Add some type file preview and more

.travis.yml

@@ -1,25 +1,26 @@
-sudo: false
+sudo: true
 language: python
 python:
+  - "3.3"
   - "3.4"
   - "3.5"
   - "3.6"
 
-addons:
-  apt:
-    packages:
-    - libreoffice
-
 install:
+  # Install specific version of exiftool
+  - wget https://sno.phy.queensu.ca/~phil/exiftool/Image-ExifTool-11.11.tar.gz && gzip -dc Image-ExifTool-11.11.tar.gz | tar -xf - && cd Image-ExifTool-11.11 && perl Makefile.PL && sudo make install
+  - cd ..
+  - sudo apt-get install xvfb libreoffice libfile-mimeinfo-perl inkscape poppler-utils -y
+  - pip install xvfbwrapper
   - python setup.py install
-  - pip install preview-generator
   - pip install pytest
   - python3 -m pip install -U mypy
 
 # command to run tests
 script:
   - "cd ${TRAVIS_BUILD_DIR} && py.test tests"
-  - "cd ${TRAVIS_BUILD_DIR} && mypy preview_generator --ignore-missing-imports --disallow-untyped-defs"
+
+#  - "cd ${TRAVIS_BUILD_DIR} && mypy preview_generator --ignore-missing-imports --disallow-untyped-defs"
 
 
-
+

README.rst

@@ -777,14 +777,47 @@ From scratch on a terminal :
      * build your virtual env (I can say that it work with python 3.4 but did not try with other versions)(env will be called "myenv", you can name it the way you want): `python3.4 -m venv myenv`
      * if it's not already, activate it : `source myenv/bin/activate`. (`deactivate` to deactivate)
   - install dependencies :
+
+     * Exiftool - Follow instruction on the main website: https://sno.phy.queensu.ca/~phil/exiftool/
      * `apt-get install zlib1g-dev`
      * `apt-get install libjpeg-dev`
+     * `apt-get install python3-pythonmagick`
+     * `apt-get install inkscape`
+     * `apt-get install xvfb`
      * `pip install wand`
-     * `pip install python-magick`
-     * `pip install pillow`
+     * `pip install Pillow`
      * `pip install PyPDF2`
+     * `pip install python-magic`
+     * `pip install pyexifinfo`
+     * `pip install packaging`
+     * `pip install xvfbwrapper`
      * if you use python 3.5 or less `pip install typing`
 
+
+.. code:: console
+
+  # general dependencies
+  apt-get install libjpeg-dev libjpeg-dev python3-pythonmagick inkscape xvfb
+  pip install wand Pillow PyPDF2 python-magic pyexifinfo packaging xvfbwrapper
+  # Exiftool
+  wget https://sno.phy.queensu.ca/~phil/exiftool/Image-ExifTool-11.11.tar.gz
+  gzip -dc Image-ExifTool-11.11.tar.gz | tar -xf -
+  cd Image-ExifTool-11.11
+  perl Makefile.PL
+  sudo make install
+
+If you need to preview scribus `.sla` files you will need scribus >= 1.5.
+If it's not available in your distribution you can use an AppImage.
+
+Download the last AppImage from the official website https://www.scribus.net/downloads/unstable-branch/
+
+.. code:: console
+
+  mv /path/to/image/scribus-x.y.appimage /usr/local/bin/scribus
+  chmod +x /usr/local/bin/scribus
+
+
+
 Running Pytest :
 ----------------
  Pytest is a motor for unit testing

preview_generator/file_converter.py

@@ -1,13 +1,10 @@
 # -*- coding: utf-8 -*-
 
-from io import BytesIO
-import json
 import logging
-from PIL import Image
 from PyPDF2 import PdfFileReader
 import typing
 from wand.image import Image as WImage
-from preview_generator.utils import PreviewGeneratorJsonEncoder
+
 
 def txt_to_txt(text: typing.IO[typing.Any]) -> typing.IO[typing.Any]:
     logging.info('Converting text to text')

preview_generator/manager.py

@@ -7,6 +7,7 @@
 import typing
 
 from preview_generator.preview.builder.office__libreoffice import OfficePreviewBuilderLibreoffice  # nopep8
+from preview_generator.preview.builder.document__scribus import DocumentPreviewBuilderScribus  # nopep8
 from preview_generator.preview.builder_factory import PreviewBuilderFactory
 from preview_generator.utils import ImgDims
 
@@ -109,7 +110,7 @@ def get_jpeg_preview(
         builder = self._factory.get_preview_builder(mimetype)
         extension = '.jpeg'
 
-        if isinstance(builder, OfficePreviewBuilderLibreoffice):
+        if type(builder) in [OfficePreviewBuilderLibreoffice, DocumentPreviewBuilderScribus]:
             file_path = self.get_pdf_preview(
                 file_path=file_path,
                 force=force,
@@ -134,11 +135,11 @@ def get_jpeg_preview(
         return preview_file_path
 
     def get_pdf_preview(
-            self,
-            file_path: str,
-            page: int = -1,
-            force: bool = False,
-            file_ext: str = ''
+        self,
+        file_path: str,
+        page: int = -1,
+        force: bool = False,
+        file_ext: str = ''
     ) -> str:
         """
         Return a PDF preview of given file, according to parameters
@@ -174,10 +175,10 @@ def get_pdf_preview(
             raise Exception('Error while getting the file the file preview')
 
     def get_text_preview(
-            self,
-            file_path: str,
-            force: bool = False,
-            file_ext: str = ''
+        self,
+        file_path: str,
+        force: bool = False,
+        file_ext: str = ''
     ) -> str:
         """
         Return a TXT preview of given file, according to parameters
@@ -208,10 +209,10 @@ def get_text_preview(
             raise Exception('Error while getting the file the file preview')
 
     def get_html_preview(
-            self,
-            file_path: str,
-            force: bool = False,
-            file_ext: str = ''
+        self,
+        file_path: str,
+        force: bool = False,
+        file_ext: str = ''
     ) -> str:
         """
         Return a HTML preview of given file, according to parameters
@@ -242,10 +243,10 @@ def get_html_preview(
             raise Exception('Error while getting the file the file preview')
 
     def get_json_preview(
-            self,
-            file_path: str,
-            force: bool = False,
-            file_ext: str = ''
+        self,
+        file_path: str,
+        force: bool = False,
+        file_ext: str = ''
     ) -> str:
         """
         Return a HTML preview of given file, according to parameters
@@ -255,7 +256,7 @@ def get_json_preview(
                 it's usefull if the extension can't be found in file_path
         :return: path to the generated preview file
         """
-        mimetype = self._factory.get_file_mimetype(file_path,file_ext)
+        mimetype = self._factory.get_file_mimetype(file_path, file_ext)
         logging.info('Mimetype of the document is :' + mimetype)
         builder = self._factory.get_preview_builder(mimetype)
         extension = '.json'
@@ -275,10 +276,10 @@ def get_json_preview(
             raise Exception('Error while getting the file preview')
 
     def _get_file_hash(
-            self,
-            file_path: str,
-            size: ImgDims=None,
-            page: int = None,
+        self,
+        file_path: str,
+        size: ImgDims=None,
+        page: int = None,
     ) -> str:
         """
         Build a hash based on the given parameters.
@@ -314,10 +315,8 @@ def _get_file_hash(
     def get_supported_mimetypes(self) -> typing.List[str]:
         return self._factory.get_supported_mimetypes()
 
-    def get_file_extensions(self, mime: str) -> typing.List[str]:
+    def get_file_extension(self, mime: str) -> str:
         return mimetypes.guess_extension(mime)
 
     def get_supported_file_extensions(self) -> typing.List[str]:
-        return [
-            ext for ext in mimetypes.guess_extension(self.get_supported_mimetypes())
-        ]
+        return [mimetypes.guess_extension(mime) for mime in self.get_supported_mimetypes()]

preview_generator/preview/builder/archive__zip.py

@@ -1,5 +1,6 @@
 # -*- coding: utf-8 -*-
 
+from datetime import datetime
 from io import BytesIO
 import json
 import logging
@@ -9,19 +10,20 @@
 from preview_generator.utils import PreviewGeneratorJsonEncoder
 from preview_generator.preview.generic_preview import OnePagePreviewBuilder
 
+
 class FileInfo(object):
     FILE = 'file'
     DIR = 'dir'
     UNDEFINED = 'undefined'
 
-    def __init__(self):
-        self.last_modification = None
+    def __init__(self) -> None:
+        self.last_modification = None  # type: datetime
         self.name = ''
         self.type = FileInfo.UNDEFINED
         self.size = 0
         self.size__compressed = 0
 
-    def to_dict(self):
+    def to_dict(self) -> dict:
         return {
             'lastModification': self.last_modification,
             'name': self.name,
@@ -31,21 +33,21 @@ def to_dict(self):
 
 
 class ArchiveInfo(object):
-    def __init__(self):
-        self.files = []  # typing.List[FileInfo]
+    def __init__(self) -> None:
+        self.files = []  # type: typing.List[FileInfo]
         self.size = 0
         self.size__compressed = 0
-        self.last_modification = None
+        self.last_modification = None  # type: datetime
 
     @property
     def compression_rate(self) -> float:
         return self.size / self.size__compressed
 
     @property
-    def file_nb(self):
+    def file_nb(self) -> int:
         return len(self.files)
 
-    def to_dict(self):
+    def to_dict(self) -> dict:
         return {
             'fileNb': self.file_nb,
             'files': [file.to_dict() for file in self.files],
@@ -55,6 +57,7 @@ def to_dict(self):
             'compressionRate': self.compression_rate
         }
 
+
 def archive_info_to_text(archive_info: ArchiveInfo) -> str:
     text = ''
     text__files = ''
@@ -181,7 +184,6 @@ def zipfile_to_infos(self, zipfile: zipfile.ZipFile) -> ArchiveInfo:
         archive_info = ArchiveInfo()
         for ziplineinfo in zipfile.infolist():
             fileinfo = FileInfo()
-            from datetime import datetime
             fileinfo.last_modification = datetime(
                 year=ziplineinfo.date_time[0],
                 month=ziplineinfo.date_time[1],