defaultApp = apps.${name};
+
+        devShell = pkgs.mkShell {
+          inputsFrom = builtins.attrValues self.packages.${system};
+          buildInputs =
+            [ pkgs.guile
+              pkgs.guile-json
+              pkgs.qpdf
+              pkgs.poppler_utils
+            ];
+
+          ADDON_DIR = self;
+          TMPDIR = "/tmp";
+          ITEM_DATA_JSON="test/item_data.json";
+          ITEM_PDF_DIR="test/pdf";
+          TMP_DIR="test/tmp";
+          CACHE_DIR="test/tmp";
+          OUTPUT_DIR="test/tmp";
+          GUILE_WARN_DEPRECATED="detailed";
+        };
+      }
+    );
+}

flake.lock

@@ -0,0 +1,30 @@
+{ stdenv, bash, qpdf, poppler_utils, guile, guile-json, lib, name }:
+
+stdenv.mkDerivation {
+  inherit name;
+  src = lib.sources.cleanSource ../.;
+
+  buildInputs = [ guile guile-json ];
+
+  patchPhase = ''
+    TARGET=src/addon.scm
+    sed -i 's,\*qpdf\* "qpdf",\*qpdf\* "${qpdf}/bin/qpdf",g' $TARGET
+    sed -i 's,\*pdftotext\* "pdftotext",\*pdftotext\* "${poppler_utils}/bin/pdftotext",g' $TARGET
+  '';
+
+  buildPhase = ''
+    guild compile -o ${name}.go src/addon.scm
+  '';
+
+  # module name must be same as <filename>.go
+  installPhase = ''
+    mkdir -p $out/{bin,lib}
+    cp ${name}.go $out/lib/
+
+    cat > $out/bin/${name} <<-EOF
+    #!${bash}/bin/bash
+    exec -a "${name}" ${guile}/bin/guile -C ${guile-json}/share/guile/ccache -C $out/lib -e '(${name}) main' -c "" \$@
+    EOF
+    chmod +x $out/bin/${name}
+  '';
+}

flake.nix

@@ -0,0 +1,109 @@
+#!/bin/sh
+exec guile -e '(rotate-pdf-addon) main' -s $0 "$@"
+!#
+;; A simple addon for Docspell for rotating pdf files.
+;;
+;; It uses "qpdf" for rotating.
+
+(define-module (rotate-pdf-addon)
+  #:use-module (json)
+  #:use-module (ice-9 rdelim)
+  #:export (main))
+
+;; Some helpers
+(define* (errln formatstr . args)
+  (apply format (current-error-port) formatstr args)
+  (newline))
+
+;; Macro for executing system commands and making this program exit in
+;; case of failure.
+(define-syntax sysexec
+  (syntax-rules ()
+    ((sysexec exp ...)
+     (let ((rc (apply system* (list exp ...))))
+       (unless (eqv? rc EXIT_SUCCESS)
+         (format (current-error-port) "> '~a …' failed with: ~#*~:*~d~%" exp ... rc)
+         (exit 1))
+       #t))))
+
+(fluid-set! %default-port-encoding "UTF-8")
+
+;; External program dependencies
+(define *qpdf* "qpdf")
+(define *pdftotext* "pdftotext")
+
+;; Getting some environment variables
+(define *output-dir* (getenv "OUTPUT_DIR"))
+(define *item-data-json* (getenv "ITEM_DATA_JSON"))
+(define *pdf-files-dir* (getenv "ITEM_PDF_DIR"))
+
+;; fail early if not in the right context
+(when (not *item-data-json*)
+  (errln "No item data json file found.")
+  (exit 1))
+
+;; The user input schema
+(define-json-type <userinput>
+  (degree))
+
+;; The itemdata record, only the fields needed here.
+(define-json-type <attachment>
+  (id)
+  (name)
+  (position)
+  (pages))
+(define-json-type <itemdata>
+  (id)
+  (attachments "attachments" #(<attachment>)))
+
+;; The output record, what is returned to docspell
+(define-json-type <itemfiles>
+  (itemId)
+  (textFiles)
+  (pdfFiles))
+(define-json-type <output>
+  (files "files" #(<itemfiles>)))
+
+(define (load-itemdata)
+  "Load the JSON file containing item data into the itemdata record."
+  (scm->itemdata (call-with-input-file *item-data-json* json->scm)))
+
+(define (load-user-input file)
+  (scm->userinput (call-with-input-file file json->scm)))
+
+(define (rotate-pdf file degree out txt)
+  (errln "Running qpdf to rotate ~s" degree)
+  (sysexec *qpdf* (format #f "--rotate=~a" degree) file out)
+  (errln "Running pdftotext to extract the text from rotatet file")
+  (sysexec *pdftotext* out txt))
+
+(define (process-file itemid degree file)
+  "Processing a single attachment."
+  (let* ((id (attachment-id file))
+         (name (attachment-name file))
+         (file (format #f "~a/~a" *pdf-files-dir* id))
+         (out (format #f "~a/~a.pdf" *output-dir* id))
+         (txt (format #f "~a/~a.txt" *output-dir* id)))
+    (errln "Processing attachment ~s" name)
+    (rotate-pdf file degree out txt)
+    (make-itemfiles itemid
+                    `((,id . ,(format #f "~a.txt" id)))
+                    `((,id . ,(format #f "~a.pdf" id))))))
+
+(define (process-all indata)
+  (let* ((item-meta (load-itemdata))
+         (item-id (itemdata-id item-meta))
+         (attachs (itemdata-attachments item-meta))
+         (degree (userinput-degree indata)))
+    (map (lambda (file)
+           (process-file item-id degree file))
+         attachs)))
+
+(define (main args)
+  (let* ((infile (load-user-input (cadr args)))
+         (out (make-output (process-all infile))))
+    (format #t "~a" (output->json out))))
+
+;; Local Variables:
+;; mode: scheme
+;; End:

nix/addon.nix

@@ -0,0 +1 @@
+{ "degree": "90" }

src/addon.scm

@@ -0,0 +1,11 @@
+{ "id": "qZDnyGIAJsXr"
+  "attachments": [
+      {"id": "BXLaDza97A",
+       "name": "my-file.converted.pdf",
+       "position": 0,
+       "content": "the whole text",
+       "language": "deu",
+       "pages": 2
+      }
+  ]
+}