Make pdfminer.six optional

Mainly since the current release of pdfminer.six lacks a sdist, blocking
homebrew packaging. Also in case other distros don't accept pdfminer.six.

.travis.yml

@@ -74,6 +74,25 @@ matrix:
             - qpdf
             - tesseract
             - unpaper
+    - os: osx
+      osx_image: xcode9.2
+      language: generic
+      env:
+        - ADD_PDFMINER=1
+      addons:
+        homebrew:
+          update: true
+          packages:
+            - exempi
+            - ghostscript
+            - jbig2enc
+            - leptonica
+            - openjpeg
+            - pngquant
+            - python
+            - qpdf
+            - tesseract
+            - unpaper
 
 before_cache:
 - rm -f $HOME/.cache/pip/log/debug.log
@@ -98,6 +117,7 @@ install:
 - pip3 install pycparser  # py3.7 workaround for https://github.com/eliben/pycparser/issues/251
 - pip3 install -r requirements/main.txt
 - pip3 install --no-deps .
+- '[[ "$ADD_PDFMINER" == "1" ]] && pip3 install .[pdfminer]'
 - pip3 install -r requirements/test.txt
 
 script:

setup.py

@@ -250,7 +250,7 @@ setup(
         'chardet >= 3.0.4, < 4',  # unlisted requirement of pdfminer.six 20181108
         'cffi >= 1.9.1',          # must be a setup and install requirement
         'img2pdf >= 0.3.0, < 0.4',       # pure Python, so track HEAD closely
-        'pdfminer.six == 20181108',
+        'pdfminer.six == 20181108 ; sys_platform != "darwin"',
         'pikepdf >= 0.10.0, < 0.11.0',
         'Pillow >= 4.0.0, != 5.1.0 ; sys_platform == "darwin"',
                                   # Pillow < 4 has BytesIO/TIFF bug w/img2pdf 0.2.3
@@ -259,7 +259,7 @@ setup(
         'ruffus >= 2.7.0',
     ],
     extras_require={
-        'fitz': [],  # Backward compatibility
+        'pdfminer': ['pdfminer.six == 20181108'],
     },
     tests_require=tests_require,
     entry_points={

src/ocrmypdf/pdfinfo/__init__.py

@@ -30,9 +30,8 @@ from pikepdf import PdfMatrix
 import pikepdf
 
 from . import ghosttext
-from .layout import get_page_analysis, get_text_boxes
 
-from ..exceptions import EncryptedPdfError
+from ..exceptions import EncryptedPdfError, MissingDependencyError
 
 
 Colorspace = Enum('Colorspace', 'gray rgb cmyk lab icc index sep devn pattern jpeg2000')
@@ -533,12 +532,12 @@ def _page_has_text(text_blocks, page_width, page_height):
     return has_text
 
 
-def simplify_textboxes(miner):
+def simplify_textboxes(miner, textbox_getter):
     """Extract only limited content from text boxes
 
     We do this to save memory and ensure that our objects are pickleable.
     """
-    for box in get_text_boxes(miner):
+    for box in textbox_getter(miner):
         first_line = box._objs[0]
         first_char = first_line._objs[0]
 
@@ -563,9 +562,17 @@ def _pdf_get_pageinfo(pdf, pageno: int, infile, xmltext):
         )
         pageinfo['bboxes'] = bboxes
     else:
+        # pdfminer required for this section
+        try:
+            from .layout import get_page_analysis, get_text_boxes
+        except ImportError:
+            raise MissingDependencyError(
+                "pdfminer is required for this feature. Your distribution "
+                "may not have installed it."
+            )
         pscript5_mode = str(pdf.docinfo.get('/Creator')).startswith('PScript5')
         miner = get_page_analysis(infile, pageno, pscript5_mode)
-        pageinfo['textboxes'] = list(simplify_textboxes(miner))
+        pageinfo['textboxes'] = list(simplify_textboxes(miner, get_text_boxes))
         bboxes = (box.bbox for box in pageinfo['textboxes'])
 
     pageinfo['has_text'] = _page_has_text(bboxes, width_pt, height_pt)

tests/conftest.py

@@ -62,6 +62,16 @@ def running_in_travis():
     return os.environ.get('TRAVIS') == 'true'
 
 
+@pytest.helpers.register
+def needs_pdfminer(fn):
+    try:
+        import pdfminer
+    except ImportError:
+        skip = pytest.mark.skipif(True, reason="pdfminer not available")
+        return skip(fn)
+    return fn
+
+
 TESTS_ROOT = os.path.abspath(os.path.dirname(__file__))
 SPOOF_PATH = os.path.join(TESTS_ROOT, 'spoof')
 PROJECT_ROOT = os.path.dirname(TESTS_ROOT)

tests/test_main.py

@@ -219,6 +219,7 @@ def test_skip_ocr(spoof_tesseract_cache, resources, outpdf):
     assert pdfinfo[0].has_text
 
 
+@pytest.helpers.needs_pdfminer
 def test_redo_ocr(spoof_tesseract_cache, resources, outpdf):
     in_ = resources / 'graph_ocred.pdf'
     before = PdfInfo(in_, detailed_page_analysis=True)

tests/test_pdfinfo.py

@@ -182,10 +182,16 @@ def test_ocr_detection(resources):
     assert pdf[0].has_text
 
 
+
 @pytest.mark.parametrize(
     'testfile', ('truetype_font_nomapping.pdf', 'type3_font_nomapping.pdf')
 )
+@pytest.helpers.needs_pdfminer
 def test_corrupt_font_detection(resources, testfile):
+    try:
+        import pdfminer
+    except ImportError:
+        pytest.skip("Needs pdfminer")
     filename = resources / testfile
     with pytest.raises(NotImplementedError):
         pdf = pdfinfo.PdfInfo(filename)